Loan and Investment in a Developing Economy : An Ethiopian Perspective [1 ed.] 9781443892285, 9781443852036

Citation preview

Loan and Investment in a Developing Economy

Loan and Investment in a Developing Economy: An Ethiopian Perspective Edited by

Arnis Vilks, Girma Tegene Demessie, Goitom Abera Baisa and Kibrom Aregawi Weldegiorgis

Loan and Investment in a Developing Economy: An Ethiopian Perspective Edited by Arnis Vilks, Girma Tegene Demessie, Goitom Abera Baisa and Kibrom Aregawi Weldegiorgis This book first published 2017 Cambridge Scholars Publishing Lady Stephenson Library, Newcastle upon Tyne, NE6 2PA, UK British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Copyright © 2017 by Arnis Vilks, Girma Tegene Demessie, Goitom Abera Baisa, Kibrom Aregawi Weldegiorgis and contributors All rights for this book reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the copyright owner. ISBN (10): 1-4438-5203-1 ISBN (13): 978-1-4438-5203-6



TABLE OF CONTENTS

Introduction by Arnis Vilks ...................................................................... viii Part I: FDI and Growth Chapter One ................................................................................................. 2 Determinants of FDI in Ethiopia: A Time Series Analysis Abiyot Dagne Chapter Two .............................................................................................. 30 The Impact of Sectoral Composition of FDI on Economic Growth in Ethiopia Gebrehiwot Hailegiorgis Chapter Three ............................................................................................ 57 FDI and the Ethiopian Economy - A Survey of Trends, Determinants, and Impact Analysis Mitiku Geberekidan Part II: Finance – Demand and Supply Chapter Four .............................................................................................. 90 Determinants of Growth in Bank Credit to the Private Sector in Ethiopia: A Supply Side Approach Million Assefa Chapter Five ............................................................................................ 118 Determinants of Trade Credit Use by Private Traders: Evidence from Mekelle City, Tigray Dereje Getacher Chapter Six .............................................................................................. 142 Determinants of the Net Interest Margin - An Empirical Study on the Ethiopian Banking Industry Misraku Molla



vi

Table of Contents

Part III: Microfinance, Poverty Reduction, and Social Inclusion Chapter Seven.......................................................................................... 172 Microfinance – Does it Support Households to Achieve an Income above Self-Sufficiency? Evidence from Rural Northern Ethiopia Achamyeleh Tamiru Chapter Eight ........................................................................................... 216 Loan Provision by Microfinance Institutions for Poverty Reduction and its Linkages with Local Economic Development Strategies in Ethiopia: An Empirical Review Muhammedamin Hussen Chapter Nine............................................................................................ 248 A Healthcare Economic Policy for Financing Social Inclusion of Hearing Impaired Artem Boltyenkov Part IV: Financing Small Enterprises Chapter Ten ............................................................................................. 272 Bank Loan and its Impact on the Growth of Small Enterprises: Empirical Evidence from Northern Ethiopia Aregawi Gebremichael Chapter Eleven ........................................................................................ 294 The Impact of Owners Financing Preference on the Growth of Small Enterprises: Empirical Evidence from Northern Ethiopia Habtamu Tefera and Aregawi Gebremichael Chapter Twelve ....................................................................................... 313 Assessment of the Factors of Loan Repayment Performance of Privately Owned Micro and Small-Scale Enterprises Financed by Oromia Credit and Saving Share Company of Robe Branch Haftu Arefe

Loan and Investment in a Developing Economy

vii

Part V: Rural Credit Issues Chapter Thirteen ...................................................................................... 348 Prescription without Laboratory Diagnosis: An Investigation into Rural Credit System in Arsi Zone, Oromia, Ethiopia Adem Kedir Chapter Fourteen ..................................................................................... 359 An Assessment of why Private Banks in Ethiopia Neglected Agriculture Evidence from 10 Selected Private Banks Ganfure Tarekegn Contributors ............................................................................................. 384



INTRODUCTION ARNIS VILKS

While “development” is a multidimensional concept, covering at least life expectancy, education, and income per capita – the three dimensions which enter the Human Development Index (UNDP, n.d.) – economists tend to focus on income per capita. Arguably, life expectancy, education, and many other aspects of human development, are unlikely to improve considerably unless there is sufficient growth of income per capita. In turn, it is a classic result in Solow’s (1956) equilibrium model of economic growth that, in the long run, income per capita in a country depends on the population growth rate, technological progress, and the rate of saving or investment. The developing countries of Sub-Saharan Africa tend to be weak in all three of these factors: population growth rates are high, technological progress is slow, and the share of investment in GDP is considerably lower than it is in the industrialized countries. It is therefore quite natural that increasing investment is one of the principal means by which developing countries try to raise growth and income per capita. As it is a national income accounting identity that investment in an economy equals the sum of private saving, public saving, and capital inflows, increasing investment may mean increasing private or public saving or else capital inflows from abroad - in particular, foreign direct investment. From a microeconomic perspective, every investment needs to be financed. Very often this poses considerable challenges for economic agents in developing countries. Even if the expected net present value of a particular project would justify the required investment, own savings may not suffice, and a loan may not be accessible. Promoting the availability of loans – be it by banks or other sources – is thus a further natural lever by which investment and thereby development can be fostered. Of course, per capita income as such is a measure of development that completely neglects issues of income distribution – poverty is not necessarily reduced by GDP growth. It is therefore loan provision to



Loan and Investment in a Developing Economy

ix

micro-investors that is of particular interest from the perspective of poverty reduction. It may be of similar importance from the perspective of social inclusion, and the prominence of microfinance institutions suggests that these special loan-providing institutions deserve particular attention and scrutiny. Throughout the developing world, the main part of GDP is produced by micro and small enterprises, and in rural areas. Again, this mandates particular attention to the special issues arising with loan and investment in these sectors. All of these general considerations are reflected in the contributions to the present volume. They are investigated and discussed from the perspective of Ethiopia, however, and in most cases based on the study of empirical data from Ethiopia. The papers in part I deal with foreign direct investment (FDI) in Ethiopia. Abiyot Dagne conducts a time series analysis in order to establish the determinants of FDI, Gebrehiwot Hailegiorgis analyses the sectoral composition of FDI in Ethiopia, and the impact it has on economic growth, while Mitiku Gebrekidan uses an Autoregressive Distributed Lag (ARDL) model to conduct a trend, determinant, and impact analysis of FDI. In part II, three papers look at aspects of the credit markets in Ethiopia: Million Assefa’s contribution analyses the short and long-term impact of bank-specific and macroeconomic variables on bank credit to the private sector (also using an ARDL model), while Dereje Getacher uses evidence from Mekelle city in the northern Ethiopian region of Tigray to establish the determinants of trade credit use, and Misraku Molla studies the determinants of the net interest margins in the Ethiopian banking industry. Microfinance, poverty reduction, and social inclusion are the topics of part III. Achamyeleh Tamiru uses evidence from rural northern Ethiopia to investigate if microfinance actually helps households to achieve an income above self-sufficiency. Muhammedamin Hussen reviews the theoretical and empirical literature on the nexus between loan provision by microfinance institutions and poverty reduction, and links it to local economic development strategies. Artem Boltyenkov discusses an economic policy for financing the social inclusion of the hearing impaired.

x

Introduction

Part IV is devoted to the financing of small enterprises. The first paper by Aregawi Gebremichael analyses the impact of bank loan, the second one, by Habtamu Tefera and Aregawi Gebremichael, the impact of owners’ financing preferences on the growth of small enterprises. Both papers use empirical evidence from northern Ethiopia. Haftu Arefe conducts a case study on the loan repayment performance of micro and small-scale enterprises, using data from Robe town in the Oromia region. Rural credit issues are the subject of the final part V: Adem Kedir investigates the rural credit system in the Arsi zone of Oromia, and argues that there is a mismatch between the demand and supply of credit which resembles “prescription without laboratory diagnosis”. Ganfure Tarekegn’s paper contains an assessment of the reasons why private banks in Ethiopia have served the agricultural sector to only a negligible extent. Preliminary versions of all contributions of the present volume were presented at an Ethiopian National Research Conference in September 2013 that was jointly organized by Mekelle University’s College of Business and Economics and the Ethiopian Public Financial Enterprises Agency.

References Solow, R. (1956). A contribution to the theory of economic growth. Quarterly Journal of Economics 70 (1): 65–94. UNDP (n.d.). Human Development Index. Retrieved on April 9, 2015 from http://hdr.undp.org/en/content/human-development-index-hdi.





PART I FOREIGN DIRECT INVESTMENT AND GROWTH 



CHAPTER ONE DETERMINANTS OF FOREIGN DIRECT INVESTMENT IN ETHIOPIA: A TIME SERIES ANALYSIS ABIYOT DAGNE BELAY

Abstract In their attempt to attract Foreign Direct Investment (FDI), most African countries have liberalized trade and attempted to create an enabling environment in recent decades. Ethiopia, like many African countries, took some steps towards liberalizing trade and the macroeconomic regime, as well as introducing some measures aimed at improving the FDI regulatory framework. This paper attempts to study determinants of foreign direct investment in Ethiopia over the period 1973-2003 E.F.Y. The study gives an extensive account for the theoretical explanation of FDI, as well as reviewing the policy regimes, and undertakes an empirical analysis to establish the determining factors of FDI in Ethiopia. Findings show that real GDP growth, export orientation, gross fixed capital formation, per capita GDP and liberalization, have a long run relationship and positive impact on FDI. On the other hand, macroeconomic instability, huge government budget deficit, poor labor productivity, and restricted financial systems also have a long run significant relationship, but negative impact on FDI. These findings simply indicate that higher GDP growth, stable macroeconomic and political environment, sufficient infrastructural development, and outward looking strategy and major improvements in infrastructure are essential to attract FDI to Ethiopia in the long run.

Acronyms AIC EEA E.F.Y.

Akaike information criterion Ethiopian Economic Association Ethiopian Fiscal Year

Determinants of Foreign Direct Investment in Ethiopia

EIA FDI GNP GDP HQIC IMF LDCs MNCs MoFED ODA OECD SBIC SNNPR SSA UNCTAD

3

Ethiopian Investment Authority Foreign direct investment Gross National Product Gross Domestic Product Hannan-Quinn information criterion International Monetary Fund Least Developed countries Multinational corporations Ministry of Finance and Economic Development Official Development Assistance Organization for Economic Cooperation and Development Schwarz’s Bayesian information criterion Southern Nations, Nationalities and Peoples Region Sub-Saharan Africa United Nations Conference on Trade and Development

1. Introduction 1.1. Background of the Study Foreign direct investment (FDI) is an alternative source of capital to bridge the gap between savings and the required investment level. The proponents of foreign direct investment point out that FDI fills savings, foreign exchange, and local revenue gaps of developing economies. FDI can also provide managerial, entrepreneurial, and technological skills, increases exports, and integrates the country’s economy into the global economic network. Conversely, the other group argues that the benefits that can be derived from FDI inflows are quite small compared to the adverse effect. The major “costs” of FDI include stifling of infant domestic industries, loss of political sovereignty, and deterioration of balance of payment due to the foreign investors’ excessive capital good importation and repatriation of profits. Consequently, most developing countries were skeptical about the virtues of FDI. Ethiopia’s domestic savings rate is low compared to the fast pace of capital accumulation observed between 2004 and 2011. Ethiopia has been experiencing single-digit domestic saving rates, while economic growth was in double digits, supported by investment rates beyond 25 percent of GDP. Consequently, Ethiopia is confronted with a persistent and wide domestic saving and investment gap which has been financed by external sources (World Bank, 2013). In recent years, Ethiopia has started encouraging the inflow of FDI by

Chapter One

4

improving the investment climate and providing different incentive packages to enhance its growth and development. This study focused on: ¾ Review of the economic and FDI performance of Ethiopia ¾ An empirical investigation of some of the determinants of FDI inflows into the country

1.2. Statement of the Problem The Ethiopian economy has grown at an annual growth rate of at least 11% for the last decade, so that the country can attain the per capita income level achieved today by average Sub-Saharan African (SSA) countries. However, Ethiopia’s gross domestic savings as a proportion of GDP is quite low, and it is unlikely to achieve this growth rate by mobilizing the meagre domestic savings (EEA, 2000 and 2007). The current government of Ethiopia has realized the inadequacy of the domestic capital and opened several economic sectors to foreign investors. Since 1992, market oriented economic reforms have taken place, and emphasis has been given to attracting FDI (Ethiopian Economics Association, 2004). As a result, there has been a significant increase in the inflow of FDI to Ethiopia. However, the gap between gross investment and domestic savings has remained wide due to the low levels of income and domestic savings. FDI as a source of capital and other business know-how is therefore desperately needed to finance growth and development. The gap between investment and savings in Ethiopia is very wide due to the low level of income and domestic savings (Getinet and Hirut, 2006). Accordingly, out of the total investment projects licensed between 1992 and 2012, FDI’s share is about 15.71 percent (MoFED, 2012). However, in 2011/12 the overall trend of investment, both by the total number of projects and capital invested, has shown a decline. The savings investment gap in Ethiopia is met through investments generated from Official Development Assistance (ODA) from abroad and Foreign Direct Investment. However, as mentioned above, the FDI inflow has remained very small. Ethiopia’s performance in attracting FDI is very poor compared to many African countries. At this juncture, identifying the determinants of FDI in Ethiopia is a key step towards knowing the factors responsible for the poor performance of Ethiopia in attracting FDI. Getinet Haile and Hirut Assefa (2006), and Solomon Mamo Woldemeskel (2008) have tried to identify the factors that are responsible for the performance

Determinants of Foreign Direct Investment in Ethiopia

5

of Ethiopia in attracting FDI. Unlike the two previously developed lines of enquiry, this study tried to analyze the determinants of FDI and whether there is the long run relationship using a Vector Error Correction Model. This study further explores, both analytically and empirically, the determinants of foreign direct investment in Ethiopia, and its linkage to the growth of the country’s economy in the long run.

1.3. Objectives of the Study The ultimate objective of this study was to identify the major determinants of inflows of FDI to Ethiopia. Therefore, this study answered the main research questions: 1. What are the major determinants of FDI inflows in Ethiopia? 2. Does a long run relationship exist between FDI and economic growth? Moreover, an objective was to forward some policy implications that can be expected to improve the contribution of FDI in order to ensure sustained economic growth.

1.4. Significance of the Study This study hopes to contribute to further research and the unsettled debate on determinants of FDI, and further inform policy makers to take prompt policy action in order to attract investments from abroad to boost economic growth.

1.5. Limitation of the Study Due to a problem in accessibility of data, particularly on effectiveness of FDI and FDI outflow, the scope of this research is limited to investigating only FDI inflow.

1.6. Organization of the Study This project has six chapters. After the above introduction, chapter two provides a review of theoretical and empirical literature related to FDI and its relationships with the determinants. Chapter three provides a brief review of the economic structure of Ethiopia, the FDI policy of the country, and the sectoral and regional distribution of FDI in Ethiopia. The

6

Chapter One

data types and sources, model specification and estimation techniques are discussed in the fourth chapter. Chapter five reports the results of the empirical analysis, and chapter six presents conclusions and recommendations

2. Literature Review The crucial role of FDI in terms of capital formation, and spillover effects on trade and technological progress has led to the development of theoretical and empirical literature that has focused on identifying the possible determinants of FDI. This section provides a survey of this literature on FDI and its linkage to economic growth.

2.1. Theoretical Explanations of FDI The theoretical explanations of FDI largely stem from traditional theories of international trade that are based on the theory of comparative advantage and differences in factor endowments between countries. The theory of portfolio investment is one of the earliest explanations of FDI. The basis for this explanation lies in interest rate differentials between countries. Capital, according to this explanation, moves in response to changes in interest rate differentials between countries/regions, and multinational companies are simply viewed as arbitrageurs of capital, from countries where its return is low to countries where it is high. This explanation, however, fails to account for the cross movements of capital between/across countries. In practice, capital moves in both directions between countries. In addition, that capital is only a complementary factor in direct investment, and this theory does not explain why firms go abroad, nor does it contribute to the criticism of the neoclassical theory of portfolio investment (Harrison et al., 2000). Vernon’s product life cycle theory is another explanation of FDI. This theory focuses on the role of innovation and economies of scale in determining trade patterns. It states that FDI is a stage in the life cycle of a new product, from its invention to maturity. A new product is first manufactured in the home country for the home market. When the home market is saturated, the product is exported to other countries. At later stages, when the new product reaches maturity and loses its uniqueness, competition from similar rival products becomes more intense. At this stage producers would then look for lower cost foreign locations. This theory shows how market seeking and cost reduction motives of

Determinants of Foreign Direct Investment in Ethiopia

7

companies lead to FDI. It also explains the behaviour of multinational companies and how they take advantage of different countries that are at different levels of development (Getinet and Hirut, 2006). Dunning (1993) identified three possible motives for FDI: Market seeking FDI: refers to FDI for the purpose of serving local and regional markets. Host countries’ characteristics that can attract marketseeking FDI include market size of the host country, per capita income, and growth (potential) of the market. Resource/asset seeking FDI: refers to FDI for the purpose of acquiring resources which are not available in the home country. Such resources include natural resources, availability of raw materials, and productivity and availability of skilled and unskilled labor. Efficiency seeking FDI: this kind of FDI occurs when the firm can gain from the common governance of geographically dispersed activities, especially in the presence of economies of scale and scope and diversification of risk. The above three motives of FDI are categorized under economic determinants of FDI. A formal definition for FDI, as a phenomenon of international business, is investment “that reflects the objective of a resident entity in one economy obtaining a lasting interest in an enterprise resident in another economy” (IMF 1993, p. 86). The resident entity (foreign investor) owns an equity capital stake of at least 10% of the ordinary shares in an incorporated enterprise, or its equivalent for an unincorporated enterprise. This reflects a long-term relationship between the investor and the enterprise, and implies a significant degree of influence by the investor in enterprise management. A direct investment enterprise can be a subsidiary (a nonresident investor owns more than 50%), associates (an investors owns 50% or less) and branches (wholly or jointly owned unincorporated enterprises) either directly or indirectly owned by the foreign investor. Foreign Direct Investment can also be defined as an investment made by a firm or an entity based in one country, into a firm or entity based in another country. According to the World Bank, foreign direct investment is defined as “an investment made to acquire a lasting management in an enterprise operating in a country other than that of the investor”.

8

Chapter One

2.2. Empirical Evidence on the Determinants of FDI Many studies have examined the determinants of foreign direct investment. Nonnenberg and Mendonca (2004) explored the determinants of foreign direct investment in developing countries. They used an econometric model based on panel data analysis for 38 developing countries (including transition economies) for the 1975-2000 period, and argued that FDI is correlated with level of schooling, economy’s degree of openness, risk and variables related to macroeconomic performance, like inflation and average rate of economic growth. Root and Ahmed (1979) empirically analyzed the determinants of nonextractive direct investment inflows for 70 developing countries over the period 1966-70. Their analysis focuses on testing the significance of the economic, social, and political variables in explaining the determinants of FDI. They conclude that developing countries that have attracted the most non-extractive direct foreign investment are those that have substantial urbanization, a relatively advanced infrastructure, comparatively high growth rates in per capita GDP, and political stability. Singh and Jun (1995) find export orientation (export as percentage of GDP) to be the strongest factor for explaining why a country attracts FDI. Chakrabarti (2001) finds openness to trade, measured by exports plus imports to GDP, being positively correlated with FDI. Bende-Nabende (2002) found market growth, export-orientation policy, and liberalization as the most dominant long run determinants of FDI. Salisu (2003) found openness to trade to have a positive and significant effect on FDI in Nigeria, while Tsikata et al. (2000) found export-orientation as a significant determinant of FDI inflows to Ghana. Asiedu (2002), using exports and imports as a percentage of GDP to proxy openness, comes to a similar conclusion for Sub-Saharan African host countries. In general, the empirical evidence supports the theoretical argument in favor of favorable government policies and liberal trade regimes as important determinants of FDI. From the theoretical point of view, market size, which is usually measured by real per capita income, plays an important role in attracting FDI, especially market seeking FDI. However, the empirical evidence for market size as a determinant of FDI has mixed results. Obwona (2001) found market size to be a significant determinant of FDI in Uganda. Investigating the determinants of FDI in developing and developed countries, Chakrabarti (2001) concludes that host country market size, measured by per capita GDP, has a positive and significant effect on FDI. The World Investment Report (1999) states that factors most frequently mentioned by foreign

Determinants of Foreign Direct Investment in Ethiopia

9

investors in Africa as having a negative influence on investment are bribery, high administrative cost of doing business, and access to capital. Human capital, both in terms of quantity and quality, is another important factor in promoting labor intensive and export oriented FDI in particular. Noorbakhsh et al. (2001), using secondary school enrolment ratio and the number of accumulated years of secondary and tertiary education in the working age population as a proxy to human capital, found human capital to be a significant determinant of FDI inflows for 36 developing countries. Lewis (1999) also provides support to the proposition that human capital in host countries is a key determinant of foreign direct investment in developing countries. He noted that education, especially in technical disciplines, provides the least developed countries with the skills that are required by the multinational companies. Nunnenkamp (2002) analyzed globalizationinduced changes in the relative importance of foreign direct investment in developing countries. His findings indicate that traditional market-related determinants are still dominant factors, but the availability of local skills has become a relevant pull factor of FDI in the process of globalization. Getinet and Hirut (2006) studied the nature and determinants of foreign direct investment in Ethiopia over the period 1974-2001. The study gave an extensive account of the theoretical explanation of FDI, as well as reviewing the policy regimes, the FDI regulatory framework, and institutional setup in the country over the study period. It also undertakes empirical analysis to establish the determining factors of FDI in Ethiopia. This paper’s findings show that the growth rate of real GDP, export orientation, and liberalization, among others, have a positive impact on FDI. On the other hand, macroeconomic instability and poor infrastructure have a negative impact on FDI.

3. Overview of Ethiopia’s Recent Economy and FDI 3.1. Macroeconomic Performance The Ethiopian economy has shifted to a higher growth trajectory since 2003/04. This has been sustained, and over the last five years, overall real GDP has grown rapidly at an average of 11% per annum. Agriculture, industry and services have registered an average annual growth rate of 8.4%, 10%, and 14.6%, respectively. By sustaining the current economic growth over the next five year period, the government aims to achieve the MDG targets by 2015, and its longer term vision of being a middle income country by 2020-2023 (MoFED, 2010).

10

Chapter One

The Ethiopian Investment Agency and regional Investment Offices licensed some 56,421 investment projects with an aggregate capital of Birr 1.1 trillion during 1992/93 – 2010/11. Of these projects, 47,420 were domestic, 8,896 foreign, and 105 public. In terms of capital, Birr 424.1 billion was attributed to domestic investors, Birr 382.2 billion to foreign investors, and Birr 272.4 billion to the public sector (NBE, 2010). Figure 1. FDI Inflows in Ethiopia

Source: UNCTAD World Investment Report 2011

FDI inflows into the agricultural sector account for 32% of the total Ethiopian FDI inflows, and it has increased heavily since 2005 according to the Ethiopian investment agency. 3.1.1. Regional Distribution of FDI The flow of FDI to Ethiopia has been unevenly distributed among the various regions. Even though the incentive system encourages foreign investors to invest in the least developed regions (Gambella, Afar, Somali and Benishangul-Gumuz) of the country by providing special benefits including provision of land free of charge, their performance in attracting FDI is very poor (EIA, 2008 and Tagesse, 2001). As shown in table 1, most of the FDI is destined for Addis Ababa, the capital. Out of the total projects (from 1992-2011), 62% were situated in Addis Ababa. This is because of the region’s better infrastructure, stable political environment, and better supply of trained manpower. Oromia Region has attracted a sizable amount of FDI with respect to the amount of capital invested. That is, of the total FDI operating in Ethiopia during

Determinants of Foreign Direct Investment in Ethiopia

11

1992-2011, 36.9% of the capital was invested in Oromia. This may be due to the region’s proximity to Addis Ababa, availability of natural resources (arable land and favorable climate), and large market size as it is the most populous region in the country. About 4% of the total FDI was invested in the Amhara region. Conversely, Harari, Gambella, Afar, Somali and Benishangul-Gumuz’s performance in attracting FDI has been very poor. For example, there is only one project in the Harari and Benishangul-Gumuz Regions each, and no foreign investments in the Somali region since the country opened its door to foreign investors. Table 1. Numbers and Capital (in Thousands of USD) of Approved Projects by Region     Regions Tigray Afar Amhara Oromia Somali BenishangulGumuz SNNPR Gambella Harari Addis Ababa Dire Dawa Multiregional Projects Grand Total

2008/09 No. of InvestProjects ment Capital 543 46 1,425 2,689 11

10,863 4,880 12,167 63,573 56

182

644

543 132 

4,488 681 

2,773 120,471 190 8,079   273 13,623 8,807 239,524

2009/10

2010/11

Percentage share to Total

No. of Invest- No. of Invest- No. of InvestProjects ment Projects ment Projects ment Capital Capital Capital 626 32 743 1,558 58

7,224 1,307 17,371 20,739 345

349 26 722 1,386 127

11,112 399 32,753 32,219 2,738

5.52 0.41 11.42 21.93 2.01

4.45 0.16 13.13 12.92 1.10

111

1,389

56

81,611

0.89

32.71

163 11 2 2,902 172  118 6496

2,020 2,675 7 29,195 1,455  12,689 96,415

160 14 48 3,221 207  6 6322

49,751 3,920 276 30,627 2,995  1,067 249,469

2.53 0.22 0.76 50.97 3.28  0.09 100

19.94 1.57 0.11 12.28 1.20  0.43 100

Source: Ethiopian Investment Agency

3.1.2. Sectoral Distribution of FDI The distribution of FDI flows to Ethiopia is fairly diversified into various sectors, ranging from the primary including all types of agricultural activities and mining and quarrying to the secondary sector or the industrial activities or to the tertiary sector including electricity generation,

12

Chapter One

construction, real estate, trade, hotel and tourism, transport service, education and health service. The manufacturing sector accounted for 42.9% of the total FDI, followed by agriculture which accounted for 26.5% from 1992-2012 and real estate, machinery and equipment rental and consultancy service constitutes 13.86% of the total FDI flows to Ethiopia. Construction contracting, including water well drilling, constitutes 11.73%. However, the mining, health, and tourism industries are areas that have not received much FDI in the country, with each accounting for less than 1% of the total inflow. 3.1.3. FDI Flows by Country of Origin During the period 1992 - July 2005, Saudi Arabia accounted for half of the FDI flows to Ethiopia. The Ethiopian Economic Association (2007) reported that one company - MIDROC group investment - highly dominates FDI flows originating from Saudi Arabia. Other than this company, Saudi was followed by the United Kingdom, accounting for 9.4%. France, USA, China and India were the other major source countries during that period. However, now, China has the largest investment in the country, followed by India, Sudan, and USA. Figure 2. Flows to Ethiopia by Country of Origin (in billion Birr) from July 1992 to July 2005



Determinants of Foreign Direct Investment in Ethiopia

13

4. Data Source and Methodology This paper was entirely dependent on secondary data. The major data sources were the Ministry of Finance and Economic Development (MoFED), world investment reports, the Ethiopian Investment Authority (EIA), and country reports published by the United Nations Conference on Trade and Development (UNCTAD). Both quantitative and qualitative methods of data analysis were employed in this study.

4.1. Definitions of Variables The World Bank defines FDI as the net amount invested or reinvested by non-residents to acquire a lasting interest (10 percent or more of voting stock) in enterprises in which they exercise significant managerial control. There are a number of FDI variables included in World Development Indicators: net FDI, BOP in current U.S. $, net FDI inflows as a percentage of gross capital formation, net FDI inflows BOP in current U.S $, and net FDI inflows as a percentage of GDP. In line with the approach used in FDI literature, the dependent variable used in this study is the net foreign direct investment inflows as a percentage of GDP. The choice of independent variables is constrained by data availability, as is mostly the case with time-series data in developing countries. For example, timeseries data on some of the factors such as tariff rates, trade taxes, real effective exchange rate, real wages, and corruption index that are used in some studies of this nature are not readily available for Ethiopia over the (entire) study period. Notwithstanding this constraint, this study uses the following variables that are commonly used in studies of FDI. The dependent variable, FDI, is measured as the net foreign direct investment inflow as a percentage of GDP, and is a widely used measure (see Adeisu, 2002; Quazi, 2005; Good speed et al., 2006). Market Size: the market size hypothesis states that multinational firms are attracted to a larger market in order to utilize resources efficiently and to exploit economies of scale (Chakrabarti, 2001). Market size has been represented by real per capita GDP and growth rate of real GDP (as market growth potential). Real GDP per capita and real GDP growth rates are included in the regression as measures of market attractiveness, and FDI is expected to be positively related to these two variables Export orientation: openness promotes FDI, and one indicator of openness is the relative size of the export sector (Singh and Jun, 1995).

14

Chapter One

Openness: It is a standard hypothesis that openness promotes FDI (Hufbauer et al. 1994). The ratio of trade to GDP is often used as a measure of openness of a country and is also often interpreted as a measure of trade restrictions. This proxy is also important for foreign direct investors who are motivated by the export market. Empirical evidence (Jun and Singh, 1996) exists to back up the hypothesis that higher levels of exports lead to higher FDI inflows. We therefore include trade/GDP in the regression to examine the impact of openness on FDI. Macro-economic stability: there is a widespread perception that macroeconomic stability shows the strength of an economy and provides a degree of certainty of being able to operate profitably (Balasubramanyam, 2001). Inflation rates and exchange rates are used as proxy variables for macro-economic stability. Low inflation and stable exchange rates are expected to have a positive impact on FDI. As pointed out earlier, data on the real exchange rate are not readily available. As a result, only the rate of inflation (based on the consumer price index) is included to capture the effect of macro-economic stability on FDI. Infrastructure: infrastructure covers many dimensions ranging from roads, ports, railways, and telecommunication systems to the level of institutional development. The availability of well-developed infrastructure will reduce the cost of doing business for foreign investors and enable them to maximize the rate of return on investment (Morriset, 2001). Therefore countries with good infrastructures are expected to attract more FDI. Taking this into account, gross fixed capital formation (percent of GDP) has been included in proxy infrastructure development. It is expected to be positively correlated with FDI. Liberalization: liberalization of trade and FDI regimes is assumed to have a positive influence on the inflow of FDI since they facilitate a freer trade and investment in conjunction with the repatriation of dividends and profits to home countries (Bende-Nabende, 2002). As explained in section three, Ethiopia has been introducing some liberalization measures since 1991, and a dummy variable is used to capture the effect of the change in policy environment on FDI. The dummy variable assumes a value of 0 for the pre-liberalization period (i.e. up to 1990) and 1 for the post liberalization period (from 1991 onwards). The dummy variable is expected to have a positive sign.

Determinants of Foreign Direct Investment in Ethiopia

15

4.2. Model Specification The general form of the model estimated has the following form: FDI = f (GDPG, GDPC, EXPO, IN, GFCF, LIB, BOP, OPP, OPLR, BD), where GDPG = Growth Rate of Gross Domestic Product GDPC = Gross Domestic Product per capita (per capita GDP) EXPO = Export orientation as percentage of GDP (measures openness) INF = Annual rate of inflation based on consumer price index GFCF = Gross Fixed Capital Formation (as percent of GDP) LIB = Measure of liberalization (dummy variable) OPP=Openness to trade (used as proxy to measure financial restrictions) OPLR=output per labor ratio (labor productivity) BD= budget deficit The model employed for the analysis can be given by ‫ܫܦܨ‬௧ ൌ ߙ ൅ ߚଵ ‫ܩܲܦܩ‬௧ ൅ ߚଶ ‫ܥܲܦܩ‬௧ ൅ ߚଷ ‫ܱܲܺܧ‬௧ ൅ ߚସ ‫ܨܰܫ‬௧ ൅ ߚହ ‫ܨܥܨܩ‬௧ ൅ ߚ଺ ‫ܤܫܮ‬௧ ൅ ߚ଻ ‫ܱܲܤ‬௧ ൅ ߚ଼ ܱܲܲ௧ ൅ ߚଽ ܱܲ‫ܴܮ‬௧ ൅ ߝ

4.3. Stationary and Integrated Stochastic Processes In general, regression models for non-stationary variables (mostly macroeconomic variables) give spurious results. A stochastic process Yt is called stationary if it has time-invariant first and second moments. In other words, Yt is stationary if: 1.‫ܧ‬ሺܻ௧ ሻ ൌ Ɋ௬ ݂‫ܶ א ݐ݈݈ܽݎ݋‬ 2.‫ܧ‬ൣሺܻ࢚ െ Ɋ࢟ ሻሺܻ௧ି௛ െ Ɋ࢟ ሻ൧ ൌ Ɋሾ݄ሿ ݂‫ ݐݐ݄ܽݐ݄ܿݑݏݏݎ݁݃݁ݐ݈݈݊݅ܽ݀݊ܽܶ א ݐ݈݈ܽݎ݋‬െ ݄ ‫ܶ א‬ Based on the above properties the following tests were employed to resolve the non-stationarity nature of the macro economic variables.

4.4. Augmented Dickey-Fuller (ADF) Test An AR (p) process is integrated when Į (1) = 1 í Į1 í ··· í Įp = 0. In other words, a hypothesis of interest is Į (1) = 0. To test this null hypothesis

16

Chapter One

against the alternative of stationarity of the process, it is useful to parameterize the model. Subtractingܻ௧ିଵ on both sides and rearranging terms results in a regression: ௣ିଵ

ο‫ ݕ‬ൌ Ȱܻ௧ିଵ ൅ ෍ ߙ௝‫ כ‬ȟܻ௧ି௝ ൅ ߤ௧ ௝ୀଵ

In this model the pair of hypotheses: H0: f=0 versus HA: f F = 0.0316 R-squared = 0.9150 Adj. R-squared = 0.7619 Root MSE = .34027

Coef.

Std. Err.

t

P>|t|

.51 .21 -.76 .24 1.44 .17 .00 -2.11 .32 -.096

.13 .10 .16 .50 1.66 .17 .02 3.16 .30 .321

3.97 2.31 -4.62 0.47 0.86 0.98 0.30 -0.67 1.05 -0.30

0.011 0.069 0.006 0.655 0.427 0.373 0.779 0.534 0.344 0.777

[95% Conf. Interval] .18 .84 -.02 .44 -1.18 -.34 -1.05 1.52 -2.83 5.69 -.035 .04 -.28 .61 -10.24 6.02 -.46 1.10 -.92 .73

Source: Stata 10 result

Based on the regression result, the primary sector FDI and tertiary sector FDI variables are statistically significant at 1% significance level, and secondary sector FDI is statistically significant at 5% significance level. The value of R-squared shows that about 91.5% of the dependent variable

46

Chapter Two

of the model (which is economic growth measured by real growth of gross domestic product RGRGDP) is explained by the selected explanatory variables, only the remaining 8.5% is explained by other variables not included in this regression model. The adjusted R-squared, the regression power of the model, indicates that about 76.19% of the model specification does not have an overestimation effect of adding other variables to the model specified. Besides, the overall significance of the model is evaluated using the Fstatistics of the regression result and P-value of the model. As a result the F-statistics (5.98) and P-value (0.0316) indicated that the model is soundly fitted at 1 percent level of significance. Regression Results of the model: 

‹ ൌ ͲǤͲͻͷͻͳͻͺ ൅ ͲǤͷͳʹʹͲ͵  ‹ ൅ ͲǤʹͳͲʹ͸  ‹ െ ͲǤ͹ͷͷ͸ͷͷ͵  ‹ ൅ ͲǤʹ͵͸ͻͳͺͺ  ൅ ͳǤͶ͵Ͳ͵͵͸  ൅ ͲǤͳ͸ͻͳ͹͹ ൅ ͲǤͲͲͶͷͲ͹͵ െ ʹǤͳͲͺͳ͹ ൅ ͲǤ͵ͳ͸͹Ͳʹ͹ ൅ İ୧ The result reveals FDI in the primary sector, FDI in the secondary sector, domestic investment, government expenditure, credit to private sector, and openness and political stability have a positive impact on the growth of the Ethiopian economy. By contrast, FDI in the tertiary sector and human capital development have a negative impact on economic growth.

4.4. Qualitative Analysis of the Impact of Sectoral FDI to Economic Growth 4.4.1. FDI and Employment in Ethiopia In emerging economies the main aim of attempts to attract FDI is to create employment opportunities, as creating employment opportunities is one way of improving the living standard of the people. The number of people employed in FDI related sectors in Ethiopia during the study period ranges from 0.003% to 0.93% of the total labour force.

Sectoral Composition of FDI and Growth

47

4.4.2. Sectoral FDI and Employment At the sectoral level, the primary sector accounts for the largest employment opportunity, employing more than 874,165 employees temporarily, and more than 301,281 people permanently, which accounts for 66.25% of the total FDI employment opportunities. The secondary sector accounts for 19.72% of the total FDI employment in the Ethiopian economy, and the tertiary sector accounts for 14.03% of the total FDI employment. 4.4.3. Spillover Impacts of Sectoral FDI on Domestic Enterprises Spillovers arise because multinational companies in general bring with them some sort of firm-specific assets that allow them to compete successfully abroad. These firm-specific assets, which can manifest themselves in a variety of forms such as superior marketing, management, or production techniques, can be conveniently described as technological advantages, i.e., foreign affiliates use better technology than domestic firms. The linkages allow domestic suppliers to produce at a more efficient scale, reducing average costs (Caves, 2008). 4.4.4. Spillover Effects 4.4.4.1. Labour Turnover According to Lesher and Miroudot (2008), in developing countries, the spillover effect from FDI is through labour turnover, by which movement of skilled labour from FDI enterprises is a main channel of spillover effects. The spillover effect works if this labour force uses knowledge acquired from working at FDI enterprises for their work in the new domestic one. In this case, the two possible mechanisms for spillover effects are: (1) When labourers from FDI establish their own businesses and (2) when the labourers recruited in domestic firms in the same industry as FDI enterprises use the acquired knowledge and skill in promoting productivity of domestic firms. Labour movement in the primary sector was 45% of labour movement from FDI to other FDI enterprises, 30% from FDI to own enterprises, 8% to domestic enterprises and 22% to others. In the secondary sector, 61% of the labour movement is to establish own businesses, 13% to other FDI, 3% to domestic enterprises, and 23% to others. In the tertiary economic sector 25% of labour from a company is moved to other FDI, 23% to establishing own businesses, 11% to domestic enterprises, and 41% to others. On the other

48

Chapter Two

hand, domestic enterprises in the secondary sector are benefiting more from the FDI enterprises as a source of labour compared to other sectors by receiving 22%, followed by the primary sector at 11%, and finally the tertiary sector at 8%. This indicates that domestic enterprises are not beneficiaries from the foreign enterprises as a source of skilled labour. 4.4.4.2 Export Improvements In the primary sector enterprises there is a great advantage working with FDI enterprises in exporting, especially in horticulture. But there is no such positive effect in secondary and tertiary sectors of working with FDI enterprises to benefit from distribution networks and learning about the regulatory frameworks with which exporters must comply. 4.4.4.3 Imitation In case of the imitation channel of spillover effects from FDI, the domestic enterprises in the secondary sector mainly benefit from imitation of different technologies and producing the products of FDI enterprises and competing in the market with the FDI products, but in the case of primary and tertiary sectors there is no imitation from FDI enterprises. 4.4.4.4 Production Linkages Production linkages are important in generating positive spillover effects in the developing economy. Foreign enterprises can be expected to have a positive spillover effect to the domestic economy by having backward or forward linkages with domestic enterprises. Only 18% of the primary sector foreign enterprises get supplies from domestic enterprises (i.e. backward linkage), in the secondary sector, 40% of foreign enterprises have backward linkages with local enterprises, and in the tertiary sector 35% have backward linkages with domestic enterprises. These linkages are the result of policies designed by the government when the FDI enterprises get licenses. 4.4.4.5 Crowding Out Effect Of the domestic enterprises in the primary sector, 61% are not affected by stiff competition. There is no crowding out effect from FDI enterprises in the sector, rather there is healthy competition and upgrading of their performance from the presence of foreign enterprises. In the secondary sector, 78% of the domestic enterprises are not affected by the crowding

Sectoral Composition of FDI and Growth

49

out effect, and similarly 54% of the tertiary sector is not affected by the crowding out effect. A possible reason for this is that domestic enterprises are working to upgrade their production capabilities in order to compete with FDI enterprises. 4.4.4.6 Training for Local Employees by FDI Enterprises Training is available and delivered to the local employees by FDI enterprises in the three sectors from simple operatives and supervisors to technically advanced professionals and top-level managers. Foreign enterprises that employ expatriates are able to replace, within a limited period, such expatriate personnel with Ethiopians by arranging the necessary training. Therefore the policy forces foreign enterprises to work in technology transfer by giving training. The various skills gained while working for an affiliate may spill over as the employees move to other firms or set up their own businesses, and this enhances productivity and thus economic growth of the country. This training to upgrade skills for employees is mainly confined to the secondary sector, and happens less often in the primary and tertiary sector.

4.5 Hypothesis Testing Economic Growth and Primary Sector FDI H1: There is a positive impact of primary sector FDI on Ethiopian economic growth. Primary sector FDI has a positive and statistically significant effect at 1% level of significance. In statistical interpretation from the result of the regression, as one percent incremental FDI in the primary sector is keeping other explanatory variables constant, the real growth rate of GDP on average increases by 0.51 percent. From the qualitative analysis part, taking employment opportunities in the primary sector to be mainly in agriculture, which is the lion’s share contributor, as well in all channels of spillover, there was no strong result found to accept the null hypothesis. From this perspective, the alternative hypothesis about positive impact of primary sector FDI on economic growth is accepted. This positive impact of primary sector FDI, especially in the agriculture sector, is consistent with Adenäuer (2011).

50

Chapter Two

Economic Growth and Secondary Sector FDI H2: There is a positive impact of secondary sector FDI on Ethiopian economic growth. The second main variable is secondary sector FDI, which measures the net inflow of FDI to the secondary sector as a percentage of GDP. It has a positive and statistically significant impact on the economic growth of the country, at a significance level of 5 percent. From the regression result (table 3), a one percent increase of FDI in the secondary sector, keeping effects of other explanatory variables constant, increases the real growth rate of GDP on average by 0.21 percent. On the other hand, from the qualitative analysis in creating employment opportunity and spillover impact, there are strong backward linkages that enhance the productivity of domestic enterprises. In all channels of spillover effects, the secondary sector has a positive impact on economic growth. As a result, the alternative hypothesis is accepted. This result is consistent with Alfaro (2003) and Aykut & Sayek (2007). Economic Growth and Tertiary Sector FDI H3: There is a positive impact by tertiary sector FDI on Ethiopian economic growth. Tertiary sector FDI has a negative and statistically significant impact on the economic growth of the country at a significance level of one percent. In statistical expression, a one percent increment of FDI in the tertiary sector, keeping effects of other explanatory variables constant, decreases the real growth rate of GDP by, on average, 0.76 percent. As a result, the alternative hypothesis is rejected. The negative effect of the tertiary sector FDI on economic growth may be explained by the nature of such investments as findings by Chakraborty and Nunnenkamp (2007) and Palmade and Anayiotos (2005) are consistent with this result. Behaviour of multinational enterprises (MNEs) and the regulatory environment (restricted FDI in the financial sector in the Ethiopian economy) could generate a dominating negative effect of the tertiary sector FDI on the economic growth of the country.

Sectoral Composition of FDI and Growth

51

5 Conclusions and Recommendations 5.1. Conclusions In the last two decades, due to changes in foreign policy and investment regime in the post reform era, FDI in Ethiopia in terms of size of capital inflow and number of projects has seen an upward trend, even though there are fluctuations in some periods. The primary sector FDI was found to have a significant positive impact on economic growth via the direct investment channel, employment creation, and spillover effects to domestic enterprises. As agriculture is the backbone of the economy, attracting FDI to the sector contributes positively to economic growth. With regard to the impact of the secondary sector on economic growth, FDI has a positive significant impact through the direct investment channel. The secondary sector contributes to employment creation, and positive spillovers by technology transfer improve the competitiveness of domestic enterprises. On the other hand, in the tertiary sector, FDI has been found to have a significant negative impact on the Ethiopian economic growth. FDI flow to this sector does not positively enhance growth of the Ethiopian economy. Finally, this study concludes that all sectoral FDI does not have the same consistent impact to improve economic growth of the Ethiopian economy. From the time series analysis in the period 1992-2012 and spill-over effect survey in the foreign and domestic enterprises, the study reveals that primary and secondary sector FDI have a positive and significant impact in promoting growth, but in the tertiary sector FDI and growth are negatively related. This may be due to the differential of the sectors in creation of employment opportunity and spillover effects, and to policies and strategies.

5.2. Recommendations ¾ Care should be taken when attracting FDI to Ethiopia and it should be directed to more productive sectors of the economy. ¾ Continuous reform of strategies to attract more FDI in the future should be designed to the tertiary sector, which is not contributing

Chapter Two

52

positively to the economy, by providing incentives and direction, especially with regard to tax incentives and other motivating factors in this sector. ¾ The government should facilitate information exchange between domestic enterprises and foreign enterprises, as well as government agencies, so as to have positive spill-over effects from foreign enterprises. ¾ The concentration of FDI in large industrial areas and cities should be reduced by encouraging FDI inflow to other regions by providing special incentives. ¾ Besides, further research is suggested in considering modes of FDI and types of FDI that may have their own impact on economic growth.

Appendix Table 4. FDI flow to Ethiopia

Source: EIA, 2013



Sectoral Composition of FDI and Growth

53

Table 5. Sectoral distribution of FDI



Source: EIA, 2013

References Adenäuer, L., (2011). Foreign direct investment in the agribusiness sector; Inaugural Dissertation. Institut für Lebensmittel- und Ressourcenökonomik der Rheinischen Friedrich-Wilhelms-Universität zu Bonn Aizhan, K., & Makaevna, M. D. (2011), Impact of foreign direct investment on econom growth in kazakhstan, 2011 International Conference on Sociality and Economics Development, IPEDR vol.10 (2011) © (2011) IACSIT Press, Singapore Alfaro, L. (2003). Foreign direct investment and growth: does the sector matter? Harvard Business School Anh,T., Hong,N., Thang,T., & Hai,M., (2006). The impact of foreign direct investment on the economic growth in Vietnam. Capacity building project for policy research to implement Vietnams socioeconomic development strategy in the period of 2001-2010. Ayanwale, B.A. (2007). Foreign direct investment and economic growth: Evidence from Nigeria. AERC Research Paper 165 African Economic Research Consortiums, Nairobi.

54

Chapter Two

Aykut, D.,and Sayek, S.(2007). The role of the sectoral composition of foreign direct investment on growth. Do multinationals feed local development and growth? London, Elsevier. Badeji, B., & Omobitan. (2011). The impact of foreign direct investment on economic growth in Nigeria. International Research Journal of Finance and Economics. Lagos State, Lagos State University. Bengos, M. & Sanchez-Robles, B. (2003). Foreign direct investment, economic freedom and growth: New evidence from Latin America. European Journal of Political Economy, 19(3): 529–45. Borensztein, E., De Gregorio, J. & Lee, J. W. (1998). How does foreign direct investment affect economic growth? Journal of International Economics, 45, 115Ǧ135. Caves, R. (2008). Multinational enterprise and economic analysis. Journal of International Economics. Cambridge University Press. Chakraborty, C. & Nunnenkamp, P. (2006). Economic reforms, foreign direct investment and its economic effects in India. Kiel Institute for the World Economy. (Working Paper No. 1272). Esubalew, A. (2008). Investment and economic growth in Ethiopia. Journal of African development studies: Ethiopian Civil Service University. Vol 1. No. 1. Ethiopian investment agency (2013). FDI flows to Ethiopia report. —. (2011). New business Ethiopia. Getinet, H. & Hirut, A. (2005). Determinants of foreign direct investment in Ethiopia: a time series analysis. Policy Studies Institute, Retrieved from http//www.wmin.ac.uk/ Westminsterresearch. Gorge, H. & Greenway, E.(2000) Multinational companies and productivity spillover; A meta analysis. Center for research globalization and labor market Gujarati, N.D. (2004). Basic econometrics. (4th ed.). International edition. The McGraw-Hill Companies Inc., Singapore. Jeleta, G. (2011). Economic growth and foreign direct investment in subSaharan African countries. (Master’s thesis, addis abeba university). Kentor, J. (2003). The longǦ–erm effects of foreign investment dependence on economic growth, The American Journal of Sociology, 103, 1024Ǧ 1046. Khaliq, A. & Noy, I. (2007). Foreign direct investment and economic growth: Empirical evidence from sectoral data in Indonesia. University of Hawaii. (Working Paper No. 26.) Khan. (2010). Greenfield investment verses foreign aquisition. London, London bussiness school.

Sectoral Composition of FDI and Growth

55

Kim, & Antony,M. (n.d.). The impact of foreign direct investments on economic growth and development in Kenya. Nairobi, University of Nairobi Koko, A. (1994). Technology, market characteristics, and spillover. Journal of development economics. Kyounga, (2008). Foreign direct investment and growth. Korea : Korea Information Strategy Development Institute Megbaru, M. (2011). Determinants of foreign direct investments in Ethiopia. (Master’s thesis, University of Mekelle). Mencinger, J. (2003). Does foreign direct investment always enhance growth: university of Ljublana, Slovania Mossa, A.I. (2002). Foreign direct investment, evidence and practice. New York: Palgrave Msuya, & Elibariki. (2007). The impact of foreign direct investment on agriculture productivity and poverty reduction in Tanzania. MPRA paper No. 3671 O‘Keef, M. & Li, Q. (n.d.). Effects of Foreign Direct Investment on Food Security in Less Developed Countries. Princeton University, Niehaus Center for Globalization Governance OECD, (2002). Foreign direct investment for development: maximizing benefits, minimizing cost structures by direct investments Palmade, T. & Anayiotos, H., (2005). The long term effects of globalization on income inequality, population growth, and economic development. Remla, K. (2012). The impact of foreign direct investment on poverty reduction in Ethiopia: Cointegrated Var approach. Master’s thesis. Addis Ababa University Rezvi, A.Z., & Nishat, M. (2009) The impact of foreign direct investment on employment opportunities: Empirical evidence from Pakistan, India and China. Institute of Business Administration (IBA), Karachi Roy, G.A., Ber, F.H. (2006). Foreign direct investment and economic growth. Lincoln: University of Nebraska Samatar, A.(1993). Structural adjustment as development strategy and poverty in Somalia economy Solomon, M. (2008). Determinants of foreign direct investment in Ethiopia. (Masters Thesis Netherlands: Maastricht University). Tonds, G., & Fornero, A.J. (2008). Sectoral productivity and spillover effects of foreign direct investment in Latin America. Wirts chafts university.

56

Chapter Two

Ullah, A., Khan, U.M., Ali, S., & Hussain, W.S. (2011). Foreign direct investment and sectoral growth of Pakistan economy. Pakistan: Institute of management science. UNCTAD, (2008). World Investment Report: Transnational Corporations, Extractive Industries and Development. New York and Geneva: United Nations. Vu, B.T., Gangnes, B., & Noy, I. (2008). Is foreign direct investment good for growth: evidence from sectoral analysis of china and Vietnam? USA: Journal of the Asia pasfic economy, vol.13 No. 4. University of Hawaii-Hilo. Wang, M. (2009). Manufacturing FDI and Economic Growth. Evidence from Asian economies. Marquette University. Weissleder, L. (2009). Foreign direct investment in the agricultural sector in Ethiopia. Discussion Paper: University of Bonn World Bank Report, (2007).World development indicators. Yared, L. (2011). The impact of foreign direct investment on technology transfer. (Master’s thesis, university of Addis Abeba).

 



CHAPTER THREE FOREIGN DIRECT INVESTMENT IN ETHIOPIA: A TREND, DETERMINANT AND IMPACT ANALYSIS MITIKU GEBREKIDAN

Abstract The main objective of this study was to assess the trends and patterns of FDI, to investigate the short run (SR) and long run (LR) determinants that constitute this flow of FDI, and to measure the SR and LR impact of FDI in the economy. The study used an Autoregressive Distributed Lag Model (ARDLM) with a bound test for cointegration for analysing the multivariate time series data from 1992-2012; Ordinary Least Squares (OLS) regression estimations were employed in exploring the SR and LR relationship of the variables specified in the econometric equation, which satisfy the different goodness-of-fit tests. In addition to this, the study used primary data collected from a sample of 76 FDIs through a five stage Likert scale type of questionnaire. Then the study took the triangulation effect of the two findings to strengthen the base of the conclusions it made. The study found that having an increasing trend, the pattern of FDI flow is highly volatile and is highly contracted in the periods of political turbulence, mainly in the period of power transitions, in the period of border war, and during the 2005 national election. In the investigation of FDI determinants, the study found that the lagged FDI, domestic investment (DI), trade liberalization, economic growth, infrastructure (road networks), and political stability attract FDI favourably. However, the macroeconomic instability (inflation rate and exchange rate), the human capital, market size, and the import/export procedure and restriction sides of the trade liberalization are unfavourable to attract FDI, with some exception in LR and SR dynamics. The study found that the economic power of the nation is not strong enough to reap the benefits of FDI in the short run. However, in the long run the economy is powerful

58

Chapter Three

enough to take advantage of FDI. The short run effect of FDI in the HCD of this nation is negative, but in the long run has a positive but insignificant effect. FDI has a SR negative effect and LR positive effect to both GDPP and DI. Addressing the HCD, creating a vertical integration among the FDIs and DIs, and a due revision on the macroeconomic policy are some of the recommendations forwarded by this study.

1. Introduction To close the gap between savings and the required investment level, countries take Foreign Direct Investment (FDI) as an important alternative source of capital. Hence, FDI is taken as an essential component of development strategy in both developed and developing nations (Ikiara, 2003). Though the developmental role of FDI is highly debatable, in the present globalized world many countries spend enormous amounts of resources and time to design policies that encourage the inflows of FDI (Hooda, 2011). Though most African countries have undertaken numerous policy measures to create a hospitable investment climate for FDI, its flow is still quite low, and declining. For instance, Africa’s share of the total FDI flows to developing economies fell from 19 percent in the 1970s to 11 percent in the 1980s, and further to 8 percent in 2000-2006 (OECD, 2005). In terms of Ethiopia, it was in 1991 that the country’s transition to a market oriented economy started. Since then, the government has made a broad range of policy reforms, and the investment code has been amended several times to increase the flow of FDI (Solomon, 2008). Despite all these reforms and amendments in the macroeconomic situation of the country in order to attract FDI, different contradicting reports have been made towards the performance of FDI flow to Ethiopia, which suggests that the flow is unhealthy and inconsistent. To note a few: as the world investment report indicates, the FDI flows to Ethiopia increased from US $255 million in 2002 to $465 million and $545 million in 2003 and 2004 respectively, but remains constant after 2005 (UNCTAD 2008). Over the past 18 years, the flow has fluctuated between $545,257,100 in 2006 and $170,000 in 1992 (UNCTAD, 2011). This report shows that the FDI flow to Ethiopia decreased almost by half in 2007 and continued constant until 2010. Over the past 18 years, FDI flow to Ethiopia has fluctuated between Birr 80,083,808 in 2008 and Birr 87,658 in 1993 (EIA, 2011). Therefore, the first task of this paper was to assess the trends and patterns of FDI.

Foreign Direct Investment in Ethiopia

59

The second task of this study was to identify the short run and long run determinant factors that constitute the flow of FDI for the period 19922012. So far, different determinant factors have been identified by different researchers and authors. Schneider and Frey’s (1985) research on 80 developing countries concludes that a country’s level of development is the major determinant of FDI, and political instability in a country leads to a sharp decline in the flow of FDI. Noorbakhsh et al. (2001) find that human capital is the chief determinant in export-oriented and labourintensive industries. Root and Ahmed (1979) study the determinants of non-extractive FDI in 70 developing countries and find that urbanization, developed infrastructure, and improved GDP per capita increase FDI inflows. Getinet & Hirut (2005) concluded that growth of real GDP, export orientation and liberalization promotes the inflow of FDI, while macroeconomic instability and poor infrastructure deter the inflow of FDI. In reviewing the previous empirical studies, the findings seem to be inconsistent and varied for the case of developing countries in general, and that of Ethiopia in particular. These inconsistencies are in part due to problems in the models used, such as using a single OLS equation (which is not adequate to establish a short run and long run causality relationship), or use of Vector Auto Regression, VAR model, (which requires only time series data that are stationary at the same level). Moreover, in using the above models, studies used a small number of observations, which aggravates the problem. Therefore the objective of this study is to assess the trend, to identify the short run and long run determinant factors, and to measure the SR and LR impacts through employing the Auto Regressive Distributive Lag, ARDL, model, which significantly reduces the above problems. In addition, unlike many international business studies, it incorporates qualitative analyses for the determinant identification to strengthen the base of its findings.

2. Methodologies 2.1. Research Strategies and Designs International business research in general and that of FDI in particular favour quantitative methods over qualitative methods. Indeed, Andersen and Skaates (2004) found that only 10% of all published international business research used qualitative methods. Paradoxically, at the same time as qualitative research continues to be marginalized in practice, calls for more research of this kind are made at regular intervals (Marschan-

Chapter Three

60

Piekkari & Welch, 2004). Even though there is limited experience using primary data sources, the research strategy adopted for this research was basically both qualitative and quantitative. In trend analysis only quantitative analysis and in description of the determinant factors both qualitative and quantitative approaches were used, and the triangulated effect of the findings of the two analyses was taken for conclusions. The research design that the researcher will use for this research is Descriptive Survey Design, which entails the collection of data on more than one case (usually quite a lot more than one) within the given study period. Through three successive sampling methods, which include purposive sampling to select the target population, which means taking operational projects found in Addis Ababa and around, stratified sampling then develops five strata, and simple random sampling to select the individual observation within each stratum; and 93 FDI operational project managers was developed to be the sample size, with a proportional distribution among the five strata through using the scientific sample size determination formula adopted by Yamane (1976). In order to collect the primary data from the identified observations, the data collection instrument that the researcher used was a structured questionnaire. The secondary data were collected from various sources; mainly from the Ministry of Finance and Economic Development (MoFED), Ethiopian Investment Agency (EIA), IMF, WTO, UNCTAD, etc. They form time series data which were collected for the period 1992 to 2012.

