The Effect of Covid-19 on Loan Loss Provisions and Earnings Management of European Banks 3658400595, 9783658400590

Citation preview

BestMasters

Merjona Lamaj

The Effect of Covid-19 on Loan Loss Provisions and Earnings Management of European Banks

BestMasters

Mit „BestMasters“ zeichnet Springer die besten Masterarbeiten aus, die an renommierten Hochschulen in Deutschland, Österreich und der Schweiz entstanden sind. Die mit Höchstnote ausgezeichneten Arbeiten wurden durch Gutachter zur Veröffentlichung empfohlen und behandeln aktuelle Themen aus unterschiedlichen Fachgebieten der Naturwissenschaften, Psychologie, Technik und Wirtschaftswissenschaften. Die Reihe wendet sich an Praktiker und Wissenschaftler gleichermaßen und soll insbesondere auch Nachwuchswissenschaftlern Orientierung geben. Springer awards “BestMasters” to the best master’s theses which have been completed at renowned Universities in Germany, Austria, and Switzerland. The studies received highest marks and were recommended for publication by supervisors. They address current issues from various fields of research in natural sciences, psychology, technology, and economics. The series addresses practitioners as well as scientists and, in particular, offers guidance for early stage researchers.

Merjona Lamaj

The Effect of Covid-19 on Loan Loss Provisions and Earnings Management of European Banks

Merjona Lamaj Wien, Austria Master Thesis Wirtschaftsuniversität Wien (WU). 2022

ISSN 2625-3577 ISSN 2625-3615 (electronic) BestMasters ISBN 978-3-658-40059-0 ISBN 978-3-658-40060-6 (eBook) https://doi.org/10.1007/978-3-658-40060-6 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2023 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer Gabler imprint is published by the registered company Springer Fachmedien Wiesbaden GmbH, part of Springer Nature. The registered company address is: Abraham-Lincoln-Str. 46, 65189 Wiesbaden, Germany

Foreword

The Covid-19 pandemic has posed a great challenge for financial institutions around the world. Due to the severe economic consequences of the pandemic, many borrowers experienced significant difficulties in servicing their liabilities and paying their debt obligations. The potential negative effects on the financial stability urged many governments and institutions to undertake a range of responsive measures such as public guarantees and suspension of debt payments. Consequently, banks have experienced significant difficulties in estimating the impact of the crisis on their loan portfolios. International Financial Reporting Standard (IFRS) 9, the current accounting standard for financial instruments requires banks to account for expected credit losses when provisioning for loan losses. Nevertheless, national and supranational prudential authorities and standard setters anticipated adverse effects of the pandemic on banks’ financial statements and their regulatory capital. Concerned about any “cliff-effect” on banks’ loan loss provisioning, standard setters and regulatory authorities advised banks to refrain from using overly pessimistic assumptions in their credit loss models and encouraged them to use the flexibility in the standard in estimating the Covid-19 impact in their portfolios. While accounting for expected credit losses requires a certain amount of judgment, it is unclear how the high level of uncertainty and the unusual response of prudential authorities during Covid-19 affect banks’ discretion when accounting for loan losses. This work provides early evidence on the effect of Covid-19 on loan loss provisions and earnings management of European banks. It shows that during Covid-19 banks use discretionary loan loss provisions to a greater extent than before. This trend is more evident for banks located in countries that have implemented strong containment measures as a response to the Covid-19 pandemic.

v

vi

Foreword

Moreover, while banks tend to overstate provisions at the beginning of the pandemic, they do, on average, understate provisions during 2021. Finally, examining the direction of earnings management, the author finds that during Covid-19 banks use upwards earnings management, whereas before Covid-19 they engage in downward earnings management. These findings are important for both standard setters and prudential authorities concerned with potential effects of the Covid-19 crisis on banks’ loan loss provisions under an expected credit loss model. They show that while concerns on procyclicality on banks’ loan loss provisions and its exacerbating effect on the crisis may justify the flexibility offered during the pandemic, the opportunistic use of discretion on accounting for loan losses may distort the banks’ reported capital and their resilience to withstand the crisis. Vienna December 2022

Univ.-Prof. Dr. Zoltán Novotny-Farkas

Abstract

This thesis examines loan loss provisions (LLP) and earnings management behavior of European banks during the Covid-19 pandemic. Specifically, I analyze how the high flexibility offered by prudential authorities and standard setters in the context of Covid-19 affects banks’ use of discretion when accounting for loan loss provisions. I find that during Covid-19 banks use discretionary LLPs to a greater extent than before Covid-19. This trend is more evident for banks located in countries that have implemented strong containment measures as a response to the Covid-19 pandemic. Moreover, while banks tend to overstate provisions at the beginning of the pandemic, they do, on average, understate them during 2021. Finally, examining the direction of earnings management I find that during Covid-19 banks use upward earnings management, whereas before Covid-19 they engage in downward earnings management. Merjona Lamaj International Accounting Group Wirtschaftsuniversitat Wien Wien, Austria

vii

Contents

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

2 IFRS 9 and the Expected Credit Loss Model . . . . . . . . . . . . . . . . . . . . . . 2.1 Why Do Banks Account for Credit Losses? . . . . . . . . . . . . . . . . . . . 2.2 “Too little, too late” Provisioning and the Transition to the ECL Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 The Main Differences Between IFRS 9 and IAS 39 . . . . . . . . . . . . 2.4 The ECL-Model Under IFRS 9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.1 The Measurement of Expected Credit Losses . . . . . . . . . . . 2.4.2 Three Stages of Credit Risk . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.3 Determining the Significant Increase in Credit Risk (SICR) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 Main Implications of the ECL Model . . . . . . . . . . . . . . . . . . . . . . . . .

5 5

12 13

3 Covid-19 and European Banks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Covid-19 at a Glance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Economic Consequences of the Pandemic . . . . . . . . . . . . . . . . . . . . . 3.3 Covid-19 Implications on Loan Loss Provisions Under IFRS 9 . . 3.3.1 Concerns Over Procyclicality . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.2 The Use of Flexibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.3 No Mechanistic Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.4 Short Versus Long Term Forecasts . . . . . . . . . . . . . . . . . . . . . 3.3.5 Transparency and Disclosure . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 Banks’ Provisioning Practices During the Pandemic . . . . . . . . . . . .

15 15 16 18 18 19 20 20 21 21

4 Hypothesis Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Banks’ Incentives to Manage Earnings . . . . . . . . . . . . . . . . . . . . . . . .

23 23

7 8 9 9 10

ix

x

Contents

4.2 Earnings Management Under IFRS 9 . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Covid-19 Effect on Income Smoothing . . . . . . . . . . . . . . . . . . . . . . .

24 25

5 Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1 Data and Sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

29 29 30

6 Research Design and Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1 Discretionary LLPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Income Smoothing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3 Cross-country Differences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4 Upwards and Downwards Earnings Management . . . . . . . . . . . . . . .

37 37 40 45 48

7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

51

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

53

Abbreviations

BCBS DELR EBA ECDC ECL EIR EM FASB FCAG FSF GDP GFC GPPCI ICL IFRS NHS IMF ITG LGD LLA LLP OCI PD SI SICR

Basel Comitte on Banking Supervision Delayed Expected Loss Recocgnition European Banking Authority European Centre for Disease Prevention and Control Expected Credit Loss Effective Interest Rate Earnings Management Financial Accounting Standard Board Financial Crisis Advisory Group Financial Stability Forum Gross Domestic Production Global Financial Crisis Global Public Policy Comitte Incurred Credit Loss International Finanical Reporting Standards National Health Service International Monetary Fond IFRS Transition Resource Group Loss Given Default Loan Loss Allowances Loan Loss Provisions Other Comprehensive Income Probability of Default Significant Institutions Significant Increase in Credit Risk

xi

List of Figures

Figure 2.1 Figure 2.2 Figure 3.1 Figure 6.1 Figure 6.2 Figure 6.3

Different Approaches for Recognition of Loan Losses . . . . . . General Approach under IFRS 9 . . . . . . . . . . . . . . . . . . . . . . . . Realized U.S. stock market volatility, January 1900 to April 2020 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discretionary LLPs over Years . . . . . . . . . . . . . . . . . . . . . . . . . Discretionary LLPs over Year-Quarters during Covid-19 . . . . Discretionary LLPs by Different Covid-19 Response Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

6 11 17 38 39 40

xiii

List of Tables

Table 5.1 Table 5.2 Table 5.3

Table 6.1 Table 6.2 Table 6.3 Table 6.4 Table 6.5

Descriptive Statistics of Bank Specific Variables and Sovereign CDSs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Covid-19 Response Measures by Countries . . . . . . . . . . . . . . . . Pairwaise Correlation Coefficients between Bank Specific Variables, Changes in Sovereign CDSs and Numbers of Covid-19 response measures . . . . . . . . . . . . . . . . . . . . . . . . . . The Effect of Covid-19 on Income Smoothing . . . . . . . . . . . . . The Effect of Covid-19 on Income Smoothing: Staging Components of Loans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cross-Country Differences on the Effect of Covid-19 on Income Smoothing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cross-Country Differences on the Effect of Covid-19 on Income Smoothing: Staging Components of Loans . . . . . . . Downwards and Upwards Earnings Management . . . . . . . . . . .

31 34

36 42 43 45 47 49

xv

1

Introduction

The Covid-19 pandemic, declared on 11 March 2020, poses a challenge in many respects. Adding to the severe health consequences and loss of human lives, the costs for the economy are unprecedented. Lock-down measures implemented by many governments in order to halt the spread of the virus have had a great effect on the economic activity. Many industries ground to a halt and consequently the unemployment numbers increased in almost every country (Coibion et al., 2020). The world GDP contracted by 4,4% in the year 2020, the sharpest decline since the Great Depression (Gopinath, 2020). As a result, many individuals and organizations experienced significant difficulties in servicing their liabilities and paying their debt obligations. Moreover, the amount of uncertainty experienced during the pandemic has exacerbated the crisis with capital markets reaching their highest level of volatility since the Global Financial Crisis in 2018 (Baker et al., 2020; Caggiano et al., 2020). The potential negative effects on the financial stability urged many governments and institutions to undertake a range of responsive measures (IMF, 2021). The response included, among others, suspension of debt payments such as public and private moratorium, issuance of public guarantees, as well as direct financial support to businesses that were directly impacted by the crisis (Skrabka, 2021). Financial institutions around the world experienced substantial difficulties in estimating the impact of the crisis on their loan portfolios. IFRS 9, the current accounting regime for financial instruments, requires banks to account for expected credit losses by considering forward-looking information when estimating the future performance of their loans. Based on a three-stage model banks are required to track changes in the credit risk of their loan portfolios and account for loan losses on a continuing basis. A financial asset is initially recognized at stage 1. If the asset experiences a significant increase in credit risk since its initial recognition, loan © The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2023 M. Lamaj, The Effect of Covid-19 on Loan Loss Provisions and Earnings Management of European Banks, BestMasters, https://doi.org/10.1007/978-3-658-40060-6_1

1

2

1

Introduction

loss provisions are recognized at the amount of lifetime credit losses and the asset is transferred to stage 2. Finally, transfers to stage 3 are triggered by objective evidence of incurred losses (i.e., the financial instrument becomes credit impaired) (EY, 2018). The implementation of IFRS 9 represents a substantial change in how banks account for loan loss provisions. While under the previous accounting regime of IAS 39, banks recognized provisions based on only incurred credit losses (i.e., provisions were built only for credit impaired assets) (EY, 2018), IFRS 9 requires banks to also incorporate expected credit losses into their loan loss provisions. Nevertheless, the Covid-19 pandemic has significantly increased the efforts of financial institutions in applying IFRS 9. In particular, the uncertainty experienced during the pandemic poses a great challenge for banks in estimating the potential amount of bad debts and loan losses in their portfolios (Barnoussi et al., 2020). Many prudential authorities and regulators have anticipated the negative effects of the pandemic on banks’ financial statements and their regulatory capital. Accordingly, they intervened by issuing guidance on the implementation of IFRS 9 in the context of Covid-19 (Barnoussi et al., 2020). Concerned about any potential “cliff-effect” of banks’ loan loss provisionings, authorities restrained banks from using procyclical assumptions in their models and encouraged them to fully use the flexibility embedded in the standard (ECB, 2020b, 2021b). Overall, the recommendations were based on the strong presumption that the financial distress experienced during Covid-19 was temporary and that the economy will rebound in the near future (EBA, 2020; ECB, 2020b). The exceptional circumstances of the Covid-19 crisis and its potential negative effects on the financial stability were the main reasons behind the unprecedented guidance and support of many supervisory authorities. However, it is unclear how this environment affects banks’ discretion when accounting for loan losses. Indeed, accounting for future losses requires a certain amount of judgment in their estimation. On the one hand, banks are purposely let certain flexibility in order to better reflect credit losses by using their private information on the future performance of the loan portfolios (FSF, 2009; IASB, 2009). On the other hand, the flexibility embedded in the standard can increase banks’ incentives to manage earnings by using discretion opportunistically (Kirschenheiter & Melumad, 2002). The high level of uncertainty during the pandemic and the unusual response of prudential authorities in the context of Covid-19 creates a different environment for banks’ financial reporting. In its supervisory review of 2020, the European Central Bank has evidenced heterogeneous practices among European institutions in implementing IFRS 9. In their financial reportings for the pandemic year, some banks introduced certain adjustments in their models, such as smoothing factors in order to ease the overall impact of the Covid-19 on their financial statements.