2.2. Description of Variables and Model Specifications 2.2.1.

Description of variables

Variables

Abbreviations

Descriptions

Market size

MS

Real GDP per capita Lagged FDI Domestic investment

GDPP LFDI DI

Human Capital Development

HCD

Population/number of consumers Real GDP per capita Previous year FDI flow Annual domestic investment Annual secondary school enrolment



Exp. signs (SR/LR) +/+ -/+ +/+/+ -/+

Foreign Direct Investment in Ethiopia Infrastructure

Macroeconomic stability

Telecomm

IS_Tele

Power

IS_EPow

Transportation

IS_Trans

Inflation rate Gov’t expenditure Exchange rate Foreign debt

MES_IR MES_govex

MES_EXR MES_Ford

Openness

OP

Net export performance Political stability

NEP DWAR

Number of telecom users per 100 people Annual production of electric energy Availability of transportation facilities (road) The average increase in price of commodities Amount of gov’t total expenses Rate of exchange of birr to foreign currency Debt of the government relative to GDP Ratio of trade to GDP in current price and in current exchange rate Difference b/n export and import Dummy variable which is1 in case of political instabilities and tribunals, otherwise 0

61 -/+ -/+ +/+

-/+/+

+/+/-/+

-/-/+

2.2.2. Model Specification Taking the above variables into consideration, the following model equations were specified. 1. First simple linear regression was conducted by treating FDI as a dependent variable and time as an independent variable to look at the average change of FDI over time. Then, to compute annual growth rates, the following formula will be used: ௑ ି௑ ‫ ܴܩܣ‬ൌ మ భ Where; X1=first value of variable X and X2 = second ௑భ

value of variable X ................................................................................. 1 2. To study the determinant factors of FDI, the following model was framed: FDI = f [lagged FDI, HCD, MS, GDPP, MES_IR, MES_EXR, MES_Ford, MES_govex, IS_Tele, IS_Elpow, IS_Trans, OP, NEP, DWAR] ............................................................................................ 2

Chapter Three

62

3. To measure the economic power to reap the benefits of FDI (GDPP*), impact of FDI on GDPP, HCD and DI, the following equations were framed respectively: GDPP*=f[lagGDPP, FDI,DI,HCD,MS,(FDI*HCD), Tele, IR, EXR, govex, OP, DWAR]........................................................................... 3 GDPP = f [lagged GDP, FDI, HCD, MS, MES_IR, MES_EXR, MES_Ford, MES_govex, IS_Tele, IS_Elpow, IS_Trans, OP, NEP, DWAR] ............................................................................................. 4 HCD = f [laggedHCD, FDI, GDPP, MS, MES_IR, MES_EXR, MES_Ford, MES_govex, IS_Tele, IS_Elpow, IS_Trans, OP, NEP, DWAR] ............................................................................................. 5 DI = f [lagged DI, FDI, HCD, MS, GDPP MES_IR, MES_EXR, MES_Ford, MES_govex, IS_Tele, IS_Elpow, IS_Trans, OP, NEP, DWAR] ............................................................................................. 6 2.2.3. Bound Test for Cointegration The two-step Engle and Granger (1987) approach and the Johansen test (1988) method are some of the instruments that are repeatedly used by different researchers to test for cointegration among the variables. However, estimations that are undertaken with the presence of a combination of different level stationary series under the Johansen procedure may lead to biased results. The other problem associated with these instruments is that both are not reliable for relatively small samples (Narayan, 2004). Pesaran and Shin (1999) develop a new ARDL testing instrument called a bounds testing approach, which has many advantages over the previous approaches. First, the test can be applied to variables with a combination of stationary series at level and at first difference (Narayan, 2004). Second, it can give us reliable estimates of studies with small observation numbers (Narayan, 2004 and Harris, 2003). Third, it reduces serial correlations and endogeneity problems, and provides unbiased estimates of the long run and short run model and valid t-statistics (Harris, 2003). Fourth, in its estimation, the bound test can use OLS to identify the long run and short run effects simultaneously (Narayan, 2004 and Harris, 2003). The ARDL model specification is as follows: ௣

௣

ܻ௧ ൌ ߚ଴ ൅ σ௜ୀ௢ ߚ௜ ܻ௧ିଵି௜ ൅ ෍

௜ୀ଴

ߚ୨ ௜ǡ௧ି௜ ൅ ɔ୲ ......................................... 7

Foreign Direct Investment in Ethiopia

63

Where Ⱦ଴ is the constant, ୲ is the endogenous variable, ୧ǡ୲ the ith of the independent variables at period t, p is the maximum lag number to be used,Ⱦ୧ andȾ୨ are coefficients of the independent variables, and ɔ୲ is the white noise error. Then by substituting equations 2 to 6 into equation 7 the following new equations will be developed. ௣

‫ܫܦܨ‬௧ ൌ ߚ଴ ൅ ෍ ߚଵ U݈݊‫ܫܦܨ‬௧ି௜ ௜ୀଵ

௣

௣

௣

൅ ෍ ߚଶ U݈݊௧ି௜ ൅ ෍ ߚଷ U݈݊ ௧ି௜ ൅ ෍ ߚସ U݈݊௧ି௜ ௜ୀଵ ௣

௣

௜ୀଵ

௜ୀଵ

൅ ෍ ߚ଺ U݈݊௧ି௜ ൅ ෍ ߚ଻ U̴݈݊  ௧ି௜ ௜ୀଵ ௣

௜ୀଵ

௣

൅ ෍ ߚ଼ U̴݈݊ ௧ି௜ ൅ ෍ ߚଵ଴ U̴݈݊ ‘”†௧ି௜ ௜ୀଵ ௣

௜ୀଵ

௣

൅ ෍ ߚଵଶ U݈݊‫ݔ݁ݒ݋̴݃ܵܧܯ‬௧ି௜ ൅ ෍ ߚଵଵ U݈݊ ̴‡Ž‡௧ି௜ ௜ୀଵ ௣

௜ୀଵ

൅ ෍ ߚଵଷ U݈݊ ̴”ƒ•௧ି௜ ൅  ൅ ܷ௧ Ǥ Ǥ ǥ ǥ ǥ ǥ ǥ Ǥ Ǥ ǥ ͺ ௜ୀଵ ௣

U‫ܲܲܦܩ‬௧‫כ‬

௣

ൌ ߚ଴ ൅ ෍ ߚଵ U݈݊‫ܲܲܦܩ‬௧ିଵ ൅ ෍ ߚଵ U݈݊‫ܫܦܨ‬௧ି௜ ௜ୀଵ

௣

௜ୀଵ

௣

൅ ෍ ߚଶ U݈݊௧ି௜ ൅ ෍ ߚଷ U݈݊ ௧ି௜ ௜ୀଵ ௣

௜ୀଵ

௣

൅ ෍ ߚସ U݈݊  ‫ כ‬௧ି௜ ൅ ෍ ߚ଺ U݈݊௧ି௜ ௜ୀଵ ௣

௜ୀଵ

௣

൅ ෍ ߚ଻ U୍݈݊ୖ ௧ି௜ ൅ ෍ ߚ଼ U݈݊୉ଡ଼ୖ ௧ି௜ ௜ୀଵ ௣

௜ୀଵ

௣

൅ ෍ ߚଵ଴ U݈݊୊୭୰ୢ ௧ି௜ ൅ ෍ ߚଵଶ U݈݊‫ܵܧܯ‬௚௢௩௘௫ ௜ୀଵ ௣

௣

௧ି௜

௜ୀଵ

൅ ෍ ߚଵଵ U݈݊ ୘ୣ୪ୣ ௧ି௜ ൅ ෍ ߚଵଷ U݈݊ ୘୰ୟ୬ୱ ௧ି௜ ൅  ௜ୀଵ

௜ୀଵ

൅ – ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ Ǥ Ǥ ǥ ǥ ǥ ǥ Ǥ Ǥ ǥ ǥ ǥ ͻ

Chapter Three

64 ௣

௣

U ௧ ൌ ߚ଴ ൅ ෍ ߚଵ U݈݊‫ܲܲܦܩ‬௧ିଵ ൅ ෍ ߚଵ U݈݊‫ܫܦܨ‬௧ି௜ ௜ୀଵ

௣

௜ୀଵ ௣

௣

൅ ෍ ߚଶ U݈݊௧ି௜ ൅ ෍ ߚଷ U݈݊ ௧ି௜ ൅ ෍ ߚସ U݈݊௧ି௜ ௜ୀଵ ௣

௣

௜ୀଵ

௜ୀଵ

൅ ෍ ߚ଺ U݈݊௧ି௜ ൅ ෍ ߚ଻ U̴݈݊  ௧ି௜ ௜ୀଵ ௣

௜ୀଵ

௣

൅ ෍ ߚ଼ U̴݈݊ ௧ି௜ ൅ ෍ ߚଵ଴ U̴݈݊ ‘”†௧ି௜ ௜ୀଵ ௣

௜ୀଵ

௣

൅ ෍ ߚଵଶ U݈݊‫ݔ݁ݒ݋̴݃ܵܧܯ‬௧ି௜ ൅ ෍ ߚଵଵ U݈݊ ̴‡Ž‡௧ି௜ ௜ୀଵ ௣

௜ୀଵ

൅ ෍ ߚଵଷ U݈݊ ̴”ƒ•௧ି௜ ൅  ൅ ܸ௧ ǥ ǥ ǥ ǥ ǥ ǥ Ǥ ǤͳͲ ௜ୀଵ ௣

U௧ ൌ ߚ଴ ൅ ෍ ߚଵ U݈݊‫ܦܥܪ‬௧ିଵ ௜ୀଵ

௣

௣

௣

൅ ෍ ߚଵ U݈݊‫ܫܦܨ‬௧ି௜ ൅ ෍ ߚଷ U݈݊ ௧ି௜ ൅ ෍ ߚସ U݈݊௧ି௜ ௜ୀଵ ௣

௜ୀଵ ௣

௜ୀଵ

൅ ෍ ߚହ U݈݊ ௧ି௜ ൅ ෍ ߚ଺ U݈݊௧ି௜ ௜ୀଵ ௣

௜ୀଵ

௣

൅ ෍ ߚ଻ U̴݈݊  ௧ି௜ ൅ ෍ ߚ଼ U̴݈݊ ௧ି௜ ௜ୀଵ ௣

௜ୀଵ

௣

൅ ෍ ߚଵ଴ U̴݈݊ ‘”†௧ି௜ ൅ ෍ ߚଵଶ U݈݊‫ݔ݁ݒ݋̴݃ܵܧܯ‬௧ି௜ ௜ୀଵ ௣

௣

௜ୀଵ

൅ ෍ ߚଵଵ U݈݊ ̴‡Ž‡௧ି௜ ൅ ෍ ߚଵଷ U݈݊ ̴”ƒ•௧ି௜ ൅  ௜ୀଵ

௜ୀଵ

൅ ܸ௧ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ Ǥ ǥ ǥ Ǥ Ǥ ǥ ǥ ǥ ͳͳ

Foreign Direct Investment in Ethiopia ௣

65

௣

U ௧ ൌ ߚ଴ ൅ ෍ ߚଷ U݈݊ ௧ିଵ ൅ ෍ ߚଵ U݈݊‫ܫܦܨ‬௧ି௜ ௜ୀଵ

௣

௜ୀଵ

௣

௣

൅ ෍ ߚଶ U݈݊௧ି௜ ൅ ෍ ߚସ U݈݊௧ି௜ ൅ ෍ ߚହ U݈݊ ௧ି௜ ௜ୀଵ ௣

௣

௜ୀଵ

ଵ

௜ୀଵ ௣

൅ ෍ ߚ଺ U݈݊௧ି௜ ൅ ෍ ߚ଻ U̴݈݊  ௧ି௜ ൅ ෍ ߚ଼ U݈݊‫ܴܺܧ̴ܵܧܯ‬௧ି௜ ௜ୀଵ ௣

௜ୀଵ

௜ୀଵ

௣

൅ ෍ ߚଵଶ U݈݊‫ݔ݁ݒ݋̴݃ܵܧܯ‬௧ି௜ ൅ ෍ ߚଵ଴ U̴݈݊ ‘”†௧ି௜ ௜ୀଵ ௣

௜ୀଵ

௣

൅ ෍ ߚଵଵ U݈݊ ̴‡Ž‡௧ି௜ ൅ ෍ ߚଵଷ U݈݊ ̴”ƒ•௧ି௜ ൅  ௜ୀଵ

௜ୀଵ

൅ ܸ௧ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ Ǥ ǥ ǥ Ǥ ǥ ͳʹ

Then, the dependent variables of the equations must take their first differences. However, the independent variables can take either their level or their first difference values. In addition to that in order to separate the short run and long run effects, a Vector Error Correction Model (VECM) is included in the equation. VECM is the residual obtained from the dependent variables of the equations (Pesaran et al., 2001). By incorporating all these issues the model is again adjusted as follows: ௣

Uଵ ‫ܫܦܨ‬௧ ൌ ߚ଴ ൅ ෍ ߚଵ U݈݊‫ܫܦܨ‬௧ିଵ ௜ୀଵ

௣

௣

௣

൅ ෍ ߚଶ U݈݊௧ି௜ ൅ ෍ ߚଷ U݈݊ ௧ି௜ ൅ ෍ ߚସ U݈݊௧ି௜ ௜ୀଵ ௣

௜ୀଵ ௣

௜ୀଵ ௣

ଵ

൅ ෍ ߚହ U݈݊ ௧ି௜ ൅ ෍ ߚ଺ U݈݊௧ି௜ ൅ ෍ ߚ଻ U̴݈݊  ௧ି௜ ௜ୀଵ ௣

௜ୀଵ

௜ୀଵ

௣

൅ ෍ ߚ଼ U̴݈݊ ௧ି௜ ൅ ෍ ߚଵ଴ U̴݈݊ ‘”†௧ି௜ ௜ୀଵ ௣

௜ୀଵ

௣

൅ ෍ ߚଵଶ U̴݈݊‰‘˜‡š௧ି௜ ൅ ෍ ߚଵଵ U݈݊ ̴‡Ž‡௧ି௜ ௜ୀଵ ௣

௜ୀଵ

൅ ෍ ߚଵଷ U݈݊ ̴”ƒ•௧ି௜ ൅  ൅ Ɂଵ ‫ܫܦܨ‬௧ିଵ ൅ Ɂଶ ୲ିଵ ௜ୀଵ

൅  Ɂଷ  ୲ିଵ ൅ Ɂସ ୲ିଵ ൅ Ɂହ ୲ିଵ ൅ Ɂ଺ ߚ଺ ୲ିଵ ൅ Ɂ଻ ̴  ౪ ൅ Ɂ଼ ̴ ౪షభ ൅ Ɂଽ ୊୭୰ୢ౪షభ ൅ Ɂଵ଴ ̴‰‘˜‡š௧ି௜ ൅ Ɂଵଵ ୘ୣ୪ୣ౪షభ ൅ Ɂଵଷ ୉୪୮୭୵౪షభ ൅ Ɂଵସ ୘୰ୟ୬ୱ౪షభ ൅ ൅ ൅ ɀ୲ିଵ ൅ ୲ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ Ǥ Ǥ Ǥͳ͵

Chapter Three

66 ௣

௣

U‫ܲܲܦܩ‬௧‫ כ‬ൌ ߚ଴ ൅ ෍ ߚଵ U݈݊‫ܲܲܦܩ‬௧ି௜ ൅ ෍ ߚଶ U݈݊‫ܫܦܨ‬௧ି௜ ௜ୀଵ

௣

௜ୀଵ

௣

௣

൅ ෍ ߚଷ U݈݊௧ି௜ ൅ ෍ ߚସ U݈݊ ௧ି௜ ൅ ෍ ߚହ U݈݊௧ି௜ ௜ୀଵ ௣

௜ୀଵ

௜ୀଵ

௣

௣

ଵ

൅ ෍ ߚ଺ U݈݊  ‫ כ‬௧ି௜ ൅ ෍ ߚ଻ U݈݊௧ି௜ ൅ ෍ ߚ଼ U̴݈݊  ௧ି௜ ௜ୀଵ ௣

௜ୀଵ ௣

௜ୀଵ

൅ ෍ ߚଽ U̴݈݊ ௧ି௜ ൅ ෍ ߚଵ଴ U̴݈݊ ‘”†௧ି௜ ௜ୀଵ ௣

௜ୀଵ ௣

൅ ෍ ߚଵଵ U݈݊ ̴‡Ž‡௧ି௜ ൅ ෍ ߚଵଶ U݈݊ ̴Ž’‘™௧ି௜ ௜ୀଵ ௣

௜ୀଵ

൅ ෍ ߚଵଷ U݈݊ ̴”ƒ•௧ି௜ ൅  ൅ Ɂଵ ‫ܫܦܨ‬௧ିଵ ൅ Ɂଶ ୲ିଵ ௜ୀଵ

൅  Ɂଷ  ୲ିଵ ൅ Ɂସ ୲ିଵ ൅ Ɂହ  ୲ିଵ ൅ Ɂ଺ ߚ଺ ୲ିଵ ൅ Ɂ଻ ̴  ౪ ൅ Ɂ଼ ̴ ౪షభ ൅ Ɂଵ଴ ୊୭୰ୢ౪షభ ൅ Ɂଵଵ ୘ୣ୪ୣ౪షభ ൅ Ɂଵଶ ୉୪୮୭୵౪షభ ൅ Ɂଵଷ ୘୰ୟ୬ୱ౪షభ ൅  ൅ ɀ୲ିଵା ୲ ǥ ǥ ǥ ǥ ǥ ǥ ǥ Ǥ Ǥ ǥ ǥ ǤͳͶ ௣

௣

U ௧ ൌ ߚ଴ ൅ ෍ ߚଵ U݈݊‫ܲܲܦܩ‬௧ି௜ ൅ ෍ ߚଶ U݈݊‫ܫܦܨ‬௧ି௜ ௜ୀଵ

௣

௜ୀଵ

௣

௣

൅ ෍ ߚଷ U݈݊௧ି௜ ൅ ෍ ߚସ U݈݊ ௧ି௜ ൅ ෍ ߚହ U݈݊௧ି௜ ௜ୀଵ ௣

௣

௜ୀଵ

௜ୀଵ

௣

ଵ

൅ ෍ ߚ଻ U݈݊௧ି௜ ൅ ෍ ߚ଼ U̴݈݊  ௧ି௜ ൅ ෍ ߚଽ U̴݈݊ ௧ି௜ ௜ୀଵ ௣

௜ୀଵ

௣

௜ୀଵ

൅ ෍ ߚଵ଴ U̴݈݊ ‘”†௧ି௜ ൅ ෍ ߚଵଵ U݈݊ ̴‡Ž‡௧ି௜ ௜ୀଵ ௣

௜ୀଵ ௣

൅ ෍ ߚଵଶ U݈݊ ̴Ž’‘™௧ି௜ ൅ ෍ ߚଵଷ U݈݊ ̴”ƒ•௧ି௜ ൅  ௜ୀଵ

௜ୀଵ

൅ Ɂଵ ‫ܫܦܨ‬௧ିଵ ൅ Ɂଶ ୲ିଵ ൅  Ɂଷ  ୲ିଵ ൅ Ɂସ ୲ିଵ ൅ Ɂ଺ ߚ଺ ୲ିଵ ൅ Ɂ଻ ̴  ౪ ൅ Ɂ଼ ̴ ౪షభ ൅ Ɂଵ଴ ୊୭୰ୢ౪షభ ൅ Ɂଵଵ ୘ୣ୪ୣ౪షభ ൅ Ɂଵଶ ୉୪୮୭୵౪షభ ൅ Ɂଵଷ ୘୰ୟ୬ୱ౪షభ ൅  ൅ ɀ୲ିଵା ୲ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ Ǥ ǥ ǥ ǥ ǥ ǥ Ǥͳͷ

Foreign Direct Investment in Ethiopia

67

௣

U௧ ൌ ߚ଴ ൅ ෍ ߚଵ U݈݊‫ܦܥܪ‬௧ି௜ ௜ୀଵ

௣

௣

௣

൅ ෍ ߚଵ U݈݊‫ܫܦܨ‬௧ି௜ ൅ ෍ ߚଷ U݈݊ ௧ି௜ ൅ ෍ ߚସ U݈݊௧ି௜ ௜ୀଵ ௣

௜ୀଵ ௣

௜ୀଵ

ଵ

௣

൅ ෍ ߚହ U݈݊ ௧ି௜ ൅ ෍ ߚ଺ U݈݊௧ି௜ ൅ ෍ ߚ଻ U̴݈݊  ௧ି௜ ௜ୀଵ ௣

௜ୀଵ

௜ୀଵ

௣

൅ ෍ ߚ଼ U̴݈݊ ௧ି௜ ൅ ෍ ߚଵ଴ U̴݈݊ ‘”†௧ି௜ ௜ୀଵ ௣

௜ୀଵ

௣

൅ ෍ ߚଵଶ U݈݊‫ݔ݁ݒ݋̴݃ܵܧܯ‬௧ି௜ ൅ ෍ ߚଵଵ U݈݊ ̴‡Ž‡௧ି௜ ௜ୀଵ ௣

௜ୀଵ

൅ ෍ ߚଵଷ U݈݊ ̴”ƒ•௧ି௜ ൅  ൅ Ɂଶ ୲ିଵ ൅ Ɂଵ ‫ܫܦܨ‬௧ିଵ ௜ୀଵ

൅  Ɂଷ  ୲ିଵ ൅ Ɂସ ୲ିଵ ൅ Ɂହ  ୲ିଵ ൅ Ɂ଺ ߚ଺ ୲ିଵ ൅ Ɂ଻ ̴  ౪ ൅ Ɂ଼ ̴ ౪షభ ൅ Ɂଵ଴ ୊୭୰ୢ౪షభ ൅ Ɂଵଵ ‫ܵܧܯ‬௚௢௩௘௫ ൅ Ɂଵଶ ୘ୣ୪ୣ౪షభ ௧ି௜ ൅ Ɂଵଷ ୉୪୮୭୵౪షభ ൅ Ɂଵସ ୘୰ୟ୬ୱ౪షభ ൅  ൅ ɀ୲ିଵ ൅ – ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ Ǥ ǥ ǥ ͳ͸ ௣

௣

U ௧ ൌ ߚ଴ ൅ ෍ ߚଷ U݈݊ ௧ି௜ ൅ ෍ ߚଵ U݈݊‫ܫܦܨ‬௧ି௜ ௜ୀଵ

௣

௜ୀଵ

௣

௣

൅ ෍ ߚଶ U݈݊௧ି௜ ൅ ෍ ߚସ U݈݊௧ି௜ ൅ ෍ ߚହ U݈݊ ௧ି௜ ௜ୀଵ ௣

௣

௜ୀଵ

ଵ

௜ୀଵ ௣

൅ ෍ ߚ଺ U݈݊௧ି௜ ൅ ෍ ߚ଻ U̴݈݊  ௧ି௜ ൅ ෍ ߚ଼ U݈݊‫ܴܺܧ̴ܵܧܯ‬௧ି௜ ௜ୀଵ ௣

௜ୀଵ

௣

௜ୀଵ

൅ ෍ ߚଵଶ U݈݊‫ݔ݁ݒ݋̴݃ܵܧܯ‬௧ି௜ ൅ ෍ ߚଵ଴ U̴݈݊ ‘”†௧ି௜ ௜ୀଵ ௣

௣

௜ୀଵ

൅ ෍ ߚଵଵ U݈݊ ̴‡Ž‡௧ି௜ ൅ ෍ ߚଵଷ U݈݊ ̴”ƒ•௧ି௜ ൅  ௜ୀଵ

௜ୀଵ

൅ Ɂଵ ‫ܫܦܨ‬௧ିଵ ൅ Ɂଶ ୲ିଵ ൅  Ɂଷ  ୲ିଵ ൅ Ɂସ ୲ିଵ ൅ Ɂ଺ ୲ିଵ ൅ Ɂ଻ ̴  ౪ ൅ Ɂ଼ ̴ ౪షభ ൅ Ɂଵ଴ ୊୭୰ୢ౪షభ ൅ Ɂଵଵ ‫ܵܧܯ‬௚௢௩௘௫ ௧ି௜ ൅ Ɂଵଶ ୘ୣ୪ୣ౪షభ ൅ Ɂଵଷ ୉୪୮୭୵౪షభ ൅ Ɂଵସ ୘୰ୟ୬ୱ౪షభ ൅  ൅ ɀ୲ିଵ ൅ – ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ Ǥ ǥ ǥ ǥ Ǥ Ǥͳ͹

Where'is the first difference of a variable, ߚଵ െ ߚଵଷ stand for the short run coefficients of the explanatory variables, Ɂଵ െ Ɂଵଷ stand for the long run coefficients of the explanatory variables, ECM stands for Error Correction Model, and J stands for the percentage speed of the adjustment process of the ECM.

Chapter Three

68

Now it is possible to test for cointegration with the help of the ARDL model of the bound test instrument. Two asymptotic critical value bounds provide a test for cointegration when the independent variables are I(d) with 0 ” d ”1. The lower bound assumes that all the regressors are I(0), i.e. are at level, and the upper bound assumes that they are I(1) i.e. at first level. Then, the ARDL bound test has these three possible decision rules. (1) failed to accept the null hypotheses, if the F-statistics lies above the upper bound of the critical value; (2) failed to reject the null hypotheses, if the F-statistics lies below the lower bound of the critical value; and (3) indifference if the F-statistics lies in between the lower and the upper bound of the critical value for a given significance level. Table 2.1. Bound test for cointegration for analysis using time series data from Ethiopia, 1992-2012

Test

K (level of the rank) and F Values

Critical values 1% 5% 10% Lower Upper Lower Upper Lower Upper bound bound bound bound bound bound

FDI determinants The GDPP* FDI impact on GDPP FDI impact on DI FDI impact on HCD

K=5, F=14.53 K=4, F=18.19 K=5, F=476.2 K=3, F=43.42 K=4, F=23.21

3.41 3.74 3.41 4.29 3.74

4.68 5.06 4.68 5.61 5.06

2.62 2.86 2.62 3.23 2.86

3.79 4.01 3.79 4.35 4.01

2.26 2.45 2.26 2.72 2.45

3.35 3.52 3.35 3.77 3.52

Source: Meta data of Ethiopia from NB, EIA, IMF and WB, 2013 Critical Values are taken from Pesaran et al. (2001), Table CI (iii), Case 111: Unrestricted intercept and no trend.

As shown in table 2.1 above, the value of the F-test is much greater than the upper bounds of the critical value, at 1% significance level. In this case the study failed to accept the null hypotheses of no long run cointegration on the explanatory variables of the model specifications. To put it differently, the study is in a position to estimate the long run as well as the short run relations of the variables. Then to estimate the long run elasticity described, the coefficient of one lagged explanatory variable multiplied by a negative sign should be divided to the coefficient of one lagged dependent variable (Bardsen, 1989).

2.3. Methods and Tools of Data Analysis The analysis of data was carried out using different data analysis tools. Both the qualitative and quantitative data are analysed concurrently to arrive at dual edged sword conclusions. The primary data, which were

Foreign Direct Investment in Ethiopia

69

collected from the questionnaire respondents, mainly focuses on the analysis of the determinants. For this reason a questionnaire of the Likert type with a five-point rating scale was prepared, and the interval for breaking the range distance in measuring the variables is computed by the formula adopted from Vichea (2005): ሺ௡ିଵሻ ௡

ൌ

ሺହିଵሻ ହ

ସ

ൌ ൌ ͲǤͺ, where n=number of rates in each questionnaire. ହ

Mean values of the variables falling within 9 4.20-5.00 are going to be taken as the most important or indicating the highest problem level. 9 3.40-4.19 are going to be taken as highly important or indicating the high problem level. 9 2.60-3.39 are going to be taken as medium important or indicating the medium problem level. 9 1.80-2.59 are going to be taken as less important or indicating the lower problem level. 9 1.00-1.79 are going to be taken as least important or indicating the lowest problem level. The time series data focused on analysing the trends, determinants and impacts. In the trend analysis the researcher evaluated the linearity of increase or decrease over the given period through simple regression, and in analysing the determinants. After satisfying the data to the requirements of the model and approach in use through different goodness-of-fit tests, the researcher used an Ordinary Least Square (OLS) estimator, and the estimations were carried out using the 11th version of STATA.

3. Results and Discussions 3.1. Assessments of Performance Trends and Patterns of FDI Flows in Ethiopia As shown in the figure below, which compares the FDI flows to 8 subSaharan countries, unlike those to the other sub-Saharan countries, the flow of FDI to Ethiopia is quite low and full of accidental ups and downs, which indicates the existence of unhealthy flows. In addition to that, the highest investment in the nation is registered in 2011, followed by 2008, 2010, and 2012. From the study period years, the lowest FDI flows were registered in 1992 up to 1995, 1998 up to 2000, and in 2005. The special

Chapter Three

70

identifiable occasions that occurred in these periods included the political power shift, the Ethio-Eritrea border war, and the national election. (NB: In the figure below, to make visible the volatility of the flow, the data for Ethiopia are expressed in $100,000; whereas for the other economies they are expressed in $1,000,000.) Figure 3.1. FDI flows to Ethiopia in comparison with seven subSaharan economies from 1992-2012 14000

Burundi

12000

Ethiopia

10000 Kenya

8000 Mozambique

6000 Rwanda

4000 Uganda

2000 Zambia

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

0

Zimbabwe

Source: UNCTAD data base, 2013

From 2005 onwards, the trend shows a general increase, with the familiar problem of accidental ups and downs. Within this period the significant change in EXR leads to patterns with different slopes, but the trend seems to be increasing in both Birr and $US. Within this increasing trend, accidental downs occurred in 2007, 2009, and 2012. From the study, one can conclude that the flow of FDI into the country is not smooth and healthy. The identified occasions in the years of accidental falls indicate

Foreign Direct Investment in Ethiopia

71

that political instability and turbulence play a major role for this instability and deterioration of the smooth flow of investment. Taking the data from the NBE, the annual increase in FDI is computed with formula 1, and the result is shown in the figure below. Just as with the flow of FDI, the annual rate of increase in FDI is also full of accidental ups and downs. With the highest rate of growth registered in 2006 (with 776% rate of increase) followed by lower increases in 1997 (344%), 1994 (259%), 2008 (255%), 1996 (231%), and 2004 (114%), were the years in which better growth rates had been registered relative to their previous years of investment. By contrast, years like 1998 (-60.57%), 2007 (58.12%), 1999 (-44.70%), 2002 (-34.73%) and 2005 (-19.21%) registered negative annual growth rates. The remaining years registered accidental ups and downs of flow that lay in between these figures. Table 3.1. OLS estimation for FDI flows using time series data from Ethiopia, 1992-2012

Source: EIA data bases, 2013

As the regression result depicted above in table 3.1 shows, there is a positive relation between FDI and year of investment, with a coefficient of birr 3.51 billion and a standard deviation of birr 534 million. To put it differently, on average, the flow of FDI to Ethiopia increases by birr 3.51 billion each year, with the possibility of plus or minus birr 534 million.

72

Chapter Three

Figure 4.2.1 Comparison of contributions of FDI and total investment to GDP annual growth rate in FDI in Ethiopia, 1992-2012

Source: UNCTAD and EIA data bases, 2013

 As illustrated in the above table, the total investment as a percentage of GDP shows a linear increase throughout the study period, with a maximum share of 28.08, 26.51, 25.52 and 24.67 percent registered in 2012, 2006, 2011, and 2010; and a minimum share of 10.62, 12.39, 13.43 15.24 and 15.25 in the years of 1992, 1995, 1997, 1998, and 1999 respectively. However, the pattern of FDI as a percentage of GDP shows a somewhat increasing trend only up to 2004. At the same time this trend indicates that the highest percentage share of FDI to GDP is registered to be 5.43 in 2003, followed by 5.42 and 4.28 in 2004 and 2001 respectively. However, from 2004 onwards, the share of FDI to GDP is continuously declining. This decline is not due to a decline in FDI flow, it is because the increase in FDI is becoming insignificant relative to the surpassing increase in GDP of the nation.

3.2. Determinants of FDI In identifying the determinants of FDI, both qualitative and quantitative analyses were undertaken. Of the designed sample size, 76 were collected. Of these, 87% were general and deputy general managers, and the rest delegated; 57 percent of the respondents have their own share, and 75 percent were non-Ethiopians.

Foreign Direct Investment in Ethiopia

73

The time series data which were collected for the identified macroeconomic variables was tested with different goodness-of-fit tests to satisfy the requirement of the model in use. As a result, the unit root test was conducted using ADF and Phillip Perron tests; their results indicated the data series are smaller than the critical values at level and their first difference; the bound test for cointegration assured that the models have short run and long run relations; the Durbin-Watson alternative test assured that the residuals of the models have no serial correlations. Using the Skewness/Kurtosis tests plus two graphical examinations, i.e., the kernel density and the normal probability plot for normality, the residuals are confirmed to be normally distributed along theoretical lines. The mean VIF is found to be less than one exception, which is treated with its tolerance rate. The linktest tests that the model specifications have no errors that can occur due to omission of relevant variables or else due to addition of irrelevant variables that are included in the model specified. Then the short run and long run dynamics discussed below are the triangulated effects of both the qualitative and quantitative analyses, which were taken as the findings for this study. 3.2.1. The Short Run Dynamics In the quantitative analysis of determinants, the study found that in the short run the lagged FDI has a positive effect in attracting FDI, which is in parallel with many empirical findings like Vichea (2005) and Andreia et al. (2011), but it is opposite to the findings of the study by Megbar (2011). However, the study by Megbar (2011) used a simple OLS estimation that cannot differentiate the long run and short run effects. In the quantitative analysis the domestic investment is found to have a positive influence in attracting FDI. Parallel to this finding, the qualitative analysis also found that it has a positive impact on attracting FDI. This is consistent with the studies of Vichea (2005), Haile (2006) and Andreia et al. (2011). Parallel to the studies by Edwards (1990) and Asidu (2002) it was shown that there is no significant impact of MS on FDI inflows; in the quantitative analysis the market size is surprisingly found to be negatively affecting FDI. However, in the qualitative analysis it is found to have a positive effect upon the decision of foreign investors. Parallel to this also, Megbar (2011) and Haile (2006) found that MS is a significant determinant of FDI. However, these studies used GDP and GDPP respectively as a proxy for MS.

74

Chapter Three

Table 3.2. Short-run OLS estimation for the determinants of FDI flow using time series data from Ethiopia, 1992-2012

Variables laglnFDID1 lnDID1 lnMSD1 lnHCDD1

The Short Run Dynamics Specification 1 Coef. (Std. Err) 1.83e+10* (1.12e+09) 7.22e+10** (7.56e+09) -4.99e+12** (1.07e+12) -9.69e+10*** (2.98e+10) -4.81e+10*** (1.37e+10) 1.29e+11** (2.26e+10) -7.54e+08*** (2.16e+08) 9.47e+10 (4.29e+10) 1.68e+08* (1.58e+07)

Specification 2 Coef. (Std. Err.) 2.28e+10* (1.84e+09) 2.89e+10*** (9.73e+09)

6.99e+09 (4.37e+10) lnTeleD1 -3.80e+09 (8.85e+09) lnEXRD1 1.50e+11** (3.24e+10) IRD1 4.10e+08 (2.93e+08) lnTransD1 -3.19e+10 (4.24e+10) OPD1 1.26e+08** (2.15e+07) lnGGDPD1 1.01e+10*** (3.08e+09) DWAR -1.19e+10** -1.02e+10*** (2.58e+09) (3.49e+09) lagECMF -28.55634* 25.62896** (2.509008) (3.260502) _cons 1.25e+11** -1.99e+10 (3.05e+10) (7.50e+09) F(11, 2) = 43.42** F(11, 2) = 23.22** R-squared = 0.9958 R-squared = 0.9922 Adj R-squared =0.9729 Adj R-squared = 0.9495 Diagnostics test durbinalt= 0.3200 estat bgodfrey= 0.7679 swilk ECM=0.99435 swilk ECM= 0.99435 Mean VIF= 6.68 Mean VIF= 4.48 hettest= 0.4509 hetttest = 0.9243 _hat= 0.0000 _hat=0.000 _hatsq= 0.369 _hatsq= 0.128 Source: Meta data of Ethiopia from NB, EIA, IMF and WB, 2013 ***significant at 10 percent, **significant at 5 percent, * significant at 1 percent.

Foreign Direct Investment in Ethiopia

75

In the two specifications of the quantitative analysis, the HCD of the nation is found to have an opposite result in which one indicates a positive (but statistically insignificant) impact, and the second specification indicates a strong negative impact. In the qualitative analysis, the availability of the fitting skills required were found to be of medium importance - which means that it does not have a significant role in attracting FDI. Then, taking these two inputs, the study is in a position to conclude that the level of the HCD is not a significant force of attraction to FDI. But the qualitative analysis also showed that the total stock and the cost sides of labour are found to have a strong attraction force to FDI. Parallel to this finding, the studies by Andria et al. (2011), and Haile (2006) found that schooling is insignificant in attracting FDI. Except for the transportation networks, the infrastructure that is proxied by four variables (telephone lines, internet, electric power, and transportation network) in the qualitative analysis is found to be less important in attracting FDI. In the quantitative analysis of the proxies of the infrastructure it is only the transportation network that plays a significant role in attracting FDI. This finding is similar with the findings of Megbar (2011), Haile (2006), Solomon (2008), etc. With respect to economic growth, which is proxied by the growth rate of the GDP in the quantitative analysis and the economic image in the qualitative analysis, the study found that they play a significant role in attracting FDI. Concerning macroeconomic stability, which is proxied by the inflation rate and the exchange rate, the quantitative analysis found that inflation has a negative effect whereas the exchange rate has a positive effect. The exchange rate, which is also analysed as a variable of macroeconomic stability in qualitative analysis, was found to have a positive effect in attracting FDI. The foreign currency reserve, which is only analysed in the qualitative analysis, was found to have a negative effect on attracting FDI. In the quantitative analysis, the trade liberalization proxied by openness, which is measured in turn by the ratio of imports and exports to GDP, was found to have a strong positive effect. However, in the qualitative analysis, where it is proxied by import and export procedures and restrictions on some investment areas, it was found to have less importance in attracting FDI. Taking these two extreme inputs, it is difficult to draw a conclusion on this variable. Taking these two scenarios, the volume of import and export as a percentage of GDP has a positive effect; however the import/export procedures and restrictions have a negative effect on attracting FDI.

76

Chapter Three

Another contradiction created in the two categories of findings is with respect to political stability. In the quantitative analysis it is found that political stability has negative effects. However, in the qualitative analysis, the political stability of the nation is found to play an important role in attracting FDIs. But both seem reasonable. Because the quantitative analysis tells us about the short run effects, it is logical to say that the political quarrels and tribunals that happened in a given year adversely affect the investment flow of that year. Then, from this standpoint the quantitative analysis is right. In the same replica it is expected that the attitude of the respondents largely did not look at these specific few years of political quarrels and tribunals; rather their mind is expected to look at the big picture of the whole system. Therefore this is also right with this assumption. The study thus concludes that even though the few years of political quarrels negatively affect the FDI flow to the country, the whole political system of the country has a positive influence on attracting FDI. 3.2.2. The Long Run Determinants of FDI in Ethiopia The long run effect of the lagged FDI is found to be negative. Theoretically this may not seem right, but there are situations that can lead to previous investments and can negatively affect the subsequent years of investment. For instance, if the FDIs are attracted by the market size of the nation and if the previous investments have the capacity to satisfy the existing demand, then other subsequent investors are not drawn to make investments in such saturated markets. The qualitative analysis of this study also supports this. It states that the market size of the nation plays a significant role in attracting FDI. Then, FDIs who are seeking such market size logically look into a market where such investment is not yet done. In both the qualitative and quantitative analysis, the domestic investment is recognized as a best catalyst of FDIs. The increase in domestic investment is taken as a reason to increase the confidence of the foreign investors in a number of directions, which include assurance for the existence of healthy investment climate; the DIs can be taken as assurance of availabilities of established plants to create a backward and forward integration and etc. In the quantitative analysis, similar to the short run effect, the MS is found to have a negative effect that is, however, statistically insignificant. In the qualitative analysis it was found to be an important element for attracting FDI.

Foreign Direct Investment in Ethiopia

77

Table 3.3 Long-run OLS estimation of determinants of FDI flow using time series data from Ethiopia, 1992-2012

Variables laglnFDI laglnDI laglnMS

Specification 2 Coef. (Std. Err.) -1.166507* (.1315112) 1.134774** (.3061675)

-5.647595* (.9710925) laglnTele 2.630546** (.4983309) laglnEXR -4.729879** (1.147636) laglnIR -.3189901** (.0871084) laglnTrans 12.11255** (2.222103) laglnOP -.6990125 (.2928498) laglnGGDP 1.030061** (.1792163) lagDWAR 0.6499 1.089583*** 0.99349 1.665625* (.4783371) (.2058987) _cons 165.3 277.1309 -20.52509 -34.41104** (197.0207) (8.159582) F(10, 6) = 5.48** F(10, 3) =41.69** R-squared = 0.9014 R-squared =0.9929 Adj R-squared =0.7370 Adj R squared=0.9690 Durbinalt=0.1829 Durbinalt=0.6116 swilk ECM= 0.99435 swilk ECM=0.99435 1/VIF= 0.076625 1/VIF=0.873415 hettest=0.0853 hettest=0.3050 _hat=0.000 _hat=0.000 _hatsq=0.929 _hatsq=0.279 Source: Meta data of Ethiopia from NB, EIA, IMF and WB, 2013 ***significant at 10 percent, **significant at 5 percent, * significant at 1 percent. Diagnostics Test

laglnHCD

The Long Run Dynamics Specification 1 Computed Coef. coefficients (Std. Err) -1.676535* (.310358) 1.05137 1.762665* 0.67686 (.45258) -14.3125 -23.99541 (14.4507) 1.06588 1.786989*** (.8070033) -3.368611 0.72547 1.21627*** 1.56904 (.6174871) -0.561778 -.9418404 -2.821223 (1.727668) -0.355428 -.5958873** -0.190267 (.2374854) 7.53861 12.63875*** 7.22475 (5.811157) -0.326754 -.5478142 -0.416939 (.3299208) 0.6144

Computed coefficients

78

Chapter Three

Taking into account the insignificance of the negative coefficient and the positive effect revealed in the qualitative analysis, the study is in a position to conclude that the MS of the nation has a positive effect on attracting FDI. The HCD in the quantitative analysis in the first specification is found to have a strong positive, and the second specification is found to have a weak negative, effect. However in the qualitative analysis the available stock of labour and the cost of this labour are found to have a positive effect. But the skill requirement, which was the proxy for the qualitative analysis, is found to have a less important effect. From this, the study can conclude that that the labour stock and its cost are found to have a positive effect in attracting FDI however. With the level of skill requirement of human capital, it is insignificant in attracting FDI. Unlike the short run analysis in which the telecom were found to have a negative effect in both specifications, the two proxies, telecom and transportation networks, are found to have a positive effect in the two long run specifications. However, as discussed earlier in the qualitative analysis for the infrastructure, except for the transportation networks, the rest (telephone, internet, electric power) are found to play less importance. But it is logical and empirically supported that the current investments made on these utilities have a significant positive effect in attracting FDI in the long run. And the study is in a position to conclude the transport side of infrastructure is found to have a positive effect in attracting FDI. Economic growth, which is proxied by the GDP growth rate in the quantitative analysis and by the economic image in the qualitative analysis, was found in both cases to have a positive effect in attracting FDI. As a result, the study found that the economic growth of the country is playing a significant role in attracting FDI. In the long run analysis for macroeconomic stability, the quantitative result indicates the proxied variables, the inflation and the exchange rates are found to have a negative effect. This can be taken as the result of rapid ups and downs in these variables adversely affecting the confidence of the investors about their investment. Over time this leads to a decrease in the investment flow into the country. The qualitative analyses on the macroeconomic stability other than the exchange rate were indicated as less important role players in attracting FDIs. Even for the exchange rate the reason why the respondents agree with the exchange rate is because they are asked to look into the valuation level of the US$ in terms of Birr.

Foreign Direct Investment in Ethiopia

79

They are not asked to look into its fluctuations. As such, the fluctuations in exchange rate can be inferred to have a negative effect in attracting FDI. Therefore, from all these scenarios, the study concludes that the level of macroeconomic stability of the nation has a negative effect on attracting FDI. Trade liberalization, proxied by the level of openness in both specifications of the quantitative analysis, is found to have a strong positive effect. On the contrary, the qualitative analysis of this study indicates that it is less important in attracting FDI, through its looking into the restrictions of the government, and the import and export procedures and policies. Looking at the implications of the two findings is important. Since the long run relation between the real imports and exports relative to GDP in relation to the FDI flow shows a positive trend, it is logical to infer the openness and FDI are positively related. But whenever there are restrictions, it means that the economy of the nation is linked to the rest of the world only with specific sectors. As such, this has a negative effect on attracting FDI. Therefore even though the restrictions in the investment proclamations of 7/1996, 37/1996, 35/1998, 36/1998 and 116/1998 are significantly reduced in the recent proclamation 769/2012, still the unrepealed restrictions will have to continue to adversely affect FDI. The political stability in both specifications of the quantitative and qualitative analysis is found to have a positive effect on attracting FDI. As such, the study provides strong evidence to conclude that the political stability of the nation plays a significant role in attracting FDI.

3.3. Impact of FDI in the Ethiopian Economy In analysing the impact, the researcher first measures the power of the Ethiopian economy to reap the benefits of FDI. Studies like Mounir (2009), Neguyen et al. (2006), Borensztein et al. (1997), etc. indicate that the impact of FDI on economic growth is highly dependent on the absorptive capacity of the human stock of the host country. Specifically, Borensztein et al. (1997) develop a proxy for the absorptive power of the economy by taking the product of FDI and human capital stock as one variable. This study takes this variable as a proxy variable for measuring the absorptive power of the economy.

Chapter Three

80

Table 3.4. The short run and long run OLS estimation of FDI flows’ impact on GDPP using time series data from Ethiopia, 1992-2012 The Short Run dynamics Variables

laglnGDPPD1

Spec. 1

Spec. 2

Coef. (Std. Err) -18.60** (.74)

Coef. (Std. Err) -9.31* (.18)

-1.07** (.047) -3.49*** (.31) 80.20** (3.30)

-.76* (.01) .29** (.06) 27.25* (.36) -1.52* (.02) -.37* (.01) -.06* (.00) -1.22* (.02) 21.76* (.17) -1.20* (.036)

The Long Run Dynamics Variables

Specification 1 Comp. coef.

laglnGDPP

Specification 2

Coef. Comp. (Std. Err) coef. -36.88** (8.52)

Coef. (Std. Err) 6816.53** (2505.90)



lnDID1 lnHCDD1 lnFDIHCDD1 lnTele

-8.88*** (.74) lnIRD1 .11** (.01) lnOPD1 -2.24** (.12) lngovexD1 32.30** (1.10) DWAR -3.15** (.15) lnEXRD1 24.50** (1.00) lnMSD1 159.02 (28.70) lagECMG -7.89e-10*** -1.68e-09* (7.06e-11) (1.84e-11) _cons -15.04 ** -4.91* (.90) (.05) F( 12,1) = 298.49** F( 11, 2) = 5039.78 R-squared = 1.00 R-squared = 0.99 Adj R-squ = 0.99 Adj R-squared= 0.99 Root MSE = .09 Root MSE = .024 Durbinalt= 0.47 Durbinalt= 0.93 swilk ECM= 0.99 swilk ECM= 0.99 1/VIF=0.047 VIF=5.07 hettest=0.11 hettest=0.44 _hat=0.00 _hat=0.000 _hatsq= 0.94 _hatsq= 0.87

laglnFDI laglnDI laglnHCD laglnFDIHCD laglnTele

0.01 0.06 -0.21

.35 (.31) 2.23** (.80) -7.92** (1.91)

0.02 0.17 -0.49 0.09

-130.47 (113.40) -1131.27** (330.19) 3361.79** (931.62) -591.94** (166.24) -1476.90*** (542.16) 57.61 (94.13) 347.64*** (125.38) -3580.22** (1061.65) -195.23 (201.10) 834.09 (854.48) 11790.33** (4291.43)

8.46** 0.23 0.22 (1.87) laglnIR 7976368** 216251 -0.01 (.25) laglnOP . -1.50** -0.04 -0.05 (.62) Laglngovex 2.00 0.05 0.53 (4.02) lagDWAR 1.63** 0.04 0.03 (.46) laglnEXR 14.30* 0.39 -0.12 (3.02) laglnMS -36.06*** -0.98 -1.73 (16.08) laglnFord -4.49* -0.12 (.62) _cons 989.54** -197336.8*** 26.83 28.95 (277.23) (74079.43) F( 12, 3) = 7.47 F( 12, 4) = 15.18* R-squared = 0.97 R-squared = 0.98 Adj R-squared = 0.91 Adj R-squared = 0.84 Root MSE = .41 Root MSE = 105.06 Durbinalt= 0.07** Durbinalt= 0.29 swilk ECM= 0.99 swilk ECM= 0.99 1/VIF=0.06 1/VIF=0.023 hettest=0.72 hettest=0.92 _hat=0.00 _hat=0.00 _hatsq= 0.03* _hatsq= 0.27

Diagnostics Tests

lnFDID1

Source: Metadata of Ethiopia from NB, EIA, IMF and WB, 2013 ***significant at 10 percent, **significant at 5 percent, * significant at 1 percent

The regression results indicate that the proxy for the power of the economy ሺ  ‫ כ‬ሻ is negatively related and statistically significant at 1 percent in the short run. However, in the long run, the relation between

Foreign Direct Investment in Ethiopia

81

these two variables is positive and statistically significant at 5 percent. As a result, the study found that the economic power of the nation is not strong enough to reap the benefits of FDI in the short run. However, in the long run the economy is powerful enough to take advantage of FDI. The study found that the short run effect of FDI to GDPP is negative and significant at 5 and 1 percent in the first and second specifications respectively. Whereas the long run dynamics illustrated in the same table indicate that the long run effect of FDI to GDPP is positive but statistically insignificant in both specifications. Table 3.5. The short run and long run OLS estimation of FDI flows’ impact on HCD using time series data from Ethiopia, 1992-2012 The Short Run Dynamics

The Long Run Dynamics

Variables

Coef. (Std. Err)

Variables

.1939688** (.0050398) 1.606778* (.0243143) -1.41e-12** (3.77e-14) -.0252096*** (.003901) .0032911*** (.0004225)

laglnHCD

laglnHCDD1 lnGDPPD1 FDID1 lnDID1 TeleD1

lnIRD1 lnOPD1 Lngovex DWAR lagECMG _cons

-.0000661 (.0009034) .0740689** (.0021782) -.0257501** (.0010969) -.0472164** (.0031635) 7.06e-11** (1.97e-12) .5643613** (.0252448)

Computed coefficient

laglnGDPP

0.222386

laglnFDI

0.019891

laglnDI

0.019891

laglnTele

0.244285

laglnTrans

2.712432

lagIR

-0.00655

Laglngovex

-0.77169

laglnOP

-0.05064

lagDWAR

0.232171

_cons

-0.4944

Coef. (Std. Err) -.6705052* (.0683558) .1491107* (.0135342) .0133369 (.0288237) .0133369** (.0341885) .1637944** (.0590198) 1.8187* (.3774969) -.0043917*** (.0023603) -.5174243* (.1468755) -.0339555 (.0234898) .1556718* (.0372614)

-.3314964 (1.46695)

Chapter Three

Diagnostics Tests

82

F( 10, 1) = 1739.27** R-squared = 0.9999 Adj R-squared = 0.9994 Root MSE = .00128 Durbinalt= 0.2881 swilk ECM= 0.99435 VIF=8.63 hettest=0.0701** _hat=0.000 _hatsq= 0.322

F( 10, 9) = 38.40* R-squared = 0.9771 Adj R-squared = 0.9517 Root MSE = .05392 Durbinalt= 0.2437 swilk ECM= 0.99435 1/VIF= .0179721 hettest=0.8125 _hat=0.000 _hatsq= 0.840

Source: Meta data of Ethiopia from NB, EIA, IMF and WB, 2013 ***significant at 10 percent, **significant at 5 percent, * significant at 1 percent

The study found that the short run effect of FDI on HCD is negative and statistically significant at 5 percent. However in the long run it has a positive but insignificant effect on the HCD. The finding of the qualitative analysis, which is largely parallel with the finding of the long run quantitative analysis, indicates that the effect of FDI on the HCD is positive. In the short run, the effect of FDI on domestic investment is negative and significant at 10 percent. However the long run effect of FDI on domestic investment is positive and significant at 10 percent. Looking at the longrun effects in which the absorptive power of the HCD is controlled, a 1% increase in FDI leads to a 2% increase in the GDPP. In this regard the effect of domestic investment on GDPP seems much more significant than that of FDI. As shown in table 3.4 above, a 1% increase in DI leads to a 17% increase in GDPP. Looking at the proxies of the qualitative analysis of the impact of FDI upon DI, except for two proxies which were identified to have a positive effect, the remaining five proxies of the seven were found to have a negative effect.

Foreign Direct Investment in Ethiopia

83

Table 3.6. The short run and long run OLS estimation of FDI flows’ impact on DI using time series data from Ethiopia, 1992-2012

The Short Run Dynamics

The Long Run Dynamics

Variables

Variables

laglnDID1 lnFDID1 lnHCDD1

MSD1 lnTele

IRD1 lnEXRD1 lnOPD1 lngovexD1 DWAR lagECMG _cons

Coef. (Std. Err) -.3564609** (.0704685) -.0938632*** (.0282178) 5.79616** (1.025603)

3.41e-06** (6.80e-07) -.5071027** (.0647096)

.0060049 (.0048105) .2450501 (.7479083) -.1909791*** (.0558443) 2.435194** (.4519548) -.2259279*** (.0761474) -2.59e-11 (4.86e-11) -7.084135** (1.189148)

Computed coefficient

laglnDI laglnFDI

1.054185

laglnHCD

5.886828

laglnGDP

-51.8596

laglnMS

-36.9469

laglnTele

11.19559

laglnTrans

34.00633

lagIR

-0.13247

laglnEXR

-7.1952

laglnOP

0.594

Laglngovex

-0.76129

lagDWAR

3.293143

_cons

1567.08

Coef. (Std. Err) -.1340397 (.1660957) .1413027*** (.0707117) .7890686** (.2577605) -6.951249* (1.189141) -4.952354** (4.6475) 1.500653** (.264095) 4.558198 (1.927933) -.0177557 (.0069723) -.9644418 (.862292) .0796196 (.0778543) -.1020425 (.6055741) .4414119** (.1743809)

210.051** (80.4548)

Chapter Three

Diagnostics Tests

84 F( 11, 2) = 27.06* R-squared = 0.9933 Adj R-squared = 0.9566 Root MSE = .06794 Durbinalt= 0.8180 swilk ECM= 0.99435 VIF= 9.34 hettest= 0.4849 _hat=0.000 _hatsq= 0.063**

F( 12, 7) = 22.63* R-squared = 0.9749 Adj R-squared = 0.9318 Root MSE = .11189 Durbinalt= 0.2010 swilk ECM= 0.99435 VIF= 0.035872 hettest= 0.4194 _hat=0.000 _hatsq= 0.355

Source: Meta data of Ethiopia from NB, EIA, IMF and WB, 2013 ***significant at 10 percent, **significant at 5 percent, * significant at 1 percent

4. Conclusions and Policy Implications 4.1. Conclusions The flow of FDI to Ethiopia is not only quite low, but also highly characterized with very high volatility, even when it is compared to the sub-Saharan countries that are thought to be found largely in similar socioeconomic conditions. With this behaviour of high volatility and frequent accidental ups and downs, the average flow showed an increasing trend over the study period, with an average annual flow of Birr 3.52 billion. And even though it shows an increasing trend, its share to GDP has never been more than 5.43 percent, and from 2004 onwards this percentage share showed a continuous decreasing trend, mainly due to the insignificancy increase of its flow in relation to the surpassing increase to GDP of the nation. In the short run the macroeconomic variables, which include lagged FDI, domestic investment, the available stock and cost of human capital, the road transport network side of the infrastructure, the growth of the economy, the exchange rate side of the macroeconomic stability, and the openness side of the trade liberalization measured by the import and export as a percentage of GDP, are found to have a strong positive effect on attracting FDI. However, the MS, the skill level of the human capital, the telecom and electric power sides of the infrastructure, the inflation and foreign currency reserve sides of the macroeconomic stability, trade liberalization measured by the import and export procedures and the restrictive policies, and political instabilities affect FDI unfavourably. In the long run the domestic investment, market size, economic image, telecom and road transportation network sides of the infrastructure, growth

Foreign Direct Investment in Ethiopia

85

rate of the economy, openness side of the trade liberalization measured by the import and export as a percentage of GDP, and the political stability of the nation have a favourable effect on FDI, whereas the lagged FDI, inflation, exchange rate and foreign exchange reserve sides of the macroeconomic stability, trade liberalization measured by the import and export procedures and the restrictive policies, affect FDI unfavourably. As a same replica of the economic power of the nation to reap the benefits of FDI is negative in the short run and positive in the long run; the impact of FDI on the economy through its effect on the improvements of real GDP per capita, human capital development, and domestic investment is negative in the short run and positive in the long run.

4.2. Policy Implications Appreciating the overall endeavours and efforts of the government in the creation of the new Ethiopia; the study forwards the following remedial issues that the government better have to take into consideration in making decisions with the subject under consideration of this study. The most important catalyst in attracting FDI in all the different specifications of the quantitative and the qualitative analysis found is the DI. As such, since this DI has a striking power in attracting FDI; any macroeconomic policy designed in order to attract FDI should have to have their effect to the DI simultaneously checked as well, and it should be to the best of interest of the DI. The other problem area found in this study was the unfitting capability of the human capital of the nation. Even though the government’s effort in developing the human capital through schooling is undiminished and highly appreciable, the quality of human capital that the nation has is not as per the requirement of the FDI, which is largely assured in both the qualitative and quantitative analyses. Then the government is highly advised to have a look for solving this gap. Two strategies are forwarded by this study for solving the problem. The first is that creating an industry school relation at least starting from the secondary schools and TVETs will have a significant role in solving the problem. Second, the education and training programs of the nation should have to be opened based on an integrated demand and market analysis. For that matter the investors and other stakeholders should participate and have their requirements significantly incorporated in the development of the curriculum.

86

Chapter Three

Another result of this study is that of the proxy variables to infrastructure; the telephone line, the internet, and the electric power are largely identified to have a negative effect on attracting FDI. This is due to the frequent ON and OFF behaviour of the utilities and/or due to shortage of the supply of the utilities. If the case is due to a shortage of supply then since the government is doing significant construction on these utilities, no additional recommendation is necessary. However, since the root cause of the problem is highly expected to be with the ON and OFF behaviour of the utilities, and since this case is directly triggered to the human capital and management of each sector; the government should take a serious measure for correction of the frequent ON and OFF behaviour of the utilities. The other important finding of this study was that the macroeconomic instabilities, mainly the inflation and exchange rates of the nation, adversely affect the flow of FDI. In the essence of attracting FDIs and getting foreign currency, the value of Birr is found to be continuously depreciating. Then, on the one hand, the exchange rate of the nation is negatively affecting both FDI and DI. On the other hand, the exchange rate policy of the nation plays a significant role in the inflation problem of the nation. Therefore, taking these truths into account, the government is seriously advised to reduce the exchange rate and appreciate the value of its domestic currency. The other recommendation goes to the issue of trade liberalization. The short run effects of this openness found in the quantitative analysis, and of the import and export procedures of the nation found in the qualitative analysis, indicate that they do have a negative effect on attracting FDI. Therefore the government should look into its systems of import and export to reap the short run benefits as well. The economy is not in a position to reap the benefits of FDI in the short run. Its effect to the identified three channels also seems to be negative. But the root cause triggered two issues. The problem is directly related to the HCD of the nation. Since the human capital did not accumulate to take advantage of the FDIs, the nation’s benefit from FDI is insignificant. Then, proper application of recommendation 2 stated above is highly expected to solve the problem. The second source for the problem is lack of integration and coordination between the FDIs and DIs in the form of backward and forward integration. Therefore the government should establish an institution that facilitates such integration through organized study-based mediations and propagations between the FDIs and DIs.

Foreign Direct Investment in Ethiopia

87

References Amabachew, M.(2011). Private Investment in Ethiopia: Trends and Prospects, Proceeding of the second regional conference of the Amhara Regional State Economic Development Andreia, A. Faria, S. et al (2011) The determinants of FDI in Portugal A Sectoral Approach Dissertation submitted in partial fulfillment of requirements for the degree of Master of Science in Economics, at the Universidade Católica Portuguesa. Asiedu, E. (2002). On the Determinants of Foreign Direct Investment to Developing Countries: Is Africa Different? World Development, Vol.30, No.1, pp.107-119. Accessed on Mar 25,2012 from http://papers.ssrn.com/sol3/papers.cfm?abstract_id=280062 Asiedu, E. (2004). Policy Reform and Foreign Direct Investment in Africa: Absolute Progress but Relative Decline. Development Policy Review, 22(1): 41-48. Retrieved on Mar25, 2012 from http://papers.ssrn.com/sol3/papers.cfm?abstract_id=513443 Bardsen, G. (1989). Estimation of Long run coefficients in error correction models. Oxford bulletin of Economics and statistics, 51, 345-350 Ethiopian Economic Association, (2000). Annual Report on the Ethiopian Economy. Vol.1. 1999/2000. Addis Ababa, Ethiopia: United printers Ethiopian Investment Agency, (2013). Data base for investment profile, Addis Ababa. Ethiopian Ministry of Industry, (2012). Data base for industry profile, Addis Ababa. Getenet, A. and Hirut A. (2005), Determinants of Foreign Direct Investment in Ethiopia: A Time Series Analysis. London: Policy Studies Institute. OECD (2002). Foreign Direct Investment for Development: Maximizing benefits and minimizing cost. Paris: OECD publishing service. Retrieved on July 25, 2012 from http://www.oecd.org/dataoecd/47/51/1959815.pdf —. (2005). Investment for African Development: Making it Happen. NEPAD/ OECD Investment Initiative. Retrieved on January 10 , 2012 from http://www.oecd.org/dataoecd/57/20/34906539.pdf Pesaran, M. Shin, Y. et al (2001), Bounds Testing Approaches To The Analysis Of Level Relationships, Journal Of Applied Econometrics J. Appl. Econ. 16: 289–326 (2001); retrived on November 24, 2012 from http://dx.doi.org/10.1002/jae.616

88

Chapter Three

Pesaran. M, & Shin Y. (1999), An autoregressive distributed lag modeling approach to cointegration analysis, Chapter 11 in Econometrics and Economic Theory in the 20th Century: The Ragnar Frisch Centennial Symposium, Strom S (ed.). Cambridge University Press: Cambridge. Sapna, H. (2011), A study of FDI and Indian economy, National Institute of Technology, Kurukshetra, Deemed University, Kurukshetra, Haryana, India. Solomon, M. (2008), Determinants of Foreign Direct Investment in Ethiopia, Maastricht Graduate School of Governance, The Netherlands UNCTAD (2006). Statistics on FDI and the operations of TNCs, country profile: Egypt. Retrieved on March 15, 2012 from http://www.unctad.org/Templates/Page.asp?intItemID=3198&lang=1 —. (2008): World Investment Report 2008, available at accessed 10 April 20012. http://www.unctad.org/Templates/webflyer.asp?docid=10502& intItemID2068&lang —. (2009). World Investment Report: FDI from Developing and Transition Economies: Implications for Development. New York and Geneva: United Nations. Retrieved on Mar 15, 2012 from http://www.unctad.org/en/docs/wir2009_en.pdf —. (2010). World Investment Report: FDI from Developing and Transition Economies: Implications for Development. New York and Geneva: United Nations. —. (2011). World Investment Report: non-equity modes of international production and development. New York and Geneva: United Nations. Vichea, S. (2005). Key Factores Affecting the Performance of foreign Direct Investment in Cambodia, a thesis submitted in partial fulfillment of Masters of Business Administrations, university of the Tai chamber of ommerce.

Acknowledgment Above all, I would like to offer my deepest thanks to almighty God. AMEN! I would like to express my profound and sincere gratitude to my principal advisor Abadi Afera (assistant professor) and my co-advisor Guesh Gebremeskel (MSc) for taking their precious time in advising me throughout my study.

PART II FINANCE – DEMAND AND SUPPLY 



CHAPTER FOUR DETERMINANTS OF GROWTH IN BANK CREDIT TO THE PRIVATE SECTOR IN ETHIOPIA: A SUPPLY SIDE APPROACH MILLION ASSEFA

Abstract Despite a general awareness of the factors determining bank credits to the private sector, there is limited empirical evidence in the Ethiopian context. In this paper, the short and long run impact of bank-specific monetary policy and macroeconomic variables on bank credit to the private sector in Ethiopia, using a supply-side approach, is empirically examined for the period 1978/79-2010/11. The methodology that was employed is based on the ARDL econometric approach using annual time-series data, and follows work by Imran and Nishat (2012) for Pakistan. The study includes bank credit to the private sector as the dependent variable, while domestic deposit, foreign liabilities, lending interest rate, reserve requirement, M2 as percentage of NGDP, RGDP, and inflation are major explanatory variables. The findings indicate that domestic deposits, foreign liabilities, real lending interest rate, M2 as percentage of NGDP, GDP and inflation have a significant impact on banks’ credit to the private sector in the long run. By contrast, the reserve requirement does not affect commercial banks’ credit to the private sector both in the long and short run. Moreover, in the short run, domestic deposit and economic growth do not influence commercial banks’ credit to the private sector. The coefficient of ECMt-1 (-0.757) shows a rapid adjustment process, and dictates that the disequilibrium of the previous period shocks is adjusted into long run equilibrium in the current period. The long and short run results do not provide strong support of the influence of banking sector reform on the growth of bank credit to the private sector, as shown by the coefficient for the dummy variables in Ethiopia. Finally, the results recommend that efforts should be geared towards keeping the inflation rate low and stable.

Determinants of Growth in Bank Credit to the Private Sector in Ethiopia

91

1. Introduction 1.1 Background of the Study To promote economic advancement, private sector investment is a vital element. Finance is one factor that influences private sector investment, and it is a backbone of every firm. A growing firm needs a source of finance to assist its operational and non-operational activities. Banks are a crucial source of credit for many families and different sectors (Carbo and Rodriguez, n.d.). Commercial banks provide a lending service (grant loans and advances) to individuals, firms, and government, which may be on a short, medium, or long term basis, bearing in mind the three principles guiding their operations - profitability, liquidity, and solvency (Olokoyo, 2011). Commercial banks mobilize funds from surplus economic units (savers) in the form of deposits and provide it to the deficit economic units (ultimate borrowers) in the form of credit, and this process leads to the introduction of the credit system. This system is initially characterized by direct financing (Akpanuko & Acha, 2010), a system in which the lenders and borrowers have to search out themselves and deal directly. However, after the innovation of financial institutions, the system currently works indirectly. This means deposits are aggregated from domestic savings by financial institutions like commercial banks for lending it back to the deficit economic units. In developing countries with no stock market - like Ethiopia – the banking industry dominates the financial sector, and this sector is underdeveloped. According to the global economy report, a country is said to have a welldeveloped financial system if its banking credit to the private sector as a percentage of GDP accounts for 70% or above. In some very advanced economies it is even higher than 200%. However, in some poor countries, the amount of credit could be lower than 15% of GDP. According to Bonis and Stacchini (2006), in 2004 the ratio of loans to GDP was 46% in the U.S., 77% in France, and 100% in Germany, while the ratio of deposits to GDP was 40% in the U.S., 68% in France, and 86% in Spain. In general, credit to the private sector in Ethiopia is limited and concentrated in urban centres, performed mainly by banking institutions. Bank credit to the private sector as a percentage of GDP in Ethiopia is below 15% even after the reform, as the graph below shows. Figures 1.1 and 1.2 below show that in the case of Ethiopia, the domestic credit by the banking sector to the private sector has increased from 5.9% of GDP in 1978/79 to 10.4% in 2010/11. According to Abuka and Egesa

92

Chapter Four

(2007), one of the main advantages of financial reform is the growth of credit expansion to the private sector. Figure 1.1 below shows that bank credit to the private sector has been increasing continuously starting from 1978/79 right up to 2010/11, in particular after banking reform in 1994. From figure 1.2, however, bank credit to the private sector as a percentage of GDP, looking at the past few years, has shown a declining trend, even though in the aggregate it shows an increment (as the trend line has an upward slope in figure 1.2). Before banking reform was introduced in Ethiopia, bank credit to the private sector as a percentage of GDP was showing a declining trend. Nevertheless, after banking reform was introduced in the country, the trend shows a mixed result. Almost from 1994-2002 bank credit to the private sector as a percentage of GDP was showing an increasing trend, but from 2003 up to 2011 it has been showing a declining trend. The banking sector’s decisions to allocate credit to the economy can be influenced by different factors, including an unstable political environment of the country, the legal risk, unstable government economic policies, and investors’ own characteristics (Imran and Nishat, 2012). So, it can be difficult to conclude that the financial reform has a positive association with bank credit growth to the private sector in the case of Ethiopia. Figure 1.1: Total domestic credit provided by the banking sector to the private sector (in Birr) from 1978/79-2010/11

Source: Own Computation based on NBE data (2013)

Determinants of Growth in Bank Credit to the Private Sector in Ethiopia

93

Figure 1.2: Domestic credit provided by the banking sector to the private sector (% GDP) from 1978/79-2010/11

Source: Own Computation based on NBE data base (2013)

Imran and Nishat (2012) suggest that it is an important issue to discuss the factors that influence bank credit supply due to the growing trend of bank loans in the world economies, especially in emerging markets from the banks’ point of view.