1

Introduction

3

Moreover, the variability of propability of default (PD) estimates used to calculate expected credit losses increased significantly during 2020. In this regard, some banks’ estimates of the PD used under IFRS 9, which should represent a point-intime measure, are lower than the through-the-cycle PD estimates used for regulatory purposes. Such practices are unusual in a context of a crisis, where point-in-time PDs are normally higher than through-the-cycle PDs, which consider a whole economic cycle into their estimation (EBA, 2021). Based on this anecdotal evidence, I examine the effect of Covid-19 on banks’ earnings management through the use of loan loss provisions under IFRS 9. Using a sample of listed European banks, I find that during Covid-19 banks use significantly more discretion in their loan loss provisions than before Covid-19. While at the beginning of the pandemic banks tend to overstate provisions, this trend decreases throughout the year, whereas during 2021 banks do, on average, understate loan loss provisions. In addition, consistent with Gebhardt and Novotny-Farkas (2011), I find that under IFRS 9 banks use loan loss provisions to smooth earnings both in the pre- and post-Covid-19 period. However, during Covid-19 banks engage in upwards earnings management by understating provisions, whereas before Covid-19 earnings management is only evident through overstatements of loan loss provisions. These findings are consistent with banks’ incentives to smooth the impact of the Covid-19 in their financial statements, as also evidenced in the EBA’s report on the implementation of IFRS 9 (EBA, 2021). In addition, I examine cross-country differences in Covid-19 response measures implemented by governments during the pandemic. Countries are coded to have a strong(weak) Covid-19 response policy if the median number of Covid-19 response measures implemented by the country is higher(lower) than the median number of Covid-19 response measures implemented by all countries throughout the pandemic. I find that income smoothing behavior of banks is only evident in countries that have implemented strong Covid-19 response measures. Given the fact that countries with strong Covid-19 response measures experience a greater financial distress during the pandemic (Deb et al., 2021), these findings are consistent with the explanations that under a stressed environment banks have greater incentives to avoid losses and earnings decreases (Kirschenheiter & Melumad, 2002). However, these results should be interpreted with caution, since they might be driven by other differences in country characteristics rather than those related to the response measures implemented during the pandemic. Finally, I examine differences in the three staging components of loans based on the ECL-model under IFRS 9. I observe that despite the anticipated significant increase in LLPs in the pandemic year, Covid-19 is characterized by a significant increase in loans transferred to stage two. In addition, I find a positive relationship

4

1

Introduction

of LLPs and changes in stage two loans during the Covid-19 period. This is mainly due to the impact of the Covid-19 crisis on the credit risk of the borrowers that led to an increase in financial instruments that have experienced a significant increase in credit risk, and are therefore transferred to stage two. Consequently, loan loss provisions incorporate lifetime credit losses, therefore increasing the amount of loan loss allowances during the pandemic, especially of those related to stage 2 loans. These findings give important insights on the potential impact of the Covid-19 pandemic on the loan loss provisions of banks under the current expected credit loss model. They show that while the flexibility provided during the pandemic may be necessary to avoid procyclicality and exacerbation of the crisis, its opportunistic use when provisioning for loan losses may result in distorted reporting of banks’ capital and their ability to withstand the crisis. These results are relevant to both standard setters and prudential authorities.

2

IFRS 9 and the Expected Credit Loss Model

This chapter aims to describe the key requirements of IFRS 9 on accounting for loan loss provisions. First, it describes the main rationale behind credit loss provisioning. It proceeds by giving a short history of the standard development and finally, the last section summarizes the main implications of the ECL-model on banks’ loan loss provisioning practices.

2.1

Why Do Banks Account for Credit Losses?

Banks are a very important part of a crucial process that enables flows of capital from savings into productive activities (Allen et al., 2004). In doing so, they invest in financial instruments (e.g., loans, debt securities and equity interests) that promise contractually specified payments at contractually specified point in times. However, most investments are expected to yield less than promised, for example, when the borrower fails to meet its contractual payments (i.e., the borrower defaults). In this case a portion of the principal and accrued interest in the investment is lost, referred to as a credit loss (Ryan, 2007). Changes in the probability of default (PD), which is the likelihood that a borrower will not meet its contractual obligations, can be driven by changes in individual characteristics of the debtor and the financial asset or by changes in the economic conditions. In order to compensate lenders for the risk inherent in their loan portfolios, the initial yield on the instrument must include an appropriate premium (Ryan, 2007). At initial recognition banks recognize financial instruments on their balance sheets at the present value of expected cash receipts, which represents the economic value of the asset and already reflects any adjustments needed to compensate for expected losses due to the probability of future default (Benston & Wall, 2005). © The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2023 M. Lamaj, The Effect of Covid-19 on Loan Loss Provisions and Earnings Management of European Banks, BestMasters, https://doi.org/10.1007/978-3-658-40060-6_2

5

6

2

IFRS 9 and the Expected Credit Loss Model

After initial recognition, the accounting value of the financial asset, which is the value that the asset is presented in the financial statements, does not necessarily reflect it’s economic value. Their resemblance depends mainly on how financial institutions account for their assets and reflects the accounting standard and approaches they use during their measurement. As a consequence, accounting for financial instruments not recognized at economic values involves estimating future expected losses in order to adjust for changes in the expectations of the borrower’s PD as well as changes in the interest rates (Novotny-Farkas, 2016; Ryan, 2007). Expected losses are calculated as follows: E Lt =

N  LG Dt (It ) [P Dt (It ) ] (1 + dr )t

(2.1)

t=1

where E L are the expected lifetime losses; P Dt (It ) is the probability of default; LG D is the loss given default; dr is the discount rate that is used to discount expected cash flows (Novotny-Farkas, 2016). Figure 2.1 illustrates different approaches for recognizing loan losses. Under fair value accounting, the accounting value of the loan corresponds to its economic value and incorporates all economic losses (including changes in PD and changes in interest rates). The expected credit loss approach recognizes only expected future

Figure 2.1 Different Approaches for Recognition of Loan Losses. (Source: Gebhardt and Novotny-Farkas (2011))

2.2 “Too little, too late” Provisioning and the Transition to the ECL Model

7

credit losses. It does not incorporate changes in the loan value due to changes in the interest rate. Lastly, the incurred loss approach requires only the recognition of incurred losses, which are expected losses, for which a credit event has already occurred as of the balance sheet date and the expected PD of the borrower is nearly one (Gebhardt and Novotny-Farkas, 2011).

2.2

“Too little, too late” Provisioning and the Transition to the ECL Model

The Global Financial Crisis in 2008–2009 underlined the importance of banks and their ability to influence the economic cycle by reinforcing the interaction between the financial sector and the real sectors of the economy (Allen et al., 2008; Allen et al., 2004). The Financial Stability Forum (FSF) in 2009 highlighted that procyclicality in the financial system had a disruptive effect in the crisis. One of the main issues, according to the FSF, was the “too little, too late” recognition of credit losses due to the bank’s loan loss provisioning practices (FSF, 2009). In this regard, the FSF asked the Financial Accounting Standard Board (FASB) and the International Accounting Standard Board (IASB) to encourage the use of judgment and the flexibility embedded in the standard in order to to ensure that loan loss provisions properly reflect credit losses inherent in the loan portfolios (FSF, 2009). Following the crisis, G20 Leaders also called for action on standard setters to improve standards on valuation and provisioning of loan losses (G20, 2009). The impairment requirements on financial instruments stated in IAS 39, the standard applicable at the time, are based on an incurred credit loss (ICL) model, under which banks recognize credit losses only when a credit loss event occurs (EY, 2018).1 While the interest rates charged by the bank captures the change in credit risk over the life of the loan, loan impairment is delayed until the credit event is known to the creditor. As a result, interest revenues are overstated in the period before credit losses are recognized (IASB, 2009). This mismatch allows bankers to create too little provisions in good times. As a consequence, they are forced to increase provisions during downturns (when many borrowers default), exacerbating therefore the crisis effect (Laeven & Majnoni, 2003). In its report of July 2009, the Financial Crisis Advisory Group (FCAG) identified delayed recognition of loan

1

According to IAS 39, “credit losses are recognized only if there is an objective evidence of impairment as a result of a loss event that occurred after the initial recognition of the financial asset and the effect of that loss event on the future cash flows can be reliably estimated ”.

8

2

IFRS 9 and the Expected Credit Loss Model

losses as a primary weakness of the accounting standard (FCAG, 2009).2 A further weakness of banks’ loan loss provisionings, which raised concerns in the context of the GFC, was the diversity in approaches used by entities when calculating loan impairments (FSF, 2009). Lastly, by excluding expected credit losses, provisions under IAS 39 create an information deficiency that disadvantages investors and bank regulators (IASB, 2009). That is, delaying the recognition of credit losses by recognizing them only when these incur, understates bank’s loan loss provisions, since provisions do not incorporate losses that are already known but expected to occur in a latter period. In response to these concerns and the recommendations of regulatory bodies and prudential authorities, in November 2009, the IASB issued the Exposure Draft— Financial Instruments: Amortized Costs and Impairment. The draft proposed an impairment model based on expected credit losses and aimed to provide a more timely recognition of credit losses (IASB, 2009). However, concerns about the operational difficulties in the implementation of the new ECL-model led IASB to issue another Exposure Draft on March 2013 that involved less operational complexity (IASB, 2013). The draft aimed to establish a threshold of significant increase in credit risk and introduce a sound approach for estimating credit losses (IASB, 2013). Finally, the IASB issued the new impairment requirements in July 2014 as part of the final version of IFRS 9 with mandatory effective date on 1 January 2018. Thereafter, the IASB established the IFRS Transition Resource Group for Impairment of Financial Instruments (ITG) in order to provide support in the implementation of the new requirements (ITG, 2015). Further guidance was provided by the Basel Committee on Banking Supervision (BCBS) (BCBS, 2015), Global Public Policy Committee (GPPCI) (Mark Rhys, 2016), as well as the European Banking Authority (EBA) (EBA, 2017). All of them aimed to support financial institutions by promoting a high standard in the implementation of IFRS 9.

2.3

The Main Differences Between IFRS 9 and IAS 39

The biggest change in the IFRS 9 relates to the impairment of financial assets. The new standard eliminates the IAS 39 threshold for recognition of credit losses and requires entities to account for expected credit losses (ECL) on a continuing basis irrespective of the objective evidence that a credit loss has incurred. The inclusion 2

FCAG was created in October 2008 as part of the joint approach of IASB and the USbased standard setter, the Financial Accounting Standard Board (FASB), in order to deal with financial reporting issues arising from the financial crisis.

2.4 The ECL-Model Under IFRS 9

9

of expected credit losses rather than of incurred losses only, makes provisions under IFRS 9 more forward looking. That is, entities shall not consider only historical information, but are now required to consider also forward-looking information when estimating credit losses.3 Furthermore, the scope of impairment requirements under IFRS 9 is much broader (EY, 2018). Previously, IAS 39 imposed different impairment models for financial assets measured at amortized cost and available-for-sale financial assets. In contrast, IFRS 9 introduces a single impairment model for all debt instruments measured at amortized costs and at fair value through other comprehensive income (EY, 2018; PWC, 2017).4 Section 2.4 describes the ECL-model in more detail.

2.4

The ECL-Model Under IFRS 9

The new standard of IFRS 9 requires a consistent approach for recognizing credit losses that includes a broader information set and more compatible rules for financial asset reporting. The expected credit loss model under IFRS 9 applies to all financial assets measured at amortized cost and at fair value through other comprehensive income (OCI) (IFRS 9 5.5.1).

2.4.1

The Measurement of Expected Credit Losses

IFRS defines credit losses as the difference between all promised cash flows specified in the contract and cash flows that the entity expects to receive discounted at the original effective interest rate (EIR).5 The expected credit loss is calculated as follows: EC L = E AD × P D × LG D (2.2) where E AD is the exposure at default that represents the amount of payments due at the time of default discounted at a reasonable rate that reflects the time value 3

IFRS 9.5.5.10 stipulates that while assessing a significant increase in credit risk “an entity shall consider reasonable and supportable information, that is available without undue cost or effort”. 4 Furthermore, loan commitments and financial guarantee contracts, previously under the scope of IAS 37 Provisions, Contingent Liabilities and Contingent Assets, are now under the scope of IFRS 9. 5 EIR is the rate that discounts all future payments considering all contractual terms of the financial instrument but not including the expected credit losses (IFRS 9, Appendix A).

10

2

IFRS 9 and the Expected Credit Loss Model

of money (IFRS 9.5.5.17 (b)). P D is the probability of the borrower’s default and LG D is the loss given default. The latter reflects the cash flow shortfalls in case of a default and is presented as a percentage of the E AD. The PD used by entities to estimate credit losses reflects a probability-weighted estimate that is determined by evaluating a range of possible scenarios (IFRS 9.5.5.17 (a)), normally defined as worst, base and best case scenario. This can include, for example, the use of multiple macroeconomic scenarios (EY, 2018). While entities are not expected to consider every possible scenario, the scenarios considered are required to reflect a representative sample of possible outcomes (ITG, 2015). The use of multiple scenarios affects not only the PD, but also the LGD and the EAD (EY, 2018). In calculating credit losses under IFRS 9, entities are required to carefully consider all contractual terms (including prepayment extension, call and similar option) over the expected life of the financial instrument (IFRS 9.5.5.19). This may also include cash flows from the sale of collateral held or other credit enhancement that are integral to the contractual terms (IFRS 9.B5.5.55). Moreover, the standard requires financial institutions to consider “all reasonable and supportable information that is available without undue cost or effort at the reporting date about past events, current conditions and forecasts of future economic conditions” (IFRS 9.5.5.17 (c)). Although IFRS 9 does not oblige financial institutions to consider external information, it requires them to consider information from a variety of sources available in order to ensure that the information used is reasonable and supportable (ITG, 2015). Finally, the incorporation of forward-looking scenarios requires entities to exercise a certain amount of judgment and discretion when estimating expected credit losses. However, this is constrained by requirements for additional disclosures, as also presented in the IFRS 7—Disclosure on Financial Instruments (ITG, 2015).