1.2 Statement of the Problem In order to identify the factors that affect the growth of bank credit to the private sector and to explain the relationship between those factors, different studies have been carried out in different parts of the world. According to Imran and Nishat (2012), the determinants of bank credit can be studied at the demand side (firms’ or individuals’ access to credit) or at the supply side (financial intermediaries, like banks). Imran and Nishat (2012), Vodova (2008), Olokoyo (2011), and Djiogap and Ngomsi (2012) are some researchers whose studies of the determinants of bank credit growth to the private sector were based on a supply-side approach, while Abuka and Egesa (2007), Qayyum (2002), Afzal and Mirza (2010), Awan (2009), Khawaja (2007), Ljubaj (2007), Fetene (2010) are researchers whose studies were based on a demand-side approach. Balazs, Backe, and Zumer (2006) suggested that the supply-side studies have considered the influence of changes in the borrowers on the availability of credit, and banks’ financial positions.

94

Chapter Four

Despite a general awareness of the factors determining the bank credit to the private sector, there is limited empirical evidence provided in the literature on Ethiopia. Fetene (2010) and Sisay (2008) are among the few studies that have been done for Ethiopia considering the demand-side approach. However, the issue of bank credit determinants failed to attract the attention of researchers in Ethiopia, especially from the supply-side approach. Thus, as per the knowledge of the researcher, no study has been undertaken considering the supply-side variables of bank credit in Ethiopia, and there is a knowledge gap in the area to be filled by the present study.

1.3 Objective of the Study Therefore, by keeping the above justification in mind, the main objective of the present study attempts to empirically identify the determinants of bank credit provision to the private sector from a supply-side approach in the Ethiopian context, using a time-series data set over the period between 1978/79 & 2010/11. The study also empirically identifies if the bank credit behaviour is different after and before the banking-reform period in Ethiopia. The rest of study is organized as follows: section 2 includes literature reviews, while section 3 gives the data and estimation techniques. Empirical results are provided in section 4, and conclusions with recommendations are drawn in section 5.

2. Literature Review Imran and Nishat (2012) conducted a study on “Determinants of bank credit in Pakistan: A supply side approach” for the period between 1971 and 2010 using an ARDL model. The study concluded that in the long run, foreign liabilities, domestic deposits, economic growth, exchange rate, and monetary conditions (proxy by M2 as percentage of GDP) have a significant and positive association with private credit, while the inflation and money market rate do not affect private credit. Likewise, in the short run, all the variables are significant and positively associated with private credit except domestic deposit and inflation, which do not influence the private credit in Pakistan. According to the authors, the reason why domestic deposits do not influence bank credit in the short run may be due to banks not issuing immediate credit from the currently deposited amount by account holders. Finally, the researchers tried to find the impact of

Determinants of Growth in Bank Credit to the Private Sector in Ethiopia

95

financial liberalization on bank credit, using it as a dummy variable. However, the result did not show any impact of financial liberalization on bank credit in Pakistan. Olokoyo (2011) conducted a study of the period between 1980 and 2005 on “determinants of commercial banks’ lending behaviour in Nigeria”. It implies that the volume of deposits and the lagged volume of commercial banks’ loan and advance, investment portfolio, GDP, and foreign exchange are significant and have a positive relationship with loans and advances. The author also includes other important variables like the lending interest rate, cash reserve requirement, and liquidity reserve. These variables are positive but do not significantly influence the loan and advance. The reason for the low influence of cash reserve requirement on loan and advance is that commercial banks may not necessarily convert a lower proportion of banks’ funds available for lending. Djiogap and Ngomsi (2012) studied “determinants of bank long-term lending behaviour in the Central African Economic and Monitory Community (CEMAC)” for the period between 2001 and 2010, using a panel data model for the six countries. The study found that bank size, bank capitalization, long-term liability and GDP have a strong and positive effect on long-term bank credit to business, but inflation has an insignificant impact. Vodova’s (2008) study, entitled the “Credit market and prediction of its future development in the Czech Republic”, considered the period between 1994 and 2006, using disequilibrium models to identify the significant variables that influence demand and supply credit. Vodova summarized the possible determinants of credit demand and credit supply functions after extensive reviewing of literature and past empirical studies in a tabular form as follows: Table 2.1: Possible determinants of credit demand and credit supply functions Determinants of credit demand function Expected fixed investments or industrial production Short-term or long-term interest rate Expected inflation GDP

Exp. Sign + _ + ?

96

Chapter Four

Indebtedness of private sector Capital market index Volume of credit in previous period Difference between interest rate of loans and corporate bonds Retained profits of companies Determinants of credit supply function Deposits Bank’s capital Interest rate for loans Market capitalization of corporate bonds and shares Expected inflation Expected industrial production Volatility of prices of banks´ shares Lending capacity of banks GDP Share of capital on assets Share of classified loans on total loans Interest rate margin Profitability of banks Competition on bank market Volume of credit in previous period Difference between interest rate of loans and corporate bonds Rate of minimum required reserves Capital market index Share of created reserves and loan loss provisions on classified Cost of banks for deposits Dummy variables for specific influences (changes in regulation,

_ + + _ ? Exp. Sign + + + + _ + _ + + + _ + + ? + + _ + _ _ ?

Source: Adapted from Vodova, 2008

Even though Vodova (2008) summarized the potential determinants of demand and supply credit in the above table, he did not incorporate all the determinants in his study, because only some of the variables are suitable for the analysis of the Czech credit market. Hence, the researcher includes

Determinants of Growth in Bank Credit to the Private Sector in Ethiopia

97

the lending capacity and GDP in the supply side. The empirical analysis on the supply credit indicates that the lending capacity of banks is positively associated with bank credit. However, GDP negatively influences the real volume of bank credit. The author justified that the negative sign for GDP is potentially due to banks behaving anti-cyclically; if they expect a decline in output in the future, they can lower the credit supply in the present.

3. Data and Estimation Techniques The study has taken 33 years’ worth of annual time-series data from 1978/79-2010/11, which is a quantitative secondary source in nature. The study conducted on commercial banks of Ethiopia included both private and government owned banks. The main sources of the data were obtained from Banking Supervision Department of National Bank of Ethiopia (NBE), Central Statistical Agency (CSA) of Ethiopia, and Ministry of Finance and Economic Development (MoFED) databases. Also, other publications such as directives and various annual bulletins published by the NBE and banking institutions were reviewed. Based on the empirical and theoretical frameworks, the most relevant factors of commercial banks’ credit to the private sector are incorporated in the model. This study used the E-views statistical software package (version 3.1). The study has adopted, specified, and developed the baseline economic model of the standard literature following Imran and Nishat (2012). The explanatory variables used in previous studies such as Imran and Nishat (2012) are invoked and also extended. All the variables included in Imran and Nishat’s study are found to be relevant also in the Ethiopian context, except for the money market rate and exchange rate. This is because the money market rate only has an existence of a short time period since it was introduced in Ethiopia. In addition, both the money market rate and exchange rate are not strongly supported by the loanable funds theory and other empirical studies. Apart from these indicators, this study also includes a number of additional variables that are potentially relevant in the Ethiopian context and strongly supported by the loanable funds theory and the previous empirical studies, such as cash reserve requirement and lending interest rate. The model used in this study is formulated in equation (1) as follows: ‫ ݎܥܤ‬ൌ ߚ଴ ൅ ߚଵ ‫ ୲ܦܦ‬൅ ߚଶ ‫ ୲ܮܨ‬൅ ߚଷ ‫ ୲ܴܫܮ‬൅ ߚସ ܴܴ୲ ൅ ߚହ ‫୲ܩʹܯ‬ ൅ߚ଺ ܴ‫ ୲ܲܦܩ‬൅ ߚ଻ ‫ ୲ܫܲܥ‬൅ ߤ୲ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ Ǥ ǥ ǥ ሺͳሻ

Chapter Four

98

where BCr is bank credit to the private sector, DD is domestic deposits with commercial banks, FL is foreign liability with commercial banks, LIR is lending interest rate, RR is reserve requirement, M2G is broad money as a percentage of nominal GDP, RGDP is real gross domestic product, CPI is consumer price index, μt is the stochastic error term capturing the left over effects. Concluding from the existing literature and according to theory, the real lending interest rate, deposits by the domestic businesses and individuals, foreign liabilities, M2, and economic performance (GDP) are expected to have a positive impact on the growth of bank credit to the private sector, whereas the reserve requirement and inflation are expected to reduce the bank credit to the private sector from the supply side. To find the long and short run equilibrium relationship between the dependent and independent variables at the same time, in terms of methodology, this study used the robust econometric technique of the bound testing approach to cointegration within the framework of the Autoregressive Distributed Lag (ARDL) model. According to Imran and Nishat (2012), the ARDL approach was developed first by Pesaran (1997), Pesaran and Pesaran (1997), Pesaran and Shin (1999), and Pesaran, Shin and Smith (2001). The first step in the ARDL approach starts with a bound test for cointegration among the variables by ordinary least squares (OLS) techniques. To test the long run relationship equilibrium among the variables, F-test is used, and the ARDL representation of the equation (1) was formulated in a general form as follows:

Determinants of Growth in Bank Credit to the Private Sector in Ethiopia ୮

୮

99

୮

ο‫ ୲ݎܥܤ‬ൌ ܽ଴ ൅ ෍ ȕ୧ ο”୲ି୧ ൅ ෍ ȡ୧ ο୲ି୧ ൅ ෍ Ȗ୧ ο ୲ି୧ ୧ୀଵ

୮

୧ୀ଴

୮

୧ୀ଴

୮

൅ ෍ ij୧ ο  ୲ି୧ ൅ ෍ į୧ ο ୲ି୧ ൅ ෍ ʌ୧ οʹ୲ି୧ ୧ୀ଴ ୮

୧ୀ଴ ୮

୧ୀ଴

൅ ෍ IJ୧ ο ୲ି୧ ൅ ෍ Ȧ୧ ο ୲ି୧ ൅ Ȝଵ ”୲ିଵ ୧ୀ଴

୧ୀ଴

൅ Ȝଶ ୲ିଵ ൅ Ȝଷ ୲ିଵ ൅ Ȝସ   ୲ିଵ ൅ Ȝହ  ୲ିଵ ൅ Ȝ଺ ʹ୲ିଵ ൅ Ȝ଻  ୲ିଵ ൅ Ȝ଼  ୲ିଵ ൅ ɂ୲ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ Ǥ Ǥ ǥ ǥ ሺʹሻ where ǻ is the first difference operator of the concerned variables P is the optimal lag length The terms with Ȝ represent long run relationships The terms with Ȉ represent the short run dynamics İt is the random error. All these variables are taken in natural logarithmic form (ln) except real lending interest rate (RLIR), because it has a negative sign and is impossible to convert into logarithmic form. The ARDL approach estimates different regressions to select the optimal order of lag length for each variable. The model can be selected based on Schwartz-Bayesian Criteria (SBC) and Akaike’s Information Criteria (AIC) before the model is estimated by OLS. At the start, the null hypothesis of no cointegration against the alternative hypothesis for existence of a long run relationship is tested by using Wald or F-statistics. To implement this technique, a joint significance test was performed as: ୭ ǣȜଵ ൌ Ȝଶ ൌ Ȝଷ ൌ Ȝସ ൌ Ȝହ ൌ Ȝ଺ ൌ Ȝ଻ ൌ Ͳ against the alternative hypothesis ଵ ǣȜଵ ് Ȝଶ ് Ȝଷ ് Ȝସ ് Ȝହ ് Ȝ଺ ് Ȝ଻ ് Ͳ The null hypothesis indicates the non-existence of long run relationships, while the alternative indicates existence of long run relationships. The

Chapter Four

100

calculated F-statistics value is compared with two sets of critical values given by Narayan (2004) or Pesaran et al. (2001) for the given level of significance, for small samples size from 30-80. One set assumes that all the variables are I(0), referred to as lower bound, and the other set assumes that they are all I(1), referred to as upper bound. If the calculated Fstatistics exceed the upper critical value, then the null hypothesis of no cointegration would be rejected, irrespectively of whether the variable is I(0) or I(1), which implies that cointegration exists. If the F-statistic lies below the lower critical bounds values, then the null hypothesis of no cointegration cannot be rejected, irrespectively of whether the variable is I(0) or I(1), which means no cointegration. If the F-statistic falls into the critical bounds, the test becomes inconclusive. At this stage of the estimation process, the researchers may have to carry out the unit root tests on variables entered into the model (Pesaran and Pesaran 1997). In case the cointegration was found, that means if a long run relationship exists in the variables, the following long run model and short run model is performed from equation (3) and equation (4) respectively. The long run model is formulated as follows: ୮

୮

୮

୮

‫ ୲ݎܥܤ‬ൌ ܽଵ ൅ ෍ ȕଵ୧ ”୲ି୧ ൅ ෍ ȡଵ୧ ୲ି୧ ൅ ෍ Ȗଵ୧ ୲ି୧ ൅ ෍ ijଵ୧   ୲ି୧ ୧ୀଵ

୮

୧ୀ଴

୮

୧ୀ଴

୮

୧ୀ଴

൅ ෍ įଵ୧  ୲ି୧ ൅ ෍ ʌଵ୧ ʹ୲ି୧ ൅ ෍ IJଵ୧  ୲ି୧ ୧ୀ଴ ୮

୧ୀ଴

୧ୀ଴

൅ ෍ Ȧଵ୧  ୲ି୧ ൅ ߤ୲ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ Ǥ ǥ ǥ ǥ ǥ ሺ͵ሻ ୧ୀ଴

The ARDL specification of the short run dynamics was derived by formulating an error correction model in the following form:

Determinants of Growth in Bank Credit to the Private Sector in Ethiopia 101 ୮

୮

୮

ο‫ ୲ݎܥܤ‬ൌ ܽଶ ൅ ෍ ȕଶ୧ ο”୲ି୧ ൅ ෍ ȡଶ୧ ο୲ି୧ ൅ ෍ Ȗଶ୧ ο ୲ି୧ ୧ୀଵ

୮

୧ୀ଴

୧ୀ଴

୮

୮

൅ ෍ ijଶ୧ ο  ୲ି୧ ൅ ෍ įଶ୧ ο ୲ି୧ ൅ ෍ ʌଶ୧ οʹ୲ି୧ ୧ୀ଴ ୮

୧ୀ଴ ୮

୧ୀ଴

൅ ෍ IJଶ୧ ο ୲ି୧ ൅ ෍ Ȧଶ୧ ο ୲ି୧ ൅ ȥ୲ିଵ ୧ୀ଴

୧ୀ଴

൅ ߭௧ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ ǥ Ǥ ǥ ǥ Ǥ Ǥ ǥ ǥ Ǥ ሺͶሻ where ECMt-1 is the error correction term, which was the lagged value of the residual of the equation of the long run relationship, equation (3) in this case, ȥ is the coefficient in disequilibrium. The error correction model indicates the speed of adjustment of returning back to long run equilibrium after a short run shock. .

Finally, to ensure the fitness of the model, diagnostic and stability tests are also conducted; the diagnostic tests examine the serial correlation, functional form, normality and heteroscedasticity associated with the selected model. Cumulative sum (CUSUM) and cumulative sum of squares recursive residuals (CUSUMSQ) tests are conducted for the stability of the model.

DD 18.9* 8.57* 112.0* 1.10* 25.4* 2.13 7.36 51.06 0.00 624.0* 2.0E+22 33

FL 1.16* 0.80* 5.72* 0.04* 1.28* 1.47 5.77 22.36 0.00 38.4* 5.2E+19 33

RLIR 2.02 3.20 21.30 -24.20 9.86 -0.49 3.31 1.46 0.48 66.60 3109.11 33

Source: Author’s Computation using EVIEWS software (2013) *in billions of Birr

Mean Median Maximum Minimum Std. Dev. Skewness Kurtosis Jarque-Bera Prob. Sum Sum Sq. Dev. Obs.

BCr 8.88* 3.51* 51.9* 0.57* 12.3* 2.05 6.82 43.18 0.00 293.0* 4.8E+21 33

RR 2.25* 0.41 20.5* 0.01* 4.65* 2.75 9.73 103.7 0.00 74.1* 6.9E+20 33

4. Empirical Results

Chapter Four

Table 4.1: Summary of descriptive statistics of the variables

102

M2/NGDP 27.15 27.46 39.57 14.42 7.07 -0.01 2.33 0.62 0.73 895.90 1598.16 33

RGDP 79.8* 59.2* 415.0* 36.5* 69.3* 3.65 17.80 374.51 0.00 2630.0* 1.5E+23 33

CPI 66.52 62.60 210.20 21.60 46.01 1.68 5.39 23.48 0.00 2195.10 67731.7 33

2

2

1

1

1

1

2

1

LBCr

LDD

LFL

RLIR

LRR

LM2G

LRGDP

LCPI

-0.4985

-4.6491*

-1.2428

-0.7188

-3.9014**

-1.7061

-1.4869

-1.6543

Level

ADF

-4.1091**

___

-5.1743*

-5.3153*

___

-5.6613*

-7.4326*

-3.6520**

1 Difference

st

-3.5614

-3.5562

-3.5614

-3.5614

-3.5562

-3.5614

-3.5614

Critical value at 5% -3.5614

-1.1528

-4.7217*

-1.2082

-0.8578

-9082**

-2.0122

-1.4308

-1.8894

Level

PP

Source: Author’s Computation using EVIEWS software (2013), ADF and PP statistics with trend and intersect, *, ** & ***: statistically significant at the 1%, 5% & 10% level, respectively

AIC lags

Variables

Table 4.2: ADF and PP unit root test results on log levels of variables

-4.1258**

___

-5.1563*

-5.3099*

___

-5.7313*

-7.3943*

-3.6978**

1 Difference

st

Determinants of Growth in Bank Credit to the Private Sector in Ethiopia

-3.561

-3.556

-3.561

-3.561

-3.556

-3.561

-3.561

Critical value at 5% -3.561

I(1)

I(0)

I(1)

I(1)

I(0)

I(1)

I(1)

I(1)

Decision

103

Chapter Four

104

4.1 Econometric Model Results 4.1.1 Unit Root Test The test result shown in table 4.2 above, shows a mixture of level I(0) and I(1) of underlying variables, hence the study can proceed by the ARDL methodology. 4.1.2 Cointegration Test The cointegration result of the calculated value of F-statistic was 5.23, greater than the upper bound critical value (4.306 at 5% level of significance) suggested by Narayan (2004) for a small sample (between 30-80 observations). It implies that the null hypothesis of no cointegration cannot be accepted at 5% level of significance and, therefore, it is concluded from F-statistics that there exists cointegration or a long run relationship among the variables. 4.1.3 Estimation of Long Run Coefficients From table 4.3 below, the calculated value of the F-statistic (340.38) is greater than the upper bound of the critical value (5.966 at 1% level of significance). This implies that there exists a long run relationship. The impact of each variable is discussed in turn below: Volume of credit in the previous period: Table 4.3 shows that the tvalue of one year lagged volume of bank credit provided to the private sector affects bank credit to the private sector positively and statistically significantly at 1% in Ethiopia, in the long run. Last year, performance of bank credit provision to the private sector affects the performance of bank credit to the private sector positively in the long run. If the one year lagged volume of bank credit to private sector increases by Birr 1 billion, it has the power to increase bank credit to the private sector by Birr 0.62 billion in the long run, ceteris paribus. This result supports the study by Olokoyo (2011) that the previous year’s banks’ lending performance significantly and positively affects the current year’s bank credit performance. Domestic deposit: Table 4.3 shows that in the long run, in line with prior expectation, the t-statistics reveal that volume of domestic deposits influences the growth of bank credit to the private sector positively and statistically significantly at 10% level. Total domestic liabilities of the

Determinants of Growth in Bank Credit to the Private Sector in Ethiopia 105

banking system from non-central government mostly consist of time deposit, saving accounts, and current accounts. Olokoyo (2011) explained that total liabilities of the banking sector, used as a major source of funds for giving credit to the private sector, produce a significant result. Therefore banks should struggle hard to manage their deposits efficiently so that their objective of profitability can be achieved and the multiplier effects maintained to the maximum. This implies that generation of more deposits is tangential to the survival of Ethiopian banks as a whole. The coefficient on the domestic deposits shows that, keeping other variables constant, Birr 1 billion increase in the domestic deposits will lead to an increase in bank credit to the private sector by about Birr 1.23 billion in the long run. This coefficient shows that domestic deposits play the major role in affecting the banking sector’s credit to the private sector in the long run. The implication for the result is that as commercial banks deposits increase, their assets and liquidity also increase, and as a result they provide credit to the private sector at a domestic level in the long run (Imran and Nishat, 2012). The result supports the loanable funds theory and the empirical results by Djiogap and Ngomsi (2012), Imran and Nishat (2012), and Olokoyo (2011). Foreign liabilities: the t-value from table 4.3 shows that foreign liabilities significantly influence bank credit to the private sector positively at 1%. Similarly, as banks get loans from foreign financial institutions, their assets as well as their liquidity goes up, as a result they can lend more at the domestic level (Imran and Nishat, 2012). If foreign liabilities increase by Birr 1 billion, bank credit to the private sector will also increase by Birr 0.28 billion in the long run, ceteris paribus. It shows that the coefficient for the foreign liabilities is very low. This result also supports the loanable funds theory and the empirical results by Guo and Stepanyan (2011), and Imran and Nishat (2012). Real lending interest rate: This study also shows that real lending interest rate determines bank credit provision to the private sector positively and statistically significantly at 5% in long run. A 1% increase in lending interest rate will cause bank credit to private sector to increase by Birr 1.035 billion1 in the long run, ceteris paribus. It depicts that

 1

BCr is in natural log form while RLIR is in the level form. An anti-natural log of the coefficient of RLIR, which is 0.034167, is 1.035

Chapter Four

106

Table 4.3: Estimated long run model coefficients using ARDL approach (1, 2, 1, 1, 0, 0, 2, 1) Dependent Variable: LBCR Method: Least Squares Date: 05/10/13 Time: 03:13 Sample (adjusted): 1981 2011 Included observations: 31 after adjusting endpoints Variable Coefficient Std. Error t-Statistic C -6.690826 4.074963 -1.641935 LBCR_1

Prob. 0.1229

0.620833*

0.208820

2.973052

0.0101

1.225931***

0.695703

1.762147

0.0999

LDD_1

0.476610

0.306185

1.556610

0.1419

LDD_2

0.118688

0.268163

0.442596

0.6648

LFL

0.344118*

0.093274

3.689328

0.0024

LFL_1

-0.172212

0.122496

-1.405858

0.1816

LDD

0.034167**

0.015052

2.269953

0.0395

RLIR_1

RLIR

-0.003134

0.003776

-0.829906

0.4205

LRR

0.211559

0.134489

1.573054

0.1380

LM2

-1.452370**

0.608569

-2.386532

0.0317

LGDP

0.212854**

0.085604

2.486491

0.0261

LGDP_1

0.159034**

0.076300

2.084320

0.0559

LGDP_2

0.040937

0.108783

0.376322

0.7123

LCPI

3.036285***

1.593776

1.905089

0.0775

LCPI_1

-3.379248**

1.647359

-2.051312

0.0594

R-squared

0.997436 Mean dependent var

22.08563

Adjusted R-squared

0.994506 S.D. dependent var

1.423317

S.E. of regression

0.105502 Akaike info criterion

-1.358337

Sum squared resid

0.155828 Schwarz criterion

-0.571957

Log likelihood

38.05423 F-statistic

340.3860

Durbin-Watson stat

2.123762 Prob(F-statistic)

0.000000

Source: author’s computation using EVIEWS software (2013) *, ** and *** are statically significant at 1, 5, and 10 percent respectively. Tab. 2tailed t-values are 2.750, 2.042, and 1.697 in that order

Determinants of Growth in Bank Credit to the Private Sector in Ethiopia 107

commercial banks in Ethiopia provide a large volume of credit to the private sector in the long run, if the lending interest rate is high, as justified by the loanable funds theory. This may be due to the fact that commercial banks still have the highest market share in Ethiopia. The result is consistent with the loanable funds theory and empirical result of Olokoyo (2011). Reserve requirement: In table 4.3 above, this variable shows a positive coefficient, but it does not significantly affect bank credit to the private sector in Ethiopia. Even though the loanable funds theory and many empirical results of previous studies found a negative relationship between reserve requirement and bank credit, nevertheless, this study does not confirm it. However, Olokoyo (2011) also came up with a positive and insignificant result for the variable, similar to this finding. Therefore, the result shows that reserve requirement change has little impact on bank credit to the private sector in the long run, or the effect of high reserve requirement on commercial banks is not pronounced (Olokoyo, 2011). The insignificant impact of reserve requirement on bank credit to the private sector does not seem strange because the reserve requirement ratio does not show a noteworthy value during the sample period. Even though reserve ratio in Ethiopia has shown changes since 2007, it has been constant from 1978/79-2006/07. That is why no significant influence on bank credit was found. However, the positive sign for reserve requirement contradicts the loanable funds theory and most empirical studies. The first reason for the contradicting sign for the variable is that the reserve requirement may not necessarily convert into a lower proportion of commercial banks’ funds available for lending (Olokoyo, 2011). Second, the implication is that monetary policies such as cash reserve requirement ratios do not impact negatively on banks’ lending behaviour. Banks should therefore always ensure compliance with these policies. Third, according to Demirgüç and Huizinga (1999) banks income could be higher if the available funds would be lent out instead of reserved. This increment in commercial banks’ income leads to an increase of their credit to the private sector. Hence, the above justification may be the reason for the contradicted sign of reserve requirement in Ethiopia. M2/NGDP: As table 4.3 shows, the M2, considered as an alternative gauge of monetary conditions measured by broad money divided by nominal GDP, as Imran and Nishat used, has negatively influenced bank credit significantly at 5%, although a positive sign was expected. This result suggests that as monetary conditions of the country are going up, the growth of bank credit to the private sector will reduce in the long run. A

108

Chapter Four

1% increase in M2/NGDP will result in a Birr 1.45 billion reduction in bank credit to the private sector in the long run, with all other variables remaining constant. The sign of broad money in this study contradicts the loanable funds theory and the empirical result of Imran and Nishat (2012). It can be concluded that when money in circulation increases, the supply of credit to the private sector will decrease, because fast growth in broad money can result in more money in circulation, which in return can cause higher inflation in the country. Hence, high inflation can reduce bank credit to the private sector, because the households’ and firms’ capacity to deposit will reduce. According to the economic review of the global economy report, the M2 measure includes the money in circulation as well as bank deposits such as demand, time, and savings accounts, and a fast growth in broad money results in higher inflation, after a few years or longer. GDP: Strong economic conditions, measured by real GDP, have produced the result as expected. The coefficient is positive and statistically significant at 5% to affect bank credit to the private sector in the case of Ethiopia. In Imran and Nishat (2012), an increase in real GDP boosts up the manufacturing sector’s income, as well as the general peoples’ earnings, which leads to higher domestic deposits, thus increasing the liquidity of banks and enabling them to lend more for investment needs. So, GDP has a positive association with the growth of bank credit to the private sector. A Birr 1 billion increase in GDP will increase bank credit to the private sector by Birr 0.21 billion in the long run, with all other variables remaining constant. This finding is consistent with the loanable funds theory and previous empirical evidence of Djiogap and Ngomsi (2012), Guo and Stepanyan (2011), Imran and Nishat (2012), and Olokoyo (2011). Inflation: Current year expected inflation significantly impacts the growth of bank credit to the private sector positively at 10%, which indicates that bank credit to the private sector also increases with expected inflation in the long run. In the current year, if the price level is expected to increase by 1%, bank credit will increase in the current year by Birr 3.04 billion. Inflation has the highest coefficient value of 3.04. This explanatory variable has the highest impact and influence on the lending behaviour of commercial banks, and a change in it will yield the highest change in bank loans and advances. This result shows that commercial banks provide a large amount of credit to the private sector in the current year, if inflation is expected to increase in the next year (Djiogap and Ngomsi, 2012). However, the credit will decrease starting from the next year (i.e., at the

Determinants of Growth in Bank Credit to the Private Sector in Ethiopia 109

time the credit actually increased), in the long run. In the previous year, if the price level was expected to increase by 1%, bank credit to the private sector would decrease in the long run by Birr 3.38 billion, with all other things remaining constant. The result indicates that commercial banks reduce their credit to the private sector because they are not willing to provide credit at a lower real lending interest rate. A real lending interest rate is the result of the nominal lending interest rate minus the inflation rate. The result of this variable is as expected, and supports the loanable funds theory and the empirical results by Guo and Stepanyan (2011) and Imran and Nishat (2012). 4.1.4 Estimation of Short Run Coefficients For the short run dynamics, equation (4) was estimated. In this equation, the error correction term ECMt-1 is the lagged value of the residual of equation (3) of the long run relationship. The results are reported in table 4.4 below. Most of the results are similar in both the long run and short run. However, in the short run some difference exists, domestic deposits and GDP do not significantly influence bank credit to the private sector in the short run. Nevertheless, the impact of each variable is discussed in turn below: Volume of credit in the previous period: from table 4.4, the t-value shows that one year lagged volume of bank credit to the private sector significantly influences the growth of bank credit to the private sector in the short run period positively at 10% level. If last year bank credit to the private sector increased by Birr 1 billion, it will result in an increase of bank credit to the private sector in the short run by Birr 0.44 billion, with all other things remaining constant. This result is the same as the long run result.

110

Chapter Four

Table 4.4: Estimated short run model coefficients using ARDL approach (2, 2, 1, 0, 1, 1, 2, 1) Dependent Variable: DLBCR Method: Least Squares Date: 05/10/13 Time: 03:17 Sample (adjusted): 1982 2011 Included observations: 30 after adjusting endpoints Variable Coefficient Std. Error t-Statistic Prob. C 0.036145 0.059628 0.606171 0.5567 DLBCR_1 0.444323*** 0.224356 1.980439 0.0732 DLBCR_2 -0.157690 0.098215 -1.605564 0.1367 DLDD 0.393978 0.447324 0.880746 0.3973 DLDD_1 0.343243 0.325610 1.054154 0.3144 DLDD_2 0.075686 0.221911 0.341066 0.7395 DLFL 0.288698* 0.064207 4.496377 0.0009 DLFL_1 -0.009527 0.116711 -0.081633 0.9364 DRLIR 0.041396* 0.011672 3.546532 0.0046 DLRR 0.080422 0.102656 0.783416 0.4499 DLRR_1 -0.045883 0.106588 -0.430474 0.6752 DLM2 -1.051384** 0.440919 -2.384529 0.0362 DLM2_1 -0.257503 0.394904 -0.652066 0.5277 DLGDP 0.125365 0.099771 1.256529 0.2350 DLGDP_1 0.067856 0.086068 0.788406 0.4471 DLGDP_2 0.033434 0.085229 0.392290 0.7023 DLCPI 3.226683** 1.263782 2.553196 0.0268 DLCPI_1 -4.318032* 1.364659 -3.164185 0.0090 ECM_1 -0.757553** 0.315700 -2.399601 0.0353 R-squared 0.946051 Mean dependent var 0.132093 Adjusted R-squared 0.857770 S.D. dependent var 0.237144 S.E. of regression 0.089435 Akaike info criterion -1.727249 Sum squared resid 0.087984 Schwarz criterion -0.839824 Log likelihood 44.90874 F-statistic 10.71643 Durbin-Watson stat 1.770078 Prob(F-statistic) 0.000146 Source: Author’s computation using EVIEWS software (2013), *, ** and *** are statically significant at 1, 5, and 10 percent respectively. Tab. 2tailed t-values are 2.750, 2.042 and 1.697 in that order.

Determinants of Growth in Bank Credit to the Private Sector in Ethiopia 111

Domestic deposit: volume of domestic deposits is insignificant to influence bank credit to the private sector in the short run. According to Imran and Nishat (2012) the reason could be that banks do not issue loans immediately from the currently deposited amount by account holders. From the perspective of the loanable funds theory, the result is correct, as the sign is positive. It means that any increase of bank funds will be transformed into credit. However, the coefficient is very low, indicating the low sensitivity of the variable in the short run. It may provide evidence that the banking industry needs a structural overhaul to make it more sensible to the deposit change (Mongid, n.d.). Foreign liability: foreign liability has a positive influence on bank credit positively and that is statistically significant at 1% in the short- run also.. A Birr 1 billion increase in foreign liability will lead to an increase in bank credit provision to the private sector byof Birr 0.288 billion, ceteris paribus. This result is also the same as in the case of the long- run. Real lending interest rate: in the short- run, as shown in table 4.4 above, the real lending interest rate influences bank credit positively and statistically significantsignificantly, at 1%. A 1% increase in the real lending interest rate will cause an increase of bank credit to the private sector by Birr 1.043 billion3 in the short- run. The result is similar with the result in the case of the long- run. Reserve requirement: in the short run, reserve requirement is also not significant and has a positive sign, as in the case of the long- run to affect bank credit provision to the private sector. This result still contradicts the theory and previous empirical results. M2/NGDP: in the short- run, the monetary condition measured by broad money as a ratio of NGDP is also significant and has a negative sign, as in the case of the long- run, at 5%. A Birr 1% increasesincrease in broad money as a percentage of NGDP will reduce bank credit to the private sector by Birr 1.051 billion in the short- run. Similarly, in the short- run also, fast growth of money in circulation may result in high inflation in the country, which also leads to reduction in bank credit provision to the private sector.

 3

BCr is in natural log form while RLIR is in the level form. An anti-natural log of the coefficient of RLIR, which is 0.041396, is 1.043

112

Chapter Four

RGDP: The coefficient of real GDP in table 4.4 above is positive, as expected, but it is not significant. This result may hold because the economic condition could not generate enough additional domestic deposits in the short run in Ethiopia. That is why it does not influence bank credit provision to the private sector significantly during the study period. Inflation: Current year expected inflation significantly impacts the bank credit to the private sector, positively at 5% in the short run, which indicates that bank credit to the private sector also increases with expected inflation. If the price level is expected to increase by 1% in the next year, bank credit will increase in the short run by Birr 3.226 billion. This result shows that commercial banks provide a large amount of credit to the private sector in the current year, if inflation is expected to increase in the next year. However, the credit will decrease starting from the next year, at the time the credit increases actually in the short run. The result indicates that commercial banks are not willing to provide credit at lower real lending interest rates, because the real lending interest rate is the result of the nominal lending interest rate minus the inflation rate. If the price level is expected to increase by 1%, it will result in a decrease in bank credit provision to the private sector by Birr 4.32 billion in the short run. Error correction mechanism (ECM): The result of the error correction model given in table 4.4 demonstrates that the lagged error correction term ECMt-1 is negative and highly significant, as expected. Its coefficient of ECMt-1 (-0.757) shows a rapid adjustment process and dictates that the disequilibrium of the previous period shocks is adjusted to long run equilibrium in the current period. Banking reform: Another factor that can affect bank credit to the private sector is financial liberalization reforms. Ethiopia adopted these reforms in 1994 to promote the financial sector. To capture the impact of financial liberalization reforms on bank credit to the private sector, the study estimates models, incorporating a dummy variable with value 1 from 1993/94 to 2010/11, and 0 otherwise. The results do not exhibit a different pattern. The long and short run results do not provide strong support of the influence of financial sector reform on private credit. This suggests that there is more to be done by the government in terms of liberalizing the financial sectors further. Finally, this study performs a number of diagnostic tests to the ECM. The results of those diagnostic tests proved that the model has no serial

Determinants of Growth in Bank Credit to the Private Sector in Ethiopia 113

correlation and heteroscedasticity (ARCH effect); the residual is normally distributed, the model has no omitted variables, and the functional form of the model is well specified. Figures 6.1 and 6.2 (in the appendix) plot the results for CUSUM and CUSUMSQ tests. The plots of CUSUM and CUSUMSQ statistics lie well within the critical bounds, implying that all coefficients of the short run model (ECM) are stable. In other words, the results indicate the absence of any instability of the coefficients because the plot of the CUSUM and CUSUMSQ statistics falls inside the critical bands of the 5% confidence interval of parameter stability.

5. Conclusions and Recommendations In Ethiopia, commercial banks remain dominant in the banking system, in terms of their shares of total assets and deposit liabilities. A major component of their total credits to the private sector is still on the increase in spite of the major constraints placed by government regulations, institutional constraints, and other macroeconomic factors. Some previous studies regarding determinants of bank credit in Ethiopia mainly focused on demand side factors. This study is an endeavour to examine empirically the determinants of bank credit growth to the private sector at the supply side. This study examines whether monetary policy variables, banking sector variables, and macroeconomic variables play an important role in determining the bank credit growth to private sector. In general, the empirical result of this study supports the loanable funds theory and the base line indicator of Imran and Nishat (2012), as well as other empirical studies, except for the variables of reserve requirement and money supply. Therefore, both government and commercial banks should be aware of the fact that the environment in which they operate is an important factor in the bank performance and behaviour. Based on the study’s findings, the researcher recommends the following points: ¾ Domestic deposits do not affect bank credit to the private sector in the short run. The reason may be that commercial banks do not generate sufficient deposits in the short period. Therefore, commercial banks in Ethiopia should focus on mobilizing more deposits by planning on how to attract and retain more deposits so as to further improve their short-period lending performance. It can be also achieved if banks expand new branches in rural areas and

Chapter Four

114

¾

¾

¾

¾

introduce new and fast banking innovations or technology to attract and retain customers. The findings of macroeconomic factors revealed that inflation negatively affects bank credit to the private sector. As inflation increases, the purchasing power of money lodged in deposit accounts is reduced, to the extent that savers are forced to pay an inflation tax. Thus, the Ethiopian government should be committed to maintaining a low and stable inflation rate, because higher inflation rate fluctuations will lead to macroeconomic instability. So, maintaining macroeconomic stability is crucial in making commercial banks increase their credit to the private sector. In addition, the government should gear its effort towards reducing inflation in order to arrest its negative impact on real interest rates. In Ethiopia, broad money as a percentage of GDP has influenced the growth of bank credit to the private sector inversely. Broad money in Ethiopia comprises money in circulation, demand deposits, and time and saving deposits. Many studies found that higher broad money (money in circulation) can cause inflation in the country. Therefore, the monetary authority should focus more on control of the broad money to stabilize inflation. The National Bank of Ethiopia’s capacity for controlling and guiding the activities of financial institutions and financial intermediaries should be strengthened and must be geared towards controlling the inflation rate and enhancing economic growth. The banking reform has no positive effect on the growth of bank credit to the private sector because there is more need for deregulation of the interest rate ceiling. This rate has been negative in recent years due to high inflation in Ethiopia. Therefore, the monetary authority should be doing more in terms of banking sector liberalization by deregulating the interest rate ceiling so that savings could be mobilized to promote the availability of loanable funds for credit.

Finally, it is necessary for future research to focus on the demand side of bank credit growth to the private sector, in addition to the supply side.

Determinants of Growth in Bank Credit to the Private Sector in Ethiopia 115

References Abuka, C. A., & Egesa, K. A. (2007). An assessment of private sector credit evolution in the East African Community: The candidates for a region wide reform strategy for the financial sector. A work in progress submitted to the CSAE Conference 2007. Afzal, A., & Mirza. (2010). Determinants of interest rate spread in Pakistan commercial banking bector (CREB Working Paper WP/01/10). Awan, A. G. (2009). Comparison of Islamic and conventional banking in Pakistan. Proceeding 2nd CBRC, Lahore. Balazs, E., Backe, P., & Zumer, T. (2006). Credit growth in Central and Eastern Europe new (over) shooting stars? (ECB Working Paper, No. 687). Bonis. R and Stacchini M. (2006). Which are the determinants of the size of banking loans in industrialized countries? Carbo, S., & Rodriguez, F. (n.d.). Microeconomic determinants of bank lending: An application to the Spanish case. Demirgüç, K. A., & Huizinga, H. (1999). Determinants of commercial bank interest margins and profitability: Some international evidence. The World Bank Economic Review, 13(2), 379-408. Djiogap, & Ngomsi. (2012). Determinants of bank long-term lending behavior in the Central African Economic and Monetary Community (CEMAC). Academic Research Centre of Canada. Engle, R. F., & Granger, C. W. J. (1987). Co-integration and error correction: Representation, estimation, and testing. Econometrica, 55(2), 251-276 Fetene, Z. (2010). Access to finance and its challenge for small business enterprises (Unpublished Master’s thesis). Mekelle University, Mekelle, Tigray, Ethiopia. Imran, K., & Nishat, M. (2012). Determinants of bank credit in Pakistan: A supply side approach. Proceedings of 2nd International Conference on Business Management. Khawaja, I. (2007). Determinants of interest spread in Pakistan (PIDE Working Paper, WP/22). Ljubaj, I., Martinis, A., Mrkalj, M., & Turkalj, K. G. (2007). Estimating credit demand in Croatia. Mohanty, M. S., Schnabel, G., & Garcia, L. P. (2006). Banks and aggregate credit: What is new? BIS Papers, No. 28.

116

Chapter Four

Mongid, A. (n.d). The determinants of bank credit growth in Indonesia during 1992 –2004: A supply side approach. Center for Research and Community Service (PPPM). Olokoyo, F. O. (2011). Determinants of commercial banks’ lending behavior in Nigeria. International Journal of Financial Research, 2(2), 61-72. doi:10.5430/ijfr.v2n2p61 Pesaran, M. H. (1997). The role of economic theory in modeling the long run. Economic Journal, 107, 178-91. Pesaran, M. H., & Pesaran, B. (1997). Working with Microfit 4.0: Interactive econometric analysis. Oxford: Oxford University Press. Pesaran, M. H., & Shin, Y. (1999). An autoregressive Distributed Lag Modeling Approach to Co-integration Analysis. Chapter 11 in econometrics and economic theory in the 20th Century: The Ragnar Frisch Centennial Symposium, Strom S. Cambridge: Cambridge University Press. Pesaran, M. H., Shin, Y., & Smith, R. J. (2001). Bound testing approaches to the analysis of the long run relationships. Journal of Applied Economic, 16, 289-326. Qayyum, A. (2002). Demand for bank lending by private business sector in Pakistan. The Pakistan Development Review, 41(2), 149-159. Sisay, Y. (2008). Determinants of Smallholder Farmers Access To Formal Credit: The Case of Metema Woreda, North Gondar, Ethiopia (Master’s thesis). Haramaya University, Vodova, P. (2008). Credit market and prediction of its future development in the Czech Republic.

Determinants of Growth in Bank Credit to the Private Sector in Ethiopia 117

Appendix Figure 6.1: Plot of CUSUM Test for equation (3) 10.0 7.5 5.0 2.5 0.0 -2.5 -5.0 -7.5 -10.0 01

02

03

04

05

06

CUSUM

07

08

09

10

11

10

11

5% Significance

Source: Author’s Computation using EVIEWS software (2013) The straight lines represent critical bounds at 5% significance level.

Figure 6.2: Plot of CUSUMSQ Test for equation (3) 1.6

1.2

0.8

0.4

0.0

-0.4 01

02

03

04

05

06

CUSUM of Squares

07

08

09

5% Significance

Source: Author’s Computation using EVIEWS software (2013) The straight lines represent critical bounds at 5% significance level.



CHAPTER FIVE DETERMINANTS OF TRADE CREDIT USE BY PRIVATE TRADERS: EVIDENCE FROM MEKELLE CITY, TIGRAY DEREJE GETACHEW

Abstract This study aims to investigate the determinants of trade credit use by taking a sample of 198 private traders in Mekelle city in the Tigray regional state of Ethiopia. A semi-structured questionnaire and interviews were used to collect data, and a binary logistic regression model was used to examine significant factors determining trade credit use. The result highlighted the fact that trade credit was widely practiced among private traders in Mekelle city. It has been found that about 58 percent of sample traders found in Mekelle city were trade credit users, and about 42 percent were non-users. The result of the binary logistic regression model shows that from owner factors, gender and education of traders significantly determined trade credit use. Similarly, business specific factors such as age of the business, length of trade relation, and frequency and volume of purchase were found to be significant variables in determining trade credit use. Therefore, private traders and government offices that are concerned with the promotion of trade and private sector development need to take these factors into consideration in order to enhance trade credit use by private traders.

1. Introduction The importance of financial sector development, where banks play a special role in providing enterprises with necessary external financing, is widely recognized. However, the question of creditworthiness for small business remains unanswered in getting access to bank loans. As an

Determinants of Trade Credit Use by Private Traders

119

alternative for such enterprises, trade credit financing is becoming a common phenomenon in developing countries, as well as in developed economies (Cheng & Pike, 2003; Gustafson, 2004; Van Horen, 2007). The use of trade credit improves investment incentives because suppliers lend the products that are less easily diverted than cash, i.e., enterprises have less diversion opportunity for unintended purposes. Trade credit amounts to 15-20% of GDP in Canada, USA, Great Britain, and 55-60% in Japan (Shaffer, 2000). Moreover, recent evidence from developing countries suggests that trade credit is the major source of finance for small and informal enterprises (Huyghebaert, 2006) because emerging economies are characterized by an undeveloped financial system, and a high proportion of enterprises in these countries are financially constrained. In Ethiopia also, a survey carried out by EDRI (2003) indicated that trade credit from suppliers was singled out as the most important source of short term finance for private businesses, representing 17% of their working capital finance. It was also rated higher than borrowing from other informal and formal sources. Moreover, Gebrehiwot and Wolday (2006) showed that more than half of sampled MSEs used trade credit for their business finance. However, many potential enterprises (i.e., many credit hungry customers) were unable to access trade credit due to many factors, among which are familiarity between the supplier and the customer on the grounds of relatives, religion, and friendship (Fatoki & Odeyemi, 2010). Several studies have investigated the reasons for the use of trade credit, focusing on either the supply or demand side. Interestingly, however, there is no consensus among prior studies about the determinants of trade credit supply and demand. For example, contrasting results were reported with respect to the effect of access to external finance and transaction cost considerations on trade credit supply and demand (Chant & Walker, 1988; Cheng & Pike, 2003; Fisman, 2003; Fisman & Raturi, 2004; McMillan & Woodruff, 1999; Petersen & Rajan, 1997). Some of these contradictory results may be explained by differences between countries (Fisman & Raturi, 2004) and/or markets (Giannetti, Burkart, & Ellingsen, 2008). Despite the potential importance of trade credit, limited attention has been paid to its role and use, especially in developing countries in general, and none in the case of Ethiopia. Hence, identifying the determinant factors affecting private traders’ use of trade credit in financially underdeveloped countries is essential given that these factors are dynamic, based on disparities in the countries’ context. Therefore, this study aimed to investigate the determinant factors of private traders’ trade credit use in Mekelle city by taking owner and business characteristics into consideration.

Chapter Five

120

2.

The Literature on Trade Credit

Trade credit from a customer point of view is defined as a loan a customer receives from its supplier in conjunction with purchase. For the buyer, it is a source of finance through accounts payable, while for the seller, trade credit is an investment in accounts receivable. It is usually extended for an intermediate period of thirty to sixty days, at which point payment is due. As pointed out by Fafchamps (1997), trade credit has an attractive feature of not being guaranteed by mortgageable assets (collateral), which is advantageous for enterprises lacking collateralizable assets.

2.1 Theories of Trade Credit Many reasons have been put forward to explain why firms may offer or accept trade credit. Below are the main arguments of trade credit theories. 2.1.1 Financing Theory The financing theory suggests that firms that do not have access to bank loans will have a higher demand for trade credit, since it may be an important source of short-term finance (Giannetti et al., 2008; Huyghebaert, 2006; Nielsen, 2002; Petersen & Rajan, 1997). The argument here is that bank loans and trade credit are considered as substitute sources of finance. Opposing views to the financing theory argued that firms may still demand more trade credit even if they have access to bank loans (Chant & Walker, 1988; McMillan & Woodruff, 1999). They considered trade credit and bank loans as complementary sources of finance. Their argument was that a firm might take more trade credit, even if it has access to bank credit, maybe because the firm uses bank credit for financing business expansion such as buying assets, and trade credit to finance the purchase of goods. 2.1.2 Transaction Cost Advantage Theory Larger quantity purchase encourages demand for trade credit due to the financial constraints customers may face (Elliehausen & Wolken, 1993). According to this theory, the firm has two choices. Either to accumulate costly inventories (which may or may not be sold in later periods) or offer trade credit to its customers who may be financially constrained. A tradeoff clearly exists between carrying inventories and offering trade credit. Therefore, the idea of this theory is that purchasing large quantities will result in economies of scale and reduced fixed costs (e.g., transport costs).

Determinants of Trade Credit Use by Private Traders

121

Moreover, buying larger quantities encourages the demand for trade credit due to the financial constraints buyers may experience (Elliehausen & Wolken, 1993). Thus, the demand for trade credit grows with increases in the volume of purchases. 2.1.3 Marketing Theory In an environment with many suppliers, customers may switch easily if there are no incentives to retain a customer of a supplier (Fisman & Raturi, 2004). Providing trade credit may be one instrument to retain customers. A supporting argument by Pike and Cheng (2003) related to the marketing theory states the importance of competitive pressure in the market as a reason for offering trade credit. 2.1.4 Liquidity Theory This theory, first suggested by Emery (1984), proposes that credit rationed firms use more trade credit than those with normal access to financial institutions. The central point of this idea is that when a firm is financially constrained, the offer of trade credit can make up for the reduction of the credit offer from financial institutions.

2.2 Empirical Literature Vaidya (2011) conducted an investigation on the determinants of trade credit, taking a sample of Indian manufacturing firms, and found an inventory management motive for offering trade credit. Firms attempt to increase sales and lower finished goods inventories by offering trade credit, both on a gross and net basis. When inventories of finished goods, semi-finished goods, and raw materials rise, firms tend to postpone payments to their supplier, and this shows up on their books of accounts as higher accounts payable. This is likely to help firms tide over negative shocks to sales. Thus, trade credit in general can be seen to arise as a financial response to variable demand for their finished goods. Highly profitable firms were found to give (on both net and gross basis) and receive less trade credit. There could be many underlying results for this finding. Firstly, more profitable firms may not face a major problem with respect to variability of demand for their product. Firms with greater access to bank credit offer less trade credit to their customers. Firms with more access to bank funds do not pass them on to their buyers as accounts receivable. On the other hand, firms with higher bank loans receive more trade credit. The empirical results on the determinants of trade credit in

122

Chapter Five

India supports the complementarity hypothesis of trade credit and bank loans. Cole and Mehran (2011) conducted a study on gender and the availability of trade credit to privately held firms with evidence from the surveys of small business finances in the USA by using multivariate analysis, and found strong evidence of significant univariate differences. Specifically, female-owned firms are significantly more likely to be credit-constrained because they are more likely to be discouraged from applying for credit, and more likely to be denied credit when they do apply. However, these differences are rendered insignificant when they control other firm and owner characteristics. This evidence suggests that observed gender differences in trade credit availability are attributed to other differences in male-owned and female-owned firms, such as the firm’s size and industry and the owner’s age, experience, and educational attainment. Cole (2010) analysed factors affecting trade credit use. The results showed that firms using trade credit were larger, more liquid, of worse credit quality and older, whereas firms that were non-trade credit users were found to be significantly smaller, more profitable, of better credit quality and hold sizable tangible assets. It was found that owners of firms that used trade credit were older, less educated, less likely to be female, and were of the same ethnicity while owners of bank credit users (non-trade credit users) were found to be younger, more educated, and ethnically diversified. The result of this study showed that the amount of trade credit used increases with firm age, which indicated that older firms were reputable and had an information advantage to heavily rely on trade credit. It also found that the amount of trade credit used was greater when the firm credit quality was worse because they hold only fewer tangible assets that can be pledged as bank loan collateral. Finally, it supported the financing advantage theory of trade credit and was consistent with the pecking order theory of capital structure. A study made by Teruel and Solano (2008) on the determinants of trade credit use in British small and medium-sized firms by using a dynamic panel data model and employing the GMM method revealed that the availability of alternative financial resources leads to reduced financing from suppliers. Larger firms use less credit from suppliers since they can go to other sources of financing as a consequence of their trade capacity and reputation. Besides, the result showed that decisions about trade credit depend on the ability of the firm to obtain other forms of funding, and confirmed a substitution effect between supplier-provided credit and other

Determinants of Trade Credit Use by Private Traders

123

sources of financing. Finally, these sorts of decisions are affected by the economic environment. It means that there is no static factor that determines whether to use trade credit or not in order to finance short-term operation. Empirical evidence indicated that trade credit has an effect on firm profitability. Sola, Teruel, and Solano (2008) found that there is a positive linear relationship between investment in trade credit and firm profitability, derived from the fact that the benefits associated with trade credit surpass the costs of vendor financing of Spanish SMEs. Niskanen and Niskanen (2006), in their empirical examination of the determinants of Finnish firms’ trade credit practices, found that creditworthiness and access to capital markets are important determinants of trade credit extended by sellers. The result also showed that neither bank relationships nor competition between banks explains the levels of trade credit that sellers extend. Instead, the importance of the availability of external funds is reflected in a firm’s location in either a rural or urban area. Additionally, the level of purchases is an important explanatory variable for trade credit. The result showed that larger and older firms use less trade credit than smaller and younger firms do. The other significant determinants of accounts payable were found to be the strength of internal financing as an alternative source of capital, asset maturity, loan restructuring, and number of banks operating in the country. It suggested that alternative financing is better available in rural areas. Moreover, they found that financially constrained firms use more trade credit as an alternative source of funding, and that relationship banking increases loan availability to firms. Levchuk (2002) carried out a study on trade credit determinants of Ukrainian enterprises and tested the substitution hypothesis of the effect between trade credit and bank loans. He found that enterprises with better access to bank lending were found to have less trade credit. This empirical evidence was similarly supported by a recent study that found the trade credit substitution hypothesis accepted, in which trade credit was found to be an alternative source of finance (Santos, Sheng, & Bortoluzzo, 2011). In conclusion, empirical evidences about developed countries regarding trade credit demand found conflicting results in their testing of trade credit theories discussed above. Thus, the inconsistency of these empirical findings shows that the factors about trade credit use are dynamic and need

124

Chapter Five

regular empirical work. The following discussion looks at empirical evidence about trade credit use in Africa. 2.2.1 Empirical Evidence with Regard to Africa The other empirical evidence with regard to trade credit is a recent study by Ojenike and Olowoniyi (2012) on determinants of trade credit use by randomly selected firms in Nigeria. The study employed fixed effect panel data for the period 2000-2009. The result revealed that several variables such as depreciation value, sales value, institutional loan, tangibility, profit, and the current assets of the firms were the main determinants of trade credit in Nigeria. The result implied that retained earnings of the sampled firms were relatively higher, leading to its non-significant influence on the use of trade credit. The operating expense incurred by the sampled firms was also high. Tangible assets were also observed for the sampled firms. There was a low demand for trade credit because the findings indicate that sampled firms in Nigeria had enough working capital. Moreover, Guy and Mazra (2012), who conducted a study on determinants of trade credit demand taking a sample of Cameroonian firms and using a fixed effect logit model, found that trade credit is used more by companies enjoying a better reputation from banks than those with a lower reputation. However, the price of a bank loan, quantified by the amount of short-term bank credit, importantly affects the need for trade credit. This study reported that the transaction cost advantage theory has less explanatory power, whereas financing theory significantly correlated with the sample companies’ behaviour. Furthermore, trade credit is more influenced by the nature, frequency, and amount of transactions. The share capital of managers positively influenced the length of the trade credit period. Fatoki and Odeyemi (2010) conducted a study on the determinants of access to trade credit by new SMEs in South Africa by using logistic regression. Their finding showed that from 417 respondents, only 71 were able to access trade credit. The results of the logistic regression indicated that managerial competency, the availability of a business plan, belonging to trade associations, previous relationships, location, business size, insurance, and incorporation were significant determinants of access to trade credit by new SMEs in South Africa. Hermes, Kihanga, Lensink, and Lutz (2010) conducted an empirical investigation on determinants of trade credit demand and supply in the

Determinants of Trade Credit Use by Private Traders

125

Tanzanian rice market by applying the structural modelling approach. This study found that being of the same ethnic group leads to higher demand; it does not influence trade credit supply, suggesting that ethnic ties may work as a signal for repaying trade credit; demanding trade credit from a wholesaler of the same ethnic group raises the social pressure to repay. Moreover, having a long-term trade relationship influences trade credit supply, but it is not related to trade credit demand. Thus, for suppliers, the length of the relationship is important as a source of information with respect to the probability of repayment of credit; for customers, establishing a long-term trade relationship with a wholesaler is not important, which seems to be in line with another finding (Fisman & Raturi, 2004), showing that customer switching is a real threat in the market. Finally, they concluded also that the frequency of purchase is an important determinant for supply, but not for demand of trade credit. The study conducted by Fafchamps (1996) on trade credit in Zimbabwean manufacturing companies found that a significant number of manufacturing firms were rationed in terms of access to input credit. It is found that controlling for firm size, age, and sector, many firms were less likely to obtain trade credit because owners’ personal relationships were very rare. Results suggested that the inability to use trade credit is an obstacle to small firms to compete successfully with large firms. It was concluded that most suppliers are significantly less likely to provide trade credit to first time customers, assuming these customers are newer and thus more vulnerable to risk, hence they are generally less reliable in repaying loans. Regarding the supplier-customer relationship, the result showed that a good and continued relationship between supplier and customer is very important, not only for the purpose of getting access to trade credit and flexibility in repayment, but that it also ensured that supplies are available, reliable, and of good quality. Kimuyu and Omiti (2000), in their study on institutional impediments to access to trade credit by MSEs in Kenya, found that many entrepreneurs invest time in developing long-term relationships and a good reputation with their suppliers. The short life expectancy of many MSEs reduces opportunities for building such relationships. Some entrepreneurs report receipt of supplier credit, and many more extend credit to their customers. But, credit-related disputes appear common and difficult to resolve due to difficulties in enforcing contracts. This reduces the use of supplier credit and the benefits associated with it.

126

Chapter Five

In summary, empirical evidence in Africa shows that trade credit is a very important source of finance, even in countries where the financial market is malfunctioning, contract enforcement is insecure, and information is scarce, unreliable, and asymmetric. In many previous studies in Africa, firm size and length of business relations were extensively studied and were found to be significant factors influencing trade credit demand. 2.2.2 With Regard to Ethiopia A study by Gebrehiwot and Wolday (2006) on MSEs’ finance in Ethiopia found that the use of (participation in) trade credit is a common practice in the MSE sector. A vast majority of the sampled MSEs (about 80 per cent) participated in trade credit in the sense of receiving or giving trade credit, or both. Twenty-two per cent and 27 per cent of the sampled MSEs reported purchases on credit (i.e. received trade credit) in 2001/02 and 2002/03, respectively. The study also found that potential borrowers were discouraged, i.e., firms that need credit but were discouraged from applying by the perceived or real high collateral requirement, high cost of borrowing, difficulty of processes involved, ineligibility, or concern about their repayment ability, while some were what may be considered uninformed (i.e., not aware of the facility, or where and how to apply, etc.). Cash flow/liquidity problems appear to be very common among these traders. In fact there has been a dearth of literature regarding empirical evidence on trade credit in Ethiopia. Thus, this study tries to contribute to the growing empirical literature on trade credit, especially in Ethiopia.

Determinants of Trade Credit Use by Private Traders

127

Table 2.1: Summary of Previous Studies on Determinants of Trade Credit Use Variable Access to bank loan

Related theory Financing

Volume of Transaction transaction cost Firm size

Financing

Firm age

Financing

Positively related with trade credit use Chant & Walker, 1988; Fisman & Raturi, 2004; McMillan & Woodruff, 1999 Elliehausen & Wolken, 1993; Summers & Wilson, 2002; Paul, 2008 Danielson & Scott, 2004; Elliehausen & Wolken, 1993; Giannetti et al., 2008; Huyghebaert, 2006; Isaksson, 2002; Petersen & Rajan, 1997; Summers & Wilson, 2002 Biggs et al., 2002; Danielson & Scott, 2004; Giannetti et al., 2008; Isaksson, 2002; Petersen & Rajan, 1997

Negatively related with trade credit use Danielson & Scott, 2004; Giannetti et al., 2008; Petersen & Rajan, 1997 -

Atanasova, 2007; Mateut &Mizen, 2003; McMillan & Woodruff, 1999

Aaronson et al., 2004; Elliehausen & Wolken, 1993; Huyghebaert, 2006; Mateut & Mizen, 2003; McMillan & Woodruff, 1999; Robb & Wolken, 2002; Summers & Wilson, 2002 Aaronson et al., 2004; Isaksson, 2002 -

Financing Biggs et al., 2002; advantage Fisman, 2003 Aaronson et al., 2004; Length of relationship Financing Biggs et al., 2002; advantage Fisman, 2003; McMillan & Woodruff, 1999 Note: Authors in bold found significant results for the relationship between a specific variable and trade credit demand Familiarity

3. Research Design and Methodology The study design is based on the positivist paradigm. Positivist ontology assumes that reality is objective and its epistemology is also based on objective understanding. In this study what determines private traders’ use of trade credit is based on objective information, independent of the researcher that utilized both quantitative and qualitative data. For the purpose of investigating the determinants of trade credit use, this study

Chapter Five

128

draws on empirical evidence from the 2013 survey covering 198 randomly selected private traders from Mekelle city, in the Tigray regional state of Ethiopia. Semi-structured questionnaires were distributed to those randomly chosen traders from the sampling frame. Moreover, interviews were conducted with some selected respondents while collecting the questionnaire. The data collected in this way were classified, summarized, and presented using text and tables, and analysed using descriptive and inferential statistics. In addition, the econometric analysis tool of the binary choice logistic regression model was used to test the literature driven hypotheses and to draw conclusions.

3.1 The Econometric Model To know the status of trade credit use by private traders, traders were asked whether they have used trade credit or not in the form of a Yes or No response question. Thus, the dependent variable in this study is limited (discrete) for which the outcome can take only two values designated by “1” for a private trader using trade credit and by “0” if not. The binary logistic regression used assumes the probability to use trade credit or not. In this study the realization of the dichotomous variable used is defined as: ௞

Yi ൌ ȕ଴ ൅ ෍

௝ୀ௜

ȕ୨ ୧୨ ൅ ୧ , where Yi is directly observable as a dummy 

variable defined as ௜  ൌ ቄ

ͳ݂݅’”‹˜ƒ–‡–”ƒ†‡”‹—•‡†–”ƒ†‡…”‡†‹– Ͳ‫݁ݏ݅ݓݎ݄݁ݐ݋‬

Therefore, to estimate a logit function in which the dependent variable is the probability of using trade credit, it is represented by: ݈‫ݐ݅݃݋‬௜ ൌ Ž ቀ

ܲ݅ ቁ. ͳെܲ݅

Finally, the empirical model of trade credit use studied in this paper is given by the following equation:

§ P · P(tcu) ln¨¨ i ¸¸ E0  E1 gen  E2edu  E3mrst  E4 szbss  E5agebss  E6 acsblo  © 1  Pi ¹ E7trship  E8 log volu  E9 fpur  ui

Determinants of Trade Credit Use by Private Traders

129

where P(tcu) = the probability that the ith private trader uses trade credit given explanatory variables, ȕ0 is the constant (intercept), ȕ1 – ȕ9 are slope coefficients,

§ P · ln¨¨ i ¸¸ is the natural logarithm of the odds ratio (logit model), and © 1  Pi ¹ ui is the error term (absorbs all unobserved factors).

3.2 Formulation of Hypotheses 3.2.1 Gender of Owner versus Trade Credit Use

Previous literature shows that gender of the borrower is an important factor for participation in informal finance. There is some evidence that women generally use more informal finance than external formal credit as compared to male counterparts (Marlow & Patton, 2005; Pham, 2006; Zimmerman, Treichel, & Scott, 2006). Similarly, Coleman (2000) found that women-owned firms are less likely to use formal external financing. Given these arguments the next hypothesis was made. H1: Female retailers are more likely to use trade credit as compared to male counterparts. 3.2.2 Education versus Trade Credit Use

The success of a business will depend on the owner’s ability to obtain the necessary financial capital, which in turn is measured by educational level (Adrich & Zimmer, 1985). On the other hand, there is also evidence that more educated and experienced entrepreneurs seek other sources of finance than trade credit (Cole, 2010; Kimuyu & Omiti, 2000). Thus, the following hypothesis is made. H2: The more the private trader is educated, the higher the probability of using trade credit.

130

Chapter Five

3.2.3 Marital Status versus Trade Credit Use

According to some studies, married couples have better access to bank loans and reduced need for trade credit (Avery, Calem, & Canner, 2004; Pham, 2006), and are also capable of generating more income as compared to unmarried traders. Similarly, the following hypothesis is formulated. H3: Married traders are less likely to demand trade credit as compared to traders of another marital status 3.2.4 Size versus Trade Credit Use

Previous studies found conflicting results for firm size as measured by number of employees or log of total assets. For instance, Atanasova (2007) found an inverse relation between the use of trade credit and firm size. On the other hand, Fafchamps, Pender, and Robinson (1995), Biggs, Raturi, and Srivastavac (1996) and Isaksson (2002) have shown, for developing countries in Sub-Saharan Africa, that the use of trade credit increases with the size of the firm. In the context of this study also, the next hypothesis was derived. H4: The larger the size of the business, the higher the chance of using trade credit. 3.2.5 Age vs. Trade Credit Use

Younger firms need capital to finance growth, and also tend to be less creditworthy, less profitable, and less diversified than older firms, so they have higher probabilities of financial distress (Fatoki & Odeyemi, 2010) and hence, new firms are found to be an applicant of trade credit. Similarly, suppliers have more concern for new firms with greater potential to grow. Petersen and Rajan (1997) found that trade credit was more used at business start-up than when they become able to generate sizable profit. In this regard the following hypothesis is formulated. H5: The younger the business, the higher the probability of using trade credit as compared to relatively older businesses. 3.2.6 Access to Bank Loans versus Trade Credit Use

Financing theory suggests that if bank loans and trade credit are substitutable, access to bank loans is positively related to trade credit (the more difficult the access to bank credit, the more firms will rely on trade

Determinants of Trade Credit Use by Private Traders

131

credit to compensate), supported by empirical evidence from Huyghebaert (2006), Giannetti et al. (2008), Nielsen (2002) and Petersen and Rajan (1997). In such a way the following hypothesis is formulated. H6: The more the business has access to bank loans, the lower the likelihood of using trade credit. 3.2.7 Length of Business Relationship with Supplier versus Trade Credit Use

Summers and Wilson (2002) note that suppliers offer trade credit to customers with whom they have a long history of a trade relationship because they assume that it is relatively less risky as compared to new customers. From this point of view the following hypothesis is drawn. H7: The higher the duration of a trade relationship with its supplier, the higher the likelihood of using trade credit by private traders. 3.2.8 Volume of Purchase versus Trade Credit Use

Trade credit demand for an individual firm will increase as the volume of purchases increases (Summers & Wilson, 2002). The transaction cost theory suggests that suppliers have an incentive to offer trade credit to customers buying large quantities because it reduces storage costs. If trade credit is offered, the customer may be stimulated to increase the quantity they purchase per transaction (Chung & Liao, 2006). This reduces the need for large storage space and storage costs for the supplier (Petersen & Rajan, 1997). Therefore the average volume of purchase order determines a customer’s use of trade credit. From the above evidence, the following hypothesis is formulated. H8: The larger the volume of purchase, the greater the probability of using trade credit. 3.2.9 Frequency of Purchase versus Trade Credit Use

Suppliers consider infrequent buyers as risky customers who need to be paid special attention when they ask for credit (Hermes et al., 2010). Customers who buy frequently indicate how good their businesses are doing and what the chances of repaying are. Thus, suppliers give trade credit to those customers because they are confident that they will get their money back.

Chapter Five

132

H9: The more frequent the purchase, the higher the likelihood of using trade credit.

4. Results and Discussion 4.1 Status of Trade Credit Use To know the status of trade credit use by private traders, traders were asked whether they have used trade credit or not. Such an objective response and direct measure of the binary dependent variable (i.e., trade credit use equal to ‘1’ if firm uses trade credit and ‘0’ otherwise) was used to determine the practice of trade credit use in the study area. Table 4.1: Status of Trade Credit Use by Private Traders Trade credit Users Non-users Total

Number of traders 115 83 198

Percent 58.08 41.92 100

Source: own survey data (2013)

As shown in table 4.1, more than 58 percent of private traders in Kedamay Woyane are found to be trade credit users, while about 42 percent are nonusers. This result is consistent with the previous studies undertaken in developed countries, which found that more than half of total sample retail traders used financial support from their supplier in terms of trade credit (Teruel & Solano, 2008). Again it is also somewhat consistent with evidence found by Fafchamps (1996) in which, out of 107 sample firms, 51.4 percent purchased on credit. Moreover, it is also similar to the evidence by Gebrehiwot and Wolday (2006), which found that about 55 and 58 percent of the sample micro and small enterprises respectively, participated in trade credit use, and they have found that supplier credit was the second most important source of financing for MSEs, next to funds from families and relatives. Basis of Trader-Supplier Relation: In this study it has been found that business relation is the basic foundation of a trader-supplier relation. Thus, contrary to the common view that mostly trade credit takes place between people with similar religion, ethnicity, relatives, and friendship, the result indicates that business relation was the top reason of trading parties. The following table shows the basis of buyer-seller relationship.

Determinants of Trade Credit Use by Private Traders

133

Table 4.2: Basis of Trader and Supplier Relation

Basis of relation

Users Response (n = 115)

Trade credit Non users

Percent of response

Total

Response (n = 83)

Percent of response

Total response (n=198)

Percent of response

74

89.16

178

89.90

3

3.61

9

4.54

17

20.48

33

16.67

20

24.09

30

15.15

Business 104 90.43 relation Blood 6 5.22 relation Similar 16 14.00 language Similar 10 8.70 religion Source: own survey data (2013)

Table 4.2 above reveals that the largest proportion of 178 (about 90 percent) of sample traders mentioned that their relationship with suppliers was business relation, thus, trade credit is a matter of trust and honesty among the transacting parties, and it is not a matter of familiarity on the basis of ethnicity, religion, or blood relationship. Reasons for Trade Credit Use: One of the main reasons, as stated by respondents, to use trade credit was that they had little money to pay in cash (liquidity problem). Accordingly, table 4.3 reveals that out of the total trade credit users (115), about 67 percent responded that a liquidity problem was the major reason for using trade credit, and inaccessibility of formal sources was mentioned as the second most important reason for using trade credit. Table 4.3: Summary of Important Reasons for Trade Credit Use Reasons of Trade Credit Use Trade credit Users Response (n = 15) Percent of response

Liquidity problem

Flexible payback

Little/no collateral

To get discount

Local availability

77

11

23

10

11

Inaccessibility of formal sources 49

66.96

9.56

20.00

8.70

9.56

42.61

Source: Own survey data (2013)

Chapter Five

134

4.2 Determinants of Trade Credit Use by Binary Logit Model As explained in part three of this paper, binary logistic regression was used to estimate the potential effect of each explanatory variable on the condition to use trade credit finance. Before applying the model, the Hosmer-Lemeshow test of goodness of fit was used to see the overall fitness of the model. Similarly, before estimating the model, various detection and diagnostics tests were done to check for the related econometric problems, such as multi-collinearity, heteroscedasticity, model specification bias, and normality of the data, as discussed in chapter three. The results of these tests indicated that the model is fitted, no severe multi-collinearity, and the normality of the data set. Table 4.4: Summary of Model Diagnostic Tests Tests

Test names

lfit

ovtest

HosmerLemeshow BreuschPagan/CookWeisberg Ramsey RESET

vif

Minimum = 1.08

hettest

Null hypothesis Model fits data

Ch2/Fvalue 7.01

Prob > ch2/F-value 0.5358

Constant variance

1.24

0.2647

17.71

0.0000

No omitted variables Maximum = 3.73

Mean = 1.71

Source: Stata result from survey data (2013)

As per the logistic regression output presented in table 4.5, the interpretations of estimation results of significant explanatory variables are presented below, followed by tests of research hypotheses.

Determinants of Trade Credit Use by Private Traders

135

Table 4.5: Logistic Regression Estimation Result Variables

Odds ratio

p>ʜzʜ

Marginal effect (dy/dx)

Owner Characteristics: Gender of private trader 0.2514 0.008* -0.3256 Education of trader 0.1967 0.017** -0.3606 Marital status (Dummy variable - reference to married) Single 0.5038 0.262 -0.1674 Divorced 1.795 0.401 0.1335 Business Characteristics: Age of business 0.482 0.000* -0.1773 Size of business (log) 0.658 0.412 -0.1015 Access to bank loan 0.542 0.439 -0.1474 Length of trade relationship 2.222 0.000* 0.1940 Frequency of purchase 4.079 0.000* -0.3415 Volume of purchase (log) 3.302 0.028** 0.2902 Statistics: Number of observations = 198 Prob > chi2 = 0.0000 Wald chi2(10) = 40.71 Pseudo R2 = 0.6758 *and ** indicate level of significance at 1 percent and 5 percent respectively Source: Own survey data (2013)

4.1.1 Gender (gen)

Gender of the traders has a negative but significant effect on trade credit use. Contrary to the expectation, the result shows the probability of trade credit use is 0.25 times higher for male-owned businesses than for femaleowned businesses. The marginal effect of this variable is -0.3256, indicating the probability of trade credit use for female-owned businesses decreases by 32.56 percent as compared to male-owned businesses, ceteris paribus. Therefore, the first research hypothesis, which said “femaleowned traders are more likely to use trade credit compared to men counterparts” is rejected. The result is inconsistent with Coleman (2000) and Fafchamps (1999). In this study it can be justified, first, that women have a dual role, i.e., domestic and reproductive responsibility, hence, are participating less in business activities than men. Secondly, women might be more risk averse, even in using trade credit than men, in order to

136

Chapter Five

maintain their welfare and survival of the household. These could be true and evident reasons in this study. 4.1.2 Education (edu)

The level of education also has a negative effect on trade credit use. The marginal effect of -0.3606 indicates that, keeping other factors constant, the probability of trade credit use decreases by 36.06 as educational level increases by one unit. Hence, the research hypothesis “the more the private trader is educated, the higher the probability of using trade credit” is rejected. It is contrary to the findings by Kaniki (2006), Robb and Wolken (2002), in which managers with post-secondary education were found to be more frequent users of trade credit. This may be due to the fact that the more educated the trader is, the more know how s/he might have about the relative costs of alternative financing sources. Since trade credit is considered as an expensive source of finance as compared to bank credit (Cole, 2010), more educated traders might seek loans from formal financial institutions, rather than using supplier credit. 4.1.3 Age of the Business (agebss)

Similar to prior expectation, this variable has a negative and significant effect. The result of this study exhibits that the odds ratio of 0.482 for age of the business indicates the probability of trade credit use decreases by 0.482 times for a year’s increase in the age of the business. Similarly, the marginal effect shows that the probability of trade credit use decreases by 17.73 percent as age of the business increases by a year, ceteris paribus. Therefore, the research hypothesis “the younger the business the higher the probability of using trade credit relatively, as compared to older businesses” is accepted. The result confirmed the views in the literature that younger firms have less access to bank credit since they lack reputation and experience in business and, hence, use trade credit, which in turn confirms the substitution hypothesis that argues trade credit and bank credit are substitutable sources of finance. 4.1.4 Length of Trade Relation with Supplier (ltrship)

The result indicates that the length of the trade relationship has a positive effect on the probability of using trade credit. Similarly, the odds ratio shows that the probability of using trade credit increases by 2.22 times as the length of the trade relationship increases by one year, with other things kept constant. The marginal effect of this variable (0.194) implies that,

Determinants of Trade Credit Use by Private Traders

137

ceteris paribus, the probability of trade credit use increases by 19.4 percent as the length of the trade relationship increases by a year. As a result, the research hypothesis “the higher the duration of the trade relationship with its supplier, the higher the chance of using trade credit” is accepted. This is due to the fact that trade credit has taken place without any written documentation, which means the business familiarity (length of trade relation) between the trader and supplier may be very important in using trade credit. 4.1.5 Volume of Purchase (logvolu)

The volume of transaction has a positive and significant effect on the probability of trade credit use. The result shows that, assuming all other factors remain constant, the probability of trade credit use increases by 30.2 percent as the volume of purchase increases by one unit. Based on this, the research hypothesis “the larger the volume of purchase, the greater the probability of using trade credit” is accepted. It is consistent with the transaction cost advantage theory and the empirical evidence of Chung and Liao (2006) and Summers and Wilson (2002). This may be due to the fact that suppliers want to offload some of their excess inventories to clients via allowing for later payments, and to gain competitive advantage. On the other hand, most traders are financially constrained and they cannot afford to pay in cash for large purchases. 4.1.6 Frequency of Purchase (fpur)

The results of the study indicate that, holding other factors constant, the probability of trade credit use decreases by 4.079 times as the frequency of purchase increases by one times of order per month. Similarly, the marginal effect (-0.3415) shows that the probability of trade credit use decreases by 34.15 percent as the frequency of purchase increases by one times of order per month, with all other factors kept constant. Thus, the research hypothesis “the more frequent purchases are, the better the chance of using trade credit” is rejected. This finding is contrary to financing advantage theory, which argues that repeated ordering allows suppliers to collect more information on customers’ creditworthiness. This might be due to the fact that infrequent purchase may show poor business operation with fewer profit and liquidity problems, and thus the business may seek trade credit. In addition, in our context, infrequent purchases may show low inventory turnover, and such businesses lack cash to order new products and fashion, since their money is already tied up in the existing inventory; thus they go for trade credit use.

138

Chapter Five

5. Conclusion and Policy Implication The study covered determinants of trade credit use by considering both business owner characteristics and firm specific factors. It has been found that about 58 percent of traders that were found in Mekelle city were trade credit users, and about 42 percent of them were non-users. It also showed that trade credit is highly based on business relation only, while familiarities between traders and suppliers on the grounds of ethnicity and religion were not important in trade credit use. The study revealed that female traders lagged behind in trade credit use. It might imply that women have a dual role as compared to men. It is found that trade credit use decreases with increases in educational status. The results showed that there is a negative relationship between the age of the firm and the probability of trade credit use, whereas there is a positive relation between trade credit use and the length of the trade relationship with the supplier. The frequency of purchases influences trade credit use negatively, whereas volume of purchase has a positive effect on trade credit use. Finally, the results showed that the effects of marital status, firm size, and access to bank loans were found to be insignificant. The results of this study offer several insights and policy implications. Accordingly, the importance of trade credit use in short-term financing, in the purchase of goods by private traders, reflects that these enterprises are capital rationed on the one hand, and face difficulties in having access to credit from formal financial institutions on the other. It suggests that banks are not willing to channel formal credit to petty traders that could not satisfy their collateral requirement. So, as an alternative, trade credit should be fostered and made efficient by establishing policies that protect suppliers and enable all traders to benefit.

References Atanasova, C. (2007). Access to institutional finance and the use of trade credit. Financial Management, 36 (1), 49-67. Avery, R., Calem, P., & Canner, G. (2004). Consumer credit scoring: Do situational circumstances matter? Journal of Banking and Finance, 28, 835-856 Biggs, T., Raturi, M., & Srivastavac, P. (2002). Ethnic networks and access to credit: Evidence from the manufacturing sector in Kenya. Journal of Economic Behavior and Organization, 49(4), 473-486.

Determinants of Trade Credit Use by Private Traders

139

Chant, M., & Walker, A. (1988). Small business demand for trade credit. Applied Economics, 20, 861-876. Cheng, S., & Pike, R. (2003). The trade credit decision: Evidence of UK firms. Management and Decision Economics, 24, 219-438 Chung, K., & Liao, J. (2006). The optimal ordering policy in a DCF analysis for deteriorating items when trade credit depends on the order quantity. International Journal of Production Economics, 20, 861-876. Cole, R. (2010). Bank credit, trade credit or no credit: Evidence from the surveys of small business finances. Retrieved from http://ssrn.com/abstract=1540221 Cole, R., & Mehran, H. (2011). Gender and the availability of credit to privately held firms: Evidence from the surveys of small business finances. Retrieved from http://ssrn.com/abstract=1799649 Coleman, S. (2000). Access to capital and terms of credit: A comparison of men- and women owned small businesses. Journal of Small Business Management, 20, 37-52. Ethiopian Development Research Institute (EDRI). (2003). Determinants of private sector growth in Ethiopia’s urban industry: The role of investment climate. Addis Ababa. Elliehausen, G., & Wolken, J. (1993). The demand for trade credit: An investigation of motives for trade use by small business. Working Paper, 165, Washington DC: Board of Governors of the Federal Reserve System Emery, G. (1984). A pure financial explanation for trade credit. Journal of Financial and Quantitative Analysis, 19, 271-285. Fafchamps, M. (1997). Trade credit in Zimbabwean manufacturing. In World Development, 25(5), 795-815. Fafchamps, M., Pender, J., & Robinson, E. (1995). Enterprise finance in Zimbabwe. In Regional Program for Enterprise Development, Africa Division, Washington, D.C: The World Bank. Fatoki, O., & Odeyemi, A. (2010). The determinants of access to trade credit by new SMEs in South Africa. African Journal of Business Management, 4(13). Fisman, R. (2003). Ethnic ties and the provision of credit: Relationshiplevel evidence from African firms. Advances in Economic Analysis and Policy, 3 (1). Fisman, R., & Raturi, M. (2004). Does competition encourage credit provision? Evidence from African trade credit relationships. Review of Economics and Statistics, 86, 345-352.

140

Chapter Five

Gebrehiwot, A., & Wolday, A. (2006). Micro and small enterprises (MSEs) finance in Ethiopia: Empirical evidence. Eastern Africa Social Science Research Review, 22, Michigan State University Press Giannetti, M., Burkart, M., & Ellingsen, T. (2008). What you sell is what you lend? Explaining trade credit contracts. Review of Financial Studies. Gustafson, R. (2004). Agribusiness trade credit: A paradox. Agribusiness and Applied Economics Report, 534. Guy, O., & Mazra, M. (2012). The determinants of trade credit demand: An empirical study from Cameroonian firms, 7(17), 43. Retried from http://dx.doi.org/105539/ijbm. Hermes, N., Kihanga, E., Lensink, R., & Lutz, C. (2010). Determinants of trade credit demand and supply in the Tanzanian rice market: A structural modelling approach. Retrieved from http://ssrn.com/abstract=1674842 Huyghebaert, N. (2006). On the determinants and dynamics of trade credit use: Empirical evidence from business start-ups. Journal of Business Finance and Accounting, 33, 305-328. Huyghebaert, N., Van de, G., & Van Hulle, C. (2007). The choice between bank debt and trade credit in business start-ups. Small Business Economics: Forthcoming. Isaksson, A. (2002). Trade Credit in Kenyan Manufacturing: Evidence from plant-level data. Statistical and Information Networks (UNIDO) Working Paper, 4. Kaniki, S. (2006). The importance of courts for trade credit in East African manufacturing firms. Policy Paper Number 9, University of the Witwatersrand. Kimuyu, P., & Omiti, J. (2000). Institutional impediments to access to trade credit by MSEs in Kenya. Institute of Policy Analysis and Research, Nairobi, Kenya Levchuk, Y. (2002). Trade credit determinants of Ukrainian enterprises. Institute of Economic Forecasting at Academy of Sciences of Ukraine. Marlow, D., & Patton, S. (2005). All credit to men? Entrepreneurship, finance, and gender. Entrepreneurship Theory and Practice, 29, 71735. McMillan, J., & Woodruff, C. (1999). Inter-firm relationships and informal credit in Vietnam, Quarterly Journal of Economics, 114, 1285-1320. Niskanen, J., & Niskanen, M. (2006). The determinants of corporate trade credit policies in a bank-dominated financial environment: The case of Finnish small firms. European Financial Management, 12(1), 81–102.

Determinants of Trade Credit Use by Private Traders

141

Nilsen, J. (2002). Trade credit and the bank lending channel. Journal of Money, Credit, and Banking, 34, 226-253. Ojenike, J., & Olowoniyi, A. (2012). Use of trade credit in Nigeria: A panel econometric approach. Research Journal of Finance and Accounting, 3(2). Petersen, M., & Rajan, R. (1997). Trade credit: Theories and evidence. Review of Financial Studies, 10, 661-691. Pham, T. (2006). Essays on borrowing and debt contracting: A theoretical analysis and empirical evidence for Vietnam. Ridderkerk: Labyrinth Publications. Pike, R., & Cheng, S. (2003). Trade credit terms, asymmetric information and price discrimination: Evidence from three continents. Journal of Business Finance and Accounting, 32, 0306-686. Robb, A., & Wolken, J. (2002). Firm, owner, and financing characteristics: Difference between female and male owned small business. Retrieved from www.federalreserve.gov/pubs/oss3/ssbf98/FEDS_robbwolken.pdf Santos, G., Sheng, H., & Bortoluzzo, A. (2011). Use of trade credit by firms: Evidence for Latin America. Insper Working Paper. Shaffer, M. E. (2000). Should we be worried about the use of trade credit and non-monetary transactions in transition economies? Economic Systems, 24(1), 51-54. Sola, C., Teruel, P., & Solano, P. (2008). Trade credit and SME profitability. A preliminary draft. Summers, B., & Wilson, N. (2002). An empirical investigation of trade credit demand. International Journal of the Economics of Business, 9, 257-270. Teruel, P., & Solano, P. (2008). A dynamic perspective on the determinants of accounts payable. Retrievedfromhttp://www.efmaefm.org/0efmameetings/efma%20annu al%20meetings/2008-athens/garcia-teruel.pdf Vaidya, R. (2011). Determinants of trade credit: Evidence from Indian manufacturing firms. Retrieved from http://www.igidr.ac.in/pdf/publication/WP-2011-012.pdf Van Horen, N. (2007). Customer market power and the provision of trade credit: Evidence from Eastern Europe and Central Asia. Policy Research Working Paper, 4284, Washington DC: The World Bank Zimmerman, Treichel, M., & Scott, J. (2006). Women owned businesses and access to bank credit: Evidence from three surveys since 1987. Venture Capital, 8, 51-67.







CHAPTER SIX DETERMINANTS OF NET INTEREST MARGIN: AN EMPIRICAL STUDY ON THE ETHIOPIAN BANKING INDUSTRY MISRAKU MOLLA

Abstract This study investigates the bank-specific, industry-specific, regulatory and macro-economic determinants of the Net Interest Margin (NIM) for a total of eight commercial banks in Ethiopia, covering the period 2002-2011. To this end, the study adopts a quantitative research approach. According to the two-step multivariate regression results, operating costs, credit risk, implicit tax, and inflation have a statistically significant and positive relationship with banks’ NIM. On the other hand, variables like bank size and management quality and concentration have a negative and statistically significant relationship with banks’ NIM. However, the relationship for equity ratio, liquidity risk, macro-economic instability and GDP with NIM is found to be statistically insignificant. Moreover, this study finds that both bank-specific and industry-specific determinants are more relevant factors that determine banks’ NIM than macro-economic determinants.

1. Introduction Commercial banks play an important role in the functioning of economies in both developed and emerging countries. Several studies conclude that banks have a positive role in improving capital allocation and corporate governance (Diamond, 1984), enhancing investment and economic growth (Bencivenga and Smith, 1991; Beck and Levine, 2004). Efficient financial intermediation is an important factor in the economic development process, as it has implications for effective mobilization of resources to be

Determinants of Net Interest Margin

143

invested. Consequently, as financial intermediaries, banks play a crucial role in the operation of most economies. With regard to this idea, several researchers have established that the efficacy of financial intermediation can also affect economic growth. Crucially, financial intermediation affects the net return to savings, and the gross return for investment (Demirgue and Huizinga, 1998). A major indicator of banking sector efficiency is the Net Interest Margin (NIM), which has been found to be higher in African, Latin American, and Caribbean countries than in OECD countries (Randall, 1998; Chirwa and Mlachila, 2004; Folawewo and Tennant, 2008). Interest rate margins indicate how efficiently banks perform their intermediation role of savings mobilization and allocation. A wide deposit-lending rate margin is not only indicative of banking sector inefficiency; it also reflects the level of development of the financial sector1. Studies by Randall (1998), Brock and Rojas-Suarez (2000), Chirwa and Mlachila (2004), Gelos (2006), Crowley (2007), and Folawewo and Tennant (2008) all show that interest margins in Sub-Saharan Africa, Latin America, and the Caribbean are wider than in OECD countries. These researchers have attributed the existence of high interest margins in developing countries to several factors, such as high operating costs, financial repression, lack of competition, and market power of a few large dominant banks, enabling them to manipulate industry variables including lending and deposit rates, high inflation rates, high risk premiums in the formal credit markets due to a widely prevailing perception relating to high risk for most borrowers, and similar other factors. This wide Net Interest Margin (NIM) or Interest Rate Spread (IRS) indicates that the banking sector in these areas is inefficient and less developed. In SubSaharan Africa there is limited knowledge about bank behaviour and efficiency. As a result, the implications of banking sector inefficiency have spurred numerous debates. There is even a clear gap in empirical work on whether financial reforms in Sub-Saharan Africa have increased the efficiency of the banking system or not. Of course, this issue is not that

 1

It is not clear whether high margins are good or bad from a social welfare perspective. On the one hand, narrow margins may be indicators of a relatively competitive banking system. On the other hand, relatively large margins may bring a degree of stability to the banking system (Saunders and Schumacher, 1997).

144

Chapter Six

surprising in a continent which represents only 0.87% of the total banking assets in the world.2 Most studies carried out on banking systems of developed countries revealed that interest margins were positively and significantly correlated to the bank’s level of capital, the loan loss provision, reserve requirements, implicit taxation and interest volatility (Dermirgüç-Kunt and Huizinga, 1999; Saunders and Schumacher, 2000)3. However, studies carried out in other parts of the globe, for instance across Latin America by Brock and Rojas-Suarez (2000) and across low and middle income countries by Tennant and Folawewo (2008), reached diverging conclusions and even contradicted some of the benchmark results whereby negative and sometimes insignificant relationships were observed between the alleged determinants and interest margins. Besides, studies have shown that there is a pervasive view amongst some stakeholders that high NIMs are caused by the internal characteristics of the banks themselves, such as their tendency to maximize profits in an oligopolistic market, while many others argue that the margins are imposed by the macro-economic, regulatory and institutional environment in which banks operate. Therefore, these debates can only be resolved through objective and quantitative analysis of the determinants of banking sector interest margins in developing countries like Ethiopia. The average interest rate spread of Ethiopian banks for the last ten years (2002 to 2011) was approximately 7.2%, which is lower than the average African countries’ spread (7.52%) but much greater than the average Eurozone countries’ spreads (3.15%, as of 2010)4. The net interest margin, which is measured by net interest income divided by total earning assets, was in between, with a minimum of 3.15% in the year 2003 and maximum of 5.12% in 2009. The average NIM during the sample period was 4.08%, which is almost equal to the average African countries’ margin, and higher than that of OECD countries.

 2 Murinde (2009) points out that African banking assets represent only 0.87% of global banking assets, compared to 58.15% for the 15 countries of the Eurozone and 15.09% for the United States. 3 Scholars usually consider such results as benchmarks since banking systems of such countries are referred to as mature and stable ones. 4 However, the average NIM of Ethiopian banks from 2002-2011 was 4.08%.

Determinants of Net Interest Margin

145

Figure 1.1: Trend analysis of IRS and NIM of Ethiopian banks over the period 2002 to 2011 14 Average deposit rate

12 10

Average lendig rate

8 6

spread

4 2

NIM

0 1

2

3

4

5

6

7

8

9 10

Source: annual report of NBE, MoFED and own computation.5

Like other African countries and the rest of the developing world, Ethiopia undertook a banking reform program in 1994, three years after the centrally planned economy was overthrown. These reforms have brought about many structural changes in the banking sector of the country, and have also encouraged private banks to enter and expand their operations in the industry (Lelissa, 2007). Despite these changes, the banking industry in Ethiopia is currently characterized by operational inefficiency, little and insufficient competition, and can perhaps be distinguished by its market concentration towards the big government owned commercial bank, and having an undiversified ownership structure (Lelisa, 2007). The existence of less efficiency and little and insufficient competition in the country’s banking industry is a clear indicator of banking sector inefficiencies. Thus, it is important to know the determinants of banks’ NIM, as it is the most important measure of financial sector efficiency. The purpose of this study is to provide an econometric account of some of the main determinants of interest rate margins of Ethiopian commercial 5

Average deposit rate is calculated using average interest rate of saving, time, and demand deposits, average lending rate is calculated as the average of the minimum and maximum lending rate of all commercial banks operating during the period, and interest rate spread is the difference between the average lending rate and average deposit rate.

146

Chapter Six

banks. Thus, this study attempts to supplement the literature by providing new evidence from an emerging African economy, namely Ethiopia, and aims at investigating the effect of bank-specific, industry-specific, regulatory and macro-economic determinants on the net interest margin (NIM) using panel data over the period 2002 to 2011. Many studies, which focus on overall profitability of Ethiopian commercial banks, were conducted. However, according to the knowledge of the researcher, no studies were conducted on the determinants of NIM of the Ethiopian banking industry. Even the few studies carried out in Sub-Saharan countries were not comprehensive enough. Therefore, to the best of the researcher’s knowledge, this paper would be the first to address the determinants of Ethiopian commercial banks’ NIM. This study fills a gap both in the banking literature about single-country NIM studies, and Ethiopian banking literature. Besides, the study draws some conclusions and identifies the factors determining banks’ net interest margins significantly. Finally, this study helps other researchers as a source of reference and as a stepping-stone for those who want to conduct further studies in the area. The remainder of this paper is organized as follows. Section 2 briefly reviews studies related to the determinants of the NIM of banks. Section 3 describes research design and methodology. Sections 4 and 5 present findings and discussion, respectively. Finally, conclusions and implications of the findings are discussed and directions for future research are offered in section 6.

2. Literature Review 2.1. Theoretical Review Among different types of measurements for banks’ intermediary efficacy, NIM is an important one. NIM is calculated using ex-post spreads, which consist of the difference between banks’ interest revenues and their actual interest expenses. Interest rate margins indicate how efficiently banks perform their intermediation role of savings mobilization and allocation. An approach used in much of the literature is to classify determinants of commercial banks’ interest rate spreads according to whether they are bank-specific, industry (market) specific, or macroeconomic in nature. Bank-specific variables refer to those factors that characterize individual banks and affect the interest margin accruing to the respective institution. Demirguc-Kunt and Huizinga (1998) note that the specific characteristics

Determinants of Net Interest Margin

147

of commercial banks that are usually assumed to have an impact on their spreads include bank size, ownership structure, quality of the loan portfolio, capital adequacy, overhead costs, operating expenses, and shares of liquid and fixed assets. Robinson (2002) further notes that the incidence of fraud, the ease with which bad credit risks survive due diligence, and the state of corporate governance within banks, all lead to higher operating costs, asset deterioration, and ultimately wider interest rate spreads. The market specific determinants of commercial banks’ interest rate margins include lack of adequate competition in the banking sector and consequent market power of commercial banks, the degree of development of the banking sector, and explicit and implicit taxation - such as profit taxes and reserve requirements. Cross-country studies have also established that banking spreads tend to fall as institutional factors improve. Such factors include the efficiency of the legal system, contract enforcement, and decreased levels of corruption, which are all critical elements of the basic infrastructure needed to support efficient banking. Macroeconomic factors are certainly among the most influential sources for variations in credit spreads (Brock and Franken, 2003). Chirwa and Mlachila (2004) concur and assert that macroeconomic instability and the policy environment have important impacts on the pricing behaviour of commercial banks. They note that the macroeconomic variables typically thought to be determinants of interest rate spreads include inflation, economic growth, and interest rates. Tennant (2006) showed that macro policy variables, such as public sector domestic borrowing, discount rates, and Treasury bill rates, are commonly perceived to impact on commercial bank spreads. In an environment where the exchange rate is volatile and interest rates are heading downwards, expectations of exchange rate depreciation will result in higher lending rates. This widens the spread.

2.2 Empirical Studies Zhou and Wong (2003) empirically investigated the determinants of Chinese commercial banks’ NIMs from 1996 to 2003. The study was an extension to the Ho and Saunders (1981) model to identify the elements affecting net interest margins. The results of the study indicated that the determinants of net interest margins in the Chinese market include market competition structure, average operating costs, degree of risk aversion, transaction size, implicit interest payments, opportunity cost of reserve, and management efficiency.

148

Chapter Six

Liebeg and Schwaiger (2005), using year-end data of 903 Austrian banks, investigated the determinants of banks’ interest rate margins in Austria. The findings of the study show that the main factors driving the reduction of Austrian banks’ interest rate margins were decreasing operating costs, the growing importance of foreign currency lending combined with a rising share of non-interest revenues, as well as increased competition. They also investigated a contrasting finding to the literature, i.e., a positive effect of relationship banking on margins, with the erosion of relationship banking being another reason for the decline in interest margins. Using all commercial banks operating in the Kenyan banking industry from the period 2000 – 2009, Chekol (2011) investigated the determinants of NIMs of Kenyan banks. Scatter plots were used to see the existence of a functional relationship between the response and predictor variables as primary evidence. In the study, a linear regression model was employed for the empirical analysis of the variables. The findings of the study were as follows: (1) Operating expense has a positive and significant effect on the net interest margin of commercial banks in Kenya; (2) credit risk tends to be positively associated with the net interest margin; (3) the study found that the higher the inflation, the larger the net interest margin in Kenya; (4) growth influences net interest margins negatively; (5) market concentration has a negative and significant influence on net interest margins. Demirgüç-Kunt and Huizinga (1999) investigate the determinants of bank interest margins using bank-level data for 80 countries in the years 19881995. The set of regressors include several variables accounting for bank characteristics, macroeconomic conditions, explicit and implicit bank taxation, deposit insurance regulation, overall financial structure, and underlying legal and institutional indicators. Demirgüç-Kunt and Huizinga report that the bank interest margin is positively influenced by the ratio of equity to lagged total assets, by the ratio of loans to total assets, by a foreign ownership dummy, by bank size, as measured by total bank assets, by the ratio of overhead costs to total assets, by the inflation rate, and by the short-term market interest rate in real terms. The ratio of non-interest earning assets to total assets, on the other hand, is negatively related to the bank interest margin. All the mentioned variables are highly statistically significant. Output growth, by contrast, does not seem to have any impact on bank spread. Saunders and Schumacher (2000) analyse the bank interest rate margins in six European countries, building on a model developed by Ho and

Determinants of Net Interest Margin

149

Saunders (1981). They follow a two-step process. First, they control for the effects on net interest margins of various imperfections that cannot be built directly into the model (i.e., implicit interest, the opportunity costs of reserves, and capital requirements) so as to isolate estimates of the pure spread in each country each year. Second, they undertake an analysis of the determinants of these pure spreads (e.g. market structure and interest rate volatility). They find that bank market structure, interest rate volatility, and bank capitalization matter for the spreads.

3. Research Design and Methodology 3.1 Research Method In order to achieve the broad research objective, a quantitative research approach was adopted.

3.2 Sample Design In this study, two criteria were used to determine the study sample. First, the nature of the banks; in the study, only commercial banks registered by NBE and currently under operation are included. The aim of this criterion is to ensure that the econometric estimations are robust; it is preferable to work on a homogeneous sample for consistency purposes. Second, availability of data: this study considers only banks that have data for at least ten years from 2002 to 2011. Based on the above two criteria, only eight6 banks are included in the study from the total of 18 banks (as of June 2012) operating in the Ethiopian banking sector.

3.3 Data Collection The data set used in this study consists of year-end data of all commercial banks that held an Ethiopian banking license between 2002 and 2011. Annual reports of these banks and macroeconomic data were collected from NBE and MoFED. Due to the following two reasons the study prefers to use data extracted from NBE and MoFED. First, getting 10

 6

The eight commercial banks included in the study are Awash International Bank (AIB), Bank of Abyssinia, Commercial Bank of Ethiopia (CBE), Construction and Business Bank (CBB), Dashen Bank (DB), Nib International Bank (NIB), United Bank (UB), and Wegagen Bank (WB)

150

Chapter Six

consecutive years’ individual data from each commercial bank would be difficult. Therefore, to overcome this problem the study has to consider the consolidated market data published by the NBE and MoFED. Second, the reported data were far more detailed throughout the sample than in commercial databases, thus allowing the researcher to find better-suited empirical variables.

3.4 Data Analysis and Model Specification The panel data collected using structured document review were analysed using descriptive statistics, correlations, multiple linear regression analysis and inferential statistics. Mean values and standard deviations were used to analyse the general trends of the data from 2002 to 2011, based on the sector sample of 8 commercial banks, and a correlation matrix was also used to examine the relationship between the dependent and explanatory variables. A multiple linear regression model and F-statistic was used to determine the relative importance of each independent variable in influencing NIM. The two-step multiple linear regressions model was run, and thus OLS was conducted, using the EVIEWS 6 econometric software package, to test the causal relationship between the firms̓NIM and their potential determinants, and to determine the most relevant explanatory variables affecting the NIM of Ethiopian banks. The rationale for choosing OLS is, as noted in Petra (2007), that OLS outperforms the other estimators when the following holds: the cross section is small, and the time dimension is short. Therefore, as far as both of the above facts hold true in this study, it is rational to use OLS. Finally, the general model for this study, as it is mostly found in the existing literature, is represented by: ܻ௜ǡ௧ = ߙ௜ + ߚܺ௜ǡ௧ + ߝ௜ǡ௧ ................................................................................. (1) ܻ௜ǡ௧ represents the dependent variable, ܺ௜ǡ௧ contains the set of explanatory variables in the model. The subscripts i and t denote the cross-sectional and time-series dimension respectively. Also ߙ௜ is taken to be constant over time t and specific to the individual cross-sectional unit i. If ߙ௜ is taken to be the same across units, then Ordinary Least Squares (OLS) provides a consistent and efficient estimate of ߙ௜ ƒ†Ⱦ.

In the light of the above model and on the basis of selected variables, the current study uses a two-step regression. In the first step, all selected variables would be regressed to investigate their effect on NIMs of all commercial banks. For this, the general econometrics model would be:

Determinants of Net Interest Margin

151

ܰ‫ܯܫ‬௜ǡ௧ ൌ ԓ௜ ൅ ߙ‫ܿ݁݌ݏܤ‬௜ǡ௧ ൅ ߚ‫ܿ݁݌ݏ݀݊ܫ‬௜ǡ௧ ൅ ɍ݉ܽܿ‫݋ݎ‬௜ǡ௧ ൅  ߝ௜ǡ௧ .......................... (2)

Where ܰ‫ܯܫ‬௜ǡ௧ †‡‘–‡•‡–‹–‡”‡•–ƒ”‰‹•ǡ ԓ௜ ‹•–Š‡…‘•–ƒ–ǡĮ is the vector of coefficients of bank-specific variables (Bspeci,t), ß is the vector of coefficients of industry-specific (Indspeci,t) variables, and y is the vector of coefficients of macro-economic variables (macroi,t) that are constant over all banks in a given year. ߝ௜ǡ௧ †‡‘–‡•–Š‡‡””‘”–‡Ǥ Based on the theory and empirical results of the determinants of NIMs of banks, the first step and the general empirical model of this study take the form: ܰ‫ܯܫ‬௜ǡ௧ ൌ  ߚ଴  ൅  ߚଵ ܵ݅‫ ݁ݖ‬൅  ߚଶ ܱܲ‫ ܥ‬൅ ߚଷ ‫ ܻܶܫܮܣܷܳܯ‬൅ ߚସ ܴ‫ ܻܶܫܷܳܧ‬൅ ߚହ ‫ܴܥ‬൅ߚ଺ ‫ ܴܮ‬൅  ߚ଺ ‫ ܫܪܪ‬൅  ߚ଼ ‫ ܺܣܶܫ‬൅ ߚଽ ‫ ܨܨܫ‬൅  ߚଵ଴ ‫ ܵܰܫ‬൅  ߚଵଵ ‫ ܲܦܩ‬൅ ߝ ....... (3)

Where NIMi,t denotes the net interest margin, ȕ0 is the constant, ȕ1 to ȕ11 are coefficients for independent variables. Size denotes size of banks, OPC–operating costs, MQUALITY–management quality, REQUITY– equity ratio, CR–credit risk, LR–liquidity risk, HHI–concentration, ITAX– implicit tax, IFF–inflation, INS–macroeconomic instability, GDP–gross domestic product growth, and İ is the error term. In the second step it was necessary to investigate most relevant factors affecting NIMs of Ethiopian commercial banks. Therefore, a set of variables that are found significant at 1%, 5%, and 10% significance level in the first step regression would be regressed again separately to investigate factors that can mostly determine the NIM of banks. Therefore, the following econometric model was developed; ܰ‫ܯܫ‬௜ǡ௧ ൌ  ߚ଴  ൅ ‫ݎܽݒ݃݅ݏܤ‬௜ǡ௧ ൅  ߝ௜ǡ௧ ............................................................ (4)

Where NIMi,t denotes the net interest margin, ȕ0 is the constant, B is the coefficient of significant variables, and sigvar denotes variables/s found significant from the first step regression.

3.5 Measurement of Study Variables and Hypothesis Development Table 3.1 summarizes the measurement of the study variables and expected signs of the independent variables.

Chapter Six

152

Table 3.1. Summary of the measurement of study variables Variable

Measurement

Dependent

Variable

Net interest margin Independent Size

Operating costs Management quality Equity ratio Credit risk Liquidity risk

NIMit =

Expected sign ோ௜௧ି஼௜௧

்௢௧௔௟௘௔௥௡௜௡௚௔௦௦௘௧ሺ௜௧ሻ

‹– ൌ ‹–‡”‡•–”‡˜‡—‡ ‹– ൌ ‹–‡”‡•–‡š’‡•‡

Variables Log of total loans and advances ܶ‫ݐݏ݋ܿ݃݊݅ݐܽݎ݁݌݋݈ܽݐ݋‬ ܶ‫ݐ݁ݏݏ݈ܽܽݐ݋‬ ‫ݐݏ݋ܥ‬ ‫݁݉݋ܿ݊ܫ‬ ܲ‫݈ܽݐ݅݌ܽܿ݀݁ݎݎ݂݁݁ݎ‬ ܶ‫ݐ݁ݏݏ݈ܽܽݐ݋‬ ܶ‫݊ܽ݋݈݈ܽݐ݋‬Ƭܽ݀‫ݏ݁ܿ݊ܽݒ‬ ܶ‫ݐ݁ݏݏ݈ܽܽݐ݋‬ ‫݈ܾ݄݁ܿ݊ܽܽݏܽܥ‬ ܶ‫ݐ݁ݏݏ݈ܽܽݐ݋‬

__________

Negative Positive Negative Positive Positive Negative

௡

Concentration

෍ ܵሺ‫݅ܦ‬ሻଶ

Positive

௜ୀଵ

HH Index for deposits

  െ   ‘–ƒŽƒ••‡–

Implicit taxation

NIE= non-interest expense OPI= other operating income

Inflation

Inflation rate Exchange rate volatility (Standard devation of three years exchange rate) Real GDP Growth

Macroeconomic instability GDP Growth

Positive

Positive Positive Negative

Determinants of Net Interest Margin

153

4. Results 4.1 Test Results for the Classical Linear Regression Model Assumptions Diagnostic tests were carried out to ensure that the data fits the basic assumptions of a classical linear regression model. 4.1.1 Test for Normality

The normality test for this study is shown in figure 4.1. The coefficient of kurtosis was close to 3, and the Bera-Jarque statistic had a P-value of 0.871, implying that the data were consistent with a normal distribution assumption. Figure 4.1: Normality test for residuals 8

Series: Standardized Residuals Sample 2002 2011 Observations 80

7 6 5 4 3 2 1

Mean Median Maximum Minimum Std. Dev. Skewness Kurtosis

1.19e-19 -0.000331 0.013063 -0.015558 0.005715 -0.049066 2.729574

Jarque-Bera Probability

0.275866 0.871157

0 -0.015

-0.010

-0.005

0.000

0.005

0.010

Source: Financial statements of banks, NBE, MoFED reports and own computation

4.1.2 Test for Multicollinearity

Table 4.1 shows the correlations among the independent variables. According to Malhotra (2007), multicollinearity exists if the correlation between two independent variables is more than 0.75. Therefore, as presented in the correlation matrix table 4.1, there was no problem of multicollinearity.

Chapter Six

0.313

0.497

0.46

-0.38

-0.444

0.168

-0.648

-0.556

-0.269

0.375

0.369

LR

CR

HHI

ITAX

IFF

GDP

-0.130

-0.167

0.236

0.145

0.449

0.038

0.228

-0.298

-0.165

0.232

0.270

0.112

0.030

1

0.105

0.185

-0.052

-0.265

-0.638

1

-0.187

-0.056

0.364

1

-0.673

0.510

1

HHI

-0.383

-0.101

-0.427

REQUITY

1

CR

0.238 0.055 0.101 -0.001 0.284 -0.477 INS Source: Financial statements of banks, NBE, MoFED reports and own computation.

0.213

-0.614

OPC

REQUITY LR

-0.659

0.532

-0.513

MQUALITY

OPC

-0.163

1

1

MQUALITY

SIZE

SIZE

Table 4.1 Correlation matrix of independent variables

154

0.439

-0.391

-0.580

1

ITAX

-0.040

0.355

1

IFF

-0.078

1

GDP

1

INS

Determinants of Net Interest Margin

155

4.1.3 Test for Heteroscedasticity

As shown in table 4.2, both the F-statistic, Chi-Square versions of the test statistic and Scaled explained SS͉ all gave the same conclusion that there is no evidence for the presence of heteroscedasticity, since the p-values were in excess of 0.05. Table 4.2 Heteroscedasticity Test: White

F-statistic Obs*R-squared Scaled explained SS

2.508245 74.66822 65.59545

Prob. F(67,12) Prob. Chi-Square(67) Prob. Chi-Square(67)

0.1706 0.2433 0.5257

Source: Financial statements of banks, NBE, MoFED reports and own computation.

4.1.4 Test for Autocorrelation

The Durbin-Watson test statistic value in table 4.5 was 1.503840. The relevant critical values for the test are dL= 1.22, dU = 1.81, and 4 - dU = 4-1.81 =2.19; 4 - dL = 4-1.22=2.78. Accordingly, the Durbin-Watson test value is clearly between the lower limit (dL), which is 1.22, and the upper limit, which is 1.81, and thus the null hypothesis is neither rejected nor accepted.

4.2 Descriptive Statistics Table 4.3 presents the results of the descriptive statistics for both dependent and independent variables involved in the regression model. The total observation for each variable was 80 (8*10). The mean of NIM was 4%, with a minimum of 0.9% and a maximum of 7.3%. Standard deviation of 1.2% indicates there was little difference between banks’ net interest income over the sample period. The highest variation with standard deviation 1.028 was found between sizes of banks. The average operating costs of Ethiopian banks was 2.2%, which indicates for the last decade Ethiopian commercial banks were cost effective, ranging from a maximum of 3.8% to a minimum of 0.8%. With regard to management quality (MQUALITY), another difference among banks was found with a standard deviation of 29%. The average cost to income ratio during the period was 59% with a maximum of 200% and a minimum of 19%.

Chapter Six

Observations 80 80 80 80 80 80 80 80 80 80 80 80

Mean 0.04 7.575 0.022 0.592 0.077 0.539 0.384 0.473 0.005 0.122 3.309 0.090

Median 0.042 8 0.022 0.500 0.074 0.572 0.375 0.449 0.003 0.107 1.824 0.113

Maximum 0.073 10 0.038 2 0.261 0.752 0.594 0.644 0.036 0.364 9.345 0.126

Source: Financial statements of banks, NBE, MoFED reports and own computation

Variables NIM SIZE OPC MQUALITY REQUITY CR LR HHI ITAX IFF INS GDP

Minimum 0.009 5 0.008 0.190 0.022 0.224 0.142 0.357 -0.003 -0.106 0.216 -0.021

Table 4.3 Summary of descriptive statistics for dependent and explanatory variables

156

Std. Dev. 0.012 1.028 0.005 0.293 0.036 0.127 0.100 0.095 0.011 0.122 3.199 0.047

Determinants of Net Interest Margin

157

Banks’ equity ratio has 7.7%, 26% and 2.2% mean, maximum, and minimum percentage respectively. Over the sample period, credit risk in the banking industry of Ethiopia was in between 75.2% maximum and 22.4% minimum. On the other hand, average liquidity risk was 38.4% with a minimum of 14.8% and a maximum of 59.4%. Over the sample period, the mean of industry concentration was 0.47, indicating that the banking industry was highly concentrated. With regard to implicit tax, there was only a mean of 0.5%, 3.6% maximum and -0.3% minimum to the industry. Real GDP growth in Ethiopia for the last ten years was 9%, with a maximum of 12.6% and a minimum of -2.1 %. Moreover, macroeconomic instability had shown the highest deviation (3.199). The result indicates that there was huge devaluation of Ethiopian birr in exchange with U.S dollar. During the sample period, inflation was galloped up to 36.4% and deflated to -10.6%.

4.3 Correlation Analysis among Variables As shown in table 4.4, a positive correlation between NIM and OPC, REQUITY and CR was found. This can be looked as banks charge higher interest to cover their operating costs, the cost of holding equity and higher credit uncertainty. The negative correlation between the size of banks and NIM is not as such surprising, because larger banks have the advantage of economies of scale to reduce interest margins, which small banks do not have. MQUALITY has a negative correlation with NIM. In fact, efficient management tries to reduce their organization operating costs, and as a result they charge lower interest and then a lower margin. The negative correlation of banks’ concentration, implicit tax and real GDP with NIM was somewhat ambiguous. Finally, inflation and exchange rate volatility have a positive correlation with NIM.

Chapter Six

-0.193 1

SIZE

-0.042 0.168

0.271

-0.432 -0.556 0.497

-0.119 -0.269 0.46

0.407

0.268

0.168

LR

CR

HHI

ITAX

IFF

GDP

INS

0.101

-0.444 -0.13 0.055

0.232

0.27

0.112

-0.001

-0.298

-0.167 -0.165

0.236

0.145

0.449

0.03

1

CR

1

HHI

0.284

0.105

0.185

1

ITAX

-0.04

-0.078 1

INS

-0.477 -0.383 0.439

GDP

1

1

IFF

-0.163 -0.659 -0.391 0.355

-0.187 -0.673 -0.58

-0.052 -0.056 0.51

-0.265 0.364

-0.638 1

1

REQUITY LR

Source: Financial statements of banks, NBE, MoFED reports and own computation

0.238

0.369

-0.38

-0.648 0.313

0.375

0.228

-0.101 0.038

-0.427 0.213

0.176

REQUI-TY

-0.614 0.532

0.401

OPC

1

MQUALITY OPC

MQUA-LITY -0.215 -0.513 1

1

SIZE

NIM

NIM

Table 4.4 Correlation matrix of dependent and independent variables

158

Determinants of Net Interest Margin

159

4.4 Results of Regression Analysis As presented in the third chapter, the empirical model used in the study in order to investigate the determinants of commercial banks’ NIM Ethiopian was provided as follows: ܰ‫ܯܫ‬௜ǡ௧ ൌ  ߚ଴  ൅  ߚଵ ܵ݅‫ ݁ݖ‬൅  ߚଶ ܱܲ‫ ܥ‬൅ ߚଷ ‫ ܻܶܫܮܣܷܳܯ‬൅ ߚସ ܴ‫ܻܶܫܷܳܧ‬ ൅ ߚହ ‫ܴܥ‬൅ߚ଺ ‫ ܴܮ‬൅  ߚ଺ ‫ ܫܪܪ‬൅  ߚ଼ ‫ ܺܣܶܫ‬൅ ߚଽ ‫ܨܨܫ‬ ൅  ߚଵ଴ ‫ ܵܰܫ‬൅  ߚଵଵ ‫ ܲܦܩ‬൅ ߝ As shown in table 4.5, the R-squared statistics and the adjusted-R squared statistics of the model was 78% and 71.6% respectively. The remaining 28.4% of change was explained by other determinants that are not included in the model. The null hypothesis of F-statistic that the R2 is equal to zero was rejected at 1% as the p-value was sufficiently low. Among the bank-specific explanatory variables included in this study, size, operating costs, management quality and credit risk had a statistically significant impact on NIM; only equity ratio and liquidity risk had none. The two industry-specific determinants included in this study (concentration and implicit tax) were statistically significant, based on traditional levels of significance1. Among the three macroeconomic determinants (INS, GDP, and IFF) included in the model, only inflation (IFF) was statistically significant. Management quality and inflation were significant at 1% significance level since the p-values to these variables were 0.0007 and 0.0021 respectively. Implicit tax was the only explanatory variable found significant at 5% significance level. Finally, variables like SIZE, OPC CR, and HHI were significant at 10% significance level. As table 4.5 shows, the coefficients of operating costs, equity ratio, credit risk, implicit tax, inflation, macroeconomic instability and GDP were positive. The respective coefficients to these variables were 0.555, 0.043, 0.033, 0.360, 0.037, 0.0003, and 0.010 respectively. In contrast, the coefficients of variables like size, management quality, liquidity risk and concentration were negative.

 1

1%, 5% and 10% level of significance are sometimes called traditional levels of significance

160

Chapter Six

Table 4.5: Regression Results for determinants of Ethiopian banks’ NIM first step model.

Variable

Coefficient

Std. Error

t-Statistic

Prob.

C SIZE OPC MQUALITY REQUITY CR LR HHI ITAX IFF INS GDP

0.065606 -0.003820 0.555878 -0.014581 0.043420 0.033392 -0.000139 -0.062327 0.360656 0.037476 0.000363 0.010557

0.036641 0.002027 0.285678 0.004075 0.033256 0.018598 0.013370 0.036765 0.173489 0.011668 0.000856 0.037901

1.790538 -1.884168 1.945819 -3.577852 1.305615 1.795477 -0.010428 -1.695272 2.078843 3.211968 0.423773 0.278539

0.0783 0.0643* 0.0563* 0.0007*** 0.1966 0.0775* 0.9917 0.0951* 0.0418** 0.0021*** 0.6732 0.7815

R-squared Adjusted R-squared S.E. of regression F-statistic Prob (F-statistic)

0.780847 0.716179 0.006503 12.07468 0.000000

Durbin-Watson stat

1.503840

***, **, and * denote significance at 1%, 5%, and 10% levels respectively. Source: Financial statements of banks, NBE, MoFED reports and own computation

To investigate the most relevant determinants of NIM of Ethiopian banks it was necessary to re-regress variables that were significant in the first and general econometric model. Now therefore, equation 4 of the previous chapter can briefly be expressed as follows:

ܰ‫ܯܫ‬௜ǡ௧ ൌ  ߚ଴  ൅  ߚଵ ܵ݅‫ ݁ݖ‬൅  ߚଶ ܱܲ‫ ܥ‬൅ ߚଷ ‫ ܻܶܫܮܣܷܳܯ‬൅ ߚସ ‫ܴܥ‬ ൅ ߚହ ‫ܫܪܪ‬൅ߚ଺ ‫ ܺܣܶܫ‬൅  ߚ଻ ‫ ܨܨܫ‬൅ ߝ Comparing the results of the two regressions, the following results were investigated. First, major differences were not found in R-squared statistics, adjusted R-squared statistics and Durbin-Watson stat in both regression results. Second, all variables included in the model remain statistically significant. However, except operating costs, management quality and inflation, other variables included in the second step regression were not significant at the level of significant similar with the first step regression results. Third, with the exceptions of ITAX and HHI, the sign

Determinants of Net Interest Margin

161

and magnitude of the coefficients were nearly similar to the first regression. The above result can be understood in the following ways. First, the inclusion of variables that were not significant in the first model had no economic account to the model. Second, little difference among the regression results may be due to the decrease in the number of regressors. Table 4.6: Regression Results for determinants of Ethiopian banks NIM second step model.

Variable

Coefficient Std. Error

t-Statistic

Prob.

C SIZE OPC MQUALITY CR HHI ITAX IFF

0.080475 -0.004180 0.549900 -0.013253 0.030341 -0.074852 0.438353 0.035635

3.502768 -2.171136 1.988128 -3.428898 3.020712 -4.132852 4.581792 4.133190

0.0008 0.0336** 0.0510* 0.0011*** 0.0036*** 0.0001*** 0.0000*** 0.0001***

R-squared 0.773811 Adjusted R-squared 0.725093 S.E. of regression 0.006400 F-statistic 15.88358 Prob(F-statistic) 0.000000

0.022975 0.001925 0.276592 0.003865 0.010044 0.018111 0.095673 0.008622

Durbin-Watson stat

1.499999

***, **, and * denote significance at 1%, 5%, and 10% levels respectively. Source: Financial statements of banks, NBE, MoFED reports and own computation

5. Analysis and Discussion This section mainly focuses on the results of the two-step regression analysis for the selected factors that have an impact on bank NIM. Size: as anticipated, in both models bank size had a negative impact on the NIM of banks. This indicates that large operation size is linked with a low interest margin. This may be explained by the phenomenon that some large banks aggressively grow their credit business at low margins. In this case, growth may be driven by different factors. First, banks that have high-impaired loan ratios may expand their credit portfolios intentionally at low margins to reduce the ratios. Second, especially large and inefficient banks may increase credit business by simply rewarding credit teams on sales performance, instead of risk-adjusted performance. Third,

162

Chapter Six

banks with a larger branch network can penetrate deposit markets and mobilize savings at a lower cost. Operating costs: A positive and significant relationship was found between bank NIM and operating costs. It has a coefficient of 0.555 and 0.549 in the two modes respectively, which is the largest coefficient among all variables included in this study. This means a 1 percent increase in operating costs enables the net interest margin to increase by 0.555 or 0.549 percent. The result confirms the well-known assertion that banks operating with high costs due to diseconomies of scale must operate with high margins to cover those costs. Therefore, a decline in the Ethiopian banks’ NIMs over the last ten years should be the result of reduction in the level of banks’ operating expenses. A good initiative and promising activities that have been carried out on the use of ATM machines and movement towards electronic banking systems must be cause for the reduction of operating costs. Enhancing operational efficiencies to exploit scale and scope economies must become an urgent priority of banks in Ethiopia. There are many possible insightful and fruitful alternative measures that commercial banks operating in Ethiopia can use for further cost reduction. For example, a move to high-speed cheque readers and cheque images systems, by offering a broader array of deposit and investment products such as money market accounts, mutual funds, and securities underwriting. Management quality: The coefficient of this variable was negative and statistically significant at 1% significance level. This result shows that the efficiency of management is important in determining interest margins, and that the poorer the management, the higher the interest margin, or vice versa. The result was as hypothesized and also supported by Afzal (2011), Schumacher and Saunders (1997), Liebeg and Schwaiger (2009). Equity ratio: A positive but insignificant relationship between the equity ratio and NIM of commercial banks in Ethiopia was found. Basically, the relationship between NIM and equity ratio remains ambiguous and prolonged in the two extremes, i.e., increasing or decreasing the margin. The result of this study does not support either of the two extremes, and was not as expected. Credit risk: Credit risk was found to have a positive coefficient, as would be expected. This indicates that banks providing credit for riskier projects require higher margins as compensation for future uncertainty. As for the

Determinants of Net Interest Margin

163

banks, a high interest margin is seen as a premium for bearing credit risk, which is perceived to be high, in view of the long default culture in the Ethiopian banking system. Therefore, a consistent decline on NIMs of commercial banks in Ethiopia would be due to a steady decline of nonperforming loans, which again resulted from the government’s significant reforms on the reduction of banks non-performing loans. Liquidity risk: A negative but insignificant relationship between liquidity risk and the NIM of banks was found in this study. Thus the hypothesis that states there is a significant relationship between the liquidity risk and NIM may be rejected, as data did not support the hypothesis. The insignificant parameter of this variable indicates that the liquidity structure does not affect Ethiopian banks’ NIMs. As far the result shows, whether on the former experience (excess liquidity) or the recent experience (shortage of liquidity), Ethiopian banks did not take adjustment on their interest margin with their level of liquidity. Banking Concentration proxied by HHI in deposit was significant and negative. The coefficient of HHI in both regressions was negative (-0.062 in the first and -0.074 in the second regression). According to the result, the higher banking concentration, which means weak competition in the industry, decreases the NIM of commercial banks in Ethiopia. The result was not as expected. According to Crowley (2007), in a free-marketoriented financial system, margins might be expected to be negatively related to factors affecting competition, including concentration, the number of banks, and the size of the market, since reduced competition would allow banks to profit from higher margins. However, for the banking industry in Ethiopia, this result can be understood as follows. First, there is the existence of small but powerful banks like CBE in our case. In countries with a small number of powerful banks, the large banks could restrict competition by keeping margins artificially low. In the case of a large public sector bank like CBE, there would be less of a profit constraint because the bank could be recapitalized by the government, and even in the case of a large private bank there could be an expectation of assistance when needed2. Second, in some countries, high concentration does not necessarily imply a lack of competition. For example, Girardone et al. (2007) found that countries that have similar concentration indexes,

 2

We do not have evidence whether large private banks in Ethiopia have government assistance or not.