2.4.2

Three Stages of Credit Risk

IFRS 9 introduces a general and a simplified approach for recognizing loan credit losses. Under the general approach the standard requires an entity to recognize a loss allowance equal to either the 12-month expected credit losses or the lifetime expected losses, dependent on whether there has been a significant increase in credit risk (SICR) of the financial instrument since its initial recognition (IFRS 9.3.3.3, 5.5.5). Figure 2.2 summarizes the general approach under the ECL-model based on three stages.

2.4 The ECL-Model Under IFRS 9

11

Figure 2.2 General Approach under IFRS 9. (Source: EY (2018))

Stage 1 includes all financial assets measured at amortized cost or at fair value through OCI that have not experienced a significant deterioration in credit risk since their initial recognition. For these assets a loss allowance that equals the 12-month ECLs should be measured. 12-month ECLs are credit losses that result from default events that are possible within the next 12 months (IFRS 9.5.5.5, Appendix A). The interest revenue on financial assets at stage 1 is calculated based on the effective interest rate on the gross carrying amount of the asset (IFRS 9.5.4.1). If after the initial recognition the financial asset experiences a significant increase in credit risk (SICR), but there is still no objective evidence that a credit event has occurred, the asset is transferred to stage 2 and the loss allowance is measured based on the lifetime expected credit losses. The interest revenue of financial assets in stage 2 is measured based on the effective interest rate on the gross carrying amount of the asset (IFRS 9.5.5.3, 5.4.1). If one or more events happen that have a detrimental impact on the estimated future cash flows (i.e., financial asset has become credit impaired), the asset is

12

2

IFRS 9 and the Expected Credit Loss Model

transferred to stage 3.6 However, there is a rebuttable assumption that a default does not occur later than when a financial asset is 90 days past due (IFRS 9.B5.5.37). Loan loss allowances for loans in stage 3 are measured at lifetime ECLs, while the interest revenue is measured based on the effective interest rate on the net carrying amount, which is the gross carrying amount less the loss allowance (IFRS 9.5.5.3, 5.4.1(b)). In contrast to the general approach, the simplified approach does not require entities to track changes in the credit risk of the financial instruments. Instead, it allows the recognition of loan allowances only on a lifetime- ECLs-basis for trade receivables and contract assets that do not contain a significant financing component (IFRS 9.5.5.15(a)(i-ii)). The simplified approach applies also for lease receivables within the scope of IFRS 16—Leases (IFRS 9.5.5.15(a)(iii)).

2.4.3

Determining the Significant Increase in Credit Risk (SICR)

One of the major challenges of the ECL-model under the general approach is the determination of the significant increase in credit risk (SICR) of the financial assets. The standard requires an entity at each reporting day to track whether there has been a SICR in its financial assets since their initial recognition (EY, 2018). This assessment is very important because it impacts different stage classifications of financial assets and, consequently, the measurement of the loan loss allowances. IFRS 9 stipulates that “generally, there will be a significant increase in credit risk before a financial asset becomes credit impaired, or an actual default occurs” (IFRS 9.B5.5.7). When assessing whether a financial asset has experienced a SICR, an entity should consider both qualitative and quantitative information about the future performance of the asset. Quantitative information normally includes the assessment of changes in credit risk of the borrower. However, the change in the credit risk cannot be assessed only by comparing the change in the absolute risk of the default over time, because the risk of default usually decreases as time passes (EY, 2018). Regarding qualitative indicators, the IFRS 9 includes a non-inclusive list of information that may indicate a SICR. This contains, for example, significant changes in internal or external ratings of the financial instrument or adverse changes in financial and economic conditions that are expected to cause a significant change in the borrower’s ability to meet his or her debt obligations (IFRS 9.B5.5.17). 6

The standard includes a non-inclusive list of events that may lead a financial asset to be credit impaired. This includes significant financial difficulties of the borrower or a breach of contract, such as default or past due events.

2.5 Main Implications of the ECL Model

13

In addition, under IFRS 9 an entity can make a rebuttable assumption that the credit risk of a financial asset has increased significantly since its initial recognition when contractual payments are more than 30 days past due (IFRS 9.5.5.11). Similarly, an entity need not assess the increase in credit risk if the financial instrument is determined to have a low credit risk at the reporting date (IFRS 9.5.5.11).

2.5

Main Implications of the ECL Model

The introduction of IFRS 9 is considered to be a big change in the financial statements of banks, mainly because it significantly changes the way banks account for loan loss provisions (LLP). Due to the incorporation of expected credit losses, the new model leads to earlier recognition of credit losses and induces larger loan loss allowances (LLA), especially during good times, therefore mitigating the procyclicality concerns under the incurred credit loss model (Novotny-Farkas, 2016). Since loan loss provision is a key accrual in the bank’s financial statements, it also impacts the regulatory capital (Benston & Wall, 2005). Bank regulators use accounting numbers to calculate regulatory capital ratios. Their man objective is to ensure financial stability by assuring that banks maintain sufficient loan loss allowances to absorb expected losses and sufficient capital for unexpected losses (Benston & Wall, 2005; Novotny-Farkas, 2016). Nevertheless, the borrower’s probability of default used in the ECL-model should represent a point-in-time measure rather than a through the cycle measure, which is used by banking regulators.7 As a consequence, ECLs measured under IFRS 9 can affect banking regulation in that they add volatility to the capital ratios in cases when expected losses under IFRS 9 exceed the expected losses estimated by the regulators, and this is especially true during a downturn (Novotny-Farkas, 2016). Importantly, the incorporation of forward-looking information requires a significant amount of judgment and discretion. On the one hand, this allows the incorporation of private information banks have on the future performance of their loan portfolios.8 On the other hand, accounting discretion can create a potential for opportunistic behavior of managers (Bushman & Williams, 2012). As a consequence, it can prevent LLAs to appropriately reflect loan losses inherent in bank’s balance 7 Point-in-time PDs fluctuate more than through-the-cycle PDs. Moreover, through-the-cycle PDs tend to overstate point-in-times PDs during good times and understate them during downturns (Novotny-Farkas, 2016). 8 This is also one of the main reasons why, following the global financial crisis, the Financial Stability Forum (FSF) asked the International Accounting Standard Board (IASB) to encourage the use of judgment and flexibility embedded in the standard.

14

2

IFRS 9 and the Expected Credit Loss Model

sheets, therefore affecting financial stability. Bushman and Williams (2012), for example, shows that bank’s discretion dampens market discipline over risk-taking. Chapter 4 in the following explains earnings management and its implications for banks in more details.

3

Covid-19 and European Banks

This chapter outlines the main implications of the Covid-19 pandemic on banks’ loan loss provisions. It begins by giving a short overview of the Covid-19 pandemic and its main economic consequences. Furthermore, it summarizes the recommendations of prudential and regulatory authorities regarding banks’ loan loss provisionings in the context of Covid-19. Finally, the last section gives an overview of the loan loss provisioning practices evidenced by European supervisory institutions during the pandemic year.

3.1

Covid-19 at a Glance

In December 2019, a new coronavirus, named Covid-19, appeared in the city of Wuhan in the Hubai province in China. The infectious disease has a very high deathrate and has spread rapidly around the world. Europe reported its first Covid-19 cases not later than January 2020 and since then the increasing numbers of infected people around the region have become hardly traceable (Kinross et al., 2020). Given the world-wide spread of the virus by the time, the Director General of the World Health Organization (WHO) declared Covid-19 a pandemic on 11 March 2020 (WHO, 2020). Within almost two years since its first appearance, global cases of the Covid-19 confirmed patients according to the WHO have reached 262,178,403 with 5,215,745 deaths (as of December 2).

Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/978-3-658-40060-6_3. © The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2023 M. Lamaj, The Effect of Covid-19 on Loan Loss Provisions and Earnings Management of European Banks, BestMasters, https://doi.org/10.1007/978-3-658-40060-6_3

15

16

3

Covid-19 and European Banks

The severe health consequences on human lives and the increased occupancy in wards and intensive care units of hospitals urged many countries around the world to implement stringent containment measures to counter the effects of the pandemic. In Europe, many countries have significantly curbed the public life in order to halt the spread of the disease. Measures like school closures, banning of public events, social distancing as well as full and partial lock-downs were taken in almost every European country. Figure A.1 in appendix A shows, for example, a trajectory of England’s Covid-19 response measures taken during the pandemic.

3.2

Economic Consequences of the Pandemic

Adding to the human toll, the economic costs of the pandemic were unprecedented. The International Monetary Fund (IMF) described the Covid-19 as “a crisis like no other” (IMF, 2020b) and the great lock-down to be the “the worst downturn since the Great Depression” (Gopinath, 2020). According to the IMF the pandemic year resulted in a decrease in global GDP of 4,4 percent, a decline three times bigger than the one experienced during the Global Financial Crisis. Figure A.2 in appendix A shows the annual growth of global GDP from the year 1995. In its World Economic Outlook of October 2020, the IMF finds lock-down measures to be an important driver of the recession (IMF, 2020b). In particular, Deb et al. (2021), using a world wide-sample of countries, finds that containment measures have had, on average, a very large impact on economic activity—equivalent to a loss of about 10 percent in industrial production over the 30-days period following the implementation of the full lock-down. Furthermore, voluntary social distancing in order to prevent infection has also exacerbated the size of the recession (IMF, 2020b). Accordingly, Eichenbaum et al. (2021) describes a two-way interaction between the epidemic and the economy, namely the aggregate demand and the aggregate supply effect. On the one hand, the supply effect arises because the epidemic exposes workers to the virus and the workers react to the risk by reducing their labor supply. On the other hand, demand is affected because consumers, also affected by the virus, react by reducing consumption. As a result, both effects together generate a larger recession. In response to the downturn many governments reacted quickly by adjusting their fiscal and monetary policies in order to ease out the negative effects on the economy. The IMF described the response measures across countries as “unprecedented in both size and the lifeline intentions to escape the direct negative impact of the pandemic” (IMF, 2021). The short-term trade-off between minimizing health risks and avoiding economic losses has corroborated the challenges governments had to

3.2 Economic Consequences of the Pandemic

17

deal with during the crisis. Deb et al. (2021) shows, for example, that stay-at-home measures and workplace closures, the two containment measures which are the most effective in curbing both infections and deaths, are also the costliest in economic terms. Nevertheless, the fiscal and monetary policy stimulus implemented during the pandemic has had a softening effect on the negative impacts of the crisis (Deb et al., 2021; IMF, 2020a). Another main feature of the Covid-19 crisis is that since its appearance and even at the time of writing it has been characterized by a great amount of uncertainty. This is mainly due to the fact that the severity of the pandemic is directly driven by the spread of the Covid-19 virus and its ability to change into more infectious variants, therefore escaping the human immunity response built either naturally or artificially through anti-Covid vaccination. Such uncertainty has a great impact on the economy. Figure 3.1 shows the evolution of the financial market volatility over time and the Covid-19 induced uncertainty with regard to the capital markets in the United States. As also pointed out by Caggiano et al. (2020) and Baker et al. (2020), the peak value of financial volatility recorded in March 2020 is the highest recorded in recent history, exceeding that of the Global Financial Crisis in 2008 and even the Great Depression. Finally, as the Covid-19 virus continues its pathway, although less intensively, the global economy is recovering. Still the uncertainty remains high and the intervention of fiscal and monetary policies becomes more complex since the long-term consequences of the crisis are not fully clear. Furthermore, in its World Economic Outlook of October 2021 IMF states that “if Covid-19 were to have a prolonged impact into the medium term, it could reduce global GDP by a cumulative $5.3 trillion over the next five years.”

Figure 3.1 Realized U.S. stock market volatility, January 1900 to April 2020. (Source: Baker et al., 2020)

18

3.3

3

Covid-19 and European Banks

Covid-19 Implications on Loan Loss Provisions Under IFRS 9

In 2018, European banks published for the first time their annual reports based on IFRS 9. At the time, financial institutions had spend a great effort in developing their models in order to fit with the requirements of the new standard. Nevertheless, the Covid-19 pandemic brought the challenge to another level and uplifted their efforts in implementing the ECL-model in a sound manner. On the one hand, the crisis led many borrowers to experience major difficulties in servicing their liabilities and paying their debts. Consequently, banks faced considerable uncertainty about the potential amount of bad debts and loan losses for which they need to account for (Barnoussi et al., 2020). On the other hand, in order to mitigate the negative economic impact of the crisis and help financially distressed households and businesses overcome the situation, many governments have undertaken a broad range of relief measures. These include among others suspension of debt payments (public moratorium), public guarantees as well as overdraft facilities and mortgages. Banks has also engaged in this process by supporting impacted borrowers in renegotiating contractual terms and providing holiday payments (private moratorium) (Skrabka, 2021). Their intention is to avoid a roll-up of negative consequences on their borrowers. These actions, however, have imposed further difficulties in implementing the requirements of IFRS 9. Despite of these supporting measures, standard setters and supervisory bodies in Europe have rushed to develop their guidance regarding Covid-19 implications on banks’ financial reporting (Barnoussi et al., 2020). The European Banking Authority (EBA), the European Securities and Markets Authority (ESMA), the European Central Bank (ECB) and the International Accounting Standard Board (IASB) have issued their coordinated guidance on the application of IFRS 9 in the context of Covid-19. The following subsections summarize their response and recommendations to the European banks.

3.3.1

Concerns Over Procyclicality

Loan loss provisions under the ECL-model should absorb much of the Covid-19 induced losses already in the early stages of the crisis. A major concern of the supervisory authorities in this regard has been the excessive procyclicality the crisis might have on the loan loss provisions of banks, affecting therefore, capital adequacy ratios, while the latter might have a considerable impact on the financial stability. On 1 April 2020, in a public letter addressed to significant European banks, Andre Enria,

3.3 Covid-19 Implications on Loan Loss Provisions Under IFRS 9

19

the Chairman of the Supervisory Board of ECB, encouraged banks to avoid using procyclical assumptions in their models when accounting for loan losses (ECB, 2020b, 2021b). In mitigating these concerns the ECB has also introduced a range of capital and operational measures by assigning capital and liquidity buffers in order for banks to withstand the stressed situation during the pandemic and also to be able to continue financing households and businesses experiencing temporary difficulties (ECB, 2021a).

3.3.2

The Use of Flexibility

Because of the high amount of uncertainty during the crisis, many supervisory authorities have anticipated the limitations of the operational capability of banks when making in-depth-assessments in the context of Covid-19. In this regard, they have encouraged reporting entities to take advantage of the flexibility embedded in the standard when incorporating Covid-19 specific conditions into their ECLmodels. In particular, the ESMA outlined the sufficient flexibility included in IFRS 9 and its importance “to faithfully reflect the specific circumstances of the Covid19 outbreak and the associated public policy measures” (ESMA, 2020). Moreover, the EBA described the use of the short-term flexibility as “warranted”, given the circumstances of the pandemic (EBA, 2020). Authorities have encouraged banks to use flexibility, especially, in the assessment of the lifetime impact of the crisis on the credit risk of the financial instruments. An implication in this regard is, for example, modifications resulting from the introduction of support measures in order to counter the negative economic effects of the crisis. While entities should assess the impact such changes may have on the expected credit risk over the expected lifetime of the borrower, they are required to take into consideration the short length of these measures. Accordingly, banks should take into account the lifetime impact versus the temporary impact such measures might have on the credit risk of the financial assets (EBA, 2020; ESMA, 2020). Lastly, the use of flexibility can make banks’ managers prone to opportunistic behaviors. In particular, the incorporation of forward looking information when making in-depth assessments in the context of the Covid-19 might be subject to significant judgments and discretion (ESMA, 2020). Nevertheless, supervisory bodies have strongly emphasized the importance of sound application of provisioning requirements under IFRS 9 and that their recommendations are merely to be seen as a guidance and a reminder of the actual standards rather than exceptions in the light of the current situation (EBA, 2020; ECB, 2020b; ESMA, 2020).

20

3.3.3

3

Covid-19 and European Banks

No Mechanistic Approach

Given the exceptional circumstances of the Covid-19 pandemic, the IASB and supervisory authorities have expressed their worries about the use of any mechanistic approach in applying IFRS 9, especially regarding the estimations of loan losses within the ECL-model. (EBA, 2020; ECB, 2020b; ESMA, 2020; IASB, 2020). This affects primarily the SICR assessments (i.e., whether an asset should be transferred to stage 2). All recommendations issued during the crisis were consistent in that public or private moratorium do not automatically mean that the borrowers have experienced a significant increase in credit risk. Indeed appropriate assessment should be made on a case-by-case basis whether this holds true and if this is not possible, then other approaches such as the top-down or the bottom-up approach can be considered. Moreover, the past-day rebuttable assumptions under the IFRS 9 require further considerations, as they may loose relevance in the context of the Covid-19 (EBA, 2020; ECB, 2020b; ESMA, 2020; IASB, 2020). Nevertheless, entities can address these issues by providing adjustments to their models in order to properly reflect the pandemic effects in their ECL calculations. These include, for example, post model adjustments or model overlays that account for the exceptional and specific conditions of the current crisis such as the extension of due payments or the issuance of public loans.

3.3.4

Short Versus Long Term Forecasts

One of the main features of the ECL-model is the incorporation of forward-looking information. Considering the high amount of the uncertainty during the pandemic, reporting entities are experiencing considerable difficulties in producing appropriate future estimates when incorporating forward-looking information into their calculations of expected loan losses. Supervisory bodies have also addressed this issue in their guidelines issued during the pandemic. Particularly, their main focus is the presumption that the economy will rebound in the near future (EBA, 2020; ECB, 2020b). This strong assumption is amplified in their recommendations that encourage entities to put more weight on long-term stable economic outlook rather than on short-term forecasts (ECB, 2020b; ESMA, 2020). Moreover, mean reversion of main indicators can be presumed earlier than normal. In particular, the ECB, in its letter to European significant institutions on 1 April 2020 states that “the ECB would not object to any judgment that this rebound might occur within 2020 given the high current level of uncertainty”. In this regard, the ECB also emphasizes its commitment to continue issuing central

3.4 Banks’ Provisioning Practices During the Pandemic

21

macroeconomic scenarios and encourages entities to incorporate these into their models (ECB, 2020b). Finally, banks should also take into consideration the relevance of the future forecasts used in their models. The ECB recommends entities to put more weight on specific-macroeconomic forecasts and reduce the weighting as the forecasts loose relevance, especially for time horizons in the more distant future (ECB, 2020b).

3.3.5

Transparency and Disclosure

The supervision of management bodies should also play a very important role in implementing the aforementioned recommendations. This aims primarily the adequate response to the Covid-19 induced challenges and the transparency of all actions and adjustments made in the context of the Covid-19 pandemic. In particular, the ESMA and the IASB, in their recommendations during 2020, have underlined the need for increased transparency. This should be mainly reflected in the special disclosures regarding Covid-19 issues such as post model adjustments as well as in the overall impact of the Covid-19 crisis in the financial statements of the bank (ECB, 2020a; ESMA, 2020; IASB, 2020).

3.4

Banks’ Provisioning Practices During the Pandemic

Despite all the guidance and the recommendations that prudential and supervisory bodies in coordination with standard setters have provided to financial institutions, the unprecedented implications of the Covid-19 pandemic has added considerable noise in the financial statements of banks. On 4 December 2020, the ECB identified heterogeneous practices among European significant institutions (SI) while implementing IFRS 9 and the recommendations issued in the context of the pandemic (ECB, 2020a). Moreover, in its supervisory review of 2020 the ECB evidences the weak capacity of SIs to monitor their loan exposures. Accordingly, many banks were unable to take adequate measures in order to recognize the deterioration of their loan loss portfolios in a timely manner (ECB, 2021b). The EBA’s findings in its monitoring report of 24 November 2021 regarding the IFRS 9 implementation by EU institutions also confirm this evidence (EBA, 2021). The following paragraphs summarize its main findings in the context of Covid-19 and its implications on loan loan provisioning practices of European banks.

22

3

Covid-19 and European Banks

The impact of Covid-19 in the credit losses is mainly captured through two main channels within the ECL-model. One is the use of the overlays and post model adjustments either at the level of the risk parameters (i.e., PD, LGD or EAD) or directly at the level of the final ECL amount. During 2020 SIs significantly increased the use of model overlays compared to the year before. Post model adjustments added especially during the first half of 2020 can be attributed to the adjustments made in the context of Covid-19. However, because banks modeled the effects of Covid-19 in very different ways, the impact that such overlays had on the financial statements also varies significantly across financial institutions. The other channel includes the incorporation of forward looking estimates such as, for example, the probability weights assigned to the different scenarios under the IFRS 9. This is generally implemented by introducing additional and more pessimistic scenarios or by assigning higher weight to the original downward scenario. However, the EBA has noticed a certain trend of European banks in trying to avoid variability in their estimates in order to reduce the impact of the incorporation of macroeconomic scenarios on their expected credit losses. Moreover, the variability of 12-month estimates of the PD under IFRS 9 increased significantly during 2020. The change in the 12-month PD varied also significantly across entities. However, banks with a lower increase in 12-month PD have generally compensated by introducing Covid-19 overlays at the ECL-level. Finally, the EBA points out the need for monitoring and further scrutiny of loan loss provisioning practices of banks from an accounting as well as a prudential perspective. Nevertheless, the consistent and prudential application of IFRS continues to be a great challenge for financial institutions and their auditors, since the long-term effects of the Covid-19 pandemic are not fully clear.

4

Hypothesis Development

In this chapter I develop my main expectations with regard to the Covid-19 effect on banks’ earnings management practices. After explaining the main rationale behind earnings management, I give an overview of prior literature related to earnings management under different accounting regimes and economic environments.

4.1

Banks’ Incentives to Manage Earnings

As profit-oriented organizations banks are committed to bringing service to their customers in exchange for a certain consideration (e.g., interest rates and fees) that aims to compensate for the costs incurred plus a return on the investment. The latter is included in the company’s earnings and is very important for satisfying its shareholders and ensuring business continuity. Moreover, earnings are a powerful indicator of a firm’s business activity and play a key role in determining its share price on the capital markets (Rahman et al., 2013). In particular, prior evidence shows that price changes in the stock market are highly associated with earnings announcements (Ball & Brown, 1968; Beaver, 1968). The pressure to meet the market expectations increases managers’ incentives to avoid losses and earning decreases by engaging in earnings management (EM) (Burgstahler & Dichev, 1997; Payne & Robb, 2000). According to Healy and Wahlen (1999) “earnings management occurs when managers use judgment in financial reporting and in structuring transactions to either mislead some stakeholders about the underlying economic performance of the company or to influence contractual outcomes that depend on reported accounting”. Earnings can be manipulated by using accounting discretion when choosing accounting methods based on which managers prepare their financial statements (Sun & Rath, 2010). This includes the © The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2023 M. Lamaj, The Effect of Covid-19 on Loan Loss Provisions and Earnings Management of European Banks, BestMasters, https://doi.org/10.1007/978-3-658-40060-6_4

23

24

4

Hypothesis Development

use of accounting accruals in order to achieve a certain objective that can be reflected in the financial statement numbers. In banks, loan loss provisions are a key and a very important accrual that explains much of the variability in the total accruals of a bank. Because they are estimates of credit losses, loan loss provisions are highly discretionary. Prior literature shows that banks use discretion in their loan loss provisions in order to manage earnings and capital (Beatty & Liao, 2014). While during the pre-BASEL period banks used LLPs to manage their reported regulatory capital (Beatty et al., 1995; Collins et al., 1995), Anandarajan et al. (2007) finds that after the implementation of BASEL banks engage in more earnings management through the use of LLPs. However, the means of earnings management can be diverse. A popular pattern of earnings management is earnings smoothing, which is defined as shifting some of the firm’s income from one period to another. In doing so, managers take advantage of high earnings periods in order to compensate for low earnings periods with the intention to achieve constant and persistent earnings through the years, also using their private information about the future performance of their firms (Scott & O’Brien, 2003; Trueman & Titman, 1988). Another tool for earnings management is the so called “big bath accounting”, which happens mainly during “bad times”, where expenses are unavoidable. In such cases managers overestimate the expenses intentionally in order to release them and realize a profit in the subsequent periods (Rahman et al., 2013). Nevertheless, earnings management does not always has negative consequences. That is, within a given framework of an accounting regime, managers can exercise their judgment in order to inform investors better about the actual performance of their company. Therefore, they can give private signals for the firm’s financial performance that cannot be otherwise reflected in the financial statement numbers (Healy and Wahlen, 1999).

4.2

Earnings Management Under IFRS 9

The accounting regime plays a significant role in manager’s ability to engage in earnings management. Ewert and Wagenhofer (2005) finds that tighter accounting rules reduce the manager’s ability to engage in accounting earnings management practices. Stricter rules can limit opportunities of managerial discretion, thus resulting in less accounting earnings management. Gebhardt and Novotny-Farkas (2011) finds that after mandatory adoption of IFRS and the imposed restriction to account only for incurred losses European banks engage in less income smoothing.

4.3 Covid-19 Effect on Income Smoothing

25

However, the introduction of IFRS 9 imposed the transition from an incurred to an expected credit loss model. It requires banks to incorporate expected losses in their estimations of loan loss provisions. As a result, the new ECL-model offers more flexibility and expects managers to use judgment when calculating their estimates of loan losses. As a consequence, the new requirements create a great potential for opportunistic behavior. Therefore, given the less stricter rules of the IFRS 9, and in analogy with Gebhardt and Novotny-Farkas (2011), I expect that under the accounting regime of IFRS 9 European banks use income smoothing through LLPs. However, I do not examine any difference with the period prior to IFRS 9. Therefore, my first hypothesis is: H1: Under IFRS 9 banks use loan loss provisions to smooth income.