164

Chapter Six

such as Argentina and Brazil, appear to have considerably different competition statistics. These findings broadly suggest that high concentration does not necessarily imply low competition. Finally, concentration in this study was measured by HHI in deposit. However, recent literature casts doubt over its propriety in serving as a proxy for competition (Fenta, 2012). Berger et al. (2004, cited in Fenta 2012) suggest that concentration may not be an appropriate measure of competition because it may not prohibit competition in a market with less regulation and more possibility for foreign bank entry. Implicit tax was found positive and significant at 5% and 1% significance level in the first and the second regressions, respectively. According to the results, a 1 percent increase in implicit tax enables the net interest margin to increase by 0.43 percent. This very considerable impact of implicit tax on the NIMs of Ethiopian commercial banks can be due to the following facts. First, strong regulated policies on reserve and liquidity requirements may lead to the emergence of tax evil in the banking sector, which can easily be transferred into lending and deposit. A reserve requirement with no interest payment has a high opportunity cost as it squeezes the excess reserve available for banks to advance credit, reducing the scope of the banks’ income-earning assets3. Second, the introduction of exceptions for special lending categories (priority sectors) to direct credit, which is a key for attaining political mission, may be one factor. Mandatory investment implies inefficient allocation of resources, where banks continue giving funds to prioritized sectors. Finally, government direct control in interest rate limits the banks’ efforts to capture high-yielding investments. Inflation: Despite the low coefficient (0.037 or 0.035), inflation was found significant at 1% level of significance in both regressions. This suggests that low inflation is a critical element in the reduction of banking margins in Ethiopia. Of course, this is not surprising, as low inflation rates reduce banks’ operating and transaction costs, particularly in countries like Ethiopia wherein the bulk of these costs are labour-related and pay scales are linked to inflation rates. In this regard, Robinson (2002) further notes that: “Low and stable inflation puts a floor on deposit rates and limits the mark-up factor on the real return on assets that banks target.”

 3 According to NBE computation of NIMs of commercial banks, reserve requirement was not assumed as an income generating asset. This clearly indicates that commercial banks reserve at NBE did not provide interest for respective banks.

Determinants of Net Interest Margin

165

Macroeconomic instability: one of the most common indicators of macroeconomic instability, the volatility of the exchange rate, does not have a significant impact on interest margins of Ethiopian banks. This result is somewhat surprising, since increased exchange rate variability should lead to greater uncertainty and higher risk premiums. Ethiopian banks’, with the exception of CBE, little international banking activities may be one reason why the banking industry margins in Ethiopia are free of influence from exchange rate volatility. However, even though it is limited, they would surely have clients in the tradable sectors whose ability to repay would be affected. Obviously, the result of this study supports the result found by Tenant and Folawowe (2009) who conclude that much of the debate on exchange rate policies and management may not be highly relevant to banking margins, as whilst exchange rate volatility may impact a country’s exports and balance of payments, there is no evidence of a transmission mechanism by which this effect is translated into a widening of banking sector margins. GDP: The result suggests that the expanded national income in Ethiopia has not been associated with higher or lower interest rate margins. Basically, it is not that surprising in a country where the contribution of banks as well as financial institutions as a whole has not been important to GDP for many years. 

6. Conclusions and Recommendations 6.1. Conclusions Despite the large volume of studies on this area, the majority were carried out in developed countries. According to the nature and purpose of each study, a number of explanatory variables have been proposed. Previous studies, both those carried out on developed economies and the few made on developing economies on the determinants of NIMs of banks, summarized that the interest margin of banks is usually a function of bankspecific, industry-specific, regulatory, and macro-economic factors. Now therefore, based on the review of previous studies, theories, and problems highlighted on interest rate margins, this study aims to investigate bank-specific, industry-specific regulatory and macroeconomic determinants of NIMs, and to identify the most relevant factors explaining NIMs of the banking industry in Ethiopia over the period 2002 - 2011. To comply with these objectives, this paper is primarily based on a

166

Chapter Six

quantitative research method. The data were mainly obtained from NBE and MoFED through structured documentary analysis, and analysed by employing a two-step multivariate OLS model using the statistical package EVIEWS 6. The results of the two-step multivariate fixed effect regression on the determinants of NIMs of sample commercial banks evince the following main insights. First, the results of this paper indicate that, with some exceptions, the NIM of Ethiopian commercial banks declined over the last ten years. Second, the econometric analysis supports evidence that the observed commercial banks’ NIMs in Ethiopia were significantly determined by the size of the banks, management quality, operating costs, credit risk, banking concentration, implicit tax, and inflation. The rest of the variables included in the model, like equity ratio, liquidity risk, GDP, and macroeconomic instability are determinants that have little or no impact on the NIM of Ethiopian banks, as all those variables were not significant even at a 10% significance level. Third, to the purpose of the second step regression, generally speaking our results indicate that NIMs of commercial banks are driven more by bank-specific (operating costs) and industry-specific variables (implicit tax and concentration) than by macroeconomic variables. Fourth, in Ethiopia there was high exchange rate volatility during the sample period. Finally, the evidence in this paper shows that the degree of market concentration in Ethiopia’s banking industry has declined constantly over time.

6.2. Recommendations and Policy Implications In the light of the major findings obtained, to end up with narrower interest margins, several policy implications can be forwarded from the study. Firstly, if Ethiopia’s commercial banks cannot reduce their operating costs, do not improve their management quality and economies of scale, and cannot continue to seriously deal with the issues of the high levels of credit risk, maintaining today’s banking stability will not be easy. Secondly, if there is to be any success in reducing commercial banks’ interest margins to support long-term economic growth, the government has to take reform on the source of implicit tax on banks, such as reserve and liquidity requirements. Thirdly, the result of this study shows that bank-specific and industryspecific variables are the most relevant factors that determine NIM in the

Determinants of Net Interest Margin

167

banking industry of Ethiopia. But it does not mean that the margin is free of influence from macroeconomic variables. For example, though the coefficient is small, inflation was found to be positively related and significant at 1% significance level. In addition, exchange rate volatility will have an influence if banks want to improve their participation in international banking. Therefore, an oversight of these variables will have to keep the margin as it is or lower it further below the existing margin. Fourthly, the variables used in the statistical analysis did not include all factors that can affect Ethiopian banks’ interest margins. Thus, future research can incorporate other factors that can determine the NIM of Ethiopian banks, like statutory reserve requirement, structure and development of the banking industry, Treasury bill rate, discount rate and ownership structure. Moreover, adopting more than one proxy to measure one variable will have the advantage of reducing ambiguous and surprising results, as observed in this study, e.g., the impact of banking concentration on NIM, which was negative and strongly statistically significant. The results of such variables may again be difficult to forward policy implications. In the end, it is necessary to see the past and present facts of the banking industry in Ethiopia. Ethiopian banks are characterized by inefficiency, poor technology adoption, weakness in international banking, etc. But, surprisingly, Ethiopian banks have the lowest margins in English-speaking African countries. Many scholars support that a lower margin is an implication of banking sector efficiency and development. If it is so, why does it fail to be true in the Ethiopian context? The researcher was really impressed by this issue. Therefore, further research has to be carried out to clarify whether a higher or lower margin is desirable in Ethiopia.

References Ayesha, Afzal, 2011, ‘Interest rate spreads, loan diversification & market discipline in Pakistan's commercial banking sector’, PhD thesis, University of Pakistan. Beck, T. & Ross Levine, 2004, ‘Stock markets, banks, & growth: Panel evidence’, Journal of Banking & Finance, Vol. 28, pp. 423-442 Bencivenga,V. R. & Bruce D. Smith, 1991, ‘Financial intermediation & endogenous growth’, The Review of Economic Studies, Vol. 58, pp. 195-209.

168

Chapter Six

Brock, P. L., Rojas Suarez L., 2000, ‘Understanding the behavior of bank spreads in Latin America’, Journal of Development Economics, vol. 63, no. 1, pp. 113-134. Brock, P. & H. Franken, 2002, ‘Bank interest margins meet interest rate spreads: How good is balance sheet data for analyzing the cost of financial intermediation?’, Working Paper, central bank of Chile. Chekol, B. Y., 2011, ‘Determinants of interest margins of commercial banks in Kenya’, Chirwa, E. & M. Mlachila, 2004, ‘Financial reforms & interest rate spreads in the commercial banking system in Malawi’, IMF Staff Papers, vol. 51, pp. 96 –122. Crowley, J.2007, ‘Interest rate spreads in English-Speaking African countries’, IMF Working Paper, vol. 07, no. 101, International Monetary Fund, Washington, Demirgüç-Kunt, A. & Harry Huizinga, 1999, ‘Determinants of commercial bank interest margins & profitability: Some international evidence’, World Bank Economic Review, Vol. 13, pp. 379-408. Demirgüç-Kunt, Asli, Luc Laeven and Ross Levine, 2003, ‘Regulations, market structure, institutions, & the cost of financial intermediation’, Working Paper, No. 9890, National Bureau of Economic Research, Diamond, D. W., 1984, ‘Financial intermediation & delegated monitoring’, The Review of Economic Studies, Vol. 51, no. 3, pp. 393414. Fanta, A. 2012,’Banking reform and SME financing in Ethiopia: Evidence from manufacturing sector ‘, African Journal of Business Management, vol. 6, no.19, pp.6057-6069 Girardone, C., et al., 2007, ‘Competition, efficiency & interest rate margins in Latin American Banking’, Research paper, Lelissa, B. T., 2007, ‘The impact of financial liberalization on the ownership, market structure & performance of the Ethiopian banking industry’, MBA thesis, Addis Ababa University, Malhotra, N., 2007, Marketing research: An applied orientation, 5th ed., Phi, New Delhi. Ndung’u, N. & Ngugi, R.W., 2000, ‘Banking sector interest rate spreads in Kenya’, Kenya Institute for Public Policy Research & Analysis, Discussion Paper, No. 5. Petra T, 2007, ‘Panel data; fixed effects random effects dynamic panel data models’, Cambridge University Press, New York. Randall, 1998, ‘Interest rate spreads in the Eastern Caribbean’, IMF Working Paper, no. 98/59.

Determinants of Net Interest Margin

169

Robinson, J. W., 2002, ‘Commercial bank interest rate spreads in Jamaica: Measurement, trend & prospects. —. 2002, ‘Commercial bank interest rate spreads in Jamaica: Measurement, trend & prospects. Saunders, A. & L. Schumacher., 2000. “The determinants of bank interest rate margins: An international Study’, Journal of International Money and Finance, vol. 19, no. 6, pp. 813–832. Schwaiger, M.S. & Liebig, D., 2009, ‘Determinants of the interest rate margins in Central and Eastern Europe’, Östereichische Nationalbank, Financial Stability Report no. 14. Ugur & Erkus, 2010, ‘Determinants of net interest margin in Turkish banks’, Research paper. Zhou & C.S. Wong, 2003, ‘The determinants of net interest margins of commercial banks in Mainland China’.

 



PART III MICROFINANCE, POVERTY REDUCTION, AND SOCIAL INCLUSION 





CHAPTER SEVEN MICROFINANCE: DOES IT SUPPORT HOUSEHOLDS TO ACHIEVE AN INCOME ABOVE SELF-SUFFICIENCY? EVIDENCE FROM RURAL NORTHERN ETHIOPIA ACHAMYELEH TAMIRU EWUNETU

Abstract This paper analyses the impact of participation in microfinance on child malnutrition and annual per capita consumption expenditure at the household level in Northern Rural Ethiopia. An important contribution of this study is that the researcher employed a high level of econometric sophistication to identify the clear effect of microfinance participation. The reason behind the use of sophisticated methods in this study is that the methods used by earlier studies suffer from standard econometrics limitations. The main challenge in the research was dealing with endogenous treatment selection. To address this challenge, the researcher used three alternative approaches: A recent approach developed by Klein and Vella (2009) that exploits restrictions on the second moment as an instrument for the endogenous treatment dummy for identification given a continuous dependent variable, the standard Heckman bivariate normal (BVN) selection model, which is identified without an instrument from the nonlinearity of inverse mills ratio, allowing for essential heterogeneity, and the ‘Minimum-Bias Bias-Corrected’ estimator, due to Millimet and Tchernis (2010), that corrects for endogeneity bias without exclusion restrictions. The estimates obtained give average effects of participation in microfinance programs, and can thus lay claim to a higher level of generality, compared to most studies that focus on one specific identification strategy and identify causal effects based on a strict assumption of conditional independence, which is often violated in reality.

Microfinance

173

The results of this study suggest that there is no strong evidence of a causal effect of participation in microfinance on reducing child malnutrition and increasing annual per capita consumption expenditure.

1. Introduction Microcredit - broadly speaking, the provision of loans to very small businesses - is an increasingly common weapon in the fight to reduce poverty and promote economic growth (Karlan and Zinman, 2009). Lack of access to credit is a key obstacle for economic development in poor countries. As a counter-offensive against poverty, therefore, several microfinance schemes have gone operational around the world, providing financial access to millions of poor people both in rural and urban areas, and the term microfinance has become a development catchword since the 1970s. The microfinance revolution got considerable momentum around the world in the last two and half decades. Marking such a peak, the UN declared 2005 a ‘microcredit’ year, and Mohamed Yunus and his Grameen Bank won the 2006 Nobel Peace Prize. This signals the ways in which microfinance has shaken up the world of international development (Berhane Tesfay, 2009). Today’s microfinance institutions have been established based on Grameen’s modality. Grameen shaped the modern industry of microfinancing (De Aghion et al., 2007). The Tigray regional state of Ethiopia is one of the regions in Ethiopia hardest hit by recurrent droughts that are characterized by food shortages, famines, and excess mortality (Webb et al., 1992). Moreover, many of Ethiopia’s historical cross-boundary wars (e.g., the 1896 and 1935 Italian invasions), recent civil wars (e.g., the protracted civil war that ended in 1991) and border conflicts (e.g., the 1998-2000 conflict with Eritrea) took place in this region. Coupled with decades of poor governance, all of this resulted in environmental and ecological imbalances in the region, which are manifested in degraded lands, poor resource bases, and population pressure, which led to further land fragmentations and mismanagement, and hence to an even poorer performance of agriculture, also relative to the national average (Woldehanna, 2000). Studies indicate that close to 50 per cent of households in the region produced less than their annual food requirements in 1997 and 2000 (Hagos, 2003). In 2005, around 48 percent of the population of Tigray was unable to meet the basic requirements of consumption (MoFED, 2006).

174

Chapter Seven

Dedebit Credit and Savings Institution (DECSI), the attention of this study1, was primarily focused on providing credit services and other financial services (e.g., saving) to poor ‘credit constrained’ farmers in Tigray. Microfinance practitioners, policymakers, and donors around the globe have ambitious goals for expanding access to credit in the march against poverty. Yet, despite often grand claims about the effects of microcredit on borrowers and their businesses, there is relatively little convincing evidence on microfinance impacts. Hence, little is known about where the impacts are the strongest (Karlan and Zinman, 2009). Non-randomized empirical evaluations of microcredit impacts are typically complicated by classic endogeneity problems; e.g., client selfselection and lender strategy based on critical unobserved inputs like client opportunity sets, preferences, and aptitude (Karlan and Zinman, 2009). Another weakness of existing impact studies, which is rarely mentioned, is the fact that they use income or expenditure as outcome variables. These measures are likely to be subject to measurement errors that are correlated with the true latent variables, that is, non-classical measurement errors. In a recent simulation study, Millimet (2010) showed that even small degrees of non-classical measurement error in the dependent variable can dramatically bias coefficients. This makes the identification of the causal effects of such programs difficult (ibid.). There is also a growing concern amongst academics that the expectations of microfinance are not being met. Rigorous research approaches and employing randomized trial designs have begun to suggest that microfinance may not be the golden bullet that many had hoped for (Stewart et al., 2010). The methodological rigor of various impact studies varies considerably. Westover (2008) in general indicates the lack of stringent, rigorous impact studies with many impact studies done by Microfinance Institutions (MFIs) themselves that are case and locale-specific, and qualitative in nature. They also tend to rely heavily on anecdotal evidence. De Aghion et al. (2007) and Cotler and Woodruff (2008) reviewed impact studies and found that those with the largest methodological flaws tended to find the strongest positive

 1

The operation of DECSI in Tigray, Ethiopia, serves as a basis for this study.

Microfinance

175

impacts of microfinance 2. The bottom line is that there is really weak academic evidence that microcredit reduces poverty (Roodman, 2009). DECSI prides itself on providing credits to the poor in Tigray living in absolute poverty. Berhane Tesfay (2009) found out that microfinance credit raised annual per capita household consumption significantly. A recent assessment of the impact of microfinance, based on survey households’ poverty indicators, reported that in notwithstanding some momentary impacts, poverty is rampant in the study areas, even in the presence of micro-finance programs (Hailai, 2010). While DECSI’s quick expansion of its network throughout the region is clear, the policy ramifications are not. In addition, this research attempts to contribute to a small, but substantive as well as methodologically significant literature in the area of microfinance impact evaluation. In light of the above discussion, the primary evaluation question to guide the inquiry is geared towards the impact of microfinance as applied to the Tigray region of Ethiopia. This paper answers an important question: Does microfinance reduce rural poverty? In order to answer this question, the researcher used the data from the 2006 and 2010 survey rounds conducted in 16 communities in the Tigray region. Instead of relying on only one estimator, the researcher proposes to use a combination of estimators (based on household survey data for 2 rounds) to arrive at the answers set in this research paper. The estimators used are the Heckman bivariate normal selection model (BVN), a new estimator called bias corrected inverse-probability of weighting estimator (IPW) or simply called biased–corrected estimator, and the Klein and Vella (KV) estimator.

 2

One piece of evidence, for example, is a study by HAGOS, F., HOLDEN, S. & PENDER, J. 2006. The Effect of Program Credit on participation in off-farm employment and welfare of rural households in Northern Ethiopia: Agricultural University of Norway. They used an instrument for the endogeneous treatment, in their case the choice to participate in the microfinance program, as the treatment dummy, by interacting the number of adult labourers in the household with the village dummy and used the interacted term as an instrument for the endogeneous treatment variable. However, adult male labour is not likely to satisfy the excludability assumption, as it directly affects household’s income and the outcome equation.

176

Chapter Seven

The researcher investigated these issues using household survey data collected in the Tigray Region of Ethiopia, where the local microfinance institution prides itself on providing credit to the poor. It can be cautiously concluded that microfinance credit has a positive impact; specifically it reduces child malnutrition and increases annual per capita consumption expenditure. The remaining parts of this paper are organized as follows. The second section of the paper presents information about DECSI’s program. The third part of the paper presents a brief overview of the underlying theory and literature relevant for the study. The fourth part describes the data and introduces the empirical identification and estimation strategy of the paper. The fifth part presents the key findings and relevant discussions, and the last part concludes.

2. Background The attention of this study is impact evaluation of a local microfinance institution called Dedebit Credit and Savings Institution (DECSI), operating in the Tigray regional state of Ethiopia. Tigray is the most northern region of Ethiopia3. Subsistence agriculture is the mainstay of the rural population in Tigray. It includes mainly crop, livestock, and mixed farming. Farming systems are characterized by traditional ways of doing things. Labour and animal power are the main inputs in production. Irrigation is limited and production depends on short-season annual rainfall. With the exception of the southern plateau that enjoys an additional short rainy season, the Belg (March-May), the principal rainy season in this region is the Kiremt (JuneSeptember) season. This season typically belongs to the monsoon rainy season of the semi-arid, Sudano-Sahelian dry land belt of Africa that extends from the west (Atlantic Ocean) to east (Ethiopia and Eritrea), which is characterized by erratic rainfall and recurrent droughts (Segele and Lamb, 2005). To reverse the poverty situation and help the poor in Tigray, the Relief Society of Tigray (REST) was established in 1978,4 and has been engaged

 3

See the map online at:http://www.maplandia.com/ethiopia/tigray/ The Relief Society of Tigray was founded as a humanitarian organization in 1978, three years after the war between the central government of Mengistu Haile 4



Microfinance

177

in development programs principally in environmental rehabilitation, agricultural development, aid, social development, rural water supply, and credit and saving services (Borchgrevink et al., 2003). The pillar objectives of these programs are to improve the economic situation of the low income and poorest people in the Tigray Region, and to accomplish independence based on bona fide participation of the people, by embarking upon and surmounting the core grounds and consequences of poverty through advancing sustainable rural development (Hailai, 2010). In 1993, REST, the main NGO in the region, launched a socio-economic poverty survey in rural areas. Lack of access to credit appeared as one of the major obstacles to the rehabilitation of the region and its development. This marked the birth of the Dedebit Credit and Savings Institution (DECSI). This program of credit is created to help increase agricultural production, stimulate the local economy, reduce the influence of moneylenders, and increase incomes of the poor. The first operations began in 1994 and the organization was legally recognized in 1996 as part of the first law on microfinance in Ethiopia enacted that year. During its growth, DECSI received financial support from Novib (Netherlands), Norwegian Peoples Aid and SOS FAIM (Belgium and Luxembourg). Today, DECSI’s operation is limited to Tigray, and it has over 460,000 customers, and as such, DECSI is regarded as one of the four largest MFIs in Africa (Wikipedia, 2010). Since its establishment in 1994, DECSI has been providing the following three loan types: regular, agricultural input, and agricultural package loans. Besides, it provides saving services such as compulsory deposit of group and centre saving, voluntary deposit from loan clients and the public at large, and pension payments. Recently, DECSI has expanded its services, particularly in the area of agricultural input and agricultural package loans (mainly individual) and enterprise loans (Hailai, 2010). Detailed treatment of these operations by DECSI is not presented here, as these operations are not the interest of the researcher in this paper.5 The broad objectives of DECSI can be summarized into three points:

 Mariam and the Tigrayan People's Liberation Front began. REST's mandate was to assist drought and war-affected people living in the areas of Tigray under the control and administration of the TPLF. 5 For detailed information related to these operations and other important aspects of DECSI, see www.decsi.com.et

178

Chapter Seven

improving food security at household levels both in the rural and urban areas of the region, creating job opportunities to the unemployed parts of the population through promoting micro, small, and medium enterprises in the region and stimulating the local economies through offering adequate and efficient financial services, and building financially sound and sustainable institutions6.

2.1 Conclusion Besides the nice institutional setting of DECSI and its endeavour to pull out the poor in Tigray from the tentacles of poverty and backwardness, studies indicated that close to 50 per cent of households in the region produced less than their annual food requirements in 1997 and 2000 (Hagos, 2003). Even after a decade of DECSI’s presence, in 2005, around 48 per cent of the population of Tigray was unable to meet the basic requirements of consumption (MoFED, 2006). This shows that poverty is part of life in Tigray, even after DECSI, and hence calls for further focused impact evaluation studies on whether DECSI, if it has an observed impact, is supporting the poorest of the poor in Tigray. The point of focus for this study is therefore to evaluate whether DECSI is supporting the poorest of the poor in Tigray.

3. Theory and Literature Review 3.1 Definitional and Conceptual Issues This section will explore the definitional and conceptual issues surrounding microfinance and poverty. In the simplest terms, the idea is that micro-credit and micro savings allow the poor to invest their money in the future, increase their incomes and ‘lift themselves out of poverty’. This simple causal chain is represented in figure 1 below. Today’s researchers and practitioners are unpacking this chain in this review, and develop a more complex evidence-based understanding of how microfinance may (or may not) have positive impacts on the poor (Stewart et al., 2010).

 6

To achieve these key objectives, DECSI has designed strategies that can be seen online at www.decsi.com.et

Microfinance

179

Figure 1. A simple causal chain from microfinance to poverty alleviation

Source: Stewart (2010)

3.2 What is Microfinance? The term ‘micro-credit’ was first coined in the 1970s to indicate the provision of loans to the poor to establish income-generating projects, while the term ‘microfinance’ has come to be used since the late 1990s to indicate the so-called second revolution in credit theory and policy that is customer-centred rather than product-centred (Elahi and Rahman, 2006). But the terms ‘micro-credit’ and ‘microfinance’ tend to be used interchangeably to indicate the range of financial services offered, specifically to poor, low-income households and micro-enterprises (CGAP, 2010; Brau and Woller, 2004).7 Like anyone else, poor people need an array of financial services to help them deal with a range of short to long-term consumption needs and the ups and downs of income and expenses, to make use of opportunities, and to cope with vulnerabilities and emergencies. The spectrum of financial services available to meet these needs include investment (savings), lending (credit services), insurance (risk management) and money transfers. But the poor’s access to formal financial services is limited, and the services available do not acknowledge the diverse requirements of the poor (Matin et al., 1999). Instead, the poor people depend on various types of formal and informal community funding, credit unions, moneylenders, co-operatives, self-help groups and associations (like accumulating savings and credit associations, rotating savings and credit associations, burial societies), and financial NGOs. In addition, with commercial

 7

Microfinance principally encompasses micro-credit, micro-savings, microinsurance and money transfers for the poor

180

Chapter Seven

financial institutions considering ways in which to provide financial services to the poor in a profitable manner, microfinance services are now provided by a whole spectrum of role players. In reality, there is a mix of financial services accessed by poor people from a variety of service providers, depending on local knowledge, history, context, and need (Matin et al., 1999).

3.3 Microfinance and its Effect on the Poor Once poor people access financial services, the question of outcome arises. One of the crucial debates in microfinance literature is expressed by Brau and Woller (2004) as the trade-off between financial self-sufficiency and sustainability, the depth of outreach, and the social welfare of service recipients. Roodman (2010) refers to the latter as ‘judging microfinance by whether it reduces poverty, increases freedom, builds industries’ (Stewart et al., 2010). With the most important goal of microfinance being reducing poverty, changes in income levels of individuals and households are often used as a measure of the impact of microfinance (Johnson and Rogaly, quoted in Makina and Malobola, 2004). But Wright (1999) highlights why income levels cannot be the only measure: increasing income does not per se mean that poverty is reduced, as it depends on what the income is used for. The outcomes used to measure the impact of microfinance on the poor have to take into account the conceptualizations of poverty and who the poor are (Stewart et al., 2010). Studies of the impact of microfinance on the poor should consider different outcome variables. These could include increased consumption, income stability and income growth, reduced inequalities, health and education outcomes, nutrition improvements, employment levels, empowerment indicators, reduced vulnerability to shocks, strengthened social networks, and strengthened local economic and social development, and can vary according to who has been reached by these microfinance services (e.g. women, the poorest) (Stewart et al., 2010). Kabeer (2003) refers to such dimensions of impact as cognitive, behavioural, material,

Microfinance

181

relational, and institutional changes.8 Since the 1970s, and especially since the new wave of microfinance in the 1990s, microfinance has come to be seen as an important development policy and poverty reduction tool (Stewart et al., 2010). Some argue that microfinance is a key tool to achieve the Millennium Development Goals (MDGs) (Littlefield and others; and World Savings Bank Institute quoted in Stewart et al., 2010). The assumption is that if one gives more microfinance to poor people, poverty will be reduced. But the evidence regarding such an impact is challenging and controversial, partly due to the difficulties of reliable and affordable measurement, of fungibility9, the methodological challenge of proving causality (i.e. attribution), and because impacts are highly context-specific (Brau and Woller, 2004).10 Questions regarding the impact of microfinance on the welfare and income of the poor have therefore been raised many times (e.g.Berhane Tesfay, 2009, Hailai, 2010, Makina and Malobola, 2004, Santen, 2010). Despite various studies, ‘the question of the effectiveness and impact on the poor of microfinance programs is still highly in question’ (Westover, 2008). Roodman and Morduch, as quoted in Stewart et al. (2010) reviewed studies on micro-credit in Bangladesh, and similarly conclude that ‘30 years into the microfinance movement we have little solid evidence that it improves the lives of clients in measurable ways’. Even the World Bank report, Finance (2007), indicates that ‘the evidence from micro-studies of favourable impacts from direct access of the poor to credit is not especially strong.’ Recently this debate became heated when the findings of two randomized controlled trials (RCTs)11 in the Philippines and India by the

 8

There are also some further issues that impact studies should not only look at individual and/or household-level impacts, but also look at impacts on the community, economy and national levels 9 This refers to the inability to tie particular funds to particular expenditure and changes in wellbeing. 10 Part four of this paper discusses in detail the methodological challenges of impact evaluation. 11 RCTs are seen by many as the gold-standard methodology for assessing impact. In RCTs, steps are taken to remove potential biases and isolate the true impact of the specific intervention (such as microfinance services). These primarily include randomisation to intervention (i.e. those who receive the service) and control (i.e. comparison) groups, the collection of data before and after the intervention is implemented, and careful consideration of sample size to ensure sufficient evidence to conclude on impact. Some argue that RCTs are the best way to



182

Chapter Seven

Massachusetts Institute of Technology’s Jameel Poverty Action Lab (Banerjee and others; Karlan and Zinman, quoted in Stewart et al., 2010) raised questions about the impact of microfinance on improving the lives of the poor. These studies did not find a strong causal link between access to microfinance and poverty reduction for the poor. The results of these first RCTs in the field of microfinance have spawned a heated debate (ibid.). The main value proposition put forward on behalf of micro-credit for the last quarter century is that it helps lift people out of poverty by raising incomes and consumption, not just smoothing them. So far, there is no very strong evidence that this particular proposition is true (Stewart et al., 2010). There is still a continued debate between researchers and practitioners on the impact of microfinance and this can be taken as a yardstick for the need of a rigorous and systematic evidence of the impact of microfinance on the poor (Stewart et al., 2010). With the micro-credit movement having its origin in Asia in the 1970s, much has been written about its thinking, practices, and impacts there. In contrast, there is relatively little known about microfinance in sub-Saharan Africa (SSA) to where the micro-credit movement spread in the 1980s, and where it became stronger in the 1990s (Stewart et al., 2010). With microfinance institutions aiming to serve the poor, SSA is an important region to consider when reviewing the impact of microfinance (ibid.). Regarding impact studies on microfinance in SSA using comparative study designs, there is only one RCT on the impact of micro-savings that has been completed so far (Dupas and Robinson, quoted in ibid.). The Poverty Action Lab is currently involved in two further impact studies for the Microfinance and Health Protection Initiative: one in Benin and the other in a village savings and loans programme in Ghana. There is also a

 measure the impact of microfinance programmes and improve product design. Nevertheless, RCTs require forward planning, with the intervention delivered as part of the study – rather than retrospective evaluation of an existing programme. Furthermore, long-term outcomes are expensive to follow up, and there can be ethical concerns about withholding interventions from the control group (KARLAN, D., GOLDBERG, N. & COPESTAKE, J. 2009. 'Randomized control trials are the best way to measure impact of microfinance programmes and improve microfinance product designs.'. Enterprise Development and Microfinance, 20, 167-176.

Microfinance

183

larger body of impact studies employing non-comparison evaluation designs – both non-experimental and quasi-experimental in nature. Yet, little systematic and rigorous evidence has been provided and assessed the nature of the evidence of the impact of microfinance on the poor in SSA (ibid.).

3.4 Childhood Malnutrition and Income In many ways, the approach taken to measure childhood undernutrition has much to recommend itself. It measures effective nutritional status, not just inputs, it is built up from individual data, usually with good and transparent sampling procedures and measurement protocols (due particularly to the standardized Demographic and Health Surveys), and it focuses on children who tend to be particularly vulnerable to nutritional deficiencies. It not only allows the production of aggregate indicators, but identification of particularly hard hit groups, and can easily be used as a monitoring device for policy purposes (Klasen, 2008). Anthropometric measures consist of three different indicators: stunting, wasting, and being underweight. These indicators offer insights into different dimensions of nutritional problems. Wasting (low weight for height) is an indicator of acute undernutrition particularly relevant in famines and to monitor acute food shortages, stunting (low height for age) is an indicator of chronic undernutrition focusing on persistent nutritional deficiencies, and underweight (low weight for age) a summary indicator combining both facets (Bloss et al., 2004). While all agree that environmental factors are much more significant than genetic differences in explaining differences in anthropometric shortfall between populations, quite a few studies suggested that genetic differences are important enough to be considered, particularly for international comparisons of anthropometric shortfalls (Gunther and Klasen, 2009). Regarding the impact of income, income poverty, and income growth on undernutrition, theoretical considerations suggest a close linkage. More resources at the household level improve the ability of household members to acquire more calories and of parents to invest more in the nutrition and health of their children. These linkages will likely be larger at an aggregate level than at household-level, as higher per capita incomes also tend to increase investments in public services in the areas of health, nutrition, water and sanitation, and social protection, which are three other important factors influencing hunger and childhood mortality (Klasen, 2008). There

184

Chapter Seven

are studies indicating that the effects of income on nutrition outcomes are stronger than mortality outcomes and stronger links between income and undernutrition of under-six children. One noticeable example using household survey data from 12 countries (Haddad et al., 2003) reported that an increase in income at the household and national level implies a reduction of child malnutrition. However, this result is contestable. There are reports showing that nutrient elasticities with respect to income may be close to zero (Behrman and Deolalikar, 1987). Therefore, whether or not higher incomes imply a reduction of child malnutrition is an empirical question. Given this paucity, this study (centred around one of the MFIs in SSA, DECSI) tries to provide (if available) evidence of impact, i.e. whether DECSI is impacting the poor people it seeks to serve and map out distributional effects across outcomes sought, and estimates the amount of poverty reduction made possible using robust identification strategies of impacts. The empirical identification and estimation strategy of the paper is the subject matter of part four.

4. Data, Empirical Identification and Estimation Strategy 4.1 Description of the Panel Data Set This study used household survey data collected in the Tigray Region in two rounds (2006 and 2010) from an average sample of 400 households. The household surveys were conducted to improve the understanding of the livelihood situation in Tigray and present facts to the attention of policymakers in their attempts to reduce poverty. The household surveys include household basic characteristics, plot level information, credit and saving information, anthropometric issues of households, information on malaria and safety nets programs. Four of the five administrative zones - Southern, Eastern, Central, and Northwestern - that cover most of the highlands of Tigray - are included in the household survey. This comprises eleven Woreda’s (districts), of which sixteen villages are sampled from each zone. From each village 25 households were randomly selected from the village list, and a standardized questionnaire, designed by Mekelle University in collaboration with the Norwegian University of Life Sciences (UMB), covering the above-mentioned issues, was administered.

Microfinance

185

The logic of the sampling scheme is as follows. The survey took place in the Tigray Regional State, which contains five administrative zones. To achieve better representation, sampling was done in two stages. First, stratified by altitude (mainly highlands), agricultural potential, population density, and access to infrastructure (mainly market, credit, and irrigation), and four tabias were selected from each zone. A tabia contains a group of villages. One village is selected from each sample tabia (Hagos et al., 2006). The study covered 16 communities purposively stratified by population density, market access, (non-) presence of an irrigation development project, and their location in the four different zones in the region: central, eastern, southern and western. The following section presents the data organization process, the variables description and limitation of the data used in this study.

4.2 Data Organization Process In this paper, the malnutrition of under-six children is used to indicate the poverty status of a particular household member. One indicator showing this is stunting (a measure of low height for age) which indicates chronic undernutrition focusing on persistent nutritional deficiencies (Klasen, 2008). There is a recommendation in public health literature that the calculation of stunting indicators (for comparison and policy purpose) is best done in a study population of less than 6 years of age12. The reason is that there are no (genetic) differences between populations in their growth and weight development between 0 and 6 years (ibid.). This rule of thumb provides a solution to, at least in this paper, the most widely discussed problem in Applied Econometrics, i.e. inconsistency of estimates. Moreover, there are recommendations in the literature concerning which anthropometric indicator is best suited for various program durations. Simondon (2010) suggested that stunting prevalence (or simply stunting) is the best-suited indicator for impact evaluation of long-term intervention programs (programs with a duration of more than two years, such as DECSI). Hence, in this paper, stunting is used as an indicator of children’s nutritional status and the z-score throughout this paper refers to the heightfor-age z-score.

 12 Even though this appeared to be the consensus view, it is not a rule universally accepted by everyone, particularly for international comparisons of anthropometric shortfalls.

186

Chapter Seven

To interpret this indicator and serve the purpose of this research, an indicator variable called z-score is generated. The z-score is calculated as the difference between the relevant anthropometric indicator of a child and the median of a reference population, divided by the standard deviation of the reference population. If this z-score is below -2, severe malnutrition is presumed; if it is below -3, severe undernutrition is presumed (UNICEF, 1998). Accordingly, the researcher calculated the z-score of all children between 0 and 6 years in the 211 households13 using Stata. After calculating the anthropometric z-score using the 2006 survey data set, only 102 out of 211 sample units of the 2006 data set remained for the final analysis.14 The dramatic reduction in the sample size to be used in the final analysis has happened after calculating the z-score of the height by age anthropometric indicator. The sample size shrunk from 211 to 102 due to the following two reasons: 1. Due to the fact that some households do not have children between 0 and 6 years of age they dropped out of the sample used for the final analysis. 2. Extreme (i.e. biologically implausible) z-scores, for height by age anthropometric indicator, are flagged according to the following system: height-for-age z-score (zlen) are biologically feasible in the -6൑ ‫ ݈݊݁ݖ‬൑ ͸ range. Accordingly, individual Z-score values out of the above allowable range are dropped, and these individuals are excluded from the sample. This process contributes to the reduction of the sample size in the final analysis, as only those observations in the above allowable stated range of z-score are included in the sample. The above process of data organization works in a similar fashion for the survey round conducted in 2010. Accordingly, even though 124

 13

The anthropometric indicators are calculated using stata on the basis of specific WHO Global database on child growth and malnutrition reference data (MUACfor-age) or 2006 CDC reference data set available online at: http://www.who.int/nutgrowthdb/ 14 In 2006, not all households were asked to provide anthropometric information, only 211 households out of the total surveyed in that round were asked for anthropometric information.

Microfinance

187

households are found for the final analysis15 that have anthropometric scores in the feasible range, only the 102 observations in 2010 are found to be matched panels with the 2006 households and hence used for the final analysis. Furthermore, it is worth mentioning that whenever there is more than one child under the age of 6 and the nutrition anthropometrics indicator of all the children falls in the desired range, the younger child’s anthropometrics information is used in the regression framework of this paper. Before leaving this section, it is appropriate to tell the total number of observations used in this paper. After organizing the data according to the procedures discussed above, the total number of observations that are obtained from the two survey rounds sum up to 204; 102 observations are obtained from the 2006 survey round and the remaining 102 observations are taken from 2010.

4.3 Description of Variables Used in the Study In order to obtain a consistent estimate of program credit intervention, the researcher tried to control variables that might have a logical connection with the dependent variable and that affect the probability of participating in program credit. Hence, the choice of the control variables is motivated by the objective of minimizing omitted variables bias in the regressions, and compliance with the key identification assumption of ignorability-oftreatment assumption. The household variables the researcher controls for in the ܿ‫ݏ݈݋ݎݐ݊݋‬௜ vector include a dummy (endogenous variable) for whether the household has participated in credit, age of the household head, age squared, sex of household head, number of family members, total owned land16, a dummy for whether the household head is literate, number of oxen possessed prior

 15 Millimet’s estimator (with missing data) has an option of reweighing the individuals with complete data to more closely approximate the distribution in all subjects included in the sample. 16 Note the difference between total owned land and total cultivable land. Inclusion of the latter as a control induces the violation identification assumption of ignorability-of-treatment assumption.

188

Chapter Seven

to participation in credit17, a dummy variable for access to irrigation, a dummy variable for participation in safety net, a dummy variable for experience of shock, distance18 (here the researcher controlled distance to road, transport centre, health centre and market centre), survey dummy and village dummy19. In addition, in this study, two outcome variables are compared: percapitacons-annual consumption expenditure per capita and _zlenlength/height-for-age z-score. Both outcome variables are alternately used as a dependent variable in the regression equations that identify the impact of microcredit in Tigray (that answered the first research question) and only _zlen is used to answer the last two research questions20.

4.4 Discussion Before leaving the section on description of variables used in the study, it is worth making some final comments on this way of looking at the variables. The first point is about the inclusion of village dummies as regressors in the study. This is done based on the recommendation from public health literature that failure to control geographic variation (geographically dispersed program) has the potential to substantially bias the effect of a program on the severity of stunting (Morris et al., 2000). Besides, the researcher tested the overall significance of the village dummies (see appendix II) and found that the village dummies are jointly significant at 5% level. Another point to mention here is that this research deviates from earlier research on the same topic, especially over the inclusion of some covariates (such as cultivated land size and number of oxen possessed by the household) that are affected by the treatment. Wooldridge (2005) showed that the inclusion of covariates that are themselves affected by the treatment violates the ignorability-of-treatment assumption (assumed in

 17 Controlling the total number of oxen without sufficiently separating the numbers before and after taking credit is likely to violate the ignorability-of-treatment assumption. 18 Distance is measured in terms of walking distance in minutes 19 The surveys are conducted in 16 villages; hence the researcher controlled 15 village dummies 20 The reason is that income measures are likely to subject to non-classical measurement error.

Microfinance

189

estimating treatment effects and that can be violated when certain factors, such as the above, are included among the covariates). Researchers such as Berhane Tesfay (2009), Zewde (2009) and Hailai (2010) controlled total operated land size (as against total owned land) and number of oxen possessed by the household (without sufficiently separating the numbers before and after taking credit) in their study of the impact of credit in Tigray. However, controlling variables (such as those in the program credit type of treatments, especially when applying propensity-score methods) that are themselves affected by the treatment, violate the ignorability-of-treatment assumption: conditional on observed covariates, treatment indicators, say T, and the counterfactual outcomes Yi are independent. A weaker version of this assumption is conditional mean independence (Wooldridge, 2005). Access to credit (increased liquidity) is likely to affect households’ propensity to rent-in land. In addition, households may use the credit to buy agricultural inputs such as oxen. Renting in land and buying of oxen in turn affect quantity of agricultural investment and then affect the outcome variable, volume of agricultural production and income/consumption at the household level. This simple linkage provides a simple demonstration of how access to credit affects households propensity to rent-in land and buy oxen, which are likely to raise agricultural production and hence income (which is a potential outcome variable). To substantiate the claim of the researcher using empirics, the following evidences can be mentioned. Boucher discovered some increase in land market activity, as a result of the access of the rural poor to credit (Boucher et al., 2005). In addition, access to credit is an important determinant for buying oxen, and the introduction of an ox-traction technology as the initial capital requirements to own theses are substantial (Hesse and Runge-Metzger, 1999). From this evidence, it is likely that inclusion of these controls in the program credit impact evaluation study generally violates the key ignorability assumption. Furthermore, the two surveys that give rise to the data set used in this study contain questions to households concerning the purpose that households obtain credit from DECSI. Accordingly, out of the total sample units considered in the study, 76 households answered that they used the money to rent in land for agricultural purposes. Moreover, 55 households

Chapter Seven

190

responded that they used the money to buy oxen. This clearly shows that including these controls violates the key ignorability assumption, as these controls themselves are affected by the treatment (access to credit)21. Hence, in this study, unlike in previous studies in Tigray on the same topic, variables such as cultivated land size and ownership of oxen are not included as an argument in estimating treatment effects, both ATE and ATT, as its inclusion is likely in violation of the key ignorability-oftreatment assumption.

4.5 Quantitative Identification Methods and Estimation Specification The following section details the econometric methodology of the study: Millimet’s Estimator, Heckman bivariate normal selection model (BVN) and the Klein and Vella (KV) estimator that are used to attain the objective of this research.

4.6 Program Evaluation Specification and Heckman bivariate normal selection (BVN) estimator Below is a simple treatment specification that distinguishes between households in the control group and households in the treatment group. This specification measures the overall impact of uptake of microcredit, comparing households in the control group to clients of DECSI in the treatment group. Equation (1) shows the regression equation: ܻ௜ ൌ Į ൅ ȕ ‫ –‡–ƒ‡”– כ‬ௗ௨௠௠௬೔ ൅ ‫ݏ݈݋ݎݐ݊݋ܿ כ ׎‬௜ ൅ ‫ݑ‬௜ ............................. (1) for i= 1, 2,…N The outcome (ܻ௜ ) measures such as consumption, income, or malnutrition status of individual i are regressed on a treatment dummy that takes the value of one if a household belongs to the treatment group, i.e., client of

 21 In addition, Pearson product-moment correlation test between the two controls and treatment dummy is included in appendix three. The test t-statistics are not large enough to reject the null hypothesis that there is no correlation between access to credit and amount of cultivated land. The test t-statistics is also not large enough to reject the null hypothesis that there is no correlation between access to credit and number of oxen possessed by the household.

Microfinance

191

DECSI, and a value of zero otherwise. Further, a set of variables controlsi account for observable characteristics. A detailed description of the controls included is given in part five of the paper. In order to estimate the effect of the treatmentdummyi, as well as test for robustness, the researcher proposed the following set of estimators or specifications as different specifications add robustness to the findings. The first estimator that is used in this research to estimate the treatment effect is the standard Heckman bivariate normal selection (BVN) estimator,22 which provides a consistent estimate of the treatment effect. Heckman made the following assumption in order to identify the BVN: potential outcomes and latent treatment assignment are additively separable in observables and unobservables (Millimet and Tchernis, 2009). Following this, BVN estimator is identified without an instrument from the nonlinearity of inverse mills ratio. Millimet (2010) showed in his simulation that BVN is identified without an instrument because the inverse mills ratio is nonlinear. In short, Heckman's BVN selection model adopts a two-stage approach. First, it estimates the probability of treatment ĭሺܺ௜ ߛሻ using a standard probit model with binary treatment as the dependent variable. Second, it estimates via OLS the following second-stage outcome equation: ܻ௜ ൌ ܺ௜ ߚ଴ ൅ ܺ௜ ܶ௜ ሺ‫ܤ‬ଵ െ ‫ܤ‬଴ ሻ ൅ ‫ܤ‬ఒ଴ ሺͳ െ ܶ௜ ሻ ቂ

థሺ௑೔ ఊሻ

ቃ ൅ ߚఒଵ ܶ௜ ቂ

ଵି஍ሺ௑೔ ఊሻ

ିథሺ௑೔ ఊሻ ஍ሺ௑೔ ఊሻ

ቃ ൅ ߤ௜ .................... (2)

Where ߶ሺǤ ሻȀĭሺǤ ሻ is the inverse Mills' ratio, and ߤ௜ is a well-behaved error term. With this approach, the estimated ATE is given by: ߬Ƹ ‫ܸܰܤ‬ǡ ‫ ܧܶܣ‬ൌ ܺതሺߚመଵ െ ߚመ଴ ሻ ........................................................................ (3) Similar expressions are available for the ATU and ATT.23

 22

This is Heckman’s self selection estimator However, if the assumptions of the BVN model do not hold, or if the BVN model is poorly identified, then a more robust specification is called for that perhaps performs better than BVN in practice. 23

192

Chapter Seven

4.7 Millimet’s Estimator The program evaluation literature has expanded rapidly over the past decade. Recently, social welfare programs and their impact have come under much scrutiny by researchers and practitioners. The knowledge concerning methods that are designed to provide consistent estimates of some measure of the causal effect of a binary treatment under conditional independence, as well as typical Instrumental Variable (IV) methods imposing an exclusion restriction, is relatively well developed. However, researchers are less informed about how to proceed when conditional independence fails and the usual type of exclusion restriction is unavailable. Moreover, the lack of experimental evidence, combined with non-random selection into these programs, makes identification of the causal effects of such programs difficult (Millimet and Tchernis, 2009). Heckman’s selection model is only identified as the sample size tends to infinity; hence it produces biased estimates if no valid exclusion restrictions are available, as shown by Sartori (2003). In addition, the Heckman procedure does not allow for estimating impacts when essential heterogeneity is present. Essential heterogeneity is likely to be present whenever there are negative or positive individual specific gains from treatment, and individuals select into treatment based on these gains (Millimet and Tchernis, 2009). Millimet argued that the assumption of essential heterogeneity is more likely to be present in an applied setting, and he showed that the bias resulted in estimating ATE, ATT and ATU in the presence of selection on unobservables and essential heterogeneity. In cases like the credit program, this difficulty is exacerbated by the apparent lack of exclusion restrictions. Millimet compared, in his estimation of treatment effects without an exclusion restriction: with an ‘Application to the Analysis of the School Breakfast Program via Monte Carlo’ study finding several existing estimators that do rely on exclusion restrictions for identification, and came up with a new estimator called bias corrected inverse-probability of weighting estimator (IPW) or simply called biased–corrected estimator (ibid). Millimet essentially illustrates the usefulness of his new estimator when analysing the causal effects of binary treatments. By applying his new estimator, Millimet found consistent evidence of causal effects in his application to a US school breakfast program. In general, Millimet proposed that his new estimation approach could be used when unconfoundedness is not likely to hold, and essential heterogeneity is likely to exist, but one lacks a valid exclusion restriction.

Microfinance

193

Armendáriz and Morduch, as quoted in Berhane Tesfay (2009), noted that borrowers select themselves into the program by joining a group of their choice on which the MFI decides to grant a loan based on its own criteria, and in some cases borrowers self-select into MFI based on observed and unobserved ‘initial’ household characteristics. According to Czura (2010), the inability of poor households to cope with negative income shocks and their vulnerability to risks differ across MFI client households. Major shocks that client households faced were drought and flooding, death of a household member, loss of employment, and theft. If households select into credit treatment as a reaction to these adverse household-specific shocks, then we have (negative) selection on unobservables, violating the conditional independence assumption (ibid). Furthermore, these unobservables are time varying, violating the assumptions of the fixed effects and first difference estimators. In light of the above discussion, in this study, the researcher uses the Millimet estimator to consistently estimate whether DECSI supports poor households in reducing their poverty and malnutrition status. The researcher’s logic behind using Millimet’s estimator to identify the impact of credit is twofold: the absence of a valid exclusion restriction for the credit dummy in the microcredit literature, and the notion that both selection on unobservables and essential heterogeneity are likely to be present among the farm households operating in Tigray. As noted above, Millimet’s estimator estimates the impact of a binary treatment when one lacks a valid exclusion restriction, and it allows us to estimate impacts in the presence of essential heterogeneity when conditional independence assumption fails, which is not possible using the Heckman estimator. Millimet´s (2010) estimator is based on the normalized inverse probability weighted estimator of Hirano and Imbens (2001). Millimet’s estimator corrects the bias of Hirano-Imbens using the BVN estimator to help estimate the bias. He derived an expression for the bias under selection on unobservable and essential heterogeneity, and minimized this bias for his first estimator (minimum biased estimator (MB)) in order to identify average treatment effect (ATE). To understand the issues involved in identification and estimation of the impact of DECSI microfinance in supporting the poor households in reducing their poverty and child malnutrition status, consider the following model:

194

Chapter Seven

ܻ௜ ൌ Į ൅ ȕ–”‡ƒ–‡– ௗ௨௠௠௬೔ ൅ ‫ݏ݈݋ݎݐ݊݋ܿ׎‬௜ ൅ ‫ݑ‬௜ ............................. (4) –”‡ƒ–‡– ௗ௨௠௠௬೔ ൌ Į଴ ൅ ୧ƍ ߛ ൅ ߝ௜ ..................................................... (5) Where (4) is the outcome equation, and (5) is the selection equation. The outcome equation shows the consumption, income, or malnutrition status of individual i as a function of a treatment dummy that takes the value of one if a household belongs to the treatment group, i.e., client of DECSI, and a value of zero otherwise, a set of variables controlsi account for observable characteristics such as household and village characteristics, and an error term‫ݑ‬௜ . The selection equation captures the two levels of selection processes: the self-selection by a household and the screening of the potential borrowers by DECSI. The endogeneity arises from the fact that there are individual, household, and village level unobserved factors that may affect both the outcome and the selection equations, and thus the correlation between the error terms is not zero, i.e.,ߩ ൌ ‫ݒ݋ܥ‬ሺ‫ݑ‬௜ ǡ ߝ௜ ሻ ് Ͳ. The most obvious individual level common unobserved heterogeneity in the context of microfinance is ability (or entrepreneurial capability). An individual with higher ability would expect higher net return from participation in the credit program, and thus would self-select into the program. But higher ability may also mean that the household will have better economic outcomes in the absence of credit availability. This creates a positive selection effect, and ߩ ൐ Ͳ on this account. However, it is also possible that the outside option for a high ability entrepreneur is much better (shadow price of time is higher), and thus the household might not be interested in high interest rate micro loans with its web of restrictions (such as group liability, and substantial time commitments for regular group meetings). If this is the case, DECSI would attract relatively low ability micro-entrepreneurs, and thus the selection would be negative implyingߩ ൏ Ͳ. There is some econometric evidence that, in fact, the selection into MFI programs in Bangladesh is negative, and thus the OLS and Probit would yield estimates that are biased downward (Schroeder, 2009, and Pitt and Khandker, 1998). The correlation between ‫ݑ‬௜ ƒ†ߝ௜ can also arise because of nonrandom placement of MFI programs across different villages. For example, to ensure high repayment rates, the MFIs have incentives to select villages with a concentration of moderate poor, and eschew the most vulnerable villages with a concentration of ultra-poor. The MFI can also use

Microfinance

195

information about a village’s economic potential to pick better-endowed villages. This would result in a positive correlation between the error terms in equations (4) and (5) above, implyingߩ ൐ Ͳ. But if the MFIs are true to their objective of poverty alleviation, then they will target relatively unfavourable villages, thus makingߩ ൏ Ͳ. The available evidence shows that the objective function of microfinance NGOs is a convex combination of poverty objective and repayment objective (Salim, 2009). In the absence of any exclusion restrictions, matching methods can be used to estimate causal effects. However, the matching methods rely on the assumption that there is no significant selection on unobservables conditional on matching observable characteristics. This is a strong assumption that is unlikely to hold in many applications, including ours. Recent advances have developed ways to minimize and correct for the bias that results from the failure of the conditional independence assumption (CIA). The researchers use the ‘Bias-Corrected Minimum Bias’ (MB-BC) estimator recently proposed by Millimet and Tchernis (2010), which provides a way to correct for the bias. The MB estimator builds on the normalized Inverse Probability Weighting (IPW) estimator of Hirano and Imbens (2001)24; it modifies the IPW by restricting observations to those with a propensity score in a “neighbourhood” around the Bias Minimizing Propensity Score (BMPS). Note that the normalized IPW relies on the CIA, and thus, estimates are biased when CIA is violated. The BMPS is the propensity score that minimizes this bias, which yields the minimum bias (MB) estimator. The ‘minimum-bias bias-corrected’ (MB-BC) estimator subtracts the estimated bias from the minimum bias estimate. Millimet and Tchernis define the neighbourhood around the BMPS as observations with a propensity score within a radiusߠ such that at leastߠ percent of both treatment and control groups are included (trimming observations with a BMPS below 0.02 and above 0.98). They set this radius at 0.05 and 0.25. The Monte Carlo simulations reported in Millimet and Tchernis (2010) show that an MB-BC estimator with a large radius (0.25) performs very well in terms of correcting for the endogeneity bias even when there is significant selection on unobservables. (Millimet and Tchernis (2010)

 24

Millimet’s estimator can be called an approach that uses BVN to correct the bias in the Hirano-Imbens estimator that requires unconfoundedness.

Chapter Seven

196

4.8 Klein & Vella (2009) Estimator For comparison to the previous estimators, the research also employs the estimator of Klein and Vella (2009), shortly called KV. Millimet’s parametric implementation of this estimator relies on a similar functional form assumption to the BVN estimator in the absence of heteroscedasticity, but effectively induces a valid exclusion restriction in the presence of heteroscedasticity (Millimet and Tchernis, 2009). To identify this estimator, Millimet develop a latent treatment assignment, now given by: ܶ ‫ כ‬ൌ ܺߛ െ ‫כݑ‬ where ‫ = כݑ‬S(X)u and u is drawn from a standard normal density. In this case, the probability of receiving the treatment conditional on X is given by: ܲሺܶ ൌ ͳȁܺ ൌ ĭሺ

௑

ௌሺ௑ሻ

ߛሻ. .............................................................. (6)

Assuming S(X) = exp(Xߜ), the parameters of (12) can be estimated by maximum likelihood (ML), with the log-likelihood function given by: ݈ࣦ݊ ൌ σ௜ሾ݈݊ĭሺ

௑ఊ

௘௫௣௑ఋ

ሻሿ்೔ ቄ݈݊ሾͳ െ ĭሺ

௑ఊ

௘௫௣ሺ௑ఋሻ

ଵି்೔

ሻሿቅ

.................................. (7)

where the element of ߜcorresponding to the intercept is normalized to zero for identification. The ML estimates are then used to obtain the predicted probability of ෣ , which may be used as an instrument for T in equation treatment, ܲሺܺሻ ෣, (3), excluding the selection correction terms. Note, even if S(X) = 1,ܲሺܺሻ remains a valid instrument since it is non-linear in X. However, since the non-linearity arises mostly in the tails, identification typically relies on a small fraction of the sample. On the other hand, if ܵሺܺሻ ് ͳ, then the KV approach effectively induces a valid exclusion restriction as ܼ ‫ܺ ؠ‬ൗܵሺܺሻ is frequently linearly independent of X (Klein and Vella (2009) as quoted in Millimet and Tchernis, 2009). ෣ ) that can be In short, the KV estimator uses constructed instruments (ሺሻ used as an instrument for the endogenous treatment dummy, T to identify the treatment effect.

Microfinance

197

5. Analysis, Key Findings and Relevant Discussions 5.1 Summary Statistics of Basic Household Information The following section presents summary statistics of basic household information such as age of the household head, annual consumption expenditure per capita, operated land size in hectares owned by the household, oxen position by the household, distance to major infrastructural facilities (such as road and health centres), and child nutritional status at the household level in both 2006 and 2010. Table 1 and 2 (see appendix I) show the pattern of poverty measured with non-income (_zlen- length/height-for-age z-score) and income (in this case consumption expenditure per capita) indicators in 2006 and 2010 between participants and non-participants of DECSI. Depending on which measures are used (nutritional status of under six versus annual consumption expenditure per capita) the researcher comes to quite different conclusions about household poverty in Tigray and program credit participation status, particularly in 2006. In 2006, the non-income (steady indicator) poverty indicator shows that on average children are more severely malnourished in households participating in program credit (z-score of -2.99) as compared to households not participating (-2.22). This result can be interpreted as the microcredit in Tigray is at least targeting amongst the poorest as against to the moderate poor. Note that the classification of the moderate poor and the poorest of the poor is based on the sheer size of z-score (the mean values of z-score), the interquantile range of z-score, and the quantile measures. The classification of the poorest of the poor and the moderate poor using the mean value of z-score and the interquantile range is straightforward. The quantile gives the measure of the centre of the distribution of the values of a variable. The lower and upper quantiles (0.25 and 0.75, respectively) and the middle value of the ordered data (Mdn), as a measure of variation, show (see table 1) that the participant households have lower values of z-score (of all quantile measures) as compared to non-participant households. This justifies the targeting by DECSI. However, this targeting of the poorest of the poor does not guarantee that the poorest of the poor are appreciating the benefits of being treated. The issue of identification as to whether the poorest of the poor are benefiting from participation in the credit program is the subject matter of the later econometric analysis part of this paper.

198

Chapter Seven

In 2010, one can see that the nutritional status of average under-six children showed improvements (from an average height-for-age z-score of -2.99 in 2006 to an average height-for-age z-score of -1.21 in 2010 for households participating in program credit). In addition, the wellbeing of children also seems to be much better in 2010 than in 2006 for households who did not take part in program credit. To rigorously deal with these findings, one can take a step ahead to statistically identify the impacts and test the validity of the estimates. Furthermore, in order to get insights into the relationship between the nutritional status of under-six children (height-for-age z-score) and annual consumption expenditure per capita, the researcher presented the relationships using the following graphs. Figure 2. The relationship between annual consumption expenditure per capita and height-for-age z-score for both 2006 and 2010 data

As can be seen in the figure above, the relationship between income and non-income measures of poverty is not as such clear for the entire set of participants and nonparticipants for the two survey rounds. To further explore the two relationships, especially in the presence of program credit, the researcher presented the following graph that shows the relationship between the two measures for non-participants and participants in the presence of program credit as follows:

Microfinance

199

Figure 3. Pattern of relationship between total consumption expenditure and height-for-age z-score for non-participants and participants

200

Chapter Seven

Using the information presented in panel A and panel B, one can understand that observations are more spread out in panel B (with program credit) than in panel A. To explain this more clearly, a Spearman and Kendall correlations test (see appendix III) is conducted. The null hypothesis, Ho: _zlen and percapitacons are independent, is rejected at a 5% level in favour of the alternative hypothesis. The conclusion is that the nutritional status of under-six children is positively correlated (under treatment) with consumption expenditure per capita. However, for nonparticipants, the test shows that there is no correlation between the nutritional status of under-six children and their consumption expenditure per capita. From the above discussion, it is clear that the link between program credit and consumption expenditure per capita and nutritional status of under-six children is not clear. This necessitates the need for a more rigorous examination of the issue to discover the linkage between the credit program and consumption expenditure per capita and nutritional status of under-six children.

5.2 Econometric Estimation This part of the paper presents the econometric analysis part of the research. In this part, answers are provided to the fundamental research issues raised in part I. The flow of the presentation is organized according to the organization of the research questions. Accordingly, the presentation starts by answering the first research question, i.e. whether program credit in Tigray is helping rural poor households in reducing poverty. As already discussed in part IV, in order to provide a reliable and dependable answer to this question, the researcher employed a set of three estimators, and in this section the estimation results of the three estimators will be reported and the implications of each estimate discussed.

5.3 Estimation Results 5.3.3 Effect of Credit on Poverty Reduction (Prevalence of Stunting) Table 3 below shows the effect of program credit on the severity of malnutrition as measured by z-score (height-by-age measure of the malnutrition status of a child) using the aforementioned estimators.

Microfinance

201

Table 3. Effect of program credit participation on stunting reduction

Estimator KV-IV BVN MB-BC 0.05 MB-BC 0.25

Using z-score as an outcome variable ATE ATT 2.755 2.755 [ -7.407, 7.737] [ -7.407, 7.737] 2.373 0.774 [ -4.975, 6.051] [ -3.561, 3.405] 2.295 1.866 [ -4.698, 6.227] [ -6.113, 5.872] 2.389 0.916 [ -4.873, 5.809] [ -5.580, 4.976]

Notes: Treatment is defined as participation in program credit. The three estimators use the same set of covariates defined in the description of variables section. 90% empirical confidence (that means at a level of 10%) intervals in brackets are obtained using 250 bootstrap repetitions. KV = Klein and Vella estimator; BVN = Heckman bivariate normal selection model; MB-BC = biascorrected estimator and the 0.0.5 or 0.25 under MB-BC gives at least 5 or 25 percent of both the treatment and control groups contained in the propensity score interval that minimizes the bias in estimating ATE and ATT.