4.3

Covid-19 Effect on Income Smoothing

During the Covid-19 pandemic, the unprecedented response of supervisory authorities and standard setters encouraged European banks to use more than ever the flexibility offered in the standards. Their primary objective while dealing with the consequences of the crisis was to ensure financial stability. Indeed, concerns over any potential exacerbating effects of the pandemic led many authorities to offer temporary reliefs for reporting banks regarding both accounting and capital regulation. Although highly needed, the high flexibility offered during the pandemic led to laxer accounting rules for banks. The outstanding response and support of the authorities came after many governments had reacted to the pandemic by implementing unprecedented measures that had a direct impact in the economy. In particular, the financial distress of the borrowers has affected first and foremost the balance sheets of the banks. Under IFRS 9 banks should be able to estimate and account for this impact through their loan loss provisions even before actual losses incur. While it is clear that more flexibility in accounting rules makes earnings managements more preferable, it is unclear how the financial distress experienced during Covid-19 would impact banks’ incentives to manage earnings through the use of loan loss provisions. Kirschenheiter and Melumad (2002) shows that, when the reporting environment permits discretion, an optimal disclosure policy exists in which the manager either takes a “big bath” or smooths earnings. In cases of “bad news” (i.e., for sufficiently low levels of cash-flows), the manager under-reports earnings by the maximum amount possible, preferring to take a big bath in the current period in order to report higher earnings in the future. When the level of cash-flows starts to increase, the manager smooths earnings. According to Kirschenheiter and Melumad (2002),

26

4

Hypothesis Development

managers over-report in cases of low levels of cash flows, while they decrease overreporting as the situation starts to improve, and in times of “good news” they tend to under-report. However, prior studies on the impact of financial crises on earnings management practices of firms are not consistent in their findings. Using a sample of European listed firms, Filip and Raffournier (2014) finds that earnings management decreased significantly during the years of the Global Financial Crisis (GFC). In contrast, Liang and Dong (2018) finds US real investment trusts to engage in income-increasing earnings management during the GFC, with the magnitude of the earnings management decreasing following the crisis. While looking at the banking sector, Rajan (1994) explains the rationale behind managers’ incentives for changing their credit policies. On the one hand, banks try to convince the market of the profitability of their loans, especially by achieving a certain industry benchmark. On the other hand, the market forgives poor performance in cases of adverse economic conditions. Consistent with these incentives, using a sample of Japanese banks, Agarwal et al. (2007) finds that banks use LLPs as means for managing earnings only during high-growth and stagnant periods, while during severe recessions banks may restrain themselves from using LLPs to smooth income. Moreover, El Diri et al. (2021) finds failing banks to engage in earnings management to a significantly greater extent than non-failing banks. Examining the consequences of accounting discretion, Bushman and Williams (2012) observes that forward-looking provision designed to smooth earnings dampens discipline over risk-taking. In the same context, Bushman and Williams (2015) tests if banks with delayed expected loss recognition (DELR), as a tool for opportunistic loan loss behavior, exhibit lower bank transparency and increased investor uncertainty. They find that during economy-wide crises, banks with high DELR experience more severe recessionary increases in stock illiquidity. The Covid-19 crisis, although different in nature, can trigger similar behavior by banks when accounting for loan loss provisions in the context of the pandemic. The European Banking Authority has observed in its report of IFRS 9 implementation by EU institutions that during the first half of 2020 some banks introduced certain adjustments (e.g., smoothing factors) to their macroeconomic variables under the IFRS 9 scenarios, therefore presenting a lower increase in their 12-month PD estimates for the period. Other observations include also countercyclical changes in the severity of the downward scenarios. Some banks assigned a higher GDP to the worst scenario for 2021 than the GDP assigned to the same baseline scenario. Moreover, some banks did not update the macroeconomic information used in their model, thus relying on pre-Covid-19 forecasts. In addition, the probability weights assigned to macroeconomic scenarios were changed in a manner that a reduced

4.3 Covid-19 Effect on Income Smoothing

27

impact of Covid-19 was presented (EBA, 2021). Based on this anecdotal evidence I expect that during Covid-19 banks use more income smoothing than before the pandemic. Formally stated: H2: During the Covid-19 pandemic banks use loan loss provisions to smooth income to a greater extent than before Covid-19. The response to Covid-19 pandemic and its economic consequences differs across countries. Countries that implemented stronger containment measures in order to halt the spread of the pandemic experienced more adverse economic consequences (Deb et al., 2021). Therefore, I expect income smoothing to be more evident in countries with stronger Covid-19 response measures. My last hypothesis is: H3: During the Covid-19 pandemic banks located in countries with strong Covid19 response measures use more income smoothing than banks located in countries with weak Covid-19 response measures.

5

Descriptive Statistics

This chapter describes the sample and the data used in the research design. Section 1 gives the main features of the sample selection and data screening, whereas Section 3 presents the descriptive statistics of the main variables used in the multivariate analyses.

5.1

Data and Sample

The sample selection starts with all listed banks from 36 European countries including EU and non-EU country members. Panel A of the table B.1 in Appendix B shows the screening criteria. The sample selection yields an initial population of 199 banks. I use data on banks’ quarterly financial statements. Because prior literature finds stock prices highly associated with quarterly earnings announcements (Skinner & Sloan, 2002), I believe quarterly financial results are more closely related to manager’s incentives for managing earnings. The sample period begins from the first quarter of 2018, the time of the IFRS 9 implementation, and ends in the third quarter of 2021. Data on bank’s financial statements are downloaded from the S&P Capital IQ Pro database, whereas data on sovereign CDSs are downloaded from S&P Global Market Intelligence database. The starting sample consists of 3.184 bank-quarter observations. After cleaning for missing variables and merging the data of financial statements with data on sovereign CDSs, as shown in Panel B of table B.1 in Appendix B, the final sample consists of 911 bank-quarter observations for 92 banks from 25 European countries. Data on Covid-19 response measures Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/978-3-658-40060-6_5. © The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2023 M. Lamaj, The Effect of Covid-19 on Loan Loss Provisions and Earnings Management of European Banks, BestMasters, https://doi.org/10.1007/978-3-658-40060-6_5

29

30

5

Descriptive Statistics

of the sample countries are downloaded from the official website of the European Center for Disease Prevention and Control (ECDC).1

5.2

Descriptive Statistics

Table 5.1 presents descriptive statistics of bank specific variables and sovereign CDSs used in the multivariate analyses. Variable definitions are shown in the table B.2 of Appendix B. Descriptive statistics are presented for the pre- and post-Covid period, whereas in the post-Covid period statistics for the year 2020 and year 2021 are displayed separately. As already anticipated, the pandemic year of 2020 is characterized by a significant increase (significant at 0,1% level) in loan loss provisions with the mean (median) increasing from 0,12% (0,04%) in pre-Covid period to 0,22% (0,12%) of average total loans in year 2020 . Regarding the staging components under IFRS 9, loans assigned to stage 2 increased significantly, from 8,13% (median 7%) prior to Covid to 9,51% (median 7,76%) of total loans for the year 2020, whereas loans in stage 1 (3) have experienced a decrease of 0,13% (increase of 0,07%) of total loans, although not significantly. The significant increase of loans in stage 2 is mainly driven by the fact that during the financial distress of 2020 the majority of banks’ loan portfolios have experienced a significant increase in credit risk, and are therefore transferred to stage 2. Accordingly, loan loss reserves of stage 2 increased significantly from 0,4% (median 0,24%) to 0,46% (median 0,3%) of total loans. However, total loan loss allowances (LLA) decreased by 0,1% of total loans for the year 2020, although not significantly. The pandemic year is also marked by a significant decrease in loans from 71,22% (median 74,46% ) to 69,34% (median 72,69%) of total assets. Moreover, regulatory capital, expressed as a percentage of total risk weighted assets, increased significantly from 15,64% (median 15,94%) to 16,16% (median 16,54%). With regard to the macroeconomic indicators, the results indicate that the pandemic year is characterized by an increase (significant at 5 % level) in sovereign CDS changes. This is also captured by figure B.1 in the Appendix B, that shows the 5-year sovereign CDS changes over the period 2018–2021. Panel C of the table 5.1 displays descriptive statistics for the first three quarters of 2021 and the differences to 2020. The year 2021 is characterized by an improvement

1

Data on Covid-19 response measures are downloaded under the link: https://www.ecdc. europa.eu/en/publications-data/download-data-response-measures-covid-19 (retrieved on 19/12/2021)

LLP

0.004

0.036

0.149

0.273

777

p25

p50

p75

sd

N

5.794

4.153

1.670

0.527

3.882

LLA

0.043

0.122

0.272

0.309

326

p25

p50

p75

sd

N

873

(–0.250)

0.086

–0.014

0.025

0.140

0.226

230

mean

p25

p50

p75

sd

N

5.089

3.745

2.011

0.591

3.496

Panel C: Post-Covid (Year 2021)

(3.931)

N

0.083***

t

b

–0.100

5.564

4.426

1.668

0.632

3.944

Diff-stat. (2020 - pre-Covid)

0.223

mean

Panel B: Post-Covid (Year 2020)

0.124

mean

Panel A: Pre-Covid

Period

13.961

80.430

71.680

57.531

68.208

(–3.464)

– 3.149***

13.088

80.702

72.695

60.149

69.342

13.310

82.555

74.466

61.021

71.223

Loans

14.869

92.858

88.213

78.486

85.248

(–0.102)

–0.128

16.145

92.632

86.960

75.448

83.412

16.000

93.154

87.499

78.486

84.103

LoanStg1

5.092

12.022

7.474

5.137

8.906

(4.444)

1.773***

5.703

11.950

7.760

5.647

9.514

4.733

10.502

6.996

4.881

8.127

LoanStg2

9.584

5.159

2.545

1.257

6.104

(0.103)

0.076

10.553

6.037

2.587

1.286

6.795

10.319

6.327

2.560

1.057

6.537

LoanStg3

0.514

0.391

0.194

0.071

0.335

(0.909)

0.036

0.543

0.396

0.184

0.089

0.368

0.521

0.398

0.174

0.064

0.333

LLRStg1

0.420

0.646

0.293

0.127

0.430

(2.220)

0.073*

0.450

0.632

0.292

0.153

0.458

0.435

0.523

0.235

0.118

0.399

LLRStg2

Table 5.1 Descriptive Statistics of Bank Specific Variables and Sovereign CDSs

4.139

2.569

0.841

0.321

2.534

(–1.364)

–0.464

4.411

2.931

0.959

0.331

2.875

4.771

3.014

1.170

0.325

3.091

LLRStg3

3.088

18.006

16.509

13.875

16.032

(3.371)

0.682***

2.962

18.055

16.544

14.117

16.158

2.873

17.365

15.939

13.756

15.641

RegCap

2.242

18.094

16.464

14.819

16.540

(0.060)

0.009

2.262

18.003

16.324

14.435

16.387

2.211

17.936

16.364

14.730

16.426

SIZE

0.384

0.364

0.265

0.184

0.320

(–1.915)

–0.060

0.440

0.413

0.289

0.183

0.356

0.446

0.429

0.302

0.209

0.388

(Continued)

0.110

0.010

–0.036

–0.085

–0.067

(2.041)

0.048*

0.406

0.045

–0.066

–0.206

0.051

0.144

0.027

–0.017

–0.070

–0.017

NEBTLLP Change5YCDS

5.2 Descriptive Statistics 31

LLP

556

(–0.982)

–0.447

LLA

(–0.968)

–1.134

Loans

(1.304)

1.835

LoanStg1

(–0.803)

–0.691

LoanStg3

(–0.683)

–0.032

LLRStg1

(–0.716)

–0.028

LLRStg2

* p < 0.05, ** p < 0.01, *** p < 0.001

(–1.253)

–0.608

LoanStg2

(–0.900)

–0.341

LLRStg3

(–0.481)

–0.126

RegCap

(0.789)

0.153

SIZE

(–1.028)

–0.036

(–4.995)

–0.118***

NEBTLLP Change5YCDS

Notes: This table presents summary statistics of banks’ specific variables and sovereign CDSs. LLP is loan loss provisions presented as a percentage of the average customer loans. LLA is loan loss allowances presented as a percentage of total customer loans. LoanStg1, LoanStg2, and LoanStg3 are customer loans at stage 1, 2, and 3 respectively presented as a percentage of total customer loans. LLRStg1 LLRStg2 LLRStg3 are loan loss reserves at stage 1, 2 and 3 presented as a percentage of total customer loans.Size is the logarithm of total assets. RegCap is the Tier 1 Capital ratio. SIZE is the logarithm of total assets. NEBTLLP is net earnings before taxes and loan loss provisions presented as a percentage of the average total assets. Change5YCDS is the change in the 5 year sovereign CDS of the bank’s country calculated as the percentage change in the median of daily sovereign CDSs over the quarter period.

(–6.077)

N

–0.138***

t

b

Diff-stat. (2021 – 2020)

Period

Table 5.1 (Continued)

32 5 Descriptive Statistics

5.2 Descriptive Statistics

33

in the macroeconomic situation with sovereign CDSs decreasing, on average, by 0,06 % quarterly. Consistently, banks’ loan loss provisions have, on average, decreased from 0,22% (median 0,12%) to 0,09% (median 0,03%) of average total loans, a decrease that falls even below the pre-pandemic levels. This significant decrease suggests that many banks released considerable parts of LLPs built during 2020, resulting in a decrease of total LLPs. Moreover, as also shown in the table, the lower quantile of LLPs in 2021 indicates that one fourth of the banks have on average reported negative LLPs, which are a result of releases of loan loss allowances that exceed new provisions for the period. Table 5.2 presents statistics of Covid-19 responsive measures implemented by the countries’ governments during the pandemic. The table shows the number of Covid19 response measures that were in force during the period 2020-2021. Table B.7 of Appendix B describes the type of measures implemented by the sample countries and their frequencies. Countries are indicated to have a strong/weak Covid-19 response policy if the median number of Covid-19 response measures implemented by the country is higher/lower than the median number of Covid-19 response measures implemented by all countries throughout the pandemic. Table 5.3 shows pairwaise correlation coefficients of banks specific variables, the changes in sovereign CDSs and the number of Covid-19 response measures. Loan loss provisions have a significantly positive correlation with net earnings before taxes and loan loss provisions (NEBTLLP). This suggests that banks use loan loss provisions to smooth income. The significantly positive correlation between LLPs and the number of Covid-19 response measures indicates that banks located in countries with higher Covid-19 response measures recognize larger loan loan provisions. The correlation between LLP and changes in sovereign CDS is positive and significant at 10% level. Countries with more Covid-19 response measures tend to have more adverse changes in sovereign CDS, as indicated by the positive and significant relationship between the number of Covid-19 measures and changes in sovereign CDS.