Table 4. Effect of program credit participation on consumption expenditure per capita Using consumption expenditure per capita as an outcome variable Estimator ATE ATT KV-IV 2946.141 2946.141 [1.9e+03, 7690.062] [1.9e+03, 7690.062] BVN 2681.906 1612.310 [849.903, 3411.771] [408.093, 2140.605] MB-BC 0.05 2650.867 2458.962 [821.550, 3452.521] [1.0e+03, 3458.538] MB-BC 0.25 2636.139 2585.279 [837.640, 3306.803] [751.770, 3345.211] Notes: The three estimators use the same set of covariates defined in the description of variables section.

From the above estimation result, the estimators give a positive impact of program credit in improving the height-for-age z-score (table 3) and in increasing consumption expenditure per capita (table 4) in Tigray. Because the BVN estimator does not allow for essential heterogeneity (which does

202

Chapter Seven

hold in reality), especially when applied to sample size, and for the KV it is harder to explain the economic meaning behind the identification (Jurajda, 2007), the researcher opted to use the results of Millimet’s estimators for an interpretation25. In addition, the KV estimator has the restrictive assumption that it identifies impacts based on the variance of heteroscedasticity of the error term of the treated,ȝଵ , and the untreated,ȝ଴ . Because of this identifying assumption, if the unobservables of the treated and non-treated are equal, the ATT and ATE parameters are all the same, as in the table above.26 The ATT above can be interpreted as a 1.866 z-score improvement in height-for-age resulting from the program credit (DECSI) in Tigray, when 5% of the treatment and control groups are contained in the propensity score interval that minimizes the bias). When 25% of the treatment and control groups are contained in the propensity score interval that minimizes the bias, participating in program credit improves the heightfor-age z-score by 0.916 z-score unit. Using the same logic, one can interpret the effect of access to credit in increasing annual consumption expenditure per capita. Accordingly, access to credit increases annual consumption expenditure per capita on average by 2458.96 Birr for the participants as compared to those who did not participate in program credit in Tigray. On the other hand, the ATE measures the average causal difference in outcomes (height-for-age z-score and annual consumption expenditure per capita) under the treatment and under the control. As can be seen from

 25

Because Millimet’s bias–corrected estimator is identified without exclusion restriction and allows essential heterogeneity in identification. 26 Allowing the random draw to differ in treated and untreated states is critical to allowing unobserved heterogeneity in how people respond to treatment. There is, however, a special case where the parameters may be equal even if ¿ଵ =¿଴ , that is, when ‫ܧ‬ሺ¿ଵ െ ¿଴ ȁ‫݈݋ݎݐ݊݋ܥ‬௜ ǡ ܶሻ ൌ ͲǤ Under this restriction, T is uninformative on ¿ଵ െ ¿଴ , but it is not necessarily the case that ¿ଵ ൌ ¿଴ . The conditional mean restriction might be satisfied if agents making the participation decisions (e.g. households) do not act on ¿ଵ ൌ ¿଴ in making the decision, perhaps because they do not know anything about their own indiosyncractic gain from participating in the program at the time of deciding whether to participate. In this special case, there is ex-post heterogeneity in how people respond to treatment, but it is not acted upon ex ante (Todd, 2007).

Microfinance

203

table 3 and table 4, ATE is positive and significant at 10% level27. This analysis demonstrates that a credit DECSI achieved gains of 1.866 zscores of height-for-age to its beneficiaries in Tigray (in terms of reducing the prevalence and severity of malnutrition (stunting)) and an average increase of annual consumption expenditure per capita by 2458.96 Birr. However, this estimate is unstable due to the wide confidence interval of the estimate, and caution should be given while interpreting the findings. This is the major limitation of this paper28.

6. Conclusion In this paper, the researcher investigated the effect of microfinance on child malnutrition and annual consumption expenditure per capita at the household level, using three best available impact assessment identification strategies. Two of the three estimators, KV and Millimet’s minimum biased estimator, are new estimation strategies of the causal effects that are designed to provide consistent estimates of the causal effect of a binary treatment when conditional independence fails. In this study, the researcher aims to reduce non-classical measurement errors that dramatically bias coefficients and are likely to arise when income or expenditure are used as outcome variables by proposing a height-for-age z-score as an outcome variable, which is less likely to be subject to measurement error. Furthermore, the researcher tried to touch the effect of microfinance credit in reducing poverty using child malnutrition outcomes, as it is obvious that poverty measures go beyond the standard welfare measure of consumption and income, which are likely subject to non-classical measurement error. In the preceding analysis of the impact of program credit, the three estimators offer a coherent picture of the causal effect of the program.

 27 The stata code for this estimator is designed at 10% level and the program does not report the level of significance. 28 Other limitations of the paper are the following: Some well-nourished children might be wrongly classified as undernourished because they have genetically short parents, while others might be misclassified as well nourished even though they are undernourished, but this does not show up in their height due to genetically tall parents. Hence, genetic variability among the families is another limitation of this study. On top of these, small sample size is another limitation of the paper.

204

Chapter Seven

Specifically, the researcher finds a positive and statistically significant association between program credit and child nutritional improvements when using estimators that require conditional independence, mainly BVN. The association remains positive, but becomes smaller when the minimum-biased estimator, which does not require conditional independence, is used. This conclusion follows using child malnutrition and annual per capita consumption expenditure outcome variables. However, the empirical results in this study suggest evidence is not strong that microfinance helps households improve child nutrition and annual per capita consumption expenditure. Hence, microfinance may not have any robust effect on improving child nutrition and annual per capita consumption expenditure due to large standard errors and huge confidence intervals of estimates presented in table columns (3) to (6). Finally, future analysis into the impact of participation in microfinance on other outcome variables, such as indicators of poverty and acute malnutrition of children under six (wasting and underweight) as well as adults body mass index indicators, will provide policymakers with better information about the potential benefits of program credit.

References Adjei, J. K., Arun, T. & Hossain, F. 2009. The Role of Microfinance in Asset-Building and Poverty Reduction: The Case of Sinapi Aba Trust of Ghana. Brooks World Poverty Institute Working Paper Series. Alkire, S. & Santos, M. E. 2010. Acute multidimensional poverty: a new index for developing countries, University of Oxford, Poverty and Human Development Initiative. Araujo, M. C., Ferreira, F. H. G., Lanjouw, P. & ÷Zler, B. 2008. Local inequality and project choice: Theory and evidence from Ecuador. Journal of Public Economics, 92, 1022-1046. Behrman, J. R. & Deolalikar, A. B. 1987. Will developing country nutrition improve with income? A case study for rural South India. The Journal of Political Economy, 95, 492-507. Bennett, D. 2009. Billions of dollars and a Nobel Prize later, it looks like Ñmicrolending" doesn" t actually do much to fight poverty. Boston Globe, 20. Berhane Tesfay, G. 2009. Econometric analyses of microfinance credit group formation, contractual risks and welfare impacts in Northern Ethiopia. Proefschrift Wageningen, s.n.].

Microfinance

205

Bloss, E., Wainaina, F. & Bailey, R. C. 2004. Prevalence and predictors of underweight, stunting, and wasting among children aged 5 and under in western Kenya. Journal of tropical pediatrics, 50, 260. Borchgrevink, Axel, Jo Helle-Valle & Woldehanna, T. (2003). Credible Credit Impact Study of the Dedebit Credit and Savings Institution (DECSI)”, Tigray, Ehtiopia. Norwegian Institute of International Affair. Boucher, S. R., Barham, B. L. & Carter, M. R. 2005. The Impact of. World Development, 33, 107-128. Brau, J. C. & Woller, G. M. 2004. Microfinance: A comprehensive review of the existing literature. Journal of Entrepreneurial Finance and Business Ventures, 9, 1-26. Cgap, W. 2010. The Challenge for Islamic Microfinance [Online]. http://www.cgap.org/. 22 March, 2011]. Collins, D. & Morduch, J. 2010. Reimagining the Unbanked. Cotler, P. & Woodruff, C. 2008. The Impact of Short-Term Credit on Microenterprises: Evidence from the Fincomun-Bimbo Program in Mexico. Economic Development and Cultural Change, 56, 829-849. Czura, K. November 2010. Impact Assessment of Microfinance in Sri Lanka: A Household Survey of Microfinance Clients in 5 Selected ProMiS Partner Microfinance Institutions. Deutsche Gesellschaft für Technische Zusammenarbeit (GTZ) GmbH German Technical Cooperation. Sri Lanka: Promotion of the Microfinance Sector (ProMiS). De Aghion, B. A., Armend·Riz, B. & Morduch, J. 2007. The economics of microfinance, The MIT Press. Economist 2009. Microcredit may not work wonders but it does help the entrepreneurial poor. The Economist. Elahi, K. Q. I. & Rahman, M. L. 2006. Micro-credit and micro-finance: functional and conceptual differences. Development in Practice, 476483. Foster, J., Greer, J. & Thorbecke, E. 1984. A class of decomposable poverty measures. Econometrica: Journal of the Econometric Society, 761-766. Gugerty, M. K. & Kremer, M. 2008. Outside funding and the dynamics of participation in community associations. American Journal of Political Science, 52, 585-602. Gunther, I. & Klasen, S. 2009. Measuring chronic non-income poverty. Poverty dynamics: interdisciplinary perspectives, 77.

206

Chapter Seven

Haddad, L., Alderman, H., Appleton, S., Song, L. & Yohannes, Y. 2003. Reducing child malnutrition: How far does income growth take us? The World Bank Economic Review, 17, 107. Hagos, F. 2003. Poverty, institutions, peasant behavior and conservation investment in northern Ethiopia. PhD Thesis, Agricultural University of Norway. Hagos, F., Holden, S. & Pender, J. 2006. The Effect of Program Credit on participation in off-farm employment and walfare of rural households in Northern Ethiopia: Agricultural University of Norway. Hailai, A. 2010. Can Microfinance Help to Reduce Poverty? Master of Science degree in Economics, Mekelle University. Hedaya, H. 2009. Perhaps microfinance isn’t such a big deal after all. Financial Times Helms, B. 2010. Microfinancing changes lives around the world – measurably. . The Seattle Times 7 April. Hesse, J. & Runge-Metzger, A. Ox traction in a long-term perspective: policy implications of a socio-economic study in Ghana. 1999. Hirano, K. & Imbens, G. W. 2001. Estimation of causal effects using propensity score weighting: An application to data on right heart catheterization. Health Services and Outcomes Research Methodology, 2, 259-278. Hulme, D., Hanlon, J. & Barrientos, A. 2010. Just Give Money to the Poor: The Development Revolution from the Global South, Kumarian Pr. Jurajda, Ä. 2007. Lecture Notes on Identification Strategies. Kabeer, N. 2003. Wider Social Impacts: Assessing the 'Wider' Social Impacts of Microfinance Services: Concepts, Methods, Findings. IDS Bulletin, 34, 106-114. Karlan, D., Goldberg, N. & Copestake, J. 2009. 'Randomized control trials are the best way to measure impact of microfinance programmes and improve microfinance product designs.'. Enterprise Development and Microfinance, 20, 167-176. Karlan, D. & Zinman, J. 2009. Expanding credit access: Using randomized supply decisions to estimate the impacts. Review of Financial Studies. Khandker, S., Koolwal, B. & Samad, H. 2009. Handbook on Impact Evaluation. Handbook on Impact Evaluation, 1, 1-239. Klasen, S. 2008. Poverty, undernutrition, and child mortality: Some interregional puzzles and their implicationsfor research and policy. Journal of Economic Inequality, 6, 89-115.

Microfinance

207

Klein, R. & Vella, F. 2009. A semiparametric model for binary response and continuous outcomes under index heteroscedasticity. Journal of Applied Econometrics, 24, 735-762. Klugman, J., Rodríguez, F. & Choi, H.-J. 2010. Human Development Report: New Controversies, Old Critiques. Human Development Report UNDP. Macfarquhar, N. 2010. Banks making big profits from tiny loans. New York Times Makina, D. & Malobola, L. 2004. Impact assessment of microfinance programmes, including lessons from Khula Enterprise Finance. Development Southern Africa, 21, 799-814. Mansuri, G. & Rao, V. 2004. Community-based and-driven development: A critical review. The World Bank Research Observer, 19, 1. Matin, I., Hulme, D. & Rutherford, S. 1999. Financial services for the poor and poorest Deepening understanding to improve provision. Millimet, D. & Tchernis, R. 2009. Estimation of Treatment Effects Without an Exclusion Restriction: with an Application to the Analysis of the School Breakfast Program. Millimet, D. L. 2010. The Elephant in the Corner: A Cautionary Tale about Measurement Error in Treatment Effects Models. Mofed, M. O. F. A. E. D. 2006. Ethiopia: building on progress, a Plan for Accelerated and Sustained Development to End Poverty (PASDEP). Morris, S. S., Flores, R. & Zniga, M. 2000. Geographic targeting of nutrition programs can substantially affect the severity of stunting in Honduras. The Journal of nutrition, 130, 2514. Platteau, J. P. 2004. Monitoring Elite Capture in Community Driven Development. Development and Change, 35, 223-246. Roodman, D. 2009 New Challenge to Studies Saying Microcredit Cuts Poverty. David Roodman's Microfinance Open Book Blog[Online]. [Accessed March 21st, 2011 at 23:36 March 21, 2011 ]. —. 2010. You can’t have it all. David Roodman's Microfinance Open Book Blog[Online]. Rutherford, S. & Actionaid 1996. A critical typology of financial services for the poor, Actionaid. Santen, R. M. V. 2010. Microfinance as a Poverty Reduction Policy. Sartori, A. E. 2003. An Estimator for Some Binary-Outcome Selection Models without Exclusion Restrictions”. Society for Political Methodology, 11, 111-138. Segele, Z. & Lamb, P. 2005. Characterization and variability of Kiremt rainy season over Ethiopia. Meteorology and Atmospheric Physics, 89, 153-180.

208

Chapter Seven

Simondon, K. B. 2010. Anthropometric indicators for impact evaluation of food security programmes. Institut de Recherche pour le Développement. Staschen, S. 1999. Regulation and Supervision of Microfinance Institutions: State of Knowledge. GTZ, Eschborn. Stewart R, Van Rooyen C, Dickson K, Majoro M & T, D. W. 2010. What is the impact of microfinance on poor people? London: Evidence for Policy and Practice Information and Co-ordinating Centre (EPPICentre), Social Science Research Unit at the Institute of Education, University of London. Todd, P. E. 2007. Evaluating social programs with endogenous program placement and selection of the treated. Handbook of development economics, 4, 3847-3894. Unicef 1998. The State of World’s Children: Focus on Nutrition. New York: UNICEF. Webb, P., Von Braun, J. & Yohannes, Y. 1992. Famine in Ethiopia: policy implications of coping failure at national and household levels, Intl Food Policy Research Inst. Westover, J. 2008. The record of microfinance: The effectiveness/ ineffectiveness of microfinance programs as a means of alleviating poverty. Electronic Journal of Sociology, 12. Wikipedia. 2010. DECSI (Dedebit Credit and Savings Institution) [Online]. Wikimedia Foundation, Inc.,. [Accessed 28 Sep 2010 2010 ]. Woldehanna, T. 2000. Economic analysis and policy implications of farm and off-farm employment : a case study in the Tigray region of Northern Ethiopia. PhD Thesis, Wageningen University. Woldehanna, T. A. O., A 2002. The development and constraints of micro and small-scale enterprises and the need for member-based rural financial intermediation in rural Tigray. Proceeding of the microfinance workshop held Mekelle: Mekelle university. Woldenhanna, T. & Oskam, A. 2001. Income diversification and entry barriers: evidence from the Tigray region of northern Ethiopia. Food Policy, 26, 351-365. Wooldridge, J. M. 2005. Violating Ignorability of Treatment by Controlling for Too Many Factors. Econometric Theory, 21, 10261028. Wright, G. A. N. 1999. Examining the impact of microfinance servicesincreasing income or reducing poverty? Small Enterprise Development, 10, 38-47.

62 62 62 62 62 62 62 62 62 62

hhhage hhsize oxen landsize distroad distranspo disthealth distmkt _zlen percapitacons1

1

48.94 6.73 1.16 1.08 75.89 72.74 62.13 142.34 -2.22 571.80

Mean 12.39 2.16 0.91 0.69 62.34 67.37 48.65 88.14 2.09 311.49

S.D. Min 28.00 2.00 0.00 0.12 5.00 5.00 2.00 15.00 -5.93 0.00

.25 39.00 5.00 0.00 0.44 30.00 25.00 30.00 60.00 -3.44 362.02

Percapitacons is measured in Ethiopian currency, Birr, where 1.00 USD = Br.16.6968



No. of Obs.

Variable

Case 1: Summary statistics of households that did not participate in program credit

Table 1. Household information by credit participation in 2006

Summary statistics of basic household information in 2006

Appendix I

Appendices

Microfinance

Quantiles Mdn 48.00 7.00 1.00 1.02 60.00 50.00 40.00 150.00 -2.15 483.24

.75 55.00 8.00 2.00 1.50 120.00 120.00 90.00 180.00 -0.78 717.77

Max 83.00 12.00 3.00 3.00 240.00 240.00 180.00 480.00 5.91 1420.49

209

Chapter Seven

40 40 40 40 40 40 40 40 40 40

No. of Obs.

Source: Survey data (2006-2010)

Hhhage hhsize Oxen landsize Distroad distranspo Disthealth distmkt _zlen percapitacons

Variable 49.00 6.62 0.93 1.15 66.25 86.50 60.50 179.90 -2.99 733.70

Mean 11.99 1.84 0.94 1.01 51.81 72.95 43.95 87.79 1.83 413.07

S.D. Min 29.00 3.00 0.00 0.09 5.00 2.00 5.00 20.00 -5.80 239.14

.25 40.50 5.00 0.00 0.48 30.00 27.50 30.00 135.00 -4.60 423.20

Case2: Summary statistics of households that participated in program credit

210

Quantiles Mdn 50.00 7.00 1.00 0.73 50.00 60.00 40.00 180.00 -3.08 622.25 .75 56.50 8.00 2.00 1.75 90.00 165.00 100.00 240.00 -1.72 845.04

Max 76.00 11.00 3.00 5.00 180.00 240.00 160.00 360.00 1.35 1959.98

No. of obs. 68 68 68 68 68 68 68 68 68 68

Source: Survey data (2006-2010)

hhhage hhsize oxen landsize distroad distranspo disthealth distmkt zlen percapitacons

Variable 46.28 6.38 1.25 1.27 45.01 52.94 48.68 162.79 -1.65 1700.84

Mean 11.60 1.90 1.00 1.15 41.92 48.90 43.88 86.17 2.07 978.46

S.D. Min 26.00 2.00 0.00 0.00 2.00 0.00 5.00 0.00 -5.62 230.3

.25 38.50 5.00 0.00 0.50 15.00 15.00 20.00 120.00 -3.21 1005.07

Case 3: Summary statistics of households who did not participate in program credit

Table 2. Household information by credit participation in 2010

Summary statistics of basic household information in 2010

Microfinance

Mdn 45.50 6.50 1.00 1.00 37.50 40.00 30.00 180.00 -2.18 1523.62

Quantiles .75 53 8 2 1.50 60 60 60 180 -0.36 2163.11

Max 81 12 4.00 5.75 210 180 180 360 3.74 6092

211

Chapter Seven

No. of Obs. 56 56 56 56 56 56 56 56 56 56

Source: Survey data (2006-2010)

Variable hhhage hhsize oxen landsize distroad distranspo disthealth distmkt zlen Percapitacons 48.45 7.18 1.18 1.11 49.80 70.62 46.70 155.89 -1.21 1700.89

Mean 12.89 1.44 1.47 0.65 43.52 53.62 31.38 80.34 2.55 722.26

S.D. Min 29.00 3.00 0.00 0.25 2.00 2.00 5.00 5.00 -5.53 528.11

.25 38.00 6.50 0.00 0.62 27.50 30.00 25.00 105.00 -2.85 1213.14

Case 4: Summary statistics of households who did participate in program credit

212

Mdn 46.50 7.00 1.00 0.98 40.00 60.00 40.00 180.00 -1.73 1574.20

Quantiles .75 54 8 2 1.50 60 105 60 180 0.24 2168.04

Max 87 10 10 2.50 240 240 120 420 5.17 3468

Microfinance

213

Appendix II Village Dummies significance test Variable OLS Heckman Estimation Samredummy -6.423 -1.103* Mahberedummy 0.178 0.731* Maialemdummy -6.423 0.255* Seretmdummy 0.761** -0.543 Kihenmdummy 1.156*** -0.32 Genfelmdummy -0.222 0.408* Embaasmenay -0.744* -0.393 Hagereselay -1.110* 1.845*** Adiselamduy 0.864** 1.006* Hadegtidummy 0.26 -0.752 Tsaedaamboy 0.371 -0.059 Adimenabrdy -0.655* 0.073 Maiadrashay 0.227 0.637 Mekonidummy 0.889*** 2.430**** Maikeyahtiy -0.257 -0.403 Prob > F 0.0001 R-squared 0.1836 Adj R-squared 0.1253 Prob > chi2 0.006 Number of Observations 226 226 *: Significant at 25%; **: significant at 10%; ***: significant at 5%; ****: significant at 1%.

Chapter Seven

214

Appendix III Spearman's rank correlation coefficients test Table I. Spearman's rank correlation coefficients test for participants

Statistics Value

t-statistics

2.1474

Prob > t 0.0328 Ho: _zlen and percapitacons are independent Table II. Spearman's rank correlation coefficients test for nonparticipants

Statistics Value Spearman's rho 0.01 Prob > t 0.88 Ho: _zlen and percapitacons are independent

Microfinance

215

Appendix IV Pearson product-moment correlation Table I. Pearson product-moment correlation test between cultivated land size and program credit

Np= 96 p= 0.42 Nq=130 q= 0.58 Coefficient t-value

P>t df

-0.0273 -0.4093 0.027 224 Np=number of observations for any credit=1, p= proportion of p Nq= number of observations for any credit=0, q= proportion of q The test is valid at 5 % level

 Table II. Pearson product-moment correlation test between cultivated land size and number of oxen possessed by the household

Np= 96 p= 0.42 Nq=130 q= 0.58 Coef t-value ǦǦ

P>t

df

-0.0605 -0.9071 0.045 224 Np = number of observations for any credit=1, p= proportion of p Nq= number of observations for any credit=0, q= proportion of q The test is valid at 5% level

 





CHAPTER EIGHT LOAN PROVISION BY MICROFINANCING INSTITUTIONS FOR POVERTY REDUCTION, AND ITS LINKAGES WITH LOCAL ECONOMIC DEVELOPMENT STRATEGIES IN ETHIOPIA: AN EMPIRICAL REVIEW MUHAMMEDAMIN HUSSEN

Abstract Microfinancing institutions’ (MFIs) loan provision to the poor is proving a key strategy for poverty alleviation. In line with this view, in Ethiopia, inadequate access to credit by the poor has been identified as one of the factors contributing to poverty. The present review of studies aims at assessing linkages between MFIs’ loans provision and local economic development (LED), and providing a basis for policy formulation at regional and national levels in regard to MFIs and LED. For attaining these objectives, studies from both Ethiopia and other countries have been reviewed. Empirical reviews also showed that these institutions provided opportunities for self-employment, improved women's security, autonomy, self-confidence, and status within the society and household, and helped in improving children’s education. Above all, as pro-poor programs, they targeted the most vulnerable groups in society, particularly women, who remain confined to households with little or almost no assets. In order to accomplish these objectives, great efforts have been made over the last two decades by expanding MFIs’ loan provision services to various groups of the people, specifically the poor, and tackling the problem of poverty reduction. Despite the increasing reliance on MFIs to reduce poverty in Ethiopia, very little work has been undertaken to examine the linkage of the microfinance expansion with the LED strategy of the particular region. The studies reviewed revealed that there has been duplication of business undertaken in various parts of the region. This is due to a lack of linkage

Loan Provision by Microfinancing Institutions

217

and synergy of the loan provision by MFIs to the LED strategy of a given local area in particular, and regions in general. The reviewed literature indicated that on account of decentralization of the development plan to optimize the potential of each region and mobilize resources, the local governments have been empowered to undertake social and economic development endeavour. It has also been found that the Growth and Transformation Plan (GTP) (2010/11-2014/15) and LED were closely aligned. The interconnection between GTP and LED existed directly through the micro and small-scale enterprises (MSE), cooperatives, and other associations. Since LED aims to create efficient and functioning local economies, as a consequence it has a direct alignment with growth and transformation efforts. Therefore, linking the microfinance loan provision to the local development priority appeared to be very critical for the sustainability of the MSE businesses to benefit from local available potential resources for the poverty alleviation program. Empirical evidence reflected that the current urban policy, the MSE strategy, and the regional development framework provided additional opportunities for the implementation of LED and creating synergy between MFIs & MSE in Ethiopia. In light of the above view, the current MFIs loan provisions and local development priority of various regions is a very important point of emphasis unseen in Ethiopia. Moreover, the findings of the study revealed that there is clearly an identified lack of synergy between MFIs, MSE, Cooperatives Agency, Local administrative apparatus and LED in various regions in Ethiopia. This necessitates the stakeholders to set policy that fills up the gap and creates strong linkage between MFIs loan provisions to LED priority of particular regions to assure the sustainability of MSE businesses, and enhances the contributions of MFIs for poverty alleviation in Ethiopia. All in all, the researcher recommends that both federal and regional government and other concerned stakeholders work towards digging deeper to find keys to success (NPOWG, 2006).

1. Introduction Ethiopia is one of the poorest countries in the world, with per capita GDP of only $357 (Plummer, 2012) and with 29.6% of its population starving below the national poverty line of US$0.6 per day (Geiger and Goh, 2012). Enrolment in primary schools has increased from 33% in 1991 to

218

Chapter Eight

95% in 2007; however, more than half of the population lives in a household where nobody has completed primary school1. Ethiopia’s GNI per capita in PPI terms is $971 which, although it is one of the fastest growing, falls below the average for sub-Saharan Africa. The income gap between the financially wealthy and the poor is also high, with the inequality adjusted HDI index standing at 0.247 as of 2010. It is also one of the most populous countries in Africa, with a population growth rate of 3% per annum. The infant and child mortality rates are 118/1000 and 173/1000 children, respectively2. Hence, as an intervention strategy, Ethiopia crafted and implemented a Growth and Transformation Plan (2011/12-2014/15), where the campaign against poverty reduction is central to the government’s development agenda. In this respect many policies, goals, and objectives are focused on targeting the most disadvantaged households. To realize the plan, microfinance (MF) is considered by the government to be one of the most important tools in fighting poverty, and thus, the expansion of micro and small scale enterprise and efforts to reach the poor and expanding selfemployment opportunities, especially for the youth and women, are the current policy direction. The objective of the MFIs in Ethiopia is basically poverty alleviation through the provision of sustainable financial services to the poor, who actually do not have access to the financial services of other formal financial institutions (Yigrem, 2010). The MF sector in Ethiopia is characterized by its rapid growth, an aggressive drive to achieve scale, a broad geographic coverage, a dominance of government backed MFIs, an emphasis on rural households, the promotion of both credit and savings products, a strong focus on sustainability, and by the fact that the sector is Ethiopian owned and driven3. The industry has a strong focus on loans to the very poor, as indicated by the relatively small loans when compared to neighbouring countries. The government is heavily involved in the sector, and has prioritized facilitating the loan access to the urban and rural poor. Most MFIs take this poverty-targeting mandate very seriously (MFTransparancy, 2011). In line with strong justification of targeting the poor, five top MFIs were established which are backed by regional governments with the view of cascading GTP plan

 1

Human Development Report, 2010. http://www.hdr.undp.org/en/reports/global/hdr2010/ 2 Human Development Report, 2010. http://www.hdr.undp.org/en/reports/global/hdr2010 3 SIDA (2003) Short Study on microfinance – Ethiopia

Loan Provision by Microfinancing Institutions

219

to regional and district levels, namely, ACSI (Amahara), DECSI (Dedebit), OCSICO (Oromia), ADCSI (Addis), and Omo. In aggregate they serve about 84.7% of the total clients of the sector, and account for more than 90% of the sector’s total assets (NBE quarterly report as at June 30, 2012) as cited in Sintayehu (2013). However, significant efforts should be made by the concerned stakeholders to sustain the services, reach out and target the pro-poor plan of the government; reducing poverty through lessening the financial constraints of the poor and enabling them to create job opportunities for themselves and other youths, with particular emphasis on local economic development priority of regions in Ethiopia. This requires aligning with regional GTP and local economic development (LED) strategies that are crafted with existing local conditions involving the concerned stakeholders of the regions. The LED champions is a development approach aimed at increasing local economic potential and sustainable employment through giving local governments the tools to devise locally tailored strategies in cooperation with local, regional, and international stakeholders and actors (Rodríguez, et al, 2009). It is the joint-efforts of all state and non-state development actors in urban and rural Ethiopia towards building the economic capacity and competitiveness of their respective urban and rural local government entities, so as to create decent jobs and improve the quality of life for their residents (Mathewos, 2006). Hence the approach is concerned about local mobilization of actors and resources, building a convergence of interest around the competitive advantages of localities and building the capacity for economic actors to take up economic opportunities and enable exploitation of the opportunities, created by new market conditions. But most local officials, until recently, do not recognize or give adequate attention to local economic development issues. Since most economic development constraints are found at the local level such as access to land, skills development, micro and small-scale enterprise mobilization, and loan provision and other infrastructure, the rationale for LED becomes imperative. Thus an all-inclusive, participatory LED with the effective leadership of local officials would be necessary (ibid). These have to be closely linked to microfinance loan provision for a poverty reduction program of the regions as pro-poor strategy to achieve the GTP target of the regions in particular and the country in general. In the Ethiopian scenario, even though various research has been conducted on performance, financial profitability, and operational viability income generation, agricultural productivity and on the micro finance outreach areas etc., which have been accessed, in terms of significance of

220

Chapter Eight

the topics and the dynamic nature of the problems on the areas, a lot of research gaps can be observed. More importantly, despite the increasing reliance on MF to reduce poverty in Ethiopia, there has been surprisingly little work undertaken to link the microfinance expansion with the LED strategy of the regional states in Ethiopia. Hence, to assess the synergy and performance of MFIs within the framework of cascaded regions’ and districts’ GTP in Ethiopia, this study tries to identify ways through which the contribution of MFIs in alleviating poverty can be enhanced, the lessons to be adopted for poor nations like Ethiopia to overcome the century long enemy, i.e., poverty, is to assess the link of microfinance expansion with LED strategies and how both serve as inputs to one another to harness the economic development of the country.

2. Objectives of the Study The general objective of this study is to critically ascertain how microfinancing can be used as an effective instrument or strategy to reduce poverty and foster the development of micro- and small-scale enterprises by linking with local development strategies and cascaded GTP of particular regions in Ethiopia, with the specific objective of: 1. Assessing the contributions of microfinance institutions in poverty reduction and its alignment with local development strategies; 2. Examining the interdependence of MFI loan provision with LED strategies and GTP in Ethiopia; 3. Examining the challenges, opportunities, and threats of micro and small-scale enterprises and the link towards LED strategies.

3. Materials and Methods To meet the objectives listed above, the researcher used the available pertinent documents, particularly works produced in recent years, which are extensively reviewed. The review includes: research reports, LED project reports, studies, reports, papers, published and unpublished documents of donors, international organizations, government agencies, researchers and conference proceedings; federal/regional level documents containing policy statements, agreements, and related instruments between donor organizations.

Loan Provision by Microfinancing Institutions

221

4. Theoretical Arguments on Microfinance and Local Economic Development Strategies In this part of the paper the researcher reviewed the theoretical perspectives of microfinance expansion as an instrument for poverty alleviation in line with local development potential and strategies of regions in developing countries including Ethiopia.

4.1 Definition and Evolution of Microfinance Institutions The emergence and evolution of microfinance in development discourse by policy makers and academics is credited back to Grameen Bank, established by Mohammed Yunus in Bangladesh in the late 1970s, which spread rapidly to other developing countries. Since then, MFIs and their loan provision services have mainly been viewed as an economic development strategy, and it is a particularly relevant approach in countries where disadvantaged groups tend not to benefit from involvement in the formal economy. With these rational justifications, for more than 30 years, microfinance has been portrayed as a key policy and programme intervention for poverty reduction and ‘bottom-up’ local economic and social development tools in most developing countries, including Ethiopia. The literature reviewed showed that academics, researchers, and policy makers have tried to define the term from various perspectives. Lashley (2004) defined microfinance as: “Lending small amounts of money for enterprise development to achieve a sustainable rise in incomes above the poverty line.” The rationale behind the definition is if one gives microloans to poor people, poverty will be reduced. However, it has been a controversial issue by the researcher to clearly attest, partly due to the difficulties of reliable and affordable measurement, the methodological challenge of proving causality, and other factors (Stewart, 2010). Furthermore, poverty alleviation efforts require an understanding of the realities of local communities’ conditions, and mobilizing the stakeholders towards the effort of poverty reduction. The rationale of MFI loan provisions seems plausible to serve as vehicle through which the poor are empowered, thereby providing a valuable tool to assist the economic development process (Akinlo and Oni, 2012). Elsewhere, microfinance is seen as a credit scheme that uses collateral substitutes to short-term working capital of micro entrepreneurs (Hubka

222

Chapter Eight

and Zaidi, 2004). But, such definitions ignore two important aspects of microfinance: saving services, and loans for consumption purposes. Similar to banks, saving mobilization is perceived as important for micro lenders to ensure the repayment of non-collateral-based loans (Adam, Graham and Pischke 1984, Hulme and Mosley, 1996). On the other hand, micro credits are not only demanded for investment activities, but also to smooth consumptions of the poor in a response to sickness, death, and harvested failures (Matin et al. 2002, Hulme and Mosley 1996). More broadly, MF can be defied in terms of the following characteristics: targeting the poor (especially poor women); promoting small business; building the capacity of the poor; extending small loans without collateral; combining credits with savings, and charging a commercial interest rate (Dejen, 1998; Ajit, 2012). Some theoretical literature has been written in line with the role and contributions of MFIs for development endeavour, which is highly needed, particularly in developing countries. However, microfinance is poorly understood, and it remains unclear whether it delivers on its promises. In this respect, the reviewed literature that assesses the effect of MFI loan provision on poverty alleviation has reported mixed and disputed results so far in developing countries including Ethiopia. Thus, there seems to be a missing element, like alignment of the loan provision with local development strategies of communities. Studies conducted show (Taiwo, 2012; Okumadewa, 1998; Sowmyan, 2011) MF loan provision serves as an effective way for poor people to increase their economic security and thus reduce poverty. Moreover, it enables poor people to manage their limited financial resources, reduce the impact of economic shocks, and increase their assets and income. On the contrary, Srinivas (2004) disproved that it facilitates the diversion of valuable aid money from untested and non-viable microfinance programs - away from vital programs on health and education that are in dire need of such funds. Similarly, Hickson (2001) witnessed that most MFIs have a long way to go in reaching the extremely poor so as to effectively achieve the goal of poverty alleviation. In the same direction, Hulme (2008) argued that MF has other indirect negative effects as it introduces households to a culture of debt that might change its wellbeing adversely. NPOWG (2006) also emphasized that although most MFIs aim to reach poor people, it has become increasingly apparent that they rarely serve very poor people. Most MFIs reach the ‘upper poor’ in much greater numbers than the ‘very poor’. To sum up, these conflicting views surrounding microfinance and its effectiveness at reducing poverty in less developed economies (LDCs) has led to several empirical studies on microfinance and poverty reduction

Loan Provision by Microfinancing Institutions

223

in developing economies (Akinlo and Oni, 2012). Furthermore, it requires re-visiting again and again the policies and strategies of MFI loan provision programs towards poverty reduction, and filling the gap in line with the local communities’ realistic conditions, potential resources, beneficiaries, and markets demands by mobilizing the concerned stakeholders under the framework of local economic development strategies in particular countries, regions, and districts that can be considered as alternatives tools to facilitate the momentum of poverty reduction.

4.2 Definition and Initiatives of Local Economic Development (LED) Strategies The term local economic development (LED) originally referred to deliberate intervention to promote economic development in a specific area that is not the national area - from a very small neighbourhood through to fairly large sub-national region. A common definition of LED is a process in which partnership between local government, the private sector, and the community is established to manage local access to external resources to stimulate the economy of a well-defined territory (Meyer, 2003). Similarly, Blakey (1994) explains it as a process in which local governments or community-based organizations engage in stimulating or maintaining business activity and/or employment because the goal is to stimulate local employment opportunities in sectors that improve the community, using existing human, natural, and institutional resources. Basically, the rationale of LED is about local people working together to achieve sustainable economic growth that brings economic benefits and quality of life improvements for all in the community. “Community” is here defined as a city, town, metropolitan area, or subnational region (World Bank, 2004). Therefore, LED as a bottom-up and endogenous development approach is used to shift responsibility for dealing with unemployment, micro and small-scale enterprise mobilization, poverty reduction, and economic growth policies to the local authority. It lays a playground for local government, the private sector, the not-for-profit sector, and the local community to work together to improve the local economy with the aims of enhancing competitiveness and thus encouraging sustainable inclusive growth (Boxer and Josie, 2007). As per Boxer and Josie, the strong justifications for LED are that each community has unique local conditions that help or hinder its economic development. Before a community can

224

Chapter Eight

commence with economic development strategies, it needs to understand the nature of the local economy. Creating an inventory and profile by collecting data, then analysing the data by evaluating the profile and inventory, provides the factual basis for economic development goal setting and strategy development. As the success rate of any poverty alleviation program depends on the availability of accurate, reliable, and up-to-date information and statistics and thus, the community must be fully informed about their own town or city, their own region, and their national economy (Boxer & Josie, 2007). Hence, micro and small-scale enterprises’ mobilization efforts and MFI loan provision should be carried out as per the road map of the particular regions and districts of LED strategies, to assure the sustainability and development endeavour. As the LED is all about strategic planning through partnerships between local governments, the business community, and NGOs by focusing on particular regions’ potential, and identifies specifically what local stakeholders can and need to do to ensure their local community reaches its potential, this helps particular regions to assess a community’s comparative advantage, identify new or existing market opportunities for businesses, and reduce obstacles to business expansion and creation (USAID, 2011). It paves the way for promoting local opportunities for businesses to grow, such as improving the business enabling environment where firms operate, creating the conditions that will generate new business or expands existing ones, encouraging value added and higher growth activity, and linking public and private leadership and funds in a common vision and implementation strategy (Mike, 2006; Rodriguiz and Tijmstra, 2006). Rodríguez-Pose, et al (2009) argued LED strategies, while no panacea, may be a valid complement to traditional top-down strategies in order to deliver sustainable development and, in many cases, may deliver greater economic efficiency by mobilizing resources that otherwise may have remained untapped, as well as a large number of social benefits, by promoting voice, participation, and sustainability across territories where institutional conditions have been far from ideal. The emphasis in LED has grown beyond a preoccupation, with local selfsufficiency towards understanding, developing, and exploiting economic linkages from the district and national, through to the global level (Idwe & Bhisho, 2010). Communities within a regional economy will need to decide upon the key programs that will become part of their strategy. It often focuses on both enhancing competitiveness, and thus increasing growth, and also on redistributing that growth through the creation of small-scale micro enterprises, and through focusing on job creation that

Loan Provision by Microfinancing Institutions

225

aligned with the local potential of a particular region (DPLG, 2004). In connection to these, the Ethiopian government has crafted GTP (2011/122014/15), which is supposed to be cascaded down to regions and districts as per the existing local conditions, to end vicious circles of poverty in the country. The GTP envisages accelerated growth in the industrial sector through the creation of an enabling environment that will make the sector the basis of future economic development. In this regard, the focus will be on the development of small-scale and micro enterprises that utilize existing potential resources of regions as competitive advantage to harness the development plan of the country. To facilitate the situations more, the government has launched MFI loan provision programs, which are backed by regional governments, micro and small-scale enterprise agencies, and vocational and technical schools, to train manpower in line with local development priorities of regions of Ethiopia. Thus, the GTP visions can be achieved by strengthening the micro and small-scale enterprises in a manner that unleashes the full growth potential of the MSEs to grow into medium and large-scale domestic enterprises by properly utilizing potential resources and addressing local development priorities. This is expected to expand the country’s industrial base and increase foreign earnings by encouraging export and import substitution activities, and to generate employment opportunities and wealth by promoting value addition to the vast raw materials, including primary agricultural products available in various parts of local regions in Ethiopia (UNDP, 2012). Therefore, these need mobilizing and synergizing for the stakeholders concerned and engaged in poverty alleviation, as per the local economic development demands of the regions, which requires strong emphasis on the journey of GTP in Ethiopia. The component of LED strategies can be summarized by the following diagram:

226

Chapter Eight

Source: USAID, 2011

4.3 Linkage between MFI Loan Provisions and LED Strategies Microfinance programs focus on expanding local economic activities and improving the standard of living of their clients by providing loans needed to establish micro and small businesses, enterprises, or cooperatives (Shannon, et al, 2005). Towards these, in both developing and transition economies, microfinance has increasingly been positioned as one of the most important poverty reduction and local economic and social development policies. Its appeal is based on the widespread assumption that simply ‘reaching the poor’ with microcredit will automatically establish a sustainable economic and social development trajectory animated by the poor themselves. MF model may well generate some positive short run outcomes for a lucky few of the ‘entrepreneurial poor’, the longer run aggregate development outcome very much remains moot. It may ultimately constitute a new and very powerful institutional barrier to sustainable local economic and social development, and thus also to sustainable poverty reduction. However, empirical research reveals that participation in microfinance did not result in increased household wealth and poverty alleviation desired by developing countries (ibid). Thus, it is clear that economic policies and strategies pursued in the past and continuing on to the present have neither produced sustainable growth, nor reduced poverty. Therefore, filling the gap as one alternative tool pursuing and mobilizing and synergizing the fight against poverty through LED strategies is essential. LED is “outcome based” on local initiative, and

Loan Provision by Microfinancing Institutions

227

driven by local stakeholders. It involves identifying and using primarily local resources, ideas, and skills, to stimulate development and harness the poverty alleviation program (DPLG, 2004). The reviewed research shows that economic policies and programs in the past have catered to the larger enterprises and have been biased toward urban areas, despite the reality that the majority residing in the rural areas and the overwhelming majority of micro enterprises are running duplicated businesses here and there copied from urban areas. Microfinance institutions loan provision is also, as suspected, more prorich than pro-poor in most developing countries. They have failed to exploit the available resource opportunities and potential in particular regions. Moreover, most economic initiatives were centrally planned and failed to effectively mobilize local resources and structures. The very essential missing factor from the process was active LED strategies that promote the following: i) Optimization of usually scarce resources in an area ii) Integration of municipal and provincial plans and priorities with regional and national plans from the bottom upwards; iii) Citizen participation and consensus building among stakeholders (Simon, 2003). Thus, linking microenterprise establishment plans and hence loan provision in line with local economic development strategies is very essential and plays a significant role for the sustainability of the institutions. Microfinance must involve the people themselves in examining the problems and creating solutions if it is to be sustainable with those upon whom development is targeted (Sowmyan, 2011). As the LED is a process that brings together resources from within and outside the community to address these challenges and promote economic growth in a systematic and organized manner at the local level and with its robust history of national and local government capacity building, it is well positioned to become a catalyst for promoting LED as a key ingredient to achieving broad-based and equitable economic growth (USAID, 2011). It is an important tool for the alleviation of poverty and the development of sustainable local economies, and serves as an essential tool that can be used to unlock economic opportunities in the local area that will create jobs and ultimately uplift the livelihoods of the people in the study area. It is a participatory process where the local role-players interact to develop, grow, and strengthen the economic base of a geographical locality. It comprises the harnessing and focusing of all resources and interventions that potentially impact upon the economic development of a local area in order to better serve local economic imperatives (Milinda, 2008). Meyer-Stamer (2003) argued that LED is

228

Chapter Eight

also about the competitive advantage of local firms, and the competitive advantage of the locality. Through the creation of competitive conditions at a specific locality, investment may be attracted which otherwise would go elsewhere and create jobs and income in another locality. LED is different from regional/national economic development in several respects. Moreover, Astia et al. (2004) indicated that mainly it focuses on achieving three inter-connected goals: (i) creation of economic growth and job markets by capitalizing existing potential local resources, (ii) reduction in the number of poor people, and in turn (iii) realization of sustainable livelihoods.

4.4 Evolution and Performance of Microfinance Institutions in Ethiopia The Ethiopian government has taken MFIs as one of the instruments in the effort towards reducing hunger, famine, youth unemployment, and to overall harness the economy of the country, which is addicted to aid and identified with a brand of deep-rooted poverty. It is one way to shift from aid dependency to self-reliance through mobilizing and engaging abundant labour and other resources towards development endeavour designed by the country. These urgently need to reach out to poor and marginalized people by facilitating their financial constraints of initiating and running entrepreneurial businesses, thus creating employment opportunities in the country. Having recognized the positive contributions of MF, legislations were adapted to license and supervise service providers in 1996 (Adeno, 2007). MF in Ethiopia is in its infant stage. At its inception, four MFIs owned by the government were established, which plays a dominant role and managed closely by regional states in Ethiopia. Based on the data of 2006, the industry’s outstanding loan to GDP was 1.7 percent and its share to loan and advances of lending banks and MFIs was 1.6 percent. Mobilize client savings by MFIs had reached 3.6 percent of gross national savings. As at the end of June 2007, twenty-seven MFIs operate in the country, obtaining license from the National Bank of Ethiopia. Most of the MFIs operate both in the rural and urban areas mainly centring their head office in Addis Ababa. Dedebit Credit and Saving Institution (DECSI) and Amhara Credit and Saving Institutions (ACSI) take a more than 65% share in serving clients in the market. Similarly, in outstanding loan provision, these institutions also take the lion share (62 percent) in the market (Befekadu, 2007).

Loan Provision by Microfinancing Institutions Key Facts: Microfinance in Ethiopia By No. of No. of CGAP 4 MFIs Borrowers 22 1,420,000 No. of No. of Active By MFIs Borrowers MIX7 22 2.3 million By No. of No. of Active MFT8 MFIs Borrowers 31 2,470,641 Source: MFTransparancy, 2011

Borrowers Population5 2% Gross Loan Portfolio 409.4 million Gross Loan Portfolio 6.9 billion

229

Borrowers/Poor6 5% Average Balance per Borrower (USD) 140 % Products with a Flat Interest Rate 70%

According to the MFTransparancy (2011) report, the number of MFIs had risen to thirty-one, with 433 branches and 598 sub branches by the end of 2011. Studies estimate that this figure serves between 10-25% of the total microfinance demand in the country. Even though the institutions have extended total credit of 6.9 billion ETB to 2,470,641 active borrowers. However, the coverage so far achieved is very insignificant relative to the existing demand for such microfinance credit and related services. It is only 2% of the total population and 5% of the poor that has been reached by the institutions since their inception. The Ethiopian microfinance market is dominated by a few large players, all of which are closely linked to regional government ownership. The three largest institutions account for 65% of the market share in terms of borrowers, and 74% by gross loan portfolio9. These are Amhara (ACSI), Dedebit (DECSI) and Oromia (OCSSCO) Credit and Savings Institutions. In contrast to many other African countries, MFIs in Ethiopia reach relatively large numbers of clients, with ACSI reaching over 650,000 borrowers. Most institutions have over 20,000 clients. The average loan per borrower for the 11 Ethiopian MFIs reporting to the mix market as of June 2010 stood at 140 USD, below half of the country’s GDP per capita (MFTransparancy 2011).

 4

Consultative Group to Assist the Poor (CGAP), 2010. http://www.cgap.org/m/africa.html 5 MFI borrowers as a percentage of the country’s overall population 6 MFI borrowers as a percentage of the poor population based on national poverty rates 7 MIX Market, 2010. http://www.mixmarket.org/mfi/country/Ethiopia 8 This row of the table is populated with data from MFTransparency’s Transparent Pricing Initiative in Ethiopia 9 MFTransparency, 2010, Transparent Pricing Initiative in Ethiopia, http://www.mftransparency.org

230

Chapter Eight

The Ethiopian microfinance sector is growing in terms of number, outreach, and major financial indicators. There are 32 microfinance institutions operating throughout the country; serving more than 2.89 million clients through 1,244 branch offices located in different regions of the country. The top five microfinance institutions backed by regional governments are ACSI (Amahara), DECSI (Dedebit), OCSICO (Oromiya), ADCSI (Addis) and Omo, which in aggregate serve about 84.7% of the total clients of the sector; and account for more than 90% of the sector’s total assets (NBE quarterly report as at June 30, 2012). The sector has been showing an alarmingly increasing growth trend since its establishment in 1996. The total assets of the microfinance sector has increased to Birr 13.3 billion as of June 30, 2012 from Birr 5.3 billion as of June, 2008; an increment of 149.2%. This was mainly financed by liabilities (71.8%), of which saving mobilized from the public constituted the major portion of 41%. About 30.8% is contributed by borrowings (10.2%) and other short-term (15%) and long-term (5.6%) liabilities. The remaining 28.2% of the total assets is financed by total capital. Gross loan grew to Birr 9.3 billion as at June 2012 from Birr 4.5 billion as at June 30, 2008; an increment by 107.6%. Total savings grew from Birr 1.5 billion to Birr 5.5 billion and capital from Birr 1.3 billion to Birr 3.8 billion in the same period (Sintayehu, 2013).

5. Empirical Studies: Nexus of MFI Loan Provisions and Poverty Alleviation Program The notional view of MF as a panacea to poverty reduction has attracted wide empirical research and public policy discourse in the past couple of decades. Though several empirical studies have been conducted to ascertain the impact of microfinance on poverty alleviation worldwide; no consensus has emerged on the impact of microfinance on poverty reduction (Akinlo and Oni, 2012). The questions of whether MFIs actually reach and empower the poor and/or whether microfinance is better than some other types of development project for the poor is not settled once and for all (Flore & François, 2011). Recently, studies showed MF has positive impacts to boost the ability of poor people to improve the conditions in which they live and enhance the development program of the countries. In line with this, many countries have designed MFIs as intervention strategies to enable poor people to increase their income by starting new enterprises or expanding existing

Loan Provision by Microfinancing Institutions

231

ones; harnessing entrepreneurship and innovation for development. Towards this, Akinlo and Oni’s (2012) studies attest that loan empowerment has a significant positive effect on beneficiaries’ welfare, as the access to credit allowed the beneficiaries to take advantage of economic opportunities by providing a fundamental basis for planning and expanding business activities. Similarly, in Zimbabwe, extremely poor clients of the Zambuko Trust increased their consumption of high protein foods at a time when food expenditures across the country as a whole were decreasing (Barnes 2001). Studies conducted in Bangladesh documented that microfinance programs are able to alleviate poverty through increasing individual and household income levels, as well as improving healthcare, nutrition, education, and helping to empower women (Khandker, 2005). A panel data analysis carried out by Flore & François (2011) indicates a positive impact by microfinance loan provision on small informal enterprises in Madagascar. Taken as a snapshot, the evaluations successively conducted in 2001 and 2004 indicate that the clients’ enterprises recorded better average performances than enterprises without funding. With a dynamic perspective however, the results are more nuanced. In Ethiopia too, some research conducted in the area attests this reality on the ground. In contrast to the above, many empirical studies reported the nexus between MF loan provision and poverty alleviation programs is insignificant and they fail to find out the direct link between microfinance and poverty reduction. Hence, MF loan provision did not make a significant change to the livelihoods of the poor (Hulme and Mosley, 1996; Majoux, 2001; Doung and Izumela, 2002; Bakhtiari, 2006). Kah et al. (2005) studied the evolution, sustainability, and management of ten microfinance institutions (MFIs) in Gossas, Senegal, using the data for a period of three years. They found that MFIs have helped to create some positive changes, but that there was no clear and marked evidence of poverty reduction, and stated that the expectations of what microfinance can do to help lift communities out of poverty is a bit too optimistic. Essentially, the general consensus from studies that reported little or no positive impact of micro credit on poverty is that the former is a necessary but not sufficient condition for poverty reduction. It is contended that basic infrastructure, coupled with capacity development of the poor in terms of skills and education, is required for making microfinance an effective tool of poverty alleviation (Akinlo and Oni, 2012). In Ethiopia, Alemu (2006) reported a positive impact of microfinance on the poor in five different zones of the Amhara region. In particular, the

232

Chapter Eight

results reported that microfinance helped the poor to smooth their income in the study area. All the same, the study reported cases of loan diversion. Some clients were found to have used their loans for unintended purposes. Similarly Asemelash (2003) also attested that loan provision to the poor by microfinance had a positive impact on the poor. Specifically, the results showed that microfinance led to increased income for the poor. Besides, it helped to increase poor people’s access to better schools and medical facilities. Nur (2006) indicated that microfinance beneficiaries have more diversified incomes than non-beneficiaries. Furthermore, these households had more income relative to non-beneficiaries. As a result, households’ access to microfinance services had higher monthly food and non-food expenditures than non-service users. This indicates that the former households have better purchasing power than the latter. Thus, they have shown better calorie consumption. Access to microfinance services enabled beneficiary households to increase their income. More importantly, the beneficiary households have employed fewer coping strategies during food shortages, which imply that adequate food is at their disposal. The result also shows that the severity of responses on the utilization of coping mechanisms was more pronounced in non-beneficiary households, and recommended that pastoralist socio-economic oriented microfinance service reduces vulnerability and minimizes the risk of becoming food insecure. Furthermore, studies conducted on microfinance outreach, sustainability, and loan repayment performance areas show that (Daba, 2004; Asmelash, 2003; Fiona, 2000; Berhanu, 1998) the credit provided to the poor has brought a positive impact on the life of the clients as compared to those who do not get access to these microfinance services. They showed that micro finance has brought a positive impact on income, asset building, and access to schools and medical facilities of the household in the study area. Results show that borrowing causally increased consumption and housing improvements. A flexible specification that takes into account repeated borrowing also suggests that borrowing has cumulative long-term effects on these outcomes, implying that short-term impact estimates may underestimate credit effects (Guush and Cornelis, 2011). Despite the prominence of microfinance in development policy, only a few rigorous impact evaluations have been conducted to date, a large majority in Asia. We describe the results of two parallel large-scale randomized controlled trials completed between 2003 and 2006 in rural Amhara and Oromia (Ethiopia). We rst document that borrowing increased substantially in communities where the programs started their operations. However, we nd only mixed evidence of improvements in economic activities,

Loan Provision by Microfinancing Institutions

233

including agriculture, animal husbandry, and non-farm self-employment. We nd instead that the introduction of micro-credit was overall associated with signicant improvements in school attendance among 1016-year-old boys, and (in Amhara only) among children of age 6-9 of either gender (Jaikishan, et al., 2011). Contrary to the above results, Getaneh (2005) indicated MF has brought very little impact in poverty reduction and enterprise development. The poor marketing situation, lack of infrastructure (road network in particular) and lack of business skill by the borrowers have challenged the success of MF on poverty reduction. Moreover, Bekele (2002) recommended that not one-fits-all, but area-specific approaches or fitting with local development strategy are important for MFIs to be successful in meeting their objectives. Despite anecdotal stories of success and testimonies from female clients and other borrowers, a number of reports argue that microfinance lacks hard, quantitative data that accurately measure significant changes in the economic conditions of the poor. Although participants improve their income, it is not clear whether these benefits accrue to the society and positively affect other people in the community and national levels (Midgley, 2008). On the other hand, Tassew (2005) showed that the most important activities that best fit into the current microcredit system in the region are petty trading, goat fattening, and poultry development. Loan size needs to be increased and the repayment period needs to be lengthened to make income diversification work better. Thus, the researcher emphasized in the research that the idea of working very well on local development priorities varies from place to place in the country. In Ethiopia most loans are provided for cobblestone construction, which is probably good in some towns and may not be in other towns. Therefore, in order to harness these problems there should be synergy among all stakeholders in the region to work to gather so that micro and small-scale enterprises established here and there can be based on the local development priorities of the region, which sustain the success of their business and help to alleviate the poverty endeavour program in the region. To substantiate Getaneh (2007) argued that MF holds a good promise as one key sector to poverty alleviation and microenterprise development in particular, where appropriate financial products and methodologies suited to local circumstances are available, considerable achievements can be registered. For this to be more effective, however, such complements have to be there, particularly those related to enterprise development, aligning with potential local development priority areas

234

Chapter Eight

including: appropriate agricultural technology and extension, business development services, marketing, entrepreneurship development, rural infrastructure, etc. Given the high proportion of people in this particular sector of the economy, a collaborative effort is required to remove all challenges that are facing the industry, as this would have a strong repercussion on the entire micro-enterprise growth, as well as on the national development at large. He further argued that demand for credit is highly constrained due, mainly, to low entrepreneurship. Many clients, as can be expected, are very much risk-averse, that even with the availability of credit, many do not like to venture into activities other than those inherited from their fathers or forefathers. In a recent survey of about 300 clients, over 78% responded that they only want to be engaged in activities that they know something about previously. Hence, though MF is prescribed as an effective input to alleviate poverty through provision of financial services to those marginalized parts of the society, it cannot be a panacea in itself. The availability of credit alone could not be a solution to the problems of the ultra-poor, and its impact on poverty reduction remains in doubt (Adeno, 2007). In spite of theoretical concerns about bottom-up development paths and scepticism about MF being over ambitious, the sector possesses a strong goodwill among development partners, and tends to be a core component of the development strategy of most developing countries. Determining how microfinance can be used as an important vehicle to make an even larger and more critical contribution to alleviating poverty is in need of more careful assessment (Chowdhury, 2012). Similarly, Hulme and Mosley (1996), while acknowledging the role MF can have in helping to reduce poverty, concluded from their research on microfinance that: “Most contemporary schemes are less effective than they might be.” They stated that microfinance is not a panacea for poverty-alleviation, and that in some cases the poorest people have been made worse-off by microfinance. In some other instances, microfinance is said to play an insignificant role towards mitigating the problem of the poor. Sintayehu (2013) said the impact MF had on poverty would have been higher if the clients had been equipped with appropriate entrepreneurial skills and, thus, aligned with potential competitive advantages in areas or regions. Therefore, it can be inferred as a research gap from the above empirical reviews that local development strategies can be a missing element that can play a significant role against poverty reduction. Thus, aligning microenterprise mobilization and microfinance loan provision programs with practical, actual potential resources of a particular region can enhance the poverty

Loan Provision by Microfinancing Institutions

235

alleviation program via microfinance institutions, and facilitate the GTP plan of the country.

6. The Linkages of MSEs Development, LED, MFIs Loan Provision and Poverty Alleviation There is consensus among policy makers, economists, and business experts that small and micro enterprises (SMEs) are drivers of economic growth. A healthy SME sector contributes prominently to the economy through creating more employment opportunities, generating higher production volumes, increasing exports, and introducing innovation and entrepreneurship skills. The dynamic role of SMEs in developing countries insures them as engines through which the growth objectives of developing countries can be achieved (NCR, 2011). Therefore, SMEs have been accepted in most developing countries, including Ethiopia, as an instrument of harnessing development endeavours and grass-root level poverty reduction efforts. The basic justification is that the SMEs could serve as a catalyst in the socio-economic development of any country through reaching out to the marginalized people or communities, and could thus play a significant role in the achievement of the national macroeconomic objective in terms of employment generation at low investment cost and enhancement of local innovation and entrepreneurship for poverty alleviation. Kombo et al. (2011) argued that SME entrepreneurs who include agriculture and rural businesses have contributed greatly to the growth of the Kenyan economy. The sector contributes to the national objective of creating employment opportunities, training entrepreneurs, generating income, and providing a source of livelihoods for the majority of low income households in the country, accounting for 12-14% of GDP (Republic of Kenya, 1982, 1989, 1992, 1994). It is estimated that SMEs employ 22% of the adult population in developing countries. In the same way, a recent study conducted by Abor and Quartey (2010) estimates that 91% of formal business entities in South Africa are SMEs, and that these SMEs contribute between 52% and 57% to GDP, and provide about 61% of employment (NCR, 2011). Similarly, the United Nations Industrial Development Organization (UNIDO, 1999) estimates that SMEs represent over 90% of private business, and contribute to more than 50% of employment and gross domestic product (GDP) in most African countries cited in NCR (2011). In light of these arguments, it is no wonder that governments, particularly in developing countries, have made tremendous efforts and established

236

Chapter Eight

policies to enhance the capacity of micro and small-scale enterprises (MSEs). However, empirical research indicates that MSEs have fallen short of expectations, and thus the loan provision program towards poverty alleviation endeavour. Therefore, these generated serious concern and scepticism on whether SMEs can bring about economic growth and national developments in developing countries. SMEs are faced with significant challenges that compromise their ability to function and to contribute optimally to the economy of developing countries, and thus add value towards poverty alleviation campaigning (Osotimehin, 2012). Junior (2006) also argued that despite their significant importance and SME contribution to economic growth, SMEs across the whole world in general, and developing countries in particular, are still faced with numerous challenges that inhibit entrepreneurial growth, as well as being a catalyst for the poverty alleviation program. The most challenging factor is failing to align with and capitalize on existing potential resources, as well as lack of synergy among stakeholders involved in training, mobilizing, and finance provision institutions of a particular region. Hence it is observed that microfinance institutions and local government extend loans to business ideas that have little or no chance of success, or those that may have a negative impact on their intended outcomes. Even if market potential is visible, local governments sometimes do not have the business expertise to develop the concept, manage the project, or even communicate and cooperate with the local economic actors that do. Apart from SME funding and access to finance, the Global Entrepreneurship Monitor (GEM) reports (2001-2010) noted that in South Africa SMEs suffer from poor management skills, which are a result of a lack of adequate training and education, as well as alignment of the business to existing local available potential resources, that harness the business and means there is a high rate of business failure. Thus, the campaigns against poverty reduction through MSEs are still facing a number of difficulties and obstacles that are impeding and complicating their operations and growth (NCR, 2011). Presented below are actors involved with the linkage between MSEs and poverty reduction efforts in line with local economic development strategies, which require significant emphasis by the concerned stakeholders and policy makers in developing countries, including Ethiopia. Therefore, if the countries need to use MSEs as an instrument for poverty reduction and development endeavour, the policies towards the mobilization of the sectors according to the concern and actual realities need re-visiting to address the problem at hand.