34

5

Descriptive Statistics

Table 5.2 Covid-19 Response Measures by Countries Country Austria (Banks:4) Belgium (Banks:1) Bulgaria (Banks:1) Cyprus (Banks:3) Czechia (Banks:2) Denmark (Banks:6) Estonia (Banks:1) Finland (Banks:4) Germany (Banks:2) Greece (Banks:4) Hungary (Banks:1) Iceland (Banks:2) Italy (Banks:6) Netherlands (Banks:2) Norway (Banks:30) Poland (Banks:7) Portugal (Banks:1)

Min

Max

Mean

Med

SD

Indicator

2

27

12.79

10

8.73

Strong

4

15

9

8.5

4.43

Strong

4

18

9.8

8

5.40

Weak

2

17

10.33

10

5.56

Strong

1

35

14.45

15

12.59

Strong

1

18

10.30

11

6.52

Strong

9

9

9

9

.

Strong

2

16

6.5

5

4.70

Weak

2

17

10.11

12

6.13

Strong

1

15

7.52

6

5.73

Weak

3

17

7.83

6.5

5.19

Weak

4

15

8.4

9

3.10

Strong

4

24

10.7

8

6.92

Weak

1

22

11.57

11

7.41

Strong

1

13

6.27

4

4.88

Weak

8

17

11.84

12

3.18

Strong

1

18

8.67

9.5

6.09

Strong (Continued)

5.2 Descriptive Statistics

35

Table 5.2 (Continued) Country

Min

Max

Mean

Med

SD

Indicator

Slovenia (Banks:1) Spain (Banks:5) Sweden (Banks:2) United Kingdom (Banks:4) Total

10

23

15.6

14

5.32

Strong

1

17

6.84

2

7.33

Weak

3

8

5.5

5.5

2.07

Weak

8

16

11

9

3.77

Strong

1

35

8.43

8

6.14

Strong

Notes: The table presents statistics of Covid-19 response measures implemented by the countries’ governments during the pandemic. It shows the number of Covid-19 responsive measures that were in force during the period 2020-2021. Countries are indicated to have a strong/weak Covid-19 response policy if the median number of Covid-19 response measures implemented by the country is higher/lower than the median number of Covid-19 response measures implemented by all countries throughout the pandemic.

–0.002

–0.351***

0.118*** –0.030

0.345***

–0.164***

–0.013

0.058*

0.127*** –0.037

LOANS

LoanStg1

LoanStg2

LoanStg3

RegCap

SIZE

Change5YCDS

Covid-19 Measures

1.000 1.000 0.025

1.000 1.000

LoanStg1

0.114**

–0.003

0.003

–0.160*** 0.023 0.108**

–0.014

0.077** 0.102**

–0.020

–0.050*

–0.096***

1.000

LoanStg3

*** p < 0.01, ** p < 0.05, * p < 0.1

–0.174*** –0.060

0.030

0.219***

1.000

LoanStg2

0.126*** –0.020

–0.801***

–0.598*** –0.022

0.198***

0.883*** –0.054*

0.163*** –0.204*** –0.230***

–0.781***

LOANS

–0.113**

–0.041

–0.379***

1.000

RegCap

0.178***

–0.001

1.000

SIZE

0.085*

1.000 1.000

Change5YCDS Covid-19 Measures

Notes: The table presents pairwaise correlation coefficients of bank specific variables, the changes in sovereign CDSs, and the numbers of Covid-19 response measures. LLP is loan loss provisions scaled by average customer loans. LLA is loan loss allowances scaled by total customer loans. LoanStg1, LoanStg2, and LoanStg3 are customer loans at stage 1, 2 and 3 respectively scaled by total customer loans. Size is the logarithm of total assets. RegCap is the Tier 1 Capital ratio. NEBTLLP is net earnings before taxes and loan loss provisions scaled by average total assets. Change5YCDS is the change in the 5 year sovereign CDS of the bank’s country calculated as the percentage change in the median of daily sovereign CDSs over the quarter period. Covid-19 Measures is the number of Covid-19 response measures implemented by the bank’s country during the period.

–0.006

–0.161***

0.038

0.013

0.000

0.092*** –0.040

0.177***

0.581***

0.500***

LLA

1.000

LLP

NEBTLLP LLA

NEBTLLP

LLP

Variables

Table 5.3 Pairwaise Correlation Coefficients between Bank Specific Variables, Changes in Sovereign CDSs and Numbers of Covid-19 response measures.

36 5 Descriptive Statistics

6

Research Design and Results

6.1

Discretionary LLPs

I examine earnings management by analyzing the discretionary use of loan loss provisions of European banks. I focus on LLPs, as one of the main accruals of banks, in order to estimate the extent to which managers use discretion when provisioning for loan losses. I estimate normal (expected) LLP following Beatty et al. (2002) using the following equation: LLPi,t = α + β1 LLAi,t−1 + β2 NPLi,t−1 + β3 Loansi,t−1 + β4 NPLi,t + β5 Loansi,t + i,t

(6.1) where for every bank i at time t(t − 1), L L P i,t is loan loss provisions for the current period. L L Ai,t−1 is loan loss allowances, N P L i,t−1 is non-performing loans and Loans is gross customer loans, each at the beginning of the period.  sign indicates changes. L L P is scaled by the average customer loans. L L A and N P L are scaled by total customer loans while Loans i,t−1 is scaled by total assets. Results of the regression are displayed in table B.3 in Appendix B. I define discretionary LLP as the residual i,t from the Equation 6.1. Figure 6.1 shows the distribution of discretionary LLPs by years in a box plot. The median of discretionary LLPs increases during the pandemic year in 2020 and decreases again in 2021. A t-test for differences in discretionary LLPs, presented in Table B.4 of Appendix B, confirms that the increase in discretionary LLP during the pandemic year of 2020 is highly significant (at 0,1% level). Moreover, figure 6.1 shows that during Covid-19 discretionary LLPs exhibit greater variability and larger outliers Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/978-3-658-40060-6_6. © The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2023 M. Lamaj, The Effect of Covid-19 on Loan Loss Provisions and Earnings Management of European Banks, BestMasters, https://doi.org/10.1007/978-3-658-40060-6_6

37

38

6

Research Design and Results

-1

-.5

Discretionary LLP 0 .5

1

1.5

than in the years before. This is consistent with the different loan loss provisioning practices demonstrated during the pandemic year. However the median and the variability of discretionary LLPs decrease again in 2021. Moreover, figure 6.1 reveals that, while the pre-Covid period is characterized by an understatement of LLPs (i.e., median of discretionary LLPs for the year 2018 and 2019 are below zero), during the pandemic year of 2020 banks tend to overstate LLPs. This trend, however, decreases in 2021, with banks, on average, understating LLPs. Furthermore, figure 6.2 shows the discretionary LLPs over the year-quarters in the post-Covid-19 period, presented as a percentage of average customer loans. While in the first two quarters of 2020 discretionary LLPs are positive with large magnitudes at 0.129% and 0.108% of average customer loans respectively, they decrease in the second half of the year to 0.018% in the third quarter and 0.053% in the fourth quarter of 2020. The decreasing proceeds even further during 2021 with discretionary LLPs below zero reaching a minimum of –0.066% of average customer loans in the second quarter of 2021.

2018

2019

Figure 6.1 Discretionary LLPs over Years

2020

2021

6.1 Discretionary LLPs

39

Figure 6.3, shows cross-country differences in discretionary LLPs by Covid-19 response measures. Countries are indicated as Strong (Weak) if the median of the numbers of Covid-19 response measures implemented by a country through the post-Covid period is higher (lower) than the median of the total response measures implemented throughout the period. Banks located in countries with strong response measures exhibit greater variability of discretionary LLPs than banks located in countries with weaker Covid-19 response measures. However, these differences tend to diminish in the year 2021. Overall, these results confirm that during Covid-19 banks use more discretion when accounting for loan losses. This trend is more amplified for banks in countries with stricter Covid-19 response. Nevertheless, the discretionary LLPs decrease throughout 2020 and more prominently in 2021, where no more significant crosscountry differences are observed. In addition, while banks tend to overstate provisions during the pandemic year of 2020, they do, on average, understate them in 2021.

.128777 .108031

.052585

.017593

-.025846 -.044688 -.0655 2020Q1

2020Q2

2020Q3

2020Q4

2021Q1

Figure 6.2 Discretionary LLPs over Year-Quarters during Covid-19

2021Q2

2021Q3

6

Research Design and Results

-.5

Discretionary LLP 0 .5

1

1.5

40

Weak

Strong

2020

Weak

Strong

2021

Figure 6.3 Discretionary LLPs by Different Covid-19 Response Measures

6.2

Income Smoothing

Examining income smoothing behavior of banks through the use of LLPs allows me to use a specific accrual approach by identifying the key factors that drive changes on loan loss provisions. First, I estimate the following regression, also used in Gebhardt and Novotny-Farkas (2011). LLPi,t = α + β1 COVID19 + β2 NPLi,t−1 + β3 NPLi,t + β4 NPLi,t+1 + β5 Loansi,t−1 + β6 Loansi,t + β7 5YCDSi,t + β8 NEBTLLPi,t  + β9 RegCapi,t−1 + β10 Sizei,t + Firmfixedeffects + i,t (6.2) where the dependent variable L L P i,t represents loan loss provisions in the current period. C O V I D19 is an indicator variable that equals 1 for the pandemic year of 2020 and 0 otherwise. The explanatory variable N E BT L L P i,t is the variable of interest and represents net earnings before taxes and loan loss provisions in the

6.2 Income Smoothing

41

current period. Its coefficient β8 captures the extent of income smoothing behavior of banks. N P L i,t−1 is the balance of non-performing loans and Loans i,t−1 is the amount of total customer loans, each at the beginning of the period. N P L i,t and Loans i,t capture the change in non-performing loans and total customer loans in the current period. L L P is scaled by average customer loans, N P L is scaled by total customer loans, and Loans is scaled by total assets. I also include the future change in NPL (N P L i,t+1 ) in order to control for any changes in the timeliness of loan loss provisions in the context of Covid-19. Si zei,t and RegCapi,t−1 represent the respective logarithm of total assets and Tier 1 common capital ratio at the beginning of the period and capture the effect of size and capital in banks’ LLPs. Lastly, in order to control for the effect of macroeconomic scenarios under the current accounting regime of IFRS 9, I include changes in the 5-year sovereign CDSs (5Y C DS i,t ) calculated as the percentage change in the median of daily sovereign CDSs over the quarter period. Loan loss provisioning under IFRS 9 offers the advantage of analyzing provisions exploiting the three stage model approach demonstrated in the standard. Therefore, in a second step, I adjust the previous model by including the three stage components of loans under the ECL-model of IFRS 9, as shown in the following equation: LLPi,t = α + β1 COVID19 + β2 LoanStg1i,t−1 + β2 LoanStg2i,t−1 + β3 LoanStg3i,t−1 + β4 LoanStg1i,t + β5 LoanStg2i,t + β6 LoanStg3i,t + β7 NPLi,t+1 + β8 5YCDSi,t + β9 NEBTLLPi,t + β10 RegCapi,t−1 + β11 Sizei,t  + Firmfixedeffects + i,t

(6.3) LoanStg1i,t−1 , LoanStg2i,t−1 , LoanStg3i,t−1 represent the amount of customer loans at stage 1, 2 and 3 respectively at the beginning of the period scaled by total customer loans. LoanStg1, LoanStg2 and LoanStg3, capture the respective changes in LoanStg1, LoanStg2, LoanStg3, defined as above, for the current period. The remaining variables are defined same as in the previous model of equation 6.2. First, I estimate both regressions for the whole sample period from 2018 until the third quarter of 2021. Furthermore, I divide the sample into two sub-samples; pre-Covid for the years 2018–2019 and post-Covid for the years 2020–2021 and estimate the same regression for each sub-sample separately. Table 6.1 presents the results of the regression model in the equation 6.2. The first column shows the regression estimates for the whole sample. β1 coefficient of COVID19 is positive and significant at 1% level, showing that, as anticipated, Covid19 has led, on average, to an increase of loan loss provisions of banks. Similarly,

42

6

Research Design and Results

Table 6.1 The Effect of Covid-19 on Income Smoothing LLP COVID19 NPL NPL

Future NPL Loans Loans 5YCDS

NEBTLLP RegCap Size Cons

Pre-Covid

Post-Covid

LLP

LLP

0.100

0.109

(6.69)**

(6.22)**

–0.003

0.010

–0.013

(0.82)

(0.64)

(1.70)

0.003

0.000

0.002

(0.31)

(0.01)

(0.28)

0.006

0.015

0.001

(0.99)

(0.88)

(0.14)

0.008

0.005

0.006

(3.42)**

(1.31)

(1.56)

0.006

0.007

0.000

(1.75)

(2.60)*

(0.13)

0.055

0.056

0.044

(3.06)**

(0.88)

(2.46)*

0.124

0.111

0.137

(2.30)*

(2.98)**

(2.97)**

0.164

0.707

0.308

(0.78)

(1.47)

(1.22)

–0.022

0.017

–0.147

(0.33)

(0.11)

(1.14)

–0.561

–2.510

1.282

(0.50)

(1.68)

(0.53)

R2

0.14

0.13

0.22

N

964

509

455

* p < 0.05, ** p < 0.01

Notes: The table shows the regression results of the equation 6.2 for the whole sample (first column of the table) as well for the pre-Covid (2018-2019) and post-Covid period (20202021). The explanatory variable N E BT L L P is the variable of interest and represents net earnings before taxes and loan loss provisions in the current period scaled by average total assets. Its coefficient captures the extent of income smoothing behavior of banks. N P L is the balance of non-performing loans and Loans is the amount of total customer loans, each at the beginning of the period. N P L and Loans capture the change in non-performing loans and total customer loans in the current period. L L P is scaled by average customer loans, N P L is scaled by total customer loans, and Loans is scaled by total assets. Futur eN P L is the future change in non-performing loans calculated as the change in NPL for the next period. Si ze and RegCap represent the respective logarithm of total assets and Tier 1 common capital ratio at the beginning of the period and capture the effect of size and capital in bank’s LLPs. 5Y C DS is the change in the 5-year sovereign CDS calculated as the percentage change in the median of daily sovereign CDSs over the period.