Loan Provision by Microfinancing Institutions

237

Source: Roggerson (2006)

In Ethiopia, the empirical studies conducted on MSEs are widely acknowledged to contribute towards promotion and development of inventions, and thereby generate employment opportunities in the country. They are particularly important in the context of the country’s povertyreduction strategy because they are a seedbed for the development of medium and large enterprises, and because they absorb agriculturally under-employed labour, and diversify the sources of income for farming families (Tiruneh, 2011). It is natural to say that every small business owner starts with high hopes of success, but it is a usual phenomenon that each year firms go out of business. Although failure is not the sole reason for enterprises to close, many enterprises do fail each year. Thus, the odds of forming a profitable venture are a critical issue for those weighing the risk of starting a business (Dennis and Fernald, 2001), and also an understanding of why firms succeed is crucial to the stability and health of the economy (Pompe and Bilderbeek, 2005). To address these problems, GTP clearly indicated that MSE development is the key industrial policy direction contributing to the envisaged structural transformation of the economy and hence poverty alleviation program among urban and rural poor people in the country. Thus, giving significant emphasis to the quality and quantity of micro enterprises established at various regional and urban parts of the country with particular attention on local development priority and potential resources of the areas, it was described that GTP targets for MSEs include the following: i) Provide comprehensive support to micro and small-scale enterprises so as to create employment opportunities for about three million people and thus enhance citizens’ income, contribute to a rise in domestic saving, and improve the benefits for women and youth from the sector so as to reduce unemployment and poverty, ii) Provide training of

238

Chapter Eight

trainers for 10,000 professionals in the sub-sector, iii) Provide capacitybuilding and basic skills training for about 3 million operators in the areas of entrepreneurship, technical, and vocational skills, iv) Prepare and develop 15,000 ha of land for working premises, and construct shade and buildings for MFIs, v) Provide microcredit and marketing information and work with producers to identify bottlenecks and provide support where solutions are identified (MoFED, 2010). As per the study conducted by Sintayehu (2013) on MF loan provision perspectives, the government of Ethiopia has given due attention to financial access in its five year (2010/11-2014/15) growth and transformation plan (GTP). In particular, the microfinance sector is expected to play its vital role in two dimensions: to contribute its part in expanding access to finance from 20% as at the base year (2009/10) to 67% at the end of the GTP period (2014/15) and in increasing Gross Domestic Saving from 5.5% as of base year to 15% as at the end of GTP period. To this end, microfinance institutions are expected to finance 50% of their gross loans through savings mobilization during the first two years, and 80% at the end of the period. Savings mobilized by the sector have shown a higher growth rate during the last two years of the GTP period than before; i.e., they increased by 43.2% on average in the last two years. As a result, the sector financed 58.7% of its gross outstanding loans by savings as of June 30, 2012; this was an increment by 13% from the base year of GTP (45.6% as at June 30, 2010); on average 6.6% in the last two years, and above the plan (to finance 50% of its gross loan with savings in the first two years of GTP period). During the last two years of the GTP period, saving has grown by a higher rate (43.2%) than gross loan (26.5%); on average, but still to catch up gross loan (to finance 80% of gross loan with savings at the end of GTP period); saving needs to grow by a higher rate, as the loan by itself shall grow by a higher rate than this to achieve the GTP plan. The practical reality in many parts of the country show there is a big problem of innovativeness and the trend of business establishment is a duplication of retail shops. There is a pressing need to induce innovation in new businesses as per the local comparative resources and demands if we are to spur sustainable development into the economy. Nevertheless, there may not be a shortcut way to see a sufficient number of innovative businesses in the economy. However, as much as possible doing a level best effort to address the problems are essential to assuring the sustainability of MSEs and hence achieving the GTP target accordingly. It

Loan Provision by Microfinancing Institutions

239

seems emphasis towards local economic development strategies is a missing element that could have been addressed clearly in the plan. On the other hand, studies conducted by Minilek and Chinnan (2012) identified that the compound annual growth rate of small and micro enterprises in employment aggregately were 3.86% and the enterprises add annually 0.18 workers on average. When looking separately, the mean annual compound growth rate of small-scale and micro enterprises were 2.4% and 4% on average respectively. Small enterprises add 0.19 workers per firm and micro enterprises add 0.14 workers per firm annually. Shortage of working capital and working space are the most important problems. It was recommended that employment growth in small and micro enterprises were low, and emphasis should be given for enterprises to create more jobs rather than promoting a large number of one-man enterprises. Towards this, the Ethiopian government has a firm belief in the pivotal role of microfinance in poverty reduction; as a result, it decentralized MFIs to regional states to ease the finance problems of MSEs. However, the MSEs mobilizations have to be carried out with the road map of GTP and local development strategies to bring a satisfactory result and overcome the complicated problems in the arena.

7. Conclusion and Policy Implications In developing countries like Ethiopia where poverty has been a serious enemy for a long period of time, various alternatives strategies have a very vital role in tackling the problems. Towards this end, Ethiopia has crafted a Growth and Transformation Plan (2011-2015), where microfinance serves as an instrument for poverty alleviation as pro-poor tools to facilitate access to loans, and hence enhance employment for young people and women. Similarly, the regional state administrations are expected to cascade the national GTP and prepare regional GTPs, which take into account the particular resources and local development strategies of the regions. This is expected to materialize by zones and districts’ accordingly. Within this framework, the country engaged in mobilizing micro and small-scale enterprises in various parts of both urban and rural regions of the country. As a very essential instrument for smoothing the problem of fiancé-microfinance institution expansions, it has been taken as an essential tool to reach the poor in urban and rural parts of the country. MFI expansions have been taken as a tool to poverty reduction. In light of these views, this research has been carried out to evaluate loan provision

240

Chapter Eight

by micro financing institutions for poverty reduction and its linkages with local economic development strategies in Ethiopia. The empirical review shows that there are mixed results as to the effect of loan provision by MFIs for poverty reduction in various parts of developing countries, including Ethiopia. On the basis of theoretical and empirical discussions of relevant literature, microfinance alone cannot alleviate the poverty from the grass-root level of society and thus, requires linking the MFIs loan provisions with local development strategies of particular regions potential resources and market demands. Thus, to maximize the benefits of MFIs loan provision and achieve the GTP target requires strong emphasis on digging deeper to find keys to success (NPOWG, 2006), and hence targeting the poor as per the local demand and development priorities of regions and districts in Ethiopia. Though regional authorities are putting in a level best effort to address poverty problems through mobilizing micro and small-scale enterprises, providing facilities including training, etc., it seems these efforts are conducted without synergy among the necessary stakeholders. Furthermore, the empirical reviews indicate there are duplications of business services provided by the MSE. Through improving the entrepreneurial initiatives of the MSE, significant efforts should be made to realize the fight against poverty and improve the saving mobilization, so as to achieve its GTP or it will be required to grow by a higher rate than the past two years. Moreover, the government should introduce a financial literacy program so as to change the mind-set of the public towards financial institutions and financial services in general, and saving in particular. Side-by-side MFIs should give due emphasis to developing entrepreneurial skill of their clients in line with local development priority and resources of particular regions, along with loan provisions. To address this, MFIs should collaborate with necessary stakeholders, who are identified as responsible organs, as per GTP. These studies have also mentioned that the impact of loan provision by MFIs would have been higher, and the country could achieve the GTP target, if the MSEs’ mobilization work was aligned with local economic development strategies, and synergy among stakeholders engaged in poverty reduction in local areas in particular, and the country in general. These further facilitate the success of the MSE business and, hence, it has a strong implication for MFIs’ sustainability and outreaches to the target poor people both in rural and urban areas of Ethiopia. Therefore, the studies provoke policy makers and development actors to revisit the loan provision, particularly by regional government backed MFIs, MSEs,

Loan Provision by Microfinancing Institutions

241

mobilizations and efforts to sustain the local development endeavour, entrepreneurship, and innovations by the concerned stakeholders both governmental and non-governmental organizations to harness development momentum towards the GTP target in 2014/15 in Ethiopia.

References Adeno Kidane Ereda (2007). Outreach and Sustainability of the Amhara Credit and Saving Institution(ACSI), Ethiopia, Norwegian University of Life Sciences, Department of International Environment and Development Studies. Admasu Abera (2012). Factors Affecting the Performance of Micro and Small Enterprises in Arada and Lideta Sub-Cities, Addis Ababa-A Thesis submitted to the school of graduate studies of Addis Ababa University in partial fulfillment of the requirements for the Master of Business Administration (MBA) degree (unpublished). Agupusi, Patricia (2007). Small Business Development and Poverty Alleviation in Alexandra, South Africa. Paper prepared for the second meeting of the Society for the Study of Economic Inequality (ECINEQ Society, Berlin; July 12–14, 2007), School of Development Studies, University of East Anglia, and Norwich, UK. Ajit Kumar, Bansal & Anu Bansal (2012). Microfinance and Poverty reduction in India. Integral Review- A Journal of Management ISSN: 2278-6120, Volume 5, No. 1, pp 31-35. Akinlo Anthony Enisan & Oni Isaac Oluwafemi (2012). Impact of Microfinance on Poverty Alleviation in Ondo State, Nigeria: Australian Journal of Business and Management Research Vol.2 No.09 [31-37]. Alemu, B. A. (2006). Microfinance and Poverty Reduction in Ethiopia. A Paper Prepared Under the Internship Program of IDRC, ESARO, Nairobi. Arsyad, L. (2005). An assessment of performance and sustainability of Microfinance Institutions: A case study of Village credit institutions in Gianyar, Bali, Indonesia, unpublished PhD thesis, Flinders University, Australia. Asemelash, (2002). The Impact of Microfinance in Ethiopia: The Case of DCSI in Ganta Afeshum Woreda of Eastern Tigray. M. A. Thesis, Department of RLDS, AAU. Asiama, J.P. & Osei, V. (2007) “Micro finance in Ghana: An Overviewெ, Research Department Working Paper Bank of Ghana 07/01. Accra.

242

Chapter Eight

Astia Dendi, H. J. Heile, Mahman, Rukyatil Hilaliyah and Rifai Saleh Haryono (2004). Alleviating Poverty through Local Economic Development: Lessons from Nusa Tenggara, Jakarta. Bakhtiari, S. (2006). Microfinance and Poverty Reduction: Some International Evidence: International Business and Economics Research Journal, 5(12). Befekadu B. Kereta (2005). Outreach and Financial Performance Analysis of Microfinance Institutions in Ethiopia: National Bank of Ethiopia Economic Research and Monetary Policy Directorate. Boxer, L. Mpontshane & Josie Rowe-Setz (2007). The DRAFT DistrictWide Integrated Local Economic Development (LED) Strategic Plan: uMgungundlovu District Municipality- Draft Final Integrated LED Strategy Blueprint. Chowdhury Abdullah, Al Mamun, Nazmul Hasan and Arif Rana (2012). The Correlation between Micro-Credit and Poverty Alleviation in Bangladesh: An Empirical Analysis. Dejene, A. (2003). “Informal Financial Institution: The Economic Importance of Iddr, Iqqub, and Loans”, Proceedings of the National Workshop on Technological Progress in Ethiopia, Agriculture, Nov. 20-30, 2001, AAU, Addis Ababa. Dennis, W.J. and Fernald, L.W. (2001). “The chances of financial success (and loss) from small business ownership”, Entrepreneurship Theory and Practice, Vol. 26. Available at www.questia.com/PM.qst?a=o&d=5002434876. Development, prosperity and Good Governance; Genius Training and Consultancy Service, Addis Ababa, Ethiopia. DPLG (2004). Local Economic Development Policy and strategy. Pretoria: Government Printer: South Africa. Dzene Richman & Aseidu K. Fred (2010). Gender Composition, Competition and Sustainability of Micro Finance in Africa: Evidence From Ghana’s Microfinance industry: Ghana Institute of Management and Public Administration. Elisabeth Rhyne and María Otero (2008). Trends in Microfinance 20102015 Flore Gubert and François Roubaud (2011). The Impact of Microfinance Loans on Small Informal Enterprises in Madagascar. A Panel Data Analysis. Geiger, Michael and Goh, Chorching (2012). Ethiopia Economic Update: Overcoming Inflation, Raising Competitiveness. The World Bank Working Paper Number 74385.

Loan Provision by Microfinancing Institutions

243

Getaneh Gobezie (2007). Successes in Expanding Microfinance Opportunities in Rural Ethiopia – Where There is Little Entrepreneurship? A paper submitted for the International Finance Corporation & Financial Times “Essay Competition”, 2007, www.ifc.org/competition]. Getaneh, G. (2005). Regulating Microfinance in Ethiopia: Making it more Effective. Essays on Regulation and Supervisions, Amhara Credit and Saving Institution (ASCI). Greeley, M. (2005). Sustainable Poverty Outreach. Money with a mission: microfinance and poverty reduction. Published by ITDG publishing, Warwickshire, UK. Guush Berhane and Cornelis Gardebroek (2011). Does Microfinance Reduce Rural Poverty? Evidence Based on Household Panel Data from Northern Ethiopia: American J. of Agricultural Economics Volume 93, Issue 1. Hulme, D. (2008). Is Micro debt good for Poor People? A Note in the Dark Side of Microfinance, in T. Dichter & M. Harper (eds.), What's Wrong with Microfinance? Practical Action Publishing. Idwe House & Bhisho (2010). Local Economic Development Information Booklet: Department of Economic Development and Environmental Affairs: South Africa, Cape Town. Taiwo, J.N. (2012). The impacts of Microfinance on welfare and Poverty alleviation in South-west Nigeria: A PhD Thesis Submitted in Partial Fulfilment of the Requirements for the Award of PhD (Banking and Finance), Covenant University, Ota. Jaikishan Desai, Kristin Johnson, and Alessandro Tarozzi (2011). On the Impact of Microcredit: Evidence from a Randomized Intervention Rural Ethiopia. John D Conroy (2003). The challenges of Microfinance in Southeast Asia: Financing Southeast Asia's Economic Development, Institute of Southeast Asian Studies: Singapore. Junior R. Davis (2006). Evaluating and Disseminating Experiences in Local Economic Development: Observations on Integrated Development Programmes of the Free State, Republic of South Africa. Karnani A (2008). Employment, not microcredit, is the solution: The Journal of Corporate Citizenship, Vol. 32, pp.23-28. Kombo, Justus, Murumba and Makworo (2011). “An Evaluation of the Impact of Risk management Strategies on Micro-Finance Institutions’ Financial Sustainability: A case of Selected Micro finance institutions in Kisii Municipality, Kenya”: Educational Research, Vol. 2 (5) pp.1149-1153.

244

Chapter Eight

Letenah Ejigu (2009). Performance Analysis of a sample Microfinance Institutions of Ethiopia University Business School, Panjab University, Chandigarh. Littlefield, E., Murdurch, J. and Hashemi, S. (2003). Is Microfinance an Effective Strategy to reach the Millennium Development Goals? CGAP Focus Note, Washington, DC. Mathewos (2006). Local Economic Development Strategy Manual: Federal Urban Planning Institute: Addis Ababa. Mayne, J. (2001). Addressing attribution through contribution analysis: Using performance measures sensibly. Canadian Journal of Program Evaluation, 16(1), 1-24. Mayne, J. (2001). Addressing attribution through contribution analysis: Using performance measures sensibly. Canadian Journal of Program Evaluation, 16(1), 1-24. —. (2008). Contribution analysis: An approach to exploring cause and effect. Institutional Learning and Change. Rome, Italy. Meyer-Stamer, J. (2003). Principles of local economic development: Options for South Africa, Community Self Reliance, 3, p.1-4. Mftransparency (2011). Promoting Transparent Pricing in the Microfinance Industry, country survey, Ethiopia. Midgley, J. (2008). Microenterprise, global poverty and social development. International Social Work, Vol. 51, no.4, pp. 467-479. Mike, Keshishian (2006). Local Economic Development: Making Cities Work Assessment and Implementation Toolkit: EGAT/PR/UP Cognizant Technical Officer USAID. Milford Bateman (2011). Microfinance as a development and poverty reduction policy: is it everything it’s cracked up to be? Overseas Development Institute, UK. Milinda Brink (2008). Sol Plaatje Local Municipality: Local Economic Development Review –Report-South Africa. Minilek Kefale and K. P. M Chinnan (2012). Employment growth and challenges in small and micro enterprises Woldiya, North East Amhara region, Ethiopia Educational Research and Essays Vol. 1(2), pp. 21 – 26. Available online at http://www.wudpeckerresearchjournals.org/ERE, Wudpecker Research Journals. Ministry of Finance and Economic Development (MoFED) (2010). Growth and Transformation Plan (GTP) 2010/11-2014/15: Addis Ababa. MOFED (Ministry of Finance and Economic Development), (2008). Dynamics of Growth and Poverty in Ethiopia (1995/96-2004/05) :

Loan Provision by Microfinancing Institutions

245

Development Planning and Research Department, Addis Ababa, Ethiopia. Morduch, J. (1999). The microfinance promise: Journal of Economic Literature. Vol. 37(4). National Credit Regulator (NCR) (2011). Literature Review on Small and Medium Enterprises’ Access to Credit and Support in South Africa: Underhill Corporate Solutions (UCS). NPOWG (Network Poverty Outreach Working Group) (2006). Microfinance and Non-Financial Services for Very Poor People: Digging Deeper to Find Keys to Success. Nur Abdi Mohammed (2006). The Role of Microfinance in Strengthening Pastoral Household Food Security A Comparative Study between Microfinance Service Beneficiary and Non-beneficiary Households in Filtu and Dollo Ado Districts of Somali Region Somali Regional State Pastoral and Agro pastoral Research Institute. Okunmadewa, F. (1998). “Domestic and international response to poverty alleviation in Nigeria.” Proceedings of the 7th Annual Conference of the Zonal Research Units Organized by Research Department, CBN at Makurdi, 8th - 12th June, pp: 296- 309. Organization for Economic Co-operation and Development (OECD) (2001). Strategies for sustainable development: guidance for development co-operation head of publications service, OECD publications service, 2, Rue André-pascal, 75775 Paris Cedex 16, France. Osotimehin, Jegede and Akinlabi (2012). An Evaluation of the Challenges and Prospects of Micro and Small Scale Enterprises Development in Nigeria: American International Journal of Contemporary Research Vol. 2, No. 4. Otero, M. (1999). Bringing development back, into microfinance. Journal of microfinance Vol. 1(1). Pfister M. W., Gesesse D., Amha W., Mommartz R., Duflos W., Steel E. (2008). Access to finance in Ethiopia: Sector Assessment study volume 2, GTZ. Plummer and Janelle (2012). Diagnosing corruption in Ethiopia: perceptions, realities, and the way forward for key sectors. Directions in development: public sector governance: Washington D.C. the Worldbank: http://documents.worldbank.org/curated//01/16380611/diagnosingcorruption-ethiopia-perceptions-realities-way-forward-key-sectors.

246

Chapter Eight

Pompe, P.M. and Bilderbeek, J. (2005). “The prediction of bankruptcy of small-andmedium sized industrial firms”, Journal of Business Venturing, Vol. 20 No. 6. Available at econpapers.repec.org/article/eeejbvent/default20.htm. Rodriguez-Pose, A. (2008). Milestones and challenges of LED practice and academic research, local. glob 5, 22-24. —. (2002). The Role of the ILO in Implementing Local Economic Development Strategies in a Globalized World, Unpublished paper, London School of Economics, London. Rodríguez-Pose, Andrés, Tijmstra, Sylvia (2009). On the emergence and significance of local economic development strategies: Department of Geography and Environment Spatial Economics, Research Centre, London School of Economics. Roggerson (2006). Small business development and poverty alleviation. Ruth Marjory Adhiambo Ocholah, Cainan Ojwang, Fredrick Aila, David Oima, Simeo Okelo and Patrick B. Ojera (2013). Effect of micro finance on performance of women owned enterprises, in Kisumu City, Kenya. Greener Journal of Business and Management Studies ISSN: 2276-7827 Vol. 3 (4). Shannon Doocy, Dan Norell, Shimeles Teffera, and Gilbert Burnham (2005). Outcomes of an Ethiopian Micronance Program and Management Actions to Improve Services: Journal of Microfinance, Volume 7 Number 1. Simon Peter Gregorio (2003). Local economic development: Stimulating Growth and Improving quality life, Philippines-Canada Local Government Support Program (LGSP). Sintayehu Desalegn (2013). Ethiopian Microfinance Sector: Growth, Socio-Economic Impacts and Challenges. Birritu No.115. Birritu is a quarterly magazine published by The National Bank of Ethiopia. Sowmyan Jegatheesan, Sakthi Ganesh, and Praveen Kumar S. (2011). Research Study about the Role of Microfinance Institutions in the Development of Entrepreneurs International Journal of Trade, Economics and Finance, Vol. 2, No. 4. Stewart R, van Rooyen C, Majoro M, de Wet T (2010). What is the impact of microfinance on poor people? A systematic review of evidence from sub-Saharan Africa: Social Science Research Unit Institute of Education 18 Woburn Square London W10 5 UJ United Kingdom. Tiruneh Abebe (2011). Analysis of the Success Factors of Micro and Small Business Enterprises in Addis Ababa: A thesis submitted to the School of Graduate Studies of Addis Ababa University in partial

Loan Provision by Microfinancing Institutions

247

fulfillment of the requirements for the Degree of Masters of Business Administration. UNDP (2009). Human Development Report: Overcoming Barriers: Human mobility development. New York, USA. —. (2012). Entrepreneurship development Programme in Ethiopia-Project document: Addis Ababa. USAID (2011). Assessing and Training a Local Economic Development (LED) Initiative: EGAT/UP and the Urban Institute: Agency for International Development. Walter Okibo, Bichanga & Lilian Aseyo (2013). Causes of Loan Default within Micro Finance Institutions in Kenya-Interdisciplinary Journal of Contemporary Research in business: Institute of Interdisciplinary business research Vol 4, No 12. Werotew, Bezabih, Assefa (2010). Entrepreneurship: An Engine for Sustainable Growth. Yigrem Kassa, March (2010). Regulation & Supervision of Microfinance Business in Ethiopia: Achievements, Challenges & Prospects. Zeller, M. and Meyer, R. (2002). The Triangle of Microfinance: Financial Sustainability, Outreach, and Impact, International Food Policy Research Institute: London.

 





CHAPTER NINE MICROFINANCE AS A TOOL FOR SOCIAL INCLUSION OF THE HEARING IMPAIRED ARTEM BOLTYENKOV

Abstract In developing countries, microfinance is the main source of financial services for poor people to improve their economic situation. It is a useful economic tool to help the poor out of poverty. Microfinance researches ways to provide small loans or savings opportunities at an affordable cost to individuals and small businesses that lack access to banking. This research focuses on how microfinance can help the hearing impaired afford aided hearing, and, in turn, increase the income and employment rates of the hearing impaired. The goal is to create social and financial inclusion of the hearing impaired in Ukraine and Russia, but the findings can be applied to other developing countries. The research method used in the study is the grounded theory approach.52 All data come from empirical research. It is produced by experiment, observation, and experience. The evidence used to conceptualize the theory is gained by conduction of interviews with the hearing impaired, consumer finance experts, hearing aid dispensers, hearing instrument and implant manufacturers, non-government organizations, and other experts. Analysing the data and learning from the experience of others in this field, certain conclusions were made. The main one is that though there are different ways to finance the purchase of hearing instruments or implants in developing countries, each one of them must include certain components of borrowing and saving to be successful. This research

 52

Glaser & Strauss, 1967.

Microfinance as a Tool for Social Inclusion of the Hearing Impaired

249

demonstrates that saving factors in microfinance are more suitable for the hearing impaired with high income and a high subjective time-value-ofmoney discount rate. At the same time, the hearing impaired with good prior hearing aid experience are more fit for borrowing.

1. Introduction Microfinance is a combination of financial services to poor and lowincome individuals in developing countries that do not have access to typical banking. It has established itself as a tool of choice for lifting people out of poverty. Unfortunately, microfinance has not yet received comparable attention from the scientific community as a means of helping people with disabilities live “normal” lives. To the author’s knowledge there is no existing literature on microfinance as a tool for social inclusion of the hearing impaired, except Boltyenkov (2015). This article is based on his work. Research and findings of Daher and Flessa (2010) and Lewis (2004) show that people with disabilities that try to get access to microfinance usually face rejections. Daher and Flessa (2010) conclude in their study: “… People who are prevented from working because they cannot afford a medical device have no access to the financial services offered on the market. Traditional banking services are not accessible to the poor, and microfinance organizations fear that these people cannot pay back their credit.”53

Lewis (2004) supports this finding: “Indeed, each of the few women, who had at some point applied for small business loans outside of disability organizations had been turned down, either explicitly on the basis of disability, or because she had been unable to produce the collateral required of disabled, but not of non-disabled, credit applicants.”54

The examples above reveal that disabled individuals face more difficulties trying to obtain microfinancial help. This research analyses methods for social inclusion of the hearing impaired through microfinancial services.

 53 54

Daher & Fessa, 2010, p. 195. Lewis, 2004, p. 34.

250

Chapter Nine

Speech is a vital part of human communication. In the course of a day we use it constantly. It is an important part of our lives and affects our relationships with people around us. Strong verbal skills can greatly improve our social life, make our work more efficient, help us develop and grow, make our life richer, and us happier overall. “Impaired hearing results in distorted or incomplete communication, leading to greater isolation and withdrawal and therefore lower sensory input. In turn, the individual's life space and social life becomes restricted. One could logically think that a constricted lifestyle would negatively impact the psychosocial well-being of people with hearing loss.” 55

Due to impaired ability to communicate with others, people with hearing loss cannot fully participate in social life. In developing countries, children with hearing loss seldom have the opportunity to receive quality education. Their disability prevents them from reaching their full potential. On the other hand, adults with hearing loss have less chance of finding gainful employment. They often experience frustration and social isolation which, in turn, creates mental and health problems such as: “… Embarrassment, fatigue, irritability, tension and stress, avoidance of social activities, withdrawal from social situations, depression, negativism, danger to personal safety, rejection by others, reduced general health, loneliness, social isolation, less alertness to the environment, paranoia, lessened ability to cope, and reduced overall psychological health. In view of this, few would disagree that hearing loss per se is a serious issue.” 56

Emotional, social, physical, and psychological problems are common for people with hearing loss. “The tragedy is that untreated hearing loss impacts the individual and his or her family for the rest of his or her life in the form of lost wages, lost promotions, lost opportunities, and unrealized dreams, not to mention lower income in their retirement.” 57

According to the World Health Organization we have around 360 million people worldwide with moderate to profound hearing impairment.58 People lose hearing with age, due to loud noises, as a result of illness,

 55

Kochkin, 2013, p. 1. Kochkin & Rogin, 2000, p. 10. 57 Kochkin, 2005, p. 9. 58 World Health Organization, 2013. 56

Microfinance as a Tool for Social Inclusion of the Hearing Impaired

251

tumours, head injuries, exposure to certain drugs etc. Though “half of all cases of hearing loss are avoidable through primary prevention”,59 cure of hearing loss depends greatly on the type and degree of hearing loss. Hearing aids are the most common treatment for individuals who suffer from hearing loss. The World Health Organization found that: “In developing countries, fewer than one out of 40 people who need a hearing aid have one. The lack of availability of services for fitting and maintaining hearing aids, and the lack of batteries, are also barriers in many low-income settings.” 60

In developing countries, the build-up of sound financial help to people with hearing loss is crucial. It can enable the hearing impaired to afford aided hearing and, in turn, improve their life quality. The Convention on the Rights of Persons with Disabilities is an international human rights treaty that protects and insures the rights and dignity of people with disabilities. It was adopted on December 13, 2006 by the United Nations. The eight guiding principles of the convention make sure that individuals with disabilities can fully participate in social life. “If development is about bringing excluded people into society, then disabled people belong in schools, legislatures, at work, on buses, at the theatre, and everywhere else that those who aren’t disabled take for granted. Unless disabled people are brought into the development mainstream, it will be impossible to cut poverty in half by 2015 or to give every boy or girl the chance to achieve a primary education by the same date, goals agreed to by more than 180 world leaders at the United Nations Millennium Summit in September, 2000.” 61.

2. Background 2.1. Conductive and Sensorineural Hearing Loss There are two main types of hearing loss:

 59

World Health Organization, 2013, p. 1. World Health Organization, 2013, pp. 5-6. 61 Wolfensohn, 2002.

60

252

Chapter Nine

x Conductive hearing loss occurs when there is disruption of sound transfer from the outer ear to the middle ear. Some possible causes are the middle ear infection, fluid in the middle ear from cold, perforated eardrum, allergies, excess of ear wax, malformation of the outer ear, malformation of ear canal or middle ear, debris in the ear, tumours of the middle ear such as a glomus tumour etc. Generally, it can be treated medically or surgically. For example, an ENT doctor can clean the outer ear from foreign bodies using an examining microscope, but in some instances, surgical removal is needed. Otitis media, a painful outer ear infection, can be treated successfully with antibiotics. Therefore, people with conductive hearing loss often do not need hearing aids as with proper treatment they can fully restore their hearing ability. x Sensorineural hearing loss occurs when there is a problem with generation or transmission of the nerve impulses from the inner ear to the brain. As the exact location of point of disruption is not yet possible, hearing loss linked to the sound sensation is called sensorineural hearing loss. As a rule, it cannot be medically or surgically corrected. Usually, sensorial hearing loss is an irreversible, permanent condition. Some possible causes are damage of the inner ear and/or the inner cells, head trauma, drugs, malformation of the inner ear, damage of the neural structures, exposure to loud noises etc. Sensorial hearing loss is the result of failure during the transformation process of vibrations, which enter cochlear, to neural patterns of excitation. The disorder of sound information can occur either in the cochlear, or in the brain itself. It changes the ability to hear quiet sounds and even reduces the quality of the sound. Most often, sensorineural hearing loss happens due to damage of hair cells on the basilar membrane. Once the cochlea hair cells are impaired, they cannot be restored. The hair cells may be damaged either at birth, or gradually lose their functionality with age. The most common reasons for hair cell damage are regular and prolonged exposure to loud noises, certain infectious diseases, aging, and deafness genes. Sensorial hearing loss is about four times more prevalent than conductive hearing loss; it ranges from mild to profound. A person with profound hearing loss will hear only loud noises. The common treatment for sensorineural hearing loss is the use of hearing instruments and implants.

Microfinance as a Tool for Social Inclusion of the Hearing Impaired

253

Sometimes people have problems in the outer or middle ear and in the cochlea or auditory nerve. This type of hearing loss is referred to as mixed hearing loss. Hearing aids can be very useful to people with mixed hearing loss. In this case, a hearing care professional must always exercise caution when assigning hearing aids, as a part of hearing loss can be due to an active ear infection. Therefore, the right diagnosis of a hearing loss is crucial for selecting the right treatment. Figure 1: Hearing loss

2.2 The Connection between Hearing Loss and Poverty “Hearing impairment is the most frequent sensory deficit in human populations, affecting more than 250 million people in the world”62. There

 62

Mathers et al., 2003, p. 1.

254

Chapter Nine

are a great number of people with an untreated hearing loss in developing countries due to a lack of funds for treatment. The impact of hearing loss on a person’s work and personal life is great. Hearing loss is not only a health issue, it affects quality of life. There is a certain link between hearing loss and poverty. Often poverty is considered to be the cause of hearing loss. In developing countries, for example, people live and/or work in environment with high-level noise pollution. It leads to noise-induced hearing loss. Besides, developing countries’ populations lack funds to protect themselves against rubella, measles, meningitis, and other illnesses that play a major role in acquiring hearing loss. A population with insufficient economic resources is not able to provide for sufficient health care.63 On the other hand, untreated hearing loss produces distorted and incomplete communication, impairs memory and ability to learn new things. Emotional distress, sadness, depression and anxiety associated with hearing loss have a direct impact not only on personal life, but also greatly affect professional life as well. All this, in turn, frequently reduces job performance and earning power. The correlation between poverty and hearing loss is illustrated by figure 2.

 63

Moeller et al., 2004.

Microfinance as a Tool for Social Inclusion of the Hearing Impaired

255

Figure 2: The vicious circle between poverty and hearing loss64

The main obstacle in adopting hearing aids is usually their price tag. In his research, Kochkin (2007) proves that even in developed countries such as US, the main reason the hearing impaired do not buy hearing aids is their price (figure 3). His study illustrates that: “Three out of four (76%) respondents mentioned financial constraints as a barrier to hearing aid adoption”.65 Furthermore, “64% indicated they cannot afford hearing aids, while nearly half (49%) indicated it is a definite reason why they don’t use them”66.

 64 65 66

Own illustration following Handicap International (2006), p. 16. Kochkin, 2007, p. 37. Kochkin, 2007, p. 37.

Chapter Nine

256

Figure 3: Reasons for non-adoption of hearing aids67

3. Qualitative Study of the Hearing Impaired in Developing Countries 3.1. Research Methodology Qualitative research uses an interpretive, naturalistic approach and studies phenomena in environments where they occur naturally.68 Besides, it looks at how the social experience is created and given meaning.69 In our case, qualitative research involves in-depth understanding of the hearing impaired behaviour in Ukraine and Russia, and the detailed analysis of their decision process. It also includes extensive inquiry into hearing prosthesis industry and technology, consumer finance, government health care policies, existing charitable organizations, etc. Besides, the qualitative research approach incorporates the collection of up-to-date information on psychological behaviour of the hearing impaired, their mental and health problems, their social status, educational level, and their surroundings. The result of such an approach is a better understanding of the hearing impaired conduct, their decision making process, and their perception of life situations and challenges.

 67 68 69

Kochkin, 2007, p. 34, fig. 4c. Denzin & Lincoln, 1994. Denzin & Lincoln, 2005.

Microfinance as a Tool for Social Inclusion of the Hearing Impaired

257

To form a theory from collected data, Glaser and Strauss (1967) suggest the grounded theory. The main goal of the grounded theory approach in our research is to generate the theory that will predict and explain the hearing impaired behaviour. Strauss and Corbin (1998) proposed to use techniques such as interviews, case study research, questionnaires, and observations to make the theory adequate. Van Maanen (1998) specifies the “flexibility and emergent character”70 and the need for “highly contextualized individual judgment”71 as important components of qualitative research. He argues that qualitative research is inductive and interpretive. Thus, this research is completed inductively by collecting data and analysing them until a strong theory has emerged.

3.2. Collection of Information and Research Frame Qualitative research involved an interactive process of data collection by means of observations, discussions, questionnaires, and interviews. Most interviews with the hearing impaired took place in developing countries, Ukraine and Russia. To gather more clear information, additional interviews with health care specialists, hearing instrument manufacturers, audiologists, and microfinance and consumer finance organizations were carried out in such countries as Germany, USA, Singapore, Zambia, Bangladesh, and India. To explore the hearing impaired behaviour more deeply, a two-week research trip to Ukraine was undertaken. Although the data came from different countries, the main focus remained on the hearing impaired that live in countries where healthcare reimbursement policies are not sufficient to ensure that the majority of the hearing impaired get hearing aids. This is exactly the case with Russia and Ukraine. The research empirical evidence was accumulated between October 2010 and April 2012. The primary data were gathered through observations, questionnaires, up-to-date articles and books on hearing impairment, and consumer finance and microfinance. Additionally, 47 hour-long interviews were conducted with different people in the hearing aid industry.

 70 71

Van Maanen, 1998, p. xi. Van Maanen, 1998, p. xi.

258

Chapter Nine

All empirical data were coded into several main categories. Using the grounded theory approach and evaluating those categories, our theory emerged and solidified.

3.3. Reasons behind the Reluctance of the Hearing Impaired to use Hearing Aids Our qualitative research showed that for successful financing of the hearing impaired’s social inclusion, the hearing impaired have to have a strong desire to integrate into the normal-hearing society with the help of hearing aids. This desire can be absent when a hearing impaired person does not see how aided hearing can improve her/his life. She/he cannot imagine how aided hearing can offer her/him more opportunities to find a better job with higher income or to make her/his social life richer. Often, the lack of interest in social inclusion with the help of aided hearing is the absence of skills or knowledge required to advance a career and enjoy a higher quality of life. In some instances, a hearing impaired individual does not even realize that she/he will benefit from better hearing as she/he can earn income and lead perfectly normal personal life despite her/his hearing loss. Some Ukrainian people with severe unaided hearing loss, for example, sell goods on the market and earn pretty good money. Their friends, relatives, or neighbours help them speak with customers and complete sales. Besides, such physical activities like house painting and renovation, carpentry, different types of construction and repair work, cleaning and many factory jobs can be done by people with unaided hearing loss. Hearing loss, in a lot of cases, does not prevent the hearing impaired to enjoy physical labour and make profit out of it. Still, ensuring safety in the work place and performing tasks that require verbal communication can be a real challenge for individuals with unaided hearing loss. In these instances, hearing aids help to acquire the missing knowledge, and to make work more effective and productive. The other reason why hearing impaired people are reluctant to use hearing aids is that they did not use their hearing sense as means of communication for a long period of time, sometimes never. They often feel left out of the normal-hearing society and usually are a part of the society that consists primarily from people with unaided hearing such as, for example, deaf society. They regularly use sign language and seldom speech to communicate with each other.

Microfinance as a Tool for Social Inclusion of the Hearing Impaired

259

For people that experience lack of hearing sense for a long time, a hearing aid adjustment period can be very demanding, not only physically but also psychologically. It is a complex process that involves a lot of hard work, as every day without normal hearing increases sensory deprivation of the human brain. When people have hearing loss and do not use hearing aids, their brain cells responsible for processing information coming from the hearing sense are not active. If a visually impaired person, even without prior experience, is fitted with glasses, she/he can immediately see the result. The person recovers her/his sight ability at once. In case of hearing loss, it is different. Brain cells need time to return back to normal once the hearing improves. This process takes weeks, sometimes months. The hearing impaired who lead a quiet secluded life and use hearing only occasionally, as they have already distanced themselves from the rest of the hearing society, are also not likely to have a strong desire to re-join it. Often, these are old people in their 70s and older. When hearing loss develops with age, they start to experience difficulties in communication and speech understanding with family, friends, and other people, and gradually withdraw from society. In the end, they spend almost all their time at home. In most cases, they do not have an interest in adoption of hearing aids. Mental health problems such as depression, paranoia, fatigue, etc. appear.72 They feel emotional turmoil and insecurity. These older people are often in a state of denial or simply too embarrassed to get help. As a result, they feel inadequate and have no desire to participate in social life activities. These people not only need proper aided hearing but also psychological guidance and advice. The last reason behind the reluctance of the hearing impaired to adopt hearing aids is a strong prejudice against hearing implants. Various health economics researchers demonstrate the improvement in quality of the hearing impaired life after the implantation.73 However, some people with profound to severe hearing loss, often deaf people, are strongly against implantation, even if they do not object to aided hearing in general. They usually associate themselves with a distinct deaf culture, and do not see hearing loss as a disability, rather as a variation of human reality. These people view their medical condition in a positive light, use sign language in everyday life, and do not feel the need to fix their hearing. Mostly, they

 72

Kochkin & Rogin, 2000. Cheng & Niparko, 1999; Wong et al., 2000; Niparko, 2009; Ruben, 2001; Joore et al., 2003; Keren at al., 2002; Mohr et al., 2000.

73

260

Chapter Nine

are opposed to cochlear implants. Besides, cochlear implantation is a difficult surgery where mistakes could lead to such serious complications as meningitis, tinnitus, infection, temporary or permanent pain, etc. In some cases, the hearing impaired may lose any remaining hearing in the implanted ear. There is also a risk that the implant may fail. The decision to undergo cochlear implant surgery is not easy and should be made with a realistic expectation and having weighed up all the odds. Anyway, the implant limitations and surgical risks still cause many hearing impaired people to hesitate before going through with the operation.

3.4. Effective Audibility Ching et al. (1998) offered to differentiate between two terms: audibility and effective audibility. Effective audibility implies that the sound should not only be audible, but also useful. It strives to achieve the best speech understanding by the hearing impaired. More audibility is not always better, and effective audibility refers to the level of amplification when the most helpful information gets extracted from the sound signal. Above this level speech understanding for a hearing impaired person will not improve, and no new information will be processed by the brain. Thus for a first time hearing aid user, a hearing aid should have a certain amplification level that allows more gain at frequencies where audibility is most useful. The effectiveness of audibility decreases with hearing loss. It usually does not make sense to give the incoming sound the level of amplification equal to hearing loss. For example, for people with profound hearing loss, a lot of amplification will not necessarily contribute to better speech understanding. When fitted with hearing aids for the first time, they will not be able to understand up to 70% of the incoming sound information. After a couple of months wearing a hearing instrument, the brain of the hearing impaired restores some ability to process incoming audial information. The need to adjust the amplification level of a hearing impaired person to her/his new audibility level arises.74 This process, together with relearning to listen and communicate efficiently, is called aural rehabilitation.

 74

Ching et al., 2001.

Microfinance as a Tool for Social Inclusion of the Hearing Impaired

261

Aural rehabilitation helps the hearing impaired to understand the limits of sound amplification, and minimize the negative effects of hearing loss on communication with other people.75 It is a combination of different procedures and therapies to help the hearing impaired to overcome the handicap. Aural rehabilitation is important for increasing hearing aid satisfaction, and addresses the communication needs of people with hearing loss. Unfortunately, aural rehabilitation services are rarely used, even in such developed countries as the US, and even less so in developing countries. Up to a point, time can serve as aural rehabilitation. It is essential for hearing aid adjustment and improvement of communication skills at work and in everyday life. The time needed for rehabilitation normally amounts to the time it took the hearing loss to reach its present level. For example, the average time between the start of hearing loss and hearing aids purchase is six years in the US. Therefore, the duration of rehabilitation in this case will take months to be complete and may even last up to six full years. For this reason, it is important to be fitted with hearing aids as soon as possible. When the hearing impaired people with sudden adult-onset hearing loss, regardless of the cause and degree of hearing loss, are treated early, they will have a very short rehabilitation period. In this case, adjustment to amplification and learning to hear again will not take long. The aural rehabilitation is often unnecessary as the brain has received hearing stimulation only recently and still remembers what it was like to hear perfectly. After all, all brain cells in charge of converting audio information are functional. If the brain has not received audio stimulation for a long time, it will reassign brain cells responsible for hearing to other tasks. In this situation, it will definitely take a while for the brain to readjust after the person is fitted with hearing aids. In some instances, for example for people who lost their hearing at an early stage of life, were not fitted with hearing aids and could not properly develop their speech (speech skills are built during the first four years of life76), hearing aids are useless as means of integration into the hearing society. Usually, these people communicate with each other using sign language and are fully assimilated with deaf and hard of hearing societies. The hearing aids for them can serve only one

 75 76

Abrahamson, 1997. Reiter et al., 2012.

262

Chapter Nine

purpose: the increase of safety due to a larger awareness of environmental sounds such as the sound of an approaching car.

3.5. Research Findings 1. Borrowing and saving or a combination of both should be used to help the hearing impaired to afford aided hearing. 2. To ensure that a person with hearing loss uses aided hearing constantly, she/he has to have a strong desire to be a part of a normal-hearing society. Dillon (2001) mentions the following: “Initial motivation to obtain hearing aids has been shown to be a key determinant of whether patients continue to use hearing aids. A patient’s motivation reflects the balance of all the advantages a patient expects hearing aids will provide, offset by all the expected disadvantages, irrespective of whether all these positive and negative expectations are realistic.”77

Other researchers 78 also came to the same conclusion. 3. To predict the willingness of the hearing impaired to adopt the hearing aids and integrate with normal-hearing people, the level of hearing loss measured objectively in dB is not sufficient. Dillon (2001) concludes that pure tone hearing loss: “… Is unreliable as a sole indicator of who will benefit from hearing aids, except in cases of normal hearing (no benefit) and severe hearing loss (substantial benefit).”79

4. To determine who of the hearing impaired will benefit the most from hearing instruments and implants, communication needs of the hearing impaired should serve as an indicator. If the hearing impaired have to utilize their communication ability daily, they are more likely to fit aided hearing in their lives. Thus, three key needs for differentiation between people with hearing loss were recognized:

 77

Dillon, 2001, p. 209. Brooks, 1990; Erdman et al., 1984; Gatehouse, 1999; Hickson et al., 1986; and Hickson et al., 1999. 79 Dillon, 2001, p. 215. 78

Microfinance as a Tool for Social Inclusion of the Hearing Impaired

263

a. The need to communicate over the phone; b. The need to lead a difficult conversation, such as conversation with a group of people or conversation in situations with loud background noise; c. The need to speak often with new people or strangers. They are not difficult to assess and it is easy to confirm the conclusions. If all three of these needs are present in the life of a person with hearing loss, she/he, in all likelihood, will be willing to integrate into the hearing society with the help of aided hearing. The last finding corresponds with Dillon (2001)’s discovery: “… The more a person is in contact with other people (or would like to be if poor hearing was not a disincentive) the more likely it is that the advantages of hearing aids will outweigh their disadvantages. A hermit with a loud TV set and radio will not need hearing aids!”80

However, Dillon (2001) also declares that hearing aids are most helpful “in places that are quiet, and where speech is at a soft level”81 and “if the primary need is to hear better in very noisy places, hearing aids may disappoint, irrespective of the degree of the patient’s hearing loss.”82 Fortunately, since he was published, hearing instruments were greatly improved. Now, they can successfully suppress background noise. Some hearing aids, for example, have microphones that automatically adapt once the sound information changes and maximize the contrast between the front sounds and sounds from the background. Besides, hearing aids manufacturers have developed numerous technologies to help solve the noise problem. These technologies include digital noise reduction, directional microphones, wind noise reduction etc.

3.6. Financing Hearing Aids Social inclusion of the hearing impaired into the normal-hearing society begins with the purchase of hearing instruments and/or implants. It can be done when a person with hearing loss has and uses her/his money to buy a

 80 81 82

Dillon, 2001, p. 219. Dillon, 2001, p. 217. Dillon, 2001, p. 217.

264

Chapter Nine

hearing aid, takes a loan or starts to save money to buy a needed device at a later date. Some hearing impaired people prefer to spend their own money to finance aided hearing. Others use collateralized loans, standard bank savings accounts, consumer loans, rent-to-own agreements, hire purchase, pawnshops, or a combination of financial services to access the funds. There is also an option to rent or lease a hearing aid. “During my research trip to Ukraine, I conducted an interview with the head of the organizational department of Ukrainian Deaf Society, Irina S.. In the course of our discussion, I have learned that in Ukraine, every five years, the hearing impaired with more than 75 dB hearing loss in both ears are eligible for a pair of free of charge hearing instruments from the government. In the year 2010, 282 people with severe hearing loss received such devices. Mostly, they were given very basic cheap analog instruments. The same year, the Ukrainian government started offering digital trimmer devices, sadly only basic ones. Since the hearing impaired with less than 75 dB of hearing loss in any of the ears are not qualified to receive government help, and those who are qualified can get only poor quality hearing instruments, many hearing impaired are left to deal with their hearing loss on their own. If a person has a profound hearing loss or even deafness in both ears, the Ukrainian government pays for one cochlear implant. In the year 2010, the official queue of the hearing impaired who wanted to get a free implant was 400 people. Unfortunately, only two people each year receive the implants from the government free of charge. At this rate, every new person with profound hearing loss in need of an implant should expect to wait around 200 years to get a free one!”83

From February 1, 2011, in Russia, the hearing impaired have had the right for partial lump-sum reimbursement of the costs related to the purchase of hearing aids.84 The size of the reimbursement depends on age of the hearing impaired, the region of residence, and the hearing loss in each ear. “During my interviews with the hearing impaired in Russia, I have discovered that many of them received reimbursement from the government in the range of 150-200 USD for hearing aid purchase per piece. This sum is sufficient to buy a very basic digital trimmer hearing instrument, but not enough to buy a computer-programmed digital hearing

 83 84

Boltyenkov, 2015, p. 65. Ministry of Health Care and Social Development of Russian Federation, 2011.

Microfinance as a Tool for Social Inclusion of the Hearing Impaired

265

instrument. It is worth mentioning that for children, the hearing instruments were often completely reimbursed.”85

Many people with hearing loss do not want to consider taking a loan for a hearing aid, as they are afraid they will not be able to pay it back. “In the middle of my research trip to Ukraine, I had a discussion with a hearing impaired woman, Olga. I met her at the Ukrainian Society of Deaf People. She is in her thirties, and has about 70 dB hearing loss in both ears. Several months prior to our meeting, she had bought a new hearing instrument for one ear (Oticon Sumo DM) for 24.000 Ukrainian Hryvnas, which is roughly equal to 2,950 USD. On the second ear, Olga still wears a 15-year-old analog instrument. She has suffered from hearing loss since she was 3 years old. As Olga was growing up in both hearing and deaf environments, she is able to speak with normal hearing people and is proficient in sign language. Olga is confident in her ability to work just as well if not better than normal hearing people do. Nevertheless, she fears she will not be able to get a better job due to rejections experienced in the past. She names stigma against the hearing impairment as the main reason why she has difficulty finding a well-paid job. Besides, Olga needs a new hearing instrument for the second ear, but does not have money to buy one. At the same time, she does not want to take a consumer loan for the purchase, because she fears not being able to make regular monthly payments.”86

Their fear is not groundless. The bank expects to be paid in a timely manner and without delays. Unfortunately, to see the benefit from aided hearing, one needs time. People often forget that in the case of hearing aids, a person with hearing loss usually needs a sufficient rehabilitation period. Some require weeks, even months to see the improvement in the quality of their work and family lives. Abrams et al. (1992) discovered that an aural rehabilitation program can significantly reduce the degree of perceived hearing handicap. Some people with hearing loss are ready to take a chance and finance their aided hearing through loans. “The rational hearing impaired will not take loans to buy hearing instruments or implants if they know they will not be able to make monthly payments out of their current income. Besides, individual borrowing in

 85 86

Boltyenkov, 2015, pp. 65-66. Boltyenkov, 2015, p. 66.

Chapter Nine

266

many developing countries requires collateral. Therefore, readiness to take a loan also depends on the availability of assets.”87

Others have the stable income that is sufficient to take on a loan, but they are still reluctant to buy hearing aids due to their previous bad experience with them. In these cases, they see the investment in aided hearing as a waste of money. There are also those people with hearing loss who have enough money to make regular loan payments and have positive prior experience with aided hearing. They are more likely to use loans in order to upgrade their hearing instruments and take advantage of the latest speech understanding technologies. Nowadays, technologies improve every day to provide outstanding listening to the hearing impaired. In cases when people with hearing loss have never used aided hearing, there is a high probability that they did not utilize their hearing ability properly for a long time. There is a risk that they will be on a long journey to recovery and require a long rehabilitation period. The qualitative research showed that it is not a sound idea to add a burden of interest and principal payment on the hearing impaired in this situation. Financing people with hearing loss through loans makes sense when they have positive prior experience with aided hearing, are not opposed to borrowing, and have necessary assets to use as collateral. Savings are another financial option for the hearing impaired to afford aided hearing. The saving ability of the hearing impaired depends greatly on their income, their consumption level, their time value of money, and their propensity to save. The saving amount needed takes into account existing government reimbursement programs and health insurance coverage. In many countries, these government programs exist only for people with profound hearing loss and deafness. For other hearing impaired people, the need to save or use other financial tools arises. Savings is a good option for people with hearing loss who have enough income left after deducting taxes and living expenses, are qualified for external help that covers a portion of aided hearing costs, and who have a

 87

Boltyenkov, 2015, pp. 66-67.

Microfinance as a Tool for Social Inclusion of the Hearing Impaired

267

personal discount rate for the time value of money that is significantly higher than 0.

4. Conclusion Hearing impairment is a complex disability that affects not only family and work relationships, but also health. It is not a harmless condition, and can cause isolation and depression. Aided hearing can significantly improve many spheres of the hearing impaired life. That is why it is especially important to succeed in the social inclusion of people with hearing loss through clever financial assistance. This assistance makes particular sense when the hearing impaired are willing to be a part of the normal-hearing society, and are ready and able to finance aided hearing themselves or with outside help. Not all people with hearing loss can equally benefit from aided hearing. The hearing impaired who utilize their communication skills daily in the workplace or at home speaking over the phone, holding group conversations and discussions, meeting a lot of new people, will gain the most. Least of all will benefit people with profound hearing loss or even deaf people who are a part of deaf societies, or people with hearing loss who do not communicate with normal-hearing people often. Savings are a good option for people with hearing loss who have good income, a high subjective time-value-of-money discount rate, and access to at least partial reimbursement. Borrowing is a better choice for the hearing impaired who are satisfied with their prior aided hearing experience, and are willing to borrow money and put up some of their assets as collateral.

References Abrams, H., Hnath-Chisolm, T., Guerreeiro, S., & Ritterman, S. (1992). The effect of intervention strategy on self-perception of hearing handicap. Ear and Hearing, 13(5), 371-377. Abrahamson, J. (1997). Patient education and peer interaction facilitate hearing aid adjustment. The Hearing Review, 4(1),19-22. Boltyenkov, A (2015). A Healthcare Economic Policy for Hearing Impairment. Wiesbaden, Germany: Springer Gabler. Brooks, D. (1990). Measures for the assessment of hearing aid provision and rehabilitation. British Journal of Audiology, 24(4), 229-233.

268

Chapter Nine

Cheng, A., & Niparko, J. (1999). Cost-utility of the cochlear implant in adults. Archives of Otolaryngology - Head and Neck Surgery, 125(11), 1214-1218. Ching, T., Dillon, H., & Byrne, D. (1998). Speech recognition of hearingimpaired liste¬ners: Predictions from audibility and the limited role of high-frequency ampli¬fi¬cation. Journal of the Acoustical Society of America, 103(2), 1128–1140. Ching, T., Dillon, H., Katsch, R. & Byrne, D. (2001). Maximizing effective audibility in hearing aid fitting. Ear and Hearing, 22(3), 212– 224. Daher, H., & Flessa, S. (2010). Microfinance as a tool for financing medical devices in Syria. An assessment of needs and a call for further research. Journal of Public Health, 18, 189-197. Denzin, N., & Lincoln, Y. (1994). Computer-aided qualitative data analysis. Thousand Oaks, USA: Sage Publications. —. (2005). The Sage Handbook of qualitative research. Third Edition. Thousand Oaks, USA: Sage Publications. Dillon, H (2001). Hearing Aids. Tarramurra, Australia: Boomerang Press. Erdman, S., Crowley, J., & Gillespie, G. (1984). Considerations in counseling for the hearing impaired. Hearing Instruments, 35(11), 5058. Gatehouse, S. (1999). Glasgow Hearing Aid Benefit Profile: derivation and validation of a client-centered outcome measure for hearing aid services. Journal of the American Academy of Audiology, 10(2), 80103. Glaser, B., & Strauss, A. (1967). The discovery of grounded theory: Strategies for qualitative research. Chicago, USA: Aldine Publishing Company. Handicap International (2006). Good Practices for the Economic Inclusion of People with Disabilities in Developing Countries. Lyon, France: Handicap International. Hickson, L., Hamilton, L., & Orange, S. (1986). Factors associated with hearing aid use. Australian Journal of Audiology, 8(2),37-41. Hickson, L., Timm, M., Worrall, L., & Bishop, K. (1999). Hearing aid fitting: outcomes for older adults. Australian Journal of Audiology, 21(1),9-21. Joore, M., Brunenberg, D., Chenault, M., & Anteunis, L. (2003). Societal effects of hearing aid fitting among the moderately hearing impaired. International Journal of Audiology, 42(3), 152-160.

Microfinance as a Tool for Social Inclusion of the Hearing Impaired

269

Keren, R., Helfand, M., Homer, C., McPhillips, H., & Lieu, T. (2002). Projected cost-effectiveness of statewide universal newborn hearing screening. Pediatrics, 110(5), 855–895. Kiessling, J. (2008). Versorgung mit Hörgeräten. In: J. Kiessling, B. Kollmeier, & G. Diller (Eds.), Versorgung und Rehabilitation mit Hörgeräten (pp. 59-130). Stuttgart, Germany: Thieme Verlag. Kochkin, S. (1993). MarkeTrak III: Higher hearing aid sales don’t signal better market penetration. The Hearing Journal, 46 (7), 47-54. —. (2005). The Impact of Untreated Hearing Loss on Household Income. Better Hearing Institute. Retrieved from http://www.hearing.org/uploadedFiles/Content/impact_of_untreated_h earing_loss_on_income.pdf —. (2007). Marke Trak VII: Obstacles to adult non-user adoption of hearing aids. The Hearing Journal, 60(4), 24-51. —. (2010). Marke Trak VIII: The efficacy of hearing aids in achieving com-pensation equity in the workplace. The Hearing Journal, 63(10), 19-26. —. (2013). The Impact of Treated Hearing Loss on Quality of Life. Better Hearing Institute. Retrieved from http://www.betterhearing.org/aural_education_and_counseling/articles _tip_sheets_and_guides/hearing_loss_treatment/quality_of_life.pdf Kochkin, S., & Rogin, C. (2000). Quantifying the obvious: the impact of hearing aids on quality of life. The Hearing Review, 7(1), 8-34. Lewis, C. (2004). Microfinance from the point of view of women with disabilities: lessons from Zambia and Zimbabwe. Gender and Development, 12(1), 28-39. Mathers, C., Smith, A., & Concha, M. (2003). Global Burden of Hearing Loss in the Year 2000, Working Paper, World Health. Retrived from http://www.who.int/healthinfo/statistics/bod_hearingloss.pdf Ministry of Health Care and Social Development of Russian Federation (2011) Ɉɛ ɍɬɜɟɪɠɞɟɧɢɢ ɩɨɪɹɞɤɚ ɜɵɩɥɚɬɵ ɤɨɦɩɟɧɫɚɰɢɢ ɡɚ ɫɚɦɨɫɬɨɹɬɟɥɶɧɨ ɩɪɢɨɛɪɟɬɟɧɧɨɟ ɢɧɜɚɥɢɞɨɦ ɬɟɯɧɢɱɟɫɤɨɟ ɫɪɟɞɫɬɜɨ ɪɟɚɛɢɥɢɬɚɰɢɢ ɢ (ɢɥɢ) ɨɤɚɡɚɧɧɭɸ ɭɫɥɭɝɭ, ɜɤɥɸɱɚɹ ɩɨɪɹɞɨɤ ɨɩɪɟɞɟɥɟɧɢɹ ɟɟ ɪɚɡɦɟɪɚ ɢ ɩɨɪɹɞɨɤ ɢɧɮɨɪɦɢɪɨɜɚɧɢɹ ɝɪɚɠɞɚɧ ɨ ɪɚɡɦɟɪɟ ɭɤɚɡɚɧɧɨɣ ɤɨɦɩɟɧɫɚɰɢɢ. ɉɪɢɤɚɡ ɨɬ 31 ɹɧɜɚɪɹ 2011 ɝ. N 57ɧ. Retrieved from http://r41.fss.ru/31291/42749/index.shtml Moeller, J., Schmidt, C., & Lasser, C. (2004). Gesundheit der Ökonomie und Ökonomie der Gesundheit: Eine kritische Auseinandersetzung mit dem WHO-Bericht „Macroeconomics and Health“. Journal of Public Health, 12(1), 3-9.

270

Chapter Nine

Mohr, P., Feldman, J., Dunbar, J., McConkey-Robbins, A., Niparko, J., Rittenhouse, R., & Skinner, M. (2000). The societal costs of severe to profound hearing loss in the United States. International Journal of Technology Assessment in Health Care, 16(4), 1120-1135. Niparko, J. (2009). Cochlear implants: principles & practices, Philadelphia, USA: Lippincott Williams and Wilkins. Reiter, R., Pickhard, A., & Brosch, S. (2012). Periphere Hörstörungen und Spracherwerb. Laryngorhinootologie, 91(9), 550-559. Ruben, R. (2001). Redefining the survival of the fittest: communication disorders in the 21st century. Laryngoscope, 111(6), 1115-1116. Strauss, A., & Corbin, J. (1998). Grounded theory methodology: An overview. In N. K. Denzin & Y. S. Lincoln (Eds.), Strategies of qualitative inquiry (pp. 158–183). Thousand Oaks, USA: Sage Publications. Van Maanen, J. (1998). Qualitative studies of organizations. Thousand Oaks, USA: Sage Publications. Wolfensohn, J. (2002, December 3). Poor, Disabled and Shut Out. The Washington Post. Retrieved from http://bvs.per.paho.org/texcom/cd048370/disabled.pdf Wong, B., Hui, Y., Au, D., and Wei, W. (2000). Economic evaluation of cochlear implantation. Advances in Oto-Rhino-Laryngology, 57, 377– 381. World Health Organization. (2013). Deafness and Hearing Impairment. Fact Sheet N. 300. Retrieved from http://www.who.int/mediacentre/factsheets/fs300/en/



PART IV FINANCING SMALL ENTERPRISES



CHAPTER TEN BANK LOAN AND ITS IMPACT ON GROWTH OF SMALL ENTERPRISES: EMPIRICAL EVIDENCE FROM NORTHERN ETHIOPIA AREGAWI GEBREMICHAEL

Abstract Purpose: The small enterprises sector has been recognized as an engine of economic growth and poverty reduction due to its role in generation of employment and income. However, their growth remained below expectations due to different external and internal problems. This study seeks to investigate the following interrelated objectives. First it examines contributions of bank loans in the development of the micro and small enterprises sector, which is considered as one of the springboards of the Growth and Transformation Plan (GTP) of the Ethiopian government. Second, it investigates if adequate access to bank loans has significant influence on the growth of small enterprises. Methodology: Cross sectional primary data were collected from 333 small enterprises operating in six major towns of the National Regional State of Tigray, Northern Ethiopia, using a structured questionnaire. While the dependent variable of the study was growth of small enterprises, the explanatory variables are financial difficulty of small enterprises and access to bank/microfinance loans, controlling other variables. Qualitative and quantitative techniques were applied for data analysis. A multiple regression model was used to empirically test the literature-driven hypotheses. Findings: Findings of the study indicate that banks and microfinance institutions contributed only 10% of the initial capital of small enterprises.

Bank Loan and its Impact on Growth of Small Enterprises

273

The econometric result shows that access to adequate financial capital has a positive significant impact on growth of small enterprises. While the financially constrained firms grow only at 6.6%, those firms without financial difficulty grow at 8.83% (significant at 10%). Further, small enterprises that used bank loans to finance their operations grow at 9.41%, but the equity-financed counterparts grow at 5.98% (significant at 5%). Conclusion and Implication: The main policy implication of this study is that the government of Ethiopia should work hard to meet the credit needs of the MSE sector for speedy economic growth of the nation.