6.2 Income Smoothing

43

in the Post-Covid period coefficient β1 is positive and significant, implying that provisions during the pandemic year of 2020 are, on average, higher than those during 2021. The β8 coefficient of NEBTLLP, which is the variable of interest, is positive for the whole sample and for both the pre- and post-Covid 19 periods, with the result being more significant for the post period. This indicates that banks do, on average, use loan loss provisions to smooth income in both pre-, as well as post-Covid 19 periods. A t-test, not presented in the paper, shows that there is no significant difference in the coefficients of both sub-samples. Moreover, the coefficient of 5YCDS is positive and significant at the 1% level for the whole sample. However, this is only the case in the post-Covid period, although at a lower significance. This indicates that during the crisis LLPs are more relying on macroeconomic variables. Table 6.2 shows the estimates of the equation 6.3. The results are consistent with the previous findings. However, the coefficient for the  LoanStg2i,t is positive and significant only for the post-Covid period. This indicates that in the post-Covid period provisions are more related to changes in loans in stage 2, whereas such relation is not evident in the pre-Covid period. These results can be explained by the anticipated effect of the pandemic in the credit risk of the loan portfolios of banks, leading to an increase in loans that have experienced a significant increase in credit risk and are therefore transferred to stage 2. As a consequence, banks build up their provisions in order to increase loan loss allowances at the level of life-time expected losses.

Table 6.2 The Effect of Covid-19 on Income Smoothing: Staging Components of Loans LLP COVID19 LoanStg1 LoanStg2 LoanStg3

0.109 (6.15)** 0.000 (0.29) –0.003 (0.90) –0.000 (0.13)

Pre-Covid LLP

Post-Covid LLP

0.000 (0.01) –0.006 (0.74) 0.001 (0.14)

0.099 (5.18)** 0.011 (1.99) 0.012 (1.74) 0.008 (0.82) (Continued)

44

6

Research Design and Results

Table 6.2 (Continued) LLP LoanStg1 LoanStg2 LoanStg3 Future NPL 5YCDS NEBTLLP RegCap Size Cons R2 N

–0.001 (0.78) 0.002 (0.41) 0.003 (0.63) 0.004 (0.99) 0.050 (2.54)* 0.161 (2.92)** 0.128 (0.46) –0.070 (0.52) 0.879 (0.44) 0.14 719

Pre-Covid LLP 0.001 (0.53) –0.021 (1.71) 0.003 (0.32) 0.002 (0.27) –0.063 (1.27) 0.193 (2.20)* 0.739 (1.60) –0.062 (0.31) –0.861 (0.34) 0.16 340 * p < 0.05, ** p < 0.01

Post-Covid LLP 0.004 (1.51) 0.023 (3.97)** 0.013 (1.56) 0.004 (0.63) 0.055 (2.40)* 0.144 (3.16)** 0.220 (0.79) –0.216 (1.35) 1.931 (0.77) 0.23 379

Notes: The table shows the regression results of the equation 6.3 for the whole sample (first column of the table) as well for the pre-Covid (2018-2019) and post-Covid period (20202021). N E BT L L P is net earnings before taxes and loan loss provisions scaled by average total assets. LoanStg1i,t−1 , LoanStg2, LoanStg3 represent the amount of customer loans at stage 1, 2 and 3 respectively at the beginning of the period scaled by total customer loans. LoanStg1, LoanStg2, and LoanStg3 capture the respective changes in LoanStg1, LoanStg2, LoanStg3 for the current period. Futur eN P L is the future change in nonperforming loans calculated as the change in NPL for the next period. Si ze and RegCap represent the respective logarithm of total assets and Tier 1 common capital ratio at the beginning of the period and capture the effect of size and capital in bank’s LLPs. 5Y C DS is the change in the 5-year sovereign CDS calculated as the percentage change in the median of daily sovereign CDSs over the period.

6.3 Cross-country Differences

6.3

45

Cross-country Differences

I examine cross-country differences in the income smoothing of banks, by analyzing the effect of Covid-19 response measures on banks’ income smoothing. To do so, I partition the post-Covid sample into countries with a Str ong response and countries with a W eak response to the Covid-19 pandemic. Results in table 6.3 show that the coefficient of NEBTLLP is positive and significant only for banks located in countries with strong Covid-19 response measures. This is consistent with my expectations that banks in countries with strong Covid-19 response measures have greater incentives to engage in income smoothing practices as opposed to banks in countries with weak implemented measures. Moreover, the coefficient for 5YCDS is also positive and significant only for the Srong sample indicating that provisions of banks in countries, where high containment measures are implemented, are more related to changes in the sovereign CDS spreads.

Table 6.3 Cross-Country Differences on the Effect of Covid-19 on Income Smoothing

NPL NPL FutureNPL Loans 5YCDS NEBTLLP RegCap Size

Post-Covid LLP

Weak LLP

Strong LLP

–0.005 (0.55) 0.008 (1.35) 0.007 (1.08) 0.004 (1.20) 0.059 (3.02)** 0.125 (2.86)** 0.190 (0.75) –0.515 (3.98)**

–0.013 (1.22) 0.002 (0.39) 0.003 (0.50) 0.003 (0.92) 0.020 (0.80) 0.056 (0.46) 0.439 (1.82) –0.364 (3.15)**

0.005 (0.26) 0.032 (2.33)* 0.030 (1.97) 0.003 (0.47) 0.115 (3.28)** 0.145 (3.47)** 0.089 (0.20) –0.959 (5.27)** (Continued)

46

6

Research Design and Results

Table 6.3 (Continued) Post-Covid LLP Cons R2 N

7.796 (3.32)** 0.16 455

Weak LLP

4.429 (2.11)* 0.11 273 * p < 0.05, ** p < 0.01

Strong LLP 16.516 (4.49)** 0.28 182

Notes: The table shows the regression results of the equation 6.2 for the post-Covid period (first column of the table). Results are presented separately for the W eak and Str ong countries. Countries are indicated to have a Str ong/W eak Covid-19 response policy if the median of Covid-19 response measures implemented by the country is higher/lower than the median of Covid-19 response measures implemented by all countries throughout the pandemic. The explanatory variable N E BT L L P is the variable of interest and represents net earnings before taxes and loan loss provisions in the current period scaled by average total assets. Its coefficient captures the extent of income smoothing behavior of banks. N P L is the balance of nonperforming loans and Loans is the amount of total customer loans, each at the beginning of the period. N P L and Loans capture the change in non-performing loans and total customer loans in the current period. L L P is scaled by average customer loans, N P L is scaled by total customer loans, and Loans is scaled by total assets. Futur eN P L is the future change in non-performing loans calculated as the change in NPL for the next period. Si ze and RegCap represent the respective logarithm of total assets and Tier 1 common capital ratio at the beginning of the period and capture the effect of size and capital in bank’s LLPs. 5Y C DS is the change in the 5-year sovereign CDS calculated as the percentage change in the median of daily sovereign CDSs over the period.

Table 6.4 shows cross-country differences in the income smoothing of banks by analyzing changes in the three-stage components of loans based on the ECL-model. The results corroborate the previous findings, and confirm that income smoothing behavior is evident only for countries that have implemented strong containment measures. Nevertheless, the coefficient for  LoanStg2i,t is positive and significant for both Strong and Weak countries. Lastly, LLPs are positively and significantly related to the changes in sovereign CDSs only in the Strong sample. Overall, these findings can be explained by the fact that the economic consequences of the pandemic are more severe in countries where strong Covid-19 response measures are implemented, since such measures have a direct impact on the business environment of a country. Borrowers in countries with high measures may be experiencing higher financial distress, leading to a systematic deterioration in the credit quality of the banks’ loan portfolios in that country. Therefore, banks in high financially stressed environments may have greater incentives to avoid earning decreases through the use of LLPs.

6.3 Cross-country Differences

47

Table 6.4 Cross-Country Differences on the Effect of Covid-19 on Income Smoothing: Staging Components of Loans

LoanStg1 LoanStg2 LoanStg3 LoanStg1 LoanStg2 LoanStg3 FutureNPL Loans 5YCDS NEBTLLP RegCap Size Cons

Post-Covid LLP

Weak LLP

Strong LLP

0.012 (1.98) 0.017 (2.25)* 0.013 (1.36) 0.003 (0.94) 0.028 (4.40)** 0.014 (1.43) 0.007 (1.15) 0.005 (1.05) 0.063 (2.45)* 0.135 (3.06)** 0.138 (0.50) –0.503 (3.67)** 6.380 (2.90)**

0.005 (0.90) 0.006 (0.43) –0.011 (0.89) 0.001 (0.24) 0.020 (2.98)** 0.011 (0.76) 0.004 (0.69) 0.003 (0.73) 0.035 (0.80) 0.030 (0.17) 0.532 (1.99) –0.380 (2.83)** 3.983 (1.88)

0.041 (1.48) 0.053 (2.00) 0.057 (1.96) 0.011 (0.81) 0.041 (2.53)* 0.022 (0.40) 0.014 (0.94) 0.010 (1.09) 0.135 (3.35)** 0.164 (3.52)** 0.086 (0.15) –0.762 (2.62)* 8.222 (1.23) (Continued)

48

6

Research Design and Results

Table 6.4 (Continued) Post-Covid LLP R2 N

0.19 379

Weak LLP

0.14 224 * p < 0.05, ** p < 0.01

Strong LLP 0.27 155

Notes: The table shows the regression results of the equation 6.3 for the post-Covid period (first column of the table). Results are presented separately for the W eak and Str ong countries. Countries are indicated to have a Str ong/W eak Covid-19 response policy if the median of Covid-19 response measures implemented by the country is higher/lower than the median of Covid-19 response measures implemented by all countries throughout the pandemic. N E BT L L P is net earnings before taxes and loan loss provisions scaled by average total assets. LoanStg1, LoanStg2, LoanStg3 represent the amount of customer loans at stage 1, 2 and 3 respectively at the beginning of the period scaled by total customer loans. LoanStg1, LoanStg2, and LoanStg3 capture the respective changes in LoanStg1, LoanStg2, LoanStg3 for the current period. Futur eN P L is the future change in non-performing loans calculated as the change in NPL for the next period. Si ze and RegCap represent the respective logarithm of total assets and Tier 1 common capital ratio at the beginning of the period and capture the effect of size and capital in bank’s LLPs. 5Y C DS is the change in the 5-year sovereign CDS calculated as the percentage change in the median of daily sovereign CDSs over the period.

I test for the robustness of these results by implementing the same setting on the pre-Covid period, where no containment measures have been implemented. Results, shown in table B.5 and B.6 of the Appendix B confirm that cross-country differences are driven to a considerable part by the differences in the containment measures implemented through the pandemic. However, these results should be interpreted with caution, since the positive coefficient (significant at 5% level) of LLP and NEBTLLP for the Strong sample in the prior period, may indicate that the higher smoothing behavior of Str ong countries is driven by other differences in country characteristics other than those related to Covid-19 implemented measures during the pandemic.

6.4

Upwards and Downwards Earnings Management

Finally, I examine the direction of earnings management by regressing the discretionary LLPs estimated from the regression model 6.1 using the following regression:

436

343

0.12

0.299

93

0.45

535

0.20

(3.54)**

6.988

(2.81)**

0.059

(0.73)

–0.040

(3.81)**

–0.428

(0.02)

–0.006

(2.83)**

0.194

(4.45)**

0.246

* p < 0.05, ** p < 0.01

(3.86)**

–9.510

(0.40)

–0.018

(4.05)**

–0.323

(3.20)**

0.552

(0.88)

0.167

(3.29)**

0.324

(2.63)*

Post-Covid

304

0.11

(1.55)

1.752

(0.39)

0.011

(0.59)

–0.018

(2.18)*

–0.147

(1.19)

0.171

(4.13)**

0.109

(2.65)*

0.082

231

0.16

(1.45)

6.784

(1.72)

0.046

(0.39)

–0.030

(1.46)

–0.360

(0.50)

–0.231

(1.39)

0.118

(2.78)**

0.212

Discretionary LLP < 0 Discretionary LLP > 0

Notes: The table shows the regression results of discretionary LLPs, positive discretionary LLPs and negative discretionary LLPs on Loss; an indicator variable that equals 1 if the bank’s net earnings before taxes (NIBT) is negative and zero otherwise, N E BT L L P; net earnings before taxes and loan loss provisions scaled by average total assets, RegCap; the logarithm of Tier 1 common capital ratio at the beginning of the period, Si ze; the logarithm of total assets at the beginning of the period, (L L P); loan loss provisions scaled by average customer loans, 5Y C DS; the change in the 5-year sovereign CDS spread calculated as the percentage change in the median of daily sovereign CDSs over the period. Results are presented separately for the pre-Covid (2018–2019) and the post-Covid (2020–2021) period.

0.39

–3.716 (1.72)

–3.247

(1.19)

–0.000 (0.01)

0.067

(0.54)

(1.42)

(1.15)

0.026

(1.60)

(0.61)

–0.225

0.187

(1.10)

(1.44)

0.114

0.197

(0.61)

(2.49)*

0.475

0.063

0.161

0.124 (1.69)

0.452

N

Pre-Covid Discretionary LLP < 0 Discretionary LLP > 0 Discretionary LLP

(3.39)**

R2

Cons

5YCDS

LLP

Size

RegCap

NEBTLLP

Loss

Discretionary LLP

Table 6.5 Downwards and Upwards Earnings Management

6.4 Upwards and Downwards Earnings Management 49

50

6

Research Design and Results

DiscretionaryLLPi,t = α + β1 Loss + β2 NEBTLLPi,t + β3 Sizei,t + β4 RegCapi,t−1 + β5 LLPi,t−1 + β6 5YCDSi,t  + Firmfixedeffects + i,t

(6.4)

Discr etionar y L L P is the residual estimated from regression 6.1, Loss is an indicator variable that equals 1 if the bank’s net earnings before taxes (NIBT) is negative and zero otherwise. The remaining variables are defined same as in the previous equations. I run the regression for both understating (ie., discretionary LLP < 0) and overstating LLPs (i.e., discretionary LLP > 0) for both prior- and post-Covid period. Results are presented in table 6.5. Both in the pre- and post-Covid period, the discretionary LLPs is positively related to Loss, indicating that banks experiencing a loss use discretion to a greater extent than those that report positive earnings for the same period. The coefficient on NEBTLLP is positive and significant for both periods, corroborating the prior findings that banks use discretion to smooth income in both prior-and post-Covid. Moreover, positive discretionary LLPs are positively related to NEBTLLP in the pre-Covid period, indicating that prior to Covid-19, banks overstate LLPs in order to smooth income (i.e., they use downwards EM). In contrast, in the post-Covid period, banks tend to engage in upwards earnings management by understating provisions (discretionary LLP < 0). These results are also consistent with the anecdotal evidence showing that during the pandemic banks apply certain smoothing adjustments in their ECL model in order to minimize the impact of the Covid-19 crisis on their financial statements.

7

Conclusion

In this thesis I examine the effect of Covid-19 on banks’ loan loss provisions and their use for earnings management. I find that during Covid-19 banks use more discretion when accounting for loan loss provisions under IFRS 9 than in the period before Covid-19. This trend is, however, more evident for banks located in countries that have implemented strong Covid-19 response measures. Moreover, while banks tend to overstate provisions during the pandemic year of 2020, the overstatement decreases throughout the year until 2021, where banks do, on average, understate provisions. Examining income smoothing behavior of banks, I find no significant effect of Covid-19 on income smoothing through the use of LLPs. However, while analyzing cross-country differences on Covid-19 responses, I observe that income smoothing behavior is only evident for banks in countries with stricter Covid-19 response measures. Nevertheless, these results should be interpreted with caution, since such differences may be driven from other country characteristics rather than those related to the response measures implemented during the pandemic. Additionally, examining the direction of earnings managements I observe that prior to the pandemic banks engage in downwards earnings management by overstating provisions, whereas after Covid-19 banks tend to use upwards earnings management by understating provisions. Lastly, analyzing the different staging components of loans under IFRS 9, I find that loan loss provisions are positively related to changes in stage 2. However, this is true only for the post-Covid period. Moreover, LLPs are positively related to changes in sovereign CDSs, especially in the post-Covid period and in particular for banks in countries with strong Covid-19 response measures.

© The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2023 M. Lamaj, The Effect of Covid-19 on Loan Loss Provisions and Earnings Management of European Banks, BestMasters, https://doi.org/10.1007/978-3-658-40060-6_7

51

52

7

Conclusion

Overall, these results are consistent with three interpretations. First, the effect of the pandemic on the credit risk of the loan portfolios have led to an increase in financial instruments that have experienced a significant increase in credit risk, and are therefore transferred to stage 2. Consequently, loan loss provisions are built to the amount of lifetime credit losses, therefore, increasing the amount of loan loss allowances during the pandemic, especially of those related to stage 2. Second, the high amount of uncertainty induced during the Covid-19 and the high flexibility offered by many prudential authorities on the implementation of IFRS 9 during Covid-19 have created a convenient environment for earnings management. Finally, the financial distress experienced during the pandemic, being more amplified in countries with strong Covid-19 response measures, has made banks more vulnerable to smooth income by reducing the impact of the Covid-19 in their financial statements.

Bibliography

Agarwal, S., Chomsisengphet, S., Liu, C., & Rhee, S. G. (2007). Earnings management behaviors under different economic environments: Evidence from Japanese banks. International Review of Economics & Finance, 16(3), 429–443. Allen, F., Carletti, E., & Gu, X. (2008). The roles of banks in financial systems. Oxford Handbook of Banking, 32–57. Allen, F., Chui, M. K., & Maddaloni, A. (2004). Financial systems in europe, the usa, and asia. Oxford Review of Economic Policy, 20(4), 490–508. Anandarajan, A., Hasan, I., & McCarthy, C. (2007). Use of loan loss provisions for capital, earnings management and signalling by australian banks. Accounting & Finance, 47(3), 357–379. Baker, S. R., Bloom, N., Davis, S. J., Kost, K., Sammon, M., & Viratyosin, T. (2020). The unprecedented stock market reaction to covid-19. The Review of Asset Pricing Studies, 10(4), 742–758. Ball, R., & Brown, P. (1968). An empirical evaluation of accounting income numbers. Journal of Accounting Research, 159–178. Barnoussi, A. e., Howieson, B., & van Beest, F. (2020). Prudential application of ifrs 9: (un) fair reporting in covid-19 crisis for banks worldwide?! Australian Accounting Review, 30(3), 178–192. BCBS. (2015, December 1). Guidance on credit risk and accounting for expected credit losses. Retrieved December 30, 2021, from https://www.bis.org/bcbs/publ/d350.pdf Beatty, A., Chamberlain, S. L., & Magliolo, J. (1995). Managing financial reports of commercial banks: The influence of taxes, regulatory capital, and earnings. Journal of Accounting Research, 33(2), 231–261. Beatty, A., Ke, B., & Petroni, K. R. (2002). Earnings management to avoid earnings declines across publicly and privately held banks. The Accounting Review, 77(3), 547–570. Beatty, A., & Liao, S. (2014). Financial accounting in the banking industry: A review of the empirical literature. Journal of Accounting and Economics, 58(2–3), 339–383. Beaver, W. H. (1968). The information content of annual earnings announcements. Journal of Accounting Research, 67–92. Benston, G. J., & Wall, L. D. (2005). How should banks account for loan losses. Journal of Accounting and Public Policy, 24(2), 81–100. Burgstahler, D., & Dichev, I. (1997). Earnings management to avoid earnings decreases and losses. Journal of Accounting and Economics, 24(1), 99–126. © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2023 M. Lamaj, The Effect of Covid-19 on Loan Loss Provisions and Earnings Management of European Banks, BestMasters, https://doi.org/10.1007/978-3-658-40060-6

53

54

Bibliography

Bushman, R. M., & Williams, C. D. (2012). Accounting discretion, loan loss provisioning, and discipline of banks’ risk-taking. Journal of Accounting and Economics, 54(1), 1–18. Bushman, R. M., & Williams, C. D. (2015). Delayed expected loss recognition and the risk profile of banks. Journal of Accounting Research, 53(3), 511–553. Caggiano, G., Castelnuovo, E., & Kima, R. (2020). The global effects of covid-19-induced uncertainty. Economics Letters, 194, 109392. Coibion, O., Gorodnichenko, Y., & Weber, M. (2020). Labor markets during the covid-19 crisis: A preliminary view (tech. rep.). National Bureau of Economic Research. Collins, J. H., Shackelford, D. A., & Wahlen, J. M. (1995). Bank differences in the coordination of regulatory capital, earnings, and taxes. Journal of Accounting Research, 33(2), 263–291. Deb, P., Furceri, D., Ostry, J. D., & Tawk, N. (2021). The economic effects of covid-19 containment measures. Open Economies Review, 1–32. EBA. (2017, December 5). Guidelines on credit institution’s credit risk management practices and accounting for expected credit losses. Retrieved December 30, 2021, from https:// www.eba.europa.eu/sites/default/documents/files/documents/10180/1965596/8a9a9df0a0cc-406e-a781-7d4fb753495d/Guidelines%20on%20Accounting%20for%20ECL%20 %28EBA-GL-2017-06%29_EN.pdf?retry=1 EBA. (2020, March 25). Statement on the application of the prudential framework regarding default, forbearance and ifrs 9 in the light of covid-19 measures. Retrieved December 30, 2021, from https://www.eba.europa.eu/sites/default/documents/files/ document_library/News%20and%20Press/Press%20Room/Press%20Releases/2020/ EBA%20provides%20clarity%20to%20banks%20and%20consumers%20on%20the %20application%20of%20the%20prudential%20framework%20in%20light%20of %20COVID-19%20measures/Statement%20on%20the%20application%20of%20the %20prudential%20framework%20regarding%20Default%2C%20Forbearance%20and %20IFRS9%20in%20light%20of%20COVID-19%20measures.pdf EBA. (2021, November 24). Ifrs 9 implementation by eu institutions. Retrieved December 30, 2021, from https://www.eba.europa.eu/sites/default/documents/files/document_ library/Publications/Reports/2021/1024609/IFRS9%20monitoring%20report.pdf ECB. (2020a). Identification and measurement of credit risk in the context of the coronavirus (covid-19) pandemic. Retrieved December 30, 2021, from https://www.bankingsuper vision.europa.eu/press/letterstobanks/shared/pdf/2020/ssm.2020_letter_credit_risk_identi fication_measurement~734f2a0b84.en.pdf ECB. (2020b, April 1). Ifrs 9 in the context of the coronavirus (covid-19) pandemic. Retrieved December 30, 2021, from https://www.who.int/director-general/speeches/ detail/who-director-general-s-opening-remarks-at-the-mediabriefing-on-covid-19---12november-2021 ECB. (2021a). Faqs on ecb supervisory measures in reaction to the coronavirus. Retrieved December 30, 2021, from https://www.bankingsupervision.europa.eu/press/publications/ html/ssm.faq_ECB_supervisory_measures_in_reaction_to_the_coronavirusa631697a4. en.html ECB. (2021b). Supervisory review 2020. Retrieved December 30, 2021, from https://www.bankingsupervision.europa.eu/banking/srep/2021/html/ssm.srep202107_ outcomesrepitriskquestionnaire.en.html Eichenbaum, M. S., Rebelo, S., & Trabandt, M. (2021). The macroeconomics of epidemics. The Review of Financial Studies, 34(11), 5149–5187.

Bibliography

55

El Diri, M., King, T., Spokeviciute, L., & Williams, J. (2021). Hands in the cookie jar: Exploiting loan loss provisions under bank financial distress. Economics Letters, 209, 110098. ESMA. (2020, March 25). Accounting implications of the covid-19 outbread on the calculation of expected credit losses in accordance with ifrs 9. Retrieved December 30, 2021, from https://www.esma.europa.eu/sites/default/files/library/esma32-63-951_ statement_on_ifrs_9_implications_of_covid-19_related_support_measures.pdf Ewert, R., & Wagenhofer, A. (2005). Economic effects of tightening accounting standards to restrict earnings management. The Accounting Review, 80(4), 1101–1124. EY. (2018). Applying ifrs impairment of financial instruments under ifrs 9. Retrieved December 30, 2021, from https://www.ey.com/en_gl/ifrs-technical-resources/impairmentof-financial-instruments-under-ifrs-9 FCAG. (2009, July 28). Report of the financial crisis advisory group. Retrieved December 30, 2021, from https://www.iasplus.com/en/binary/crunch/0907fcagreport.pdf Filip, A., & Raffournier, B. (2014). Financial crisis and earnings management: The european evidence. The International Journal of Accounting, 49(4), 455–478. FSF. (2009, April 2). Report of the financial stability forum on addresing procyclicality in the financial system. Retrieved December 30, 2021, from https://www.fsb.org/wp-content/ uploads/r_0904a.pdf G20. (2009, April 2). London summit: Leader’s statement. Retrieved December 30, 2021, from http://www.g20.utoronto.ca/2009/2009communique0402.pdf Gebhardt, G. u., & Novotny-Farkas, Z. (2011). Mandatory ifrs adoption and accounting quality of european banks. Journal of Business Finance & Accounting, 38(3-4), 289–333. Gopinath, G. (2020). The great lockdown:Worst economic downturn since the great depression. Healy, P. M., & Wahlen, J. M. (1999). A review of the earnings management literature and its implications for standard setting. Accounting Horizons, 13(4), 365–383. IASB. (2009). Ed finacial instruments: Amortised costs and impairment. Retrieved December 30, 2021, from https://www.assb.gov.sg/docs/default-source/exposure-draft/basic-ofconclusion.pdf IASB. (2013, July 5). Ed finacial instruments