1. Background of the Study The micro and small enterprises sector has been considered by academics and policy makers as an engine of economic growth, poverty reduction, and social development due to its effect on employment and income generation, import substitution, its role as a springboard to entrepreneurship and industrialization, input distribution for large industries and distribution of their products through linkage and subcontacting, and income distributions among different sections of the society (Bekele and Worku, 2008; Kabongo and Okpara, 2009). For instance, the sector takes 48% of the labour force in North Africa, 51% in Latin America, 65% in Asia, 72% in Sub-Saharan African Countries (ILO, 2002). According to the Ethiopian Central Statistical Authority (2004), almost 50% of all new jobs created in Ethiopia are attributable to the MSE sector. According to Aregash (2005), cited in Bekele and Worku (2008), 98% of business firms in Ethiopia are micro and small enterprises, out of which small enterprises account for 65% of all businesses. The report by the FeMSEDA, released in April 2013, indicated that the MSE sector created 1.5 million new job opportunities, and about 4 billion birr worth of loans were provided by microfinance institutions during the years 2006-2010. Recognizing the significance of this sector as a key for rapid economic development, the Government of Ethiopia issued a Micro and Small enterprises Strategy in FDRE, MTI (1997). Besides, the Growth and Transformation Plan (GTP) of Ethiopia has envisaged the promotion of micro small enterprises as one important tool of poverty reduction (FDRE, MoFED, 2010). To this end, the government has established several agencies and offices that support the operation and development of micro and small enterprises: the Federal Micro and Small Enterprises

274

Chapter Ten

Development Agency (FeMSEDA) at federal level; Regional Micro and Small Enterprises Development Agencies (ReMSEDA) at regional levels, and Wereda (district) small enterprises promotion offices. In addition to creating an enabling institutional and policy environment, the government has also developed different assistance packages so as to streamline the provision of credit, vocational training, technical and managerial advice, specific business tailored short-term training to the operators (owners) and their employees, facilitating marketing services and forging marketing linkages, and development of appropriate infrastructures. Despite their contributions to economic growth, small enterprises in the world in general and those in less developed countries like Ethiopia face different barriers both at the time of their start-up and in operation. Lack of adequate finance, especially inadequate access to bank loans, was cited as one of the pressing problems to the growth of SEs. Previous literature (e.g. Rosmary, 2001; Kavanamur, 2002 cited in Bekele and Worku, 2008) reported that formal financial institutions are reluctant to lend money to small-scale enterprises due to the associated high risk. As a result of these constraints, small enterprises in developing countries, including Ethiopia, reported a shortage of financial capital to be the most critical bottleneck for their survival and growth (Goldmark & Nicher, 2009; Mulu, 2008; Bekelle & Worku, 2008; Ageba & Amaha, 2006a; Ageba & Amaha, 2006b; Beccetti &Trovato, 2002). The researcher observed that most of the related reviewed studies in Ethiopia (e.g. Habtewold, 2005; Negash, 2006; Geeyesus, 2007; Bekele & Worku, 2008; Mulugeta, 2008; Mohamodnur, 2009; Beyene, 2010; Tezera, n.d.) were mainly descriptive in nature and focused on assessing the status of the sector, identifying its challenges and opportunities. Findings of earlier research in Ethiopia are not only inconsistent and contradictory in identifying the critical challenges of small enterprises, but also none of them explained how and to what extent that growth was explained by the stated business constraints. Therefore, the primary purpose of this study is to empirically investigate the contribution of bank loans in financing small enterprise sector, and to what extent growth of the sector is affected by access to credit. As part of the PhD paper of the author, this study examines additional explanatory variables that were either not considered or might have been tested separately in earlier studies. For example, entrepreneurial orientation, motivational factors, human capital (level of education and prior start-up experience), location, size and age of enterprise, source of finance (mainly

Bank Loan and its Impact on Growth of Small Enterprises

275

access to formal credit) were incorporated into one regression to test the effect of bank loans or access to credit on the growth of small enterprises in the study area.

2. Literature: Theories and Empirical Evidence 2.1 Capital Structure Theories and Growth of Small Enterprises Financial capital is one of most liquid assets/resources that can be converted easily into other types of assets. Small enterprises need finance to invest in new productive activities, enter into new markets, develop new products, engage in innovative activities through research and development, cope with temporary cash flow shortage, and modernize and expand their business (Wiklund, Patzel & Shephered, 2009). These activities enable firms to expand and enhance firm growth. However, growth of small enterprises has been constrained by limited access to formal financial resources, especially bank credit (Ageba & Amaha, 2006a; Negash, 2006). The subject of capital structure choice (capital structure) and firm growth and performance lies at the centre of attention in finance. Firms should determine their optimum mix of debt and equity capital, defined as capital structure, in order to maximize return to owners and enhance their ability to deal with the competitive environment. According to Van Praag (2003), financial capital includes debt and equity. This is known as capital structure. It has been pointed out that the most relevant capital structure theories that explain the capital structure of small and medium enterprises (SMEs) are those related to static trade-off, adverse selection and moral hazard (agency theory), and the pecking order theory. Andree and Kallberg (2008) point out that the genesis of modern capital structure theory lies in the work of Modigliani and Miller (1958) in their famous proposition I – often referred to as the “irrelevance theorem”. The theorem suggests that, under certain perfect market assumptions, such as absence of taxes, bankruptcy costs, agency costs and asymmetric information, the value of the firm is unaffected by how the firm is financed. This implies that the choice of capital structure does not affect a firm’s market value. It is the assets of a firm that determine the value of the firm, not the way by which these assets are financed. The initial perfect market assumptions, on which the 1958 theory of Modigliani and Miller was based, were later reviewed in 1963 with the introduction of the tax benefits of debt. This is attributed to the fact that a perfect market does not

276

Chapter Ten

exist in the real world. Since interest on debt is tax-deductible, thereby creating tax savings for the borrower, it becomes possible for firms to minimize their costs of capital and maximize shareholders’ wealth by using debt. This is known as the leverage effect of debt (Modigliani and Miller 1963). According to Miller and Modigliani (1963), a firm should have 100% debt in its capital structure. This way the firm can take absolute advantage of the tax shield. Scott (1972) and Kraus and Litzenberger (1973) point out that theoretically a 100% tax shield does not exist in reality because of distress costs. Therefore, the optimization of capital structure involves a trade-off between the present value of the tax rebate associated with a marginal increase in leverage, and the present value of the costs of bankruptcy. In general the capital structure of a firm could be explained by four theories: the static trade-off theory, the pecking order theories, agency theory, and growth cycle theory. According to the static trade-off theory, optimum capital structure could be determined through optimizing the trade-off between the benefits and costs associated with debt financing. Static trade theory, one of the dominant capital structure theories, argues that as a firm’s capital structure has both benefits and costs, a firm can borrow up to the point where the tax benefit from an extra debt is exactly offset by the cost that comes from the increased probability of financial distress. Debt benefits include tax shields (saving) advantage induced by the deductibility of interest expenses from pre-tax income of the firm (Modigliani & Miller, 1963). On the other hand debt has both direct and indirect bankruptcy costs. While direct costs are those costs associated with periodic interest and principal payments, default and bankruptcy costs arise when periodic payment obligations increase. The probability of bankruptcy increases with the debt level since it increases the fear that the company might not be able to generate earnings to pay back the interest and principal of the loan. The costs of bankruptcy may be direct or indirect. For example, the direct bankruptcy costs are the legal and administrative costs incurred in the bankruptcy process. The indirect bankruptcy costs are the loss in profits incurred by the firm as a result of the refusal of stakeholders to do business with them (Titman, 1984). According to the trade-off theory (or theory of optimum leverage), the cost of debt is less than the cost of equity because of differences in associated risks and costs. Creditors’ funds are less risky than owners’ funds because (i) creditors have fixed (known) preferential rights on their claims, and (ii)

Bank Loan and its Impact on Growth of Small Enterprises

277

claims of creditors are legally protected and secured by collateral. The cost of debt is less than the cost of equity, due to the tax deductibility of periodic interest payments. Thus, according to trade of theory, the use of leverage can increase the rate of return to equity. Although excessive leverage can also be harmful because acquiring too much debt may subject enterprises to financial risk due to the variability in interest rates and net income. Therefore, the owners of small enterprises must weigh the tradeoff between debts and own saving (equity capital) and determine an optimum mix of debt and equity capital to efficiently operate and grow.

2.2. Theories on Growth of Small Enterprises Though there is no universally accepted theory on the determinants of growth of SEs, (1) stochastic theory, (2) learning model, (3) industrial organization (IO) model, and (4) resource based theory are some of the commonly discussed theories. The theoretical framework of this paper is the resource-based view. According to the resource-based model, differences observed in firms’ performance are primarily due to the heterogeneous distribution of unique resources and capabilities across firms, rather than to the characteristics of the industry (Barney, 1991). There has been much debate in the management strategy literature as to the source of sustained growth of firms and questions, such as why firms differ in their growth rate (some firms grow while others remain inactive or even dissolve) and how they choose strategies, have been raised. From 1960 up to the 1980s, the external environment was considered the primary determinant factor for the successful achievement of organizational objectives (Hitt et al. 2009). According to this approach, which is referred to as the industrial organization (I/O) model of aboveaverage return, the external environment or industry in which a company operates dictates a firm’s strategic actions, rather than internal factors. The IO model has four underlying assumptions (Hitt et al. 2009). First, the external environment imposes pressures and constraints that determine the strategies that would result in above-average returns. Second, most firms in the industry control similar strategically relevant resources and pursue similar strategies in light of those resources. Third, those resources are highly mobile across firms so that any resource heterogeneity across firms that may lead to sustainable advantage will disappear due to their mobility characteristics because others can acquire or learn the resources. Fourth,

278

Chapter Ten

organizational decision makers are assumed to be committed to the profit maximization objectives of the firm. In the 1990s, the focus of researchers shifted from the industry approach to the resource-based approach with regard to the source of growth of firms (Bridoux, n.d). As a result of this, the resource-based view (RBV) has become one of the dominant modern approaches to the analysis of sustained competitive advantage and growth of firms. Edith Penrose was one of the first scholars to recognize the importance of resources to a firm’s competitive position in 1959. Aside from Penrose (1959), Rubin (1973) is argued to be one of the few scholars to conceptualize firms as resource bundles prior to the formal origins of the resource-based view (Newbert, 2007). According to the resource-based theory, firms are heterogeneous in terms of the strategic resources they own and control, and these internal resources are recognized as the fundamental determinants of competitive advantage and performance (Barney, 1991). In 1991, Barney presented a comprehensive framework to identify the required characteristics of firm resources in order to generate sustainable competitive advantage. According to Barney (1991), for a firm to enjoy a competitive advantage, the firm’s resources must have four attributes: valuable, rare, inimitable, and non-substitutable, which are briefly described in the following paragraphs.

2.3. Empirical Evidence on the Effect of Finance on the Growth of Small Enterprises Research findings show mixed results on the effect of finance on the growth of small enterprises. Findings of Beccetti & Trovato (2002), Tushabomwe-Kazzoba (2006), Ishengoma & Kappel (2008), and Wiklund & Dess (2005) show strong evidence that loan and internal finance are important factors in stimulating the growth of small firms. Goldmark and Nichter (2009), on the other hand, argue that credit access is not a significant determinant of firm performance. They said: “Like many of the factors discussed in this article, access to finance may be necessary, but it is not a sufficient condition for MSE growth” (Goldmark & Nichter, 2009: 1456-1457).

Bank Loan and its Impact on Growth of Small Enterprises

279

Besides, McPherson and Rouse (n.d.) and Masakur et al. (2009) show that access to credit was not found to be a critical determinant of firm performance. As earlier researchers could not come to a consensus on the effect of finance on the growth of small enterprises, the researcher intended to investigate to what extent firm growth is affected by firm specific physical resources, with specific reference to financial condition and access to credit, following the trade-off theory of capital structure and the resource based view of firm growth. He hypothesized that a strong financial condition and adequate access to credit significantly and positively affect the growth of small enterprises.

3. Research Methodology 3.1. Source of Data, Data Collection Techniques, and Sampling To empirically investigate the effect of the financial condition and access to credit on the growth of small enterprises, using a structured questionnaire, the researcher collected primary data from 333 small enterprises operating in six major towns of the Tigray regional state Mekelle, Wukro, Adigrat, Adwa, Axum, and Endasilassie-Shire - which were selected based on their SE population and active business operation compared to others. According to the reports of the Micro and Small Enterprise Agency of the region, there were 2,765 SEs, and 354 sample respondents were selected using systematic random sampling and a programmed sample size calculator88 at a 5% significance level.

3.2 Measures of Variables and Data Analysis There is no agreement in the existing literature on how firm growth should be measured, and different authors have used a variety of measures, such as change in total assets or investment capital, sales, employees, profits, and market share (Davidson & Wiklund, 2000). Because it is more objective and more easily accessible than other measures, change in employment was used to measure the growth of small enterprises, as the dependent variable. The second commonly used growth indicator is the change in number of employees (Evans, 1987; Mead, 1994; McPherson,

 88

http://www.surveysystem.com/sscalc.htm

280

Chapter Ten

1996; Liedholm & Mead, 1999; Liedholm, 2002). Delmar (2006) and Davidson et al., (2005) observed that 29.1% of the reviewed studies used employment as a growth indicator of firms, just after sales (used by 30.9% of the studies). Strength of employment as the best measure of growth is justified because: (i) it is easily accessible data that is easily remembered by small enterprises (McPherson, 1996; USAID, 2002). Since most owners of small enterprises do not keep records, they would be unable to remember and accurately report their firms’ historical sales level, the number of employees should be used as a growth measure, (ii) unlike sales, employment is not sensitive to change in inflation or exchange rates (USAID, 2002; Wiklund & Shepherd, 2005), (iii) it is the preferred measure when the interest of policy makers is fostering employment growth (USAID, 2002; Davidson et al., 2005). Therefore, the researcher prefers to use employment to measure growth of small enterprises, (iv) Penrose (1959; in Delmar et al., 2003) suggests to use employment as a measure of growth to be applied for a resource and knowledge-based view of the firm, (v) Moreover, studies found that growth in sales and growth in the number of workers are highly correlated. For example, McPherson (1996) said that estimates using employment figures are similar to those using sales. Delmar and Wiklund (2008) found growth measured in terms of sales and employment showed the same result. This may be justified for the fact that the higher the sales, the higher the profit will be. Higher profit in turn would result in higher retained earnings and higher expansion of facilities and activities, which demands a higher number of employees. In a cross-sectional research, if a researcher uses first and last year approaches he or she would miss real fluctuations, which may result in a weak model or misspecified results and interpretations. In order to minimize such errors and obtain a higher fit, Delmar (1997) and Evans (1987) advised to apply the logarithm of the dependent. Accordingly, the growth rate used in this study was measured as the logarithmic change in employment between the date of establishment and the date/time of survey. Controlling other internal and external factors that may affect growth (such as educational attainment of owners, firm age, prior work experience of owners, gender of owners, entrepreneurial orientation, motivation, government policies and support services, BDS, infrastructure, etc.), the explanatory variables of this study were the financial condition (whether the firm had any financial constraint or not) and access to bank loans.

Bank Loan and its Impact on Growth of Small Enterprises

281

In this study the researcher applied descriptive statistics, statistical difference tests, and regression analysis for the purpose of data analysis. The model: emgrr = ln(noemp1/noemp0)/entage =ȕ0 + ȕ1owedle + ȕ2owedle2 + ȕ3owexpc + ȕ4findiff + ȕ5loctn + ȕ6entage + ȕ7entage2 +ȕ noemp0 + ȕ9capam0 + ȕ10 avoaeo +ȕ11avomot +ȕ12sectr + ȕ13ageow +ȕ14caplst + ȕ15mrk+ ȕ16genow + ȕ17avinfr + ȕ18avgovss+ ȕ19avbds+ H where: ¾ emgrr = log of change in number of employees at two points in time (beginning and survey time) in percentage, ¾ entage = enterprise’s age in years, ¾ owedle = owners’ years of schooling, ¾ owexpc = category of owners’ prior work experience (1 = had prior work experience; 0= no work experience), ¾ findiff = financial condition of SEs (1= had financial constraints, 0 = no financial problem), ¾ loctn = location of SEs (1 = far from commercial district and else = 0), ¾ noemp0 = initial number of employees, ¾ capam0 = initial amount of capital (size in initial capital), ¾ avoaeo = average EO, ¾ avomot = average motivation, ¾ sectr = sector of SE (1 = Manufacturing, else = 0), ¾ agow = age of owners in years, ¾ caplst = capital structure (debt equity ratio), ¾ mrk = average of market related problems, ¾ genow = gender of owners (1= male; else = 0), ¾ avinf = average of access and cost of infrastructure, ¾ avgovss = average government policies and strategies, ¾ avbds = average BDS, and ¾ H = error term.

4. Discussion and Analysis of Results Financial capital is the most liquid asset that can be relatively easily converted into other types of resources, i.e., it is with the help of finance

282

Chapter Ten

(especially cash and cash equivalent) that every firm conducts its transactions. For example, financial capital provides resource slack, allows firms to undertake innovative projects, expand their business, and acquire new assets such as plants and equipment that might not be possible for a financially constrained firm. However, growth of small enterprises in the world in general and in developing countries in particular has been challenged by greater financial constraints than that of larger firms because of a bias of lenders against small firms, market imperfections, the underdeveloped nature of financial markets of developing countries, inability of SEs to fulfil such requirements as collateral, business plans, and financial statements that banks and financial institutions require when they extend credit to borrowers. As a result of these and other constraints, small enterprises start their business using their meagre personal savings and informal sources such as loan from families and relatives and moneylenders. The current study tried to examine growth of small enterprises vis-à-vis the following issues: (i) size and source of initial capital, (ii) financial positions of SEs, and (iii) the adequacy of bank loan (credit) extended to this sector.

4.1. Effect of Size and Sources of Initial Capital on Growth of Small Enterprises Due to the aforementioned constraints, small enterprises in developing countries rarely apply for and receive bank loans. Though microfinance institutions are considered an important source of capital to small enterprises, their outreach and loan size are very limited. Findings of this paper and empirical evidence (e.g. Beck and Demirguc-Kunt, 2006) demonstrated the contribution of formal financial institutions in financing initial investment of small firms to be negligible. As a result of these problems, small enterprises in developing countries, including Ethiopia, rely on their meagre personal sources and other informal sources such as trade credit, family sources, and moneylenders. Respondents of this paper reported that due to the reluctance of formal financial institutions and other internal reasons they had been forced to start their current business with very small amounts of capital. The average amount of initial capital was found to be Birr 73,879 (equivalent to approximately US $3,940, at the current exchange rate of Birr 18.75 per US dollar). When the SEs are disaggregated by different categories of initial capital, about 64% of them started with less than Birr 50,000, while 30% of them had initial capital ranging from Birr 50,001-250,000 (US$ 2,667 to US$13,333) and only 6%

Bank Loan and its Impact on Growth of Small Enterprises

283

had initial capital of above 250,000 Birr. It was not only the beginning capital that was too small, but the role of banks in financing this little amount was negligible. It was found that out of the 333 small enterprises, 203 SEs (61%) financed their initial investment using a single source, mainly own savings, and the share of multiple sources (combination of two or more than two of the aforementioned sources) accounts for 39% of the initial investment. The proportion of SEs that used a single source and their related growth rate is depicted in the table below. Proportion of single sources and related growth rate Category of Single Source

No of SE (percent)

Mean Growth rate

Std Dev

Min growth rate

Max growth rate

Own saving Family Bank loan Others* Total

125 (62%)

6.06%

0.1111074

-0.0229

0.7611

55 (27%) 21 (10%) 2 (1%) 203 (100%)

6.34% 13.27% 7.70% 7.85%

0.1191449 0.2024050 0.1333962

-0.0785 0 0

0.4621 0.7324 0.2310

* includes trade credit, lease financing etc.

Of those 203 SEs that used a single source, initial investment of the 125 small enterprises (62%) had been financed from personal savings of the owners, while 28% of the initial investment was financed from informal sources. Banks and microfinance institutions contributed only 10% of the initial capital, which is similar to findings of earlier writers. For example, Carpenter and Petersen (2002), cited in Fatoki (2011), found that growth of SMEs was constrained due to their reliance on internal finance. The study of over 14,000 MSEs in Mexico also showed that owners mostly used their own resources and savings (61%) or those of their family and friends (14%) to launch their firms due to reluctance of formal financial institutions (Goldmark and Nicher, 2009). A firm with very limited alternative sources of capital is less likely to engage in innovative and expansion activities, which in turn may limit its growth opportunities. In order to substantiate this, the researcher compared the growth rate of those single source SEs with that of multiple source

284

Chapter Ten

ones and found that the growth rate of the former SEs is 1.78% less than that of the latter (6.39% against 8.17%). His quantitative models (OLS) and t-tests also provide him enough evidence to accept his research hypothesis that single source SEs grow less rapidly than those SEs with multiple sources (P < 0.10). This implies that as wider alternative sources of capital enable small enterprises to achieve better growth opportunities, policy makers and other stakeholders need to introduce accessible and broader alternative sources of capital to this sector so that it can play its expected role in income and employment generation.

4.2. Relationship between Financial Position and Growth of Small Enterprises In this paper the researcher used the financial position to refer to the financial condition of the enterprises, that is, whether or not they had faced any financial constraint. As discussed in the literature chapter, empirical evidence showed that growth of SEs has been hindered by acute financial trouble. However, it is unclear whether lack of adequate financial capital represents a critical constraint for the growth of small enterprises. In order to substantiate this, the researcher asked the respondents if they had ever faced any financial difficulty both at the start-up time and any time after establishment. Accordingly, out of the total respondents 261 SEs (78%) indicated that they had been exposed to severe financial difficulty due to lack of adequate credit from banks/micro finance institutions, and only 72 SEs (22%) were found to be capital self-sufficient. Inability of SEs to fulfil the requirements of banks (collateral, business plan, financial statements), lack of information, and inadequacy of loan amount received were cited as the most critical causes that hinder them from obtaining credit from these institutions. Though this is consistent with the previous descriptive literature, it does not indicate to what extent growth of small enterprises had been constrained as a result of such financial shortages. Therefore, the next question that must be answered is whether growth was delayed due to this financial constraint. For this purpose the researcher applied descriptive statistical analysis, multiple regression model, and ttests to examine if the growth rate of small enterprises is affected by their financial condition. In order to capture the influence of the financial position on the growth of small businesses, two dummy variables were used (1 represented SEs with no financial shortage and 0 SEs with financial shortage). The following table shows that while the financially self-reliant firms grow at 8.83%, those SEs with financial difficulty grow

Bank Loan and its Impact on Growth of Small Enterprises

285

at 6.6%, which implies that financial difficulty slows down the growth rate of small enterprises by 2.23%. Employment Growth Rate by Enterprises’ Financial Position Financial Obs Mean Std. Dev Min Max Position Growth -----------------------------------------------------------------------------------SEs with fin. diff. 261 0.066 0.119543 -0.1386 0.7611 SEs without fin. diff.

72

0. 0883 0.1310607

-0.07847

0.4621

The multiple regression result also showed that a strong financial condition/position significantly and positively affects growth of small enterprises. The growth rate of financially constrained enterprises was found to be 2.77% less than that of financially strong SEs, statistically significant at a 10% level. This is in line with the perceived hypothesis of this writer that a “financial constraint has a significant negative effect on the growth of small enterprise” and findings of many authors (Ishengoma & Kapppel, 2008; Ageba and Amaha, 2006; Tushabomwe-Kazzoba, 2006; Wiklund & Dess, 2005; and Beccetti &Trovato, 2002).

4.3. Relationship between Access to Bank Credit and Growth of Small Enterprises Small enterprises can benefit more from the use of debt financing than from equity capital (or personal saving). This is because if they finance their investment using debt capital, as the interest they pay on their loan is tax deductible, they can save their cash outflows. This means that it shields part of income of the business from taxes, and lowers tax liability and increases cash inflows of the firm as the result of this tax saving. The ultimate effect of this may be reflected in expansion of business that demands more employment, and in a higher growth rate. The growth rate of SMEs with better credit access was by far greater than of those SMEs without any chance of access to credit. Ahiawodzi and Adade (2012) said that a unit increase in access to credit leads to growth of SMEs by 10.5 units. Note that in the previous sections of this paper the researcher only discovered that the growth rate of financially constrained small enterprises was slower than those of unconstrained counterparts. But this does not indicate that all financially constrained small enterprises had been neglected by formal financial institutions, because some of them might not

286

Chapter Ten

apply for a bank loan. Besides, it is unclear whether lack of access to credit represents a critical constraint for the growth of small enterprises. For this reason the researcher further examined the number of SEs that had applied for a bank loan, the proportion of applications accepted/rejected, and the effect of adequate access to bank loans on growth of small enterprises. Of the 261 financially deficient SEs, 200, or 77%, had applied for a bank loan, and only 58 applications (29%) were accepted. Not only did banks accept a very small proportion of the applications, the amount of loan they actually dispersed was also inadequate. Only 19 of the eligible applicants (33%) received an adequate loan. In other words, about 91% of the requirements of the deficient enterprises were neglected by banks/microfinance, or remained unsatisfied, as a result of which growth was affected negatively. While the SEs with adequate access grew at 7.58%, the growth rate of the constrained SEs was only 6.45%, which has been confirmed by the regression model, and t-tests confirm these results (p < 0.05).

5. Summary and Conclusions 5.1. Summary Modigliani and Miller (1963) advised firms to use debt before equity (own saving) because the former is less costly both from the firm’s and creditor’s points of view. Fatoki (2011) also indicated that internal sources are very limited and less productive (as they are more expensive than debt). While 78% of the respondents reported that they were exposed to financial constraints due to the stringent bank requirements (collateral, business plan, and historical financial statements), only 10% of the respondents had financed their investment using bank loans. As a result of inadequate access to bank credit, small enterprises are forced to rely heavily on limited and more expensive internal sources, which in turn constrain growth. While a financial constraint has a statistically significant negative effect on growth at a 10% significance level, access to credit significantly and positively influences growth (P < 0.05). Therefore, though access to credit could have facilitated growth of small enterprises, the insignificant role of banks and other financial institutions has been retarding the employment growth rate of small enterprises in the country.

Bank Loan and its Impact on Growth of Small Enterprises

287

5.2. Implications and Suggestions The research findings imply that any additional access to credit (loan) has a significant positive impact on enhancing the growth of small enterprises, though the majority of them had inadequate access to bank loans. According to the Growth and Transformation Plan (GTP) of Ethiopia, the MSE sector is considered as a springboard of industrialization, and an engine for economic development of the country. However, such a plan cannot be achieved unless growth of the sector is facilitated by solving its financial constraint. The main policy implication of this study is that the government of Ethiopia should work hard to meet the credit need of the SE sector for speedy economic growth of the nation. The financial market should be promoted as an alternative source of capital for SEs and business in the private sector for effective mobilization of domestic capital. The regulatory and institutional framework needs to be developed and strengthened because a well-regulated and functioning financial market helps the sector not only as an alternative source of funds, but also as an alternative investment opportunity and income source for those enterprises with surplus capital. Therefore, consulting empirical evidence of many developing countries in Africa and Asia, the writer provides the following recommendations. 5.2.1. National Credit Guarantee Funds Respondents of this study reported that a lack of tangible assets to be used as collateral by banks was one of the most critical causes for their financial constraint. Thus, as one remedy, the Ethiopian government is advised to introduce and strengthen a credit guarantee fund as a risk-sharing scheme among those parties that participate in financing the SE sector. Many African and Asian countries (Ahiawodzi & Adade, 2012; Fakoti, 2011; Beck and Kunt, 2006; Kang, n.d.) adopted credit guarantee fund mechanisms as a fundamental mechanism to increase the availability of credit from banking and other financial institutions for the establishment, expansion, and improvement of SEs. Support from such a mechanism helps SEs that do not have tangible collateral to obtain bank loans. 5.2.2. Promotion of Non-Bank Financial Services As Kyaw (2008) argues, non-bank financial services and institutions such as leasing companies, saving and mutual funds, trade credit, factoring, and venture capital financing are best suited for small enterprise financing. For

288

Chapter Ten

example, leasing is considered well suited to SE financing because, unlike banks, there are neither collateral nor financial statement requirements in lease contracts. Besides, leasing contracts can more easily be structured to match the cash flow generation of the lessee’s business. In Ethiopia, however, such financial services are either totally absent, or they are used traditionally and in an unstructured, unregulated way (Ageba and Amaha, 2006). Therefore, the government of Ethiopia should assess the potential of said financial institutions/services and develop guidelines or regulations for smooth functioning of these institutions to participate in SEs lending. A lease provides the lessee the following advantages: No collateral or financial statement requirements: Banks require borrowers to prove financial soundness of their business enterprises by presenting a standardized financial statement, which could not be produced by SEs. Leasing is well suited to SEs’ activities because as the lessor-financier retains ownership of the asset during the term, the lessee is not required to provide financial statements nor assets for collateral. Simpler security arrangements and less stringent requirements for historical financial statements mean that new small enterprises can access lease finance more easily than bank loans. Simple contracts: Leasing contracts can more easily be structured to match the cash flow generation of the lessee’s business. 5.2.3. Mandatory Minimum Ratio of Bank Loan to SEs As a means of priority lending system, the government needs to initiate some guidelines so that banks are directed to make loans to potentially growing SEs. For example, in the Republic of Korea, all commercial banks are required to provide more than 45% of the increase in loans to SMEs (Kang, n.d.) 5.2.4. Easily Accessible Credit Easy accessibility of credit through development of specialized or development-oriented banking or financial institutions that specialize in financing SEs, like the SMEs Bank of Thailand, and JASME in Japan (Kyaw, 2008), should be encouraged. This fund can be made available to the MSEs at a reduced interest rate. NGOs and the government can earmark funds in order to subsidize the financial institutions.

Bank Loan and its Impact on Growth of Small Enterprises

289

5.2.5. Improve the Internal Capacity of Small Enterprises Results of this research and empirical evidence in Ethiopia (e.g. Bekele and Worku, 2008; Ageba and Amaha, 2006; Negash, 2006) revealed that small enterprise owners are also responsible for their problem on access to bank credit. First, as discussed above, they could not satisfy the collateral requirement of banks. Second, due to the lack of skilled and professionally trained personnel, they rarely keep accounting records of their business in an acceptable manner. This aggravates the problems of information asymmetry between the banks and their business. For this reason most of the accounting records submitted to banks for loan applications are not accepted by banks because of reliability problems (Fesseha and Aregawi). Third, many SEs in the study area lack the skills and knowledge to prepare a business plan that can be used to assess the feasibility of a project. Most of the small enterprises reported that they started their business without a formal business plan. As a result of this, banks rely mainly on the viability of their collateral in extending credit (Ageba and Amha, 2006). Therefore, in order to get better access to credit SE owners, government and relevant stakeholders should strive to overcome these internal problems. The first solution is to upgrade knowledge and skill of owners and/or employees of the SEs, in order to prepare financial statements that can be used to assess the financial condition and operating result of their businesses. For this purpose, the concerned body (e.g. the Regional Micro and Small Enterprise Agency) should develop an easily understandable financial manual that helps to properly record and control daily transactions and prepare acceptable financial statements. In addition, tailor-made training should be given in order to solve knowledge deficiency in accounting and preparation of tax returns. Also, in addition to short-term on-the-job training, such courses as entrepreneurship and small business management need to be given in schools and training centres. This helps the entrepreneurs not only start their business with a professionally developed business plan, but also to objectively and systematically evaluate their business. The ultimate effect of this can be manifested in adequate access of SEs to bank credit.

References Ageba, G. and Amaha, W. (2006a). ‘Micro and Small Enterprise (MSEs) Finance in Ethiopia: Empirical Evidence’, Eastern Africa Social Science Research Review, Vol 22, No 1: 63-86

290

Chapter Ten

—. (2006b). ‘Business Development Services (BDS) in Ethiopia: Status, Prospects and challenges in Micro and Small Enterprise Sector’, International Journal of Emerging Markets, Vol 1, No 4: 305-328 Ahiawodzi, A. K. & Adade, T. C. (2012). ‘Access to credit and Growth of Small and Medium Scale Enterprises in Ho Municipality of Ghana’. British Journal of Economics, Finance and Management Sciences. Vol 6 (2): 34-51 Andree, C. & Kallberg, C. (2008). The Capital Structure of SMEs: Evidence from the Swedish Security Industry. http://www.student.se/uppsok/search2.php? allasmes> (Retrieved August 17, 2013). Arinaitwe, S.K. (2006). ‘Factors Constraining the Growth and Survival of Small Scale Business, A Developing Countries Analysis’, Journal of the American Academy of Business, 8, 167-178 Atieno, R. (2009). ‘Linkage, Access to Finance and the Performance of Small Scale Enterprises in Kenya’, United Nations University-UNUWIDER, Kenya, Nairobi. Atsede W. Patricia L. Adebimpe A. (2008). Factors influencing small and medium enterprises (SMEs): an exploratory study of owner/manager and firm characteristics. Banks and Bank Systems, Volume 3, Issue 3, 2008 Beck,T., Demiriguc-Kunt, Laeven L., and Levine, R. (2004). ‘Finance, Firm Size, and Grwoth’, National Bureau of Economic Research, Working Paper 0983, Cambridge. Bekele, E. and Worku, Z. (2008). ‘Factors That Affect the Long-Term Survival of Micro, Small and Medium Enterprises in Ethiopia’, South African Journal of Economics, Vol. 76 (3) (September): 548-568. BoTIT (2011). Census of Micro and Small Enterprises of Tigray Regional State, Mekelle, 2011, Unpublished. Kallberg, C. (2008). The Capital Structure of SMEs: Evidence from the Swedish Security Industry. http://www.student.se/uppsok/search2.php? allasmes> (Retrieved August 17, 2013). Central Statistics Agency (CSA). (2003). Survey of Urban Informal Sector operations, Central Statistics Agency, Addis Ababa, Ethiopia. —. (2003a). Report on Small Scale manufacturing Industries Survey. Addis Ababa, Ethiopia. —. (2003b). Report on Urban informal Sector Sample Survey, Central Statistics Agency, Addis Ababa —. (2008). Summary and Statistical Report of the 2007 Population and Housing Census: Population size by age and Sex, Federal Democratic

Bank Loan and its Impact on Growth of Small Enterprises

291

Republic of Ethiopia (FDRE), Population Census Commission, Addis Ababa. —. (2010). Report on Small Scale Manufacturing Industries Survey, FDRE, Addis Ababa, Ethiopia. Chandler, G. N. (1993). “Measuring the performance of emerging Business: A validation study ”, Journal of Business Venturing, Vol 8: 392-408. Colombo, M. G., and Grilli, L. (2005). ‘Founders’ human capital and the growth of new technology-based firms: A competence-based view’, Reseach policy, Vol 34: 795-816. Cook, P. (2001). ‘Finance and Small and Medium-sized enterprise in Developing Countries’, Journal of Development Entrepreneurship, Vol 6, No 1: 17-40. Ethiopian Development Research Institute (EDRI). (2003). Micro and Small Enterprises Survey, EDRI, Addis Ababa. Evans, D. S. (1987). ‘Test of Alternative Theories of Firm Growth’, Journal of Political Economy, Vol 95 No 4 : 657_675 —. (1987). ‘The relationship between firm growth, size and age: estimates for 100 manufacturing industries’, Journal of Industrial Economics, Vol. 35 No. 4: 567-81. Fatoki, O.O. (2011). ‘The Impact of Human, Social and Financial Captial on the performance of Small and Medium-Sized Enterprieses (SMEs) in South Africa’. Journal of Social Science Vol 29 (3): 193-204. FDRE, MoFED. (2010). Growth and Transformation Plan, November 2010, Addis Ababa. FDRE, MTI (1997). Micro and Small Enterprises Development Strategy, Addis Ababa. Ghosh, C., Nag, R. & Sirmans, C. (2000). The pricing of seasoned equity offerings: Evidence form REITs. Real Estate Economics, 28, pp. 36384 Glancey, K. (1998), ‘Determinants of growth and profitability in small entrepreneurial firms, International Journal of Entrepreneurial Behavior and Research, Vol. 4 No. 1,:18-25. Hitt, M.A, Ireland, P.D. and Hoskin, R.E. (2009). Strategic Management: Concepts and cases-Competencies and Globalization. 8th edition. USA: South-West Cengage Learning. Hofsteen, E. (2006). Constructing a Good Dissertation: A Practical Guideline for Finishing a Master’s, MBA, or PhD Schedule, Published by EPE, South Africa. Ishengoma, E.K. and Kappel, R. (2008). ’Business Constraints and Growth of Micro and Small Manufacturing Enterprises in Uganda’,

292

Chapter Ten

German Institute of Global and Area Studies (GIGA), Working paper No 78. —. (2008). ‘Business Constraints and Growth potential of Micro and Small Manufacturing Enterprises in Uganda’ GIGA Research paper Program: Transformation in the Process of Globalization. Liedholm, C. and Mead, C. (1999). ‘Small Enterprises and Economic Development: the Dynamics of Micro and Small Enterprises’, Rutledge Studies in Envelopment Economics, New York. McPherson, M. A. (1996). ‘Growth of micro and small enterprises in southern Africa’, Journal of Development Economics, Vol 48: 253277. McPherson, M.A. and Rouse, J.J. ‘Access to Finance and Small Enterprise Growth: Evidence from East Java’ University of North Texas, USA. Mead, D. C. (1994). ‘The Contribution of Small Enterprises to Employment Growth in Southern and Eastern Africa’, World Development, Vol 22 No 12: 1881-1894. Ministry of Trade and Industry (MoTI). (1997). Micro and Small Enterprises Development Strategy, FDRE, Addis Ababa. Modigliani, F., Miller, M. H. (1958). The cost of capital, corporation finance and the theory of investment. American Economic Review, 48(3):261-295. Niskanen, M. & Niskanen J. (2007). ‘The Determinants of Firm Growth in Small and Micro Firms Evidence on relationship lending effects’ Orser, B.J., Hogarth-Scott, S. and Riding, A.L. (2000), ‘Performance, firm size, and management problem solving’, Journal of Small Business Management, Vol. 38 No. 4, pp. 42-58. Papadaki, E. and Chami, B. E. (2002). ‘Growth Determinants of MicroBusinesses in Canada’, Small Business Policy Branch Industry Canada People at Work, New York: The Free Press Politis, D. (2008). ‘Does prior Start-up experience matter for Entrepreneurs’ Learning’, Journal of Small Business and Enterprise Development,Vol 15,No 3: 472-489. Scott, D. F. (1972). Evidence on the importance of financial structure. J on Financial Management, 1(3): 45-60. Soh, P. H. (2003). ‘The role of networking alliance in information acquisition and its implication for new product performance’, Journal of Business VenturingVol 18: 727-744. Stiglitz, J. and Weiss, A. (1981). Credit rationing in markets with imperfect information. American Economic Review, 71(3): 393-410.

Bank Loan and its Impact on Growth of Small Enterprises

293

Stiglitz, E. J. (1993). ‘The role of the state in financial markets’, World Bank Research observer, Annual conference on Development Economics Supplement, (1193): 19-61. Tarmidi, L.T. (2005). The importance of MSEs in Economic Development of Developing APEC Countries, paper presented at the APEC Study Centre Consortium Conference 2005, Jeji, Korea, 22-25 May 2005. Tigrai Regional State, Bureau of Planning & Finance. (2011). Exploring the Status and Prospects of Micro and Small Enterprises (MSEs) in Tigrai, June 2011 Aksum, Tigrai, Ethiopia. Wiklund, J. and Shephared, D. (2003). ‘Knowledge based resource, Entrepreneurial Orientation, and the performance of Small and Medium-sized business’, Strategic Management Journal, Vol 24, Iss 3. Wiklund, J., Davidson, P. and Delmar, F. (2003). ‘What do they think and feel about growth? An Expectancy-value approach to small business managers’ attitudes towards growth’ , Entrepreneurship theory and practice, vol 27, No 3: 247-270. Wiklund, J., Patzelt, H., & Shepherd, D. (2009). ‘Building an integrative model of small business growth’, Small Business Economics, Vol 32, No 4: 351-374.







CHAPTER ELEVEN THE IMPACT OF OWNERS’ FINANCING PREFERENCE ON THE GROWTH OF SMALL ENTERPRISES: EVIDENCE FROM TIGRAY REGIONAL STATE, ETHIOPIA HABTAMU TEFERA AND AREGAWI GEBREMICHAEL

Abstract Small Enterprises (SEs) are recognized as a principal source of innovation and experimentation in which small savings are being transformed into investment. To this end, SEs are making various remarkable contributions to the overall growth and structural transformation of the economy through job creation, import substitution, income and value addition, input provision for large industries, and ultimately through poverty reduction among the poorest of the poor. However, most SEs were unable to grow and provide employment since start-up due to absence of access to formal finance particularly in developing countries like Ethiopia where efficient financial markets are lacking and there is high information asymmetry. Moreover, several previous studies found that access to formal finance is a key impediment for the growth of SEs. Thus, this study aims to empirically investigate the impact of owners financing preference on the growth of SEs based on a primary data collected from six major towns in the Tigray regional state of Ethiopia using a Propensity Score Matching (PSM) model. Thus, SEs that have used debt capital show higher growth than those SEs that have used equity capital. The PSM result proves that financing preference growth rate differentials are highly significant for SEs that have used debt capital, compared to SEs that have used equity

The Owners’ Financing Preference on the Growth of Small Enterprises 295

capital. This implies that debt financing has a significant positive impact on the growth of SEs. Therefore, policy makers and donor agencies have to design new intervention strategies to eradicate the access to formal finance problem of SEs and create an alternative means of finance for SEs to leverage their growth and potential contribution to the economy.

1. Introduction Small Enterprises (SEs) by virtue of their nature are recognized as a principal source of innovation and experimentation in which small savings are being transformed into investment. To this end, the sector is known for making various remarkable contributions to the overall growth and structural transformation of the economy through job creation, import substitution, income and value addition, input provision for large industries, and distribution of products (Porter, 2006; Storey, 1994). Moreover, the sector is playing a significant role in ultimately reducing poverty among the poorest of the poor, and provides a large segment of the poor and middle-income population with low priced consumer goods and services. The sector is also seen as an essential means of job creation and economic transformation in most developing countries, as it takes the lion’s share of the fast-growing labour force in the world, with the sector taking 48% of the labour force in North Africa, 51% in Latin America, 65% in Asia, and 72% in Sub-Saharan African countries in particular (ILO, 2002). Moreover, the study made by Mead and Liedholm (1998) in five Eastern and Southern African countries (Botswana, Kenya, Malawi, Swaziland and Zimbabwe) shows that the number of people engaged in the MSE sectors is nearly twice the level of employment in large-scale enterprises and public sectors. In Ethiopia, it is the largest employment generating sector after agriculture, hence the government formulated a National MSE Development and Promotion Strategy in 1997 with a major objective of creating sustainable employment and providing a basis for medium and large-scale enterprises, thereby facilitating growth and structural transformation of the country’s economy. However, a large number of MSEs are unable to grow and provide employment in the country (Gebreeyesus, 2007; Habtamu, 2012; Wasihun & Paul, 2010) due to absence of access to formal finance. Besides, most SEs fail to make a significant contribution to growth and economic transformation,

296

Chapter Eleven

particularly in developing countries where the financial markets are underdeveloped. Moreover, SEs are more dependent on external financing to start and expand operations, develop new products, and invest in new production facilities (Ateino, 2009). SEs in general have smaller financial reserves as compared to larger firms, due to the difficulty and cost of attracting new equity capital. As a result, these enterprises bear a higher financial distress risk. Financial institutions tend to respond to this risk through adopting a capital-gearing approach to lending. Therefore, most financial institutions focus more on the value of collateral available at the event of financial depression than on evaluating income streams flowing from an investment project. These lending approaches aggravate the problem of access to formal finance in SEs. As most SEs operate in trading and service sectors (Habtamu, 2012), the demand for new investment in fixed assets is also relatively low, hence they fail to secure loans in their early years of establishment. Several previous studies in this regard also suggest that access to formal finance is one of the key impediments of SEs’ growth, regardless of their size, location, and type of economic activity (Gregory et al., 2005; Van Auken, 2005, Westhead and Wright, 2000). SEs are facing a relative intricacy to raise finances from formal financial institutions like banks. Hence, access to sufficient finance for operation and expansion of their enterprises is limited to the saving ability of the enterprise owners (retained earnings), even though firms having access to external finance have a relative advantage of growth (Wasihun & Paul, 2010). The availability of finance for investment in projects having positive net present value is vital to the sustainability and growth of SEs since access to finance has a positive impact on firm performance (Clarcke et al., 2010). The theory of optimal leverage also suggests that cost of debt capital is less than the cost of equity capital due to differences in risk and tax deductibility of debt, so that when the cost of equity approaches the cost of debt, equity financing should be decreased. Thus, the use of leverage can increase the rate of return on equity. However, debt capital can also be harmful if the firm is excessively leveraged. Moreover, acquiring too much debt may subject the enterprise to financial risk due to the variability in interest rates and net incomes. Thus, the owners of SEs must weigh the trade-offs between debt and equity capital and determine an optimal level of debt and equity to operate efficiently and grow.

The Owners’ Financing Preference on the Growth of Small Enterprises 297

SEs’ growth is often closely associated with firm overall success and survival (Johannisson, 1993). Growth has been used as a simple measure of success in business (Storey, 1994), and it is the most appropriate indicator of performance for surviving SEs (Brush & Vanderwerf, 1992). Moreover, growth is an important precondition for the achievement of other financial goals of business (Reynolds, 1993; Storey, 1994). Hence, growth is usually a critical precondition for survival (Storey, 1994), in which access to formal finance is a key for business growth. The implications of access to external finance and capital structure choice has been investigated since the seminal work of Modigliani and Miller (1958) in the developed world, but much less is known about the financing preference of owners and its implications in Sub-Saharan African countries like Ethiopia in which the financial markets are weakly efficient and there is high information asymmetry between the agents, the SEs, and the principals, the financial institutions. Moreover, the extensive review of literature revealed that no study has been conducted on the impact of owners’ financing preference on the growth of SEs in this country. There are also diverging views in both theoretical and empirical literature concerning the impact of financing preference on the growth of SEs; as a result the issue of financing preference becomes imperative in most management research. So taking these all into account, it is very crucial to empirically explore the impact of owners’ financing preferences on the growth of SEs. Thus, this study aims to investigate the owners’ financing preference in light of its impact on the growth of SEs in the Tigray regional state of Ethiopia, in which major emphasis is given to examining the growth status, the owners’ financing preferences, and the financing mix in SEs. The remainder of this paper is organized as follows: First, the review of theoretical and empirical literature is presented. Second, the research methodology is outlined. After that, research results and discussions are presented followed by results-based conclusions and implications.

2. Review of Theoretical and Empirical Literature The issue of financing preference has been the subject of many studies. It has been argued that profitable firms were less likely to depend on debt in their capital structure than less profitable ones. It has also been argued that firms with a high growth rate have a high debt to equity ratio. Nonetheless, large firms and firms with low growth rates prefer to issue long-term debt

298

Chapter Eleven

(Barclay & Smith, 1995). Similarly, less risky firms usually make greater use of long-term debt (Stohs & Mauer, 1996). Hence, the subject of financing preference and firm growth becomes the main concern in many pieces of finance, strategy, organizational, and entrepreneurship research. In general the capital structure choice or financing preference of a firm could be explained by four overriding theories. These are: static trade-off theory, pecking order theories, agency theory, and growth cycle theory of small business finance. According to static trade-off theory, optimal capital structure could be determined through optimizing the trade-off between benefits and costs associated with debt financing. Debt benefits mainly include tax saving induced by the deductibility of interest expenses from operating income of the firm (Modigliani & Miller, 1963) and the reduction of agency costs through the threat of liquidation. High leverage can also enhance firm performance by mitigating conflicts between the agents and the shareholders concerning free cash flows (Jensen, 1986), optimal investment strategy (Myers, 1977), and amount of risk (Jensen & Meckling, 1976). Thus, a positive relationship could be expected between debt financing and a firm’s growth. A number of studies provide empirical evidence supporting this positive relationship between debt financing and firm growth (performance) (Abor, 2005; Ghosh, Nag & Sirmans, 2000; Hadlock & James, 2002). On the other hand, debt financing brings commitment for future cash outflows in terms of periodic interest and principal payments, and these commitments increase the probability of firm’s financial default and bankruptcy. However, several studies suggest that bankruptcy cost does exist, but it is reasonably small relative to tax saving associated with debt financing (Miller, 1977; Warner, 1977). Thus, according to static trade-off theory, more profitable firms have a higher income to shield and thus should borrow more to take the tax advantages. The pecking order theory, developed by Myers and Majluf (1984), suggests that firm growth is negatively related to the owners’ financing preference i.e. debt financing. This theory points out that information asymmetry between the agents and the principals demands an additional cost for firms to raise funds from external sources. Moreover, the issuance of debt results in an additional cost that is reflected in the higher required rate of yield. As a result, the owner may prefer financing new investments by internal sources (retained earnings), and if this source of finance is not enough, they may seek external sources of finance (debt capital) and finally seek their own contributions (Hussien, Millman & Matlay, 2006).

The Owners’ Financing Preference on the Growth of Small Enterprises 299

According to this theory, firms that are profitable and generate high earnings to be retained are expected to use less debt in their capital structure than those that do not generate high earnings due to the use of internal finance. Hence, a negative relationship could be expected between debt financing and firm growth. Moreover, high-growth firms may find it too costly to rely on debt to finance growth. Agency theory (Jenson & Meckling, 1976) explains the principal-agent relationship between the agent and the principals. This theory asserts there is a negative relationship between financing preference and firm growth. Myers (1977) argued that high growth firms might have more options for future investment than low growth firms. Thus, highly leveraged firms are more likely to bypass profitable investment opportunities, because such investment will effectively transfer wealth from the agents to the principals. As a result, firms with high growth opportunities may not use debt in the first place, and leverage is expected to be negatively related to growth opportunities. However, Myers (1977) argued that the agency problem could be mitigated if long-term debt is replaced by short-term debt. This suggests that the short-term debt ratio might actually be positively related to firm growth, if growing firms substitute short-term financing for long-term financing. A number of studies provide empirical evidence supporting this negative relationship between debt financing and firm growth (Booth et al., 2001; Fama & French, 2002; Gleason & Mathur, 2000; Titman & Wessels, 1988; Rajan & Zingales, 1995; Wald, 1999). Therefore, high levels of debt in the capital structure would decrease firm growth. The growth cycle theory of small business finance initiated by Berger and Udell (1998) suggests that there is a positive relationship between firm growth and access to formal finance. This theory illustrates the dynamic financial needs as the small business becomes more experienced, and enhanced informational transparency. In this theory, the firm gets better access to venture capital as a source of equity and midterm-loans (Gregory et al., 2005) as the firm gets older and more transparent information leaks out. This implies that the firm tends to have better access to public equity and long-term financing in which it enhances its growth. Taking these theories, one may argue that financing preference is influenced by many factors, and explaining this decision by a single theory may be sort of providing incomplete diagnosis of the decision. In fact, each capital structure theory works under its own assumptions and conditions so that it does not offer a complete explanation of firms’

300

Chapter Eleven

financing decisions. This means that searching for optimal capital structure is a multidimensional and complex process (Myers, 2001; Eldomiaty, 2007). This could explain the mixed and contradictory results of the studies that empirically tested the predictions of these theories (financing preference versus firm’s growth). Jermias (2008) argued that prior studies have examined only the direct effect of financial leverage on performance, where this leverage-performance relationship may be contingent upon some factors that may be specific to owner characteristics, sector of economic activity, and macroeconomic environment of the country. In other words, the choice of capital structure could have an impact on both growth and failure of the enterprises. On the other hand, the concept of firm growth is a controversial issue in finance, largely due to its multidimensional phenomena, so there is no comprehensive theory that explains firm growth. Hence, different scholars’ (Evans, 1987; Storey, 1994) tests came up with different firm growth proxies. However, the most frequently used measure for growth has been change in the firm’s sales turnover (Hasnu & Amjam, 2007). Another typical measure for growth has been change in the number of employees (Cheng, 2006; Eshetu & Mammo, 2007; Gebreeyesus, 2007; Mead & Liedholm, 1998; Wasihun & Paul, 2010). These measures have been found the most frequently used in the context of SEs. Thus, this study also used the change in employment size and sales turnover as a proxy to the growth of SEs. The growth rate in employment is computed by dividing the change in the natural logarithm of employment size since start-up by the enterprise age, i.e. ‫ ܲܯܧܴܩ‬ൌ ሺ݈݊ܵଵ െ ݈݊ܵ଴ ሻȀ‫ ;ܽܧ‬similarly the growth rate in average annual sales revenue is computed by dividing the change in average annual sales revenue since start-up by the enterprise age, that is ‫ ܴܵܣܴܩ‬ൌ ሺ‫ܴܵܣܣ‬ଵ െ ‫ܴܵܣܣ‬଴ ሻȀ‫ܽܧ‬, following Evans (1987), Cheng (2006) and Gebreeyesus (2007).

3. Research Methodology 3.1 Data and Methods To empirically investigate the impact of owners’ financing preference on the growth of SEs, this study has used primary data collected from six major towns i.e. Adwa, Axum, Adigrat, Mekelle, Shire, and Wukro in the Tigray regional state of Ethiopia, covering 333 randomly selected SEs. To this end, several statistical tools were combined together to examine the

The Owners’ Financing Preference on the Growth of Small Enterprises 301

financing preferences of owners and the status of growth of SEs, and the econometric analysis tool, which is the Propensity Score Matching (PSM) model, was used to empirically examine the impact of owners financing preference on the growth of SEs in the region.

3.2 Specifying Variables and Measurements There is little agreement in the existing literature concerning how to measure firm growth. Thus different studies have used a variety of proxies such as total assets, sales, employees, profit, capital, and others (Berkham et al., 1996; Davidsson & Wiklund, 2000; Holmes & Zimmer, 1994). Perhaps the most common means of measuring firm growth is through relatively objective and measurable characteristics, such as growth in sales turnover, total assets, and employment size. These measures are relatively uncontroversial, the data tend to be easily available, and it increases the scope for cross study comparability (Freel & Robson, 2004). Hence in this study, change in employment size since start-up and annual sales turnover were used as a proxy to firm growth, which is the dependent variable. The explanatory variable is the owners’ financing preference in SEs, which is a dummy variable that takes 1 for debt financed SEs (treated group) and have used multiple sources of finance, and takes 0 for firms that were equity financed (controlled groups) and rely on a single source of finance. Moreover, many variables that are related to the owners’ characteristics such as age, education, ex-ante business experience, and others that can potentially affect the owners’ financing preference were controlled.

3.3 Econometric Modelling and Estimation Strategy To empirically investigate the growth rate differentials that result from owners’ financing preference of SEs, this study follows the potentialoutcome framework for causal inference of Rubin (1974) in which Y1 is the outcome if the unit is exposed to treatment T = 1, and Y0 is the outcome if the unit is not exposed to treatment T = 0. In this study, Yj is the growth rate if the SEs used debt capital (treated) and Yi is the growth rate if the SEs used equity capital (non-treated). Hence, the average treatment effect on treated (ATET) is defined as:

E Y j  Yi / T

1 ................................................................................... (1)

Chapter Eleven

302

In this case, the Propensity Score Matching (PSM) technique has gained popularity for its potential to remove a substantial amount of bias from non-experimental data (Dehejia & Wahba, 1999). This technique helps to adjust for initial differences between cross-sections of treated and controlled groups by matching each treated unit to a controlled unit based on ‘similar’ observable characteristics. It summarizes the differences in a single dimension, the propensity score, which is then used to compute treatment effects non-parametrically. The propensity score conveniently summarizes the conditional probability of treatment given pre-treatment or exogenous characteristics (Rosenbaum & Rubin, 1983). The probability of participation conditional on X will be balanced such that the distribution of observables X will be the same for both participants and non-participants. Consequently, the differences between the groups are reduced to only the attribute of treatment assignment, and unbiased impact estimates can be produced (Rosenbaum & Rubin, 1983). The counterfactual group can be identified if potential outcomes, Yj(Yi) of participants (non-participants) are independent of treatment assignment, conditional on observables X:

Yi , Y j A T / X , X

......................................................................... (2)

This is the Conditional Independence Assumption (CIA) that the PSM technique builds on. It states that selection is solely based on observable characteristics, and potential outcomes are independent of treatment assignment. The behavioural assumption behind it is that the potential outcome in case of no treatment (Yi) does not influence treatment assignment, and the outcome of control subjects can be used to estimate the counterfactual outcome of the treated in the case of no treatment. However, with a high dimensional vector X, this task may be problematic. To deal with the dimensionality problem, one can use the propensity score by Rosenbaum and Rubin (1983). Therefore, the multidimensional matching problem is reduced to a one-dimensional problem, and the distribution of potential outcomes will be balanced among participants and counterfactuals (Rosenbaum & Rubin, 1983; Heckman, Ichimura & Todd, 1998). The propensity score is the individual probability of receiving the treatment given the observed covariates:

P X

P T

1 / X .......................................................................... (3)

If the potential outcome Yi is independent of treatment assignment conditional on X, it is also independent of treatment assignment

The Owners’ Financing Preference on the Growth of Small Enterprises 303

conditional on P(X). The propensity score can thus be used as a univariate summary of all observable variables. Consequently, if P(X) is known, the ATET can be consistently estimated as:

>

@

ATET EYj Yi / T 1 E^p X / T 1` EYj / p X , T 1 EYi / p X , T 0 ....... (4) In practice, P(X) is usually unknown and has to be estimated through some probabilistic model i.e. probit or logit (Todd, 1995). However, the estimation of the propensity score is not enough to estimate the ATET using equation (4), since the probability of finding two observations with exactly the same value of the score is almost non-existent so that various methods have been proposed in the literature to overcome this problem and match treated and controlled units on the basis of the estimated propensity score. Thus, there are different algorithms: nearest neighbour, kernel, radius, and stratification matching. These methods differ from each other with respect to the way they select the control units that are matched to the treated, and with respect to the weights they attribute to the selected controls when estimating the counterfactual outcome of the treated: E(Yi|P(X), T = 1). However, they all provide consistent estimates of the ATET under the CIA ((Yj, Yi) AT |X) and the overlap condition (0 < Pr (T = 1|X) < 1). Therefore, this study used a logit model for comparative simplicity, and the propensity score can then be defined as: PX

Pr T

1/ X

F E 1 X 1  ,  ,  E i X i

F  XE

e XE ................ (5)

where F(‫ )ڄ‬produces response probabilities strictly between zero and one. Moreover, this study considers all matching techniques to assess the ATET, since they have their own advantages and disadvantages. Thus, their joint consideration may offer a way to assess the robustness of the results. 4. Results and Discussion

4.1 Sources of Finance and Financing Preference SE sectors are mainly argued as a constrained sector to access formal (external) finance. Hence, their sources of finance are mostly limited to internal sources of finance. As table I summarizes, the majority of the SE owners (68%) raised finance from internal sources (equity capital) i.e. owners’ contribution or retained earnings, whereas about 32 percent of the

Chapter Eleven

304

SE owners used external sources of finance (debt capital). Thus, equity capital appears the primary choice of finance for SEs owners. Moreover, the majority of SE owners (61%) were a single source (either equity or debt capital) of finance dependant, and the remaining (39%) relied on multiple sources of finance, in which they have an alternative source of debt and equity capital, in which most of the multiple sources of finance users are debt financed SEs. Table 1: Shows SE owners’ financing preference Source of Finance Single Source Multiple Source Total Percent (%)

Financing Preference Debt Equity 24 179 83 47

Total

Percent (%)

203 130

60.96 39.04

107 32.13

333 100

100

226 67.87

Source: Stata result from survey data (2013)

This result proves that most SEs are still constrained by the absence of access to formal finance (external sources of finance) and remain with their primary choice of equity capital (internal source dependent), which is consistence with most previous studies (Gebreeyesus, 2007; Hassan, Millman and Matlay, 2006; Wasihun and Paul, 2010). Most SEs that have used debt capital also have access to multiple sources of finance, but few SEs that relied on equity capital have access to multiple sources of finance.

4.2 SEs Growth and Owners’ Financing Preference The growth of SEs that were computed using the change in employment size (GREMP) since start-up ranges from downsizing of 13.86 percent up to an increment in employment of 76 percent per year, as table II shows. Similarly, the growth rate in annual sales turnover (GRASR) also shows that the SEs’ growth ranges on average from a cut-off in annual sales of 76.75 percent, up to an increment of 176.61 percent in annual sales revenue since start-up. Moreover, the growth rate varied between the SEs that used debt finance and equity finance, with the debt financed SEs revealing on average 9.41 percent growth rate, whereas the equity financed SEs showed a 5.98 percent growth rate in employment. Correspondingly, the growth rate in annual sales revenue also varied for the SEs that raised

The Owners’ Financing Preference on the Growth of Small Enterprises 305

capital from external and internal sources, which is 20.97 percent for debt financed, and 16.35 percent for equity financed SEs per year. Hence, the SEs that used debt capital have shown a greater tendency of growth compared to SEs that used equity capital. Table 2: Shows growth rates of SEs and owners’ financing preference Growth rate of SEs

Financing Preference of SEs owners Debt Financed Equity Financed Min

Max

Mean

Min

Max

Mean

Min

GREMP GRASR Dif

-0.14 -0.73 -0.59

0.73 1.69 0.95

0.09 0.21 0.12

-0.08 -0.77 -0.69

0.76 1.77 1.01

0.06 0.16 0.10

-0.14 -0.77 -0.63

Overall Max 0.76 1.77 1.0

Mean 0.71 0.18 0.53

Source: Stata result from survey data (2013)

This result also suggests that there is a significant positive relationship between debt financing and growth of SEs that is consistent with previous studies of Abor (2005), Champion (1999), Ghosh et al. (2000), and Hadlock and James (2002). Moreover, it shows that the static trade-off theory of capital structure choice and the growth cycle theory of small business finance could hold for SEs that were considered in the study area.

4.3 SEs’ Growth and Source of Finance SEs may have an alternative source of debt or equity capital. Hence, they may use a different combination of sources of capital, and most frequently SEs seek to use multiple sources of finance even though they are constrained by access to formal finance from banks or other financial institutions, as table III shows. SEs that have raised capital from multiple sources of finance revealed on average the highest growth rate of 8 percent in employment and 21 percent in sales turnover, whereas SEs that have raised capital from a single source of finance show 6 percent in employment and 16 percent in sales turnover on average, which is relatively low. Table 3: Shows growth rates of SEs and sources of finance Growth rate of SEs GREMP GRASR Dif

Source of Finance Single Source Multiple Source Min Max Mean Min Max Mean -0.08 0.76 0.06 -0.14 0.73 0.08 -0.77 1.77 0.16 -0.73 1.69 0.21 -0.69 1.01 0.10 -0.59 0.95 0.12

Source: Stata result from survey data (2013)

Min -0.14 -0.77 -0.63

Overall Max 0.76 1.77 1.01

Mean 0.71 0.18 0.53

Chapter Eleven

306

This shows that SEs that use multiple sources of finance (equity and debt capital) can grow at a relatively higher rate than those SEs that use equity capital only. Thus, access to multiple sources of finance may play a significant role in the growth of SEs.

4.4 The Impact of Owners’ Financing Preference on the Growth of SEs The impact of owners’ financing preference that is determined using the PSM model shows there is a significant owner financing preference growth rate differential both in employment and sales turnover using nearest-neighbour, radius, kernel, and stratification matching techniques, as shown in table IV. As this table shows, there is at least a 3.4 percent growth rate differential in employment size between the SEs that have used debt capital and equity capital using all matching techniques. There is also a 4.5 percent growth rate differential in annual sales turnover per year between the owners that have used debt capital and equity capital to start operations in all matching techniques, except the kernel matching technique. This shows that the growth rate differential using employment growth rate is more robust than the sales turnover. Moreover, this result shows that the pecking order theory and agency theory may not hold in the context of SEs in the study area. Table 4: Shows PSM result of financing preference of owner on the growth of SEs Variables

Gr. Rate GREMP GRASR Bootstr. No. Obs. Debt fin. Equity fin.

Impact of owners’ financing preference on the growth of SEs NearestRadius Matching Kernel Matching Stratification Neighbour Matching ATET t-value ATET t-value ATET t-value ATET t-value 0.049 0.051

107 90

2.054** 0.034 1.827*** 1.534 0.045 1.777*** 0.032 0.026 107 226

0.034 0.044

107 226

2.069** 0.034 2.231** 1.436 0.045 1.687*** 0.031 0.027 107 226

Source: Stata result from survey data (2013), ** shows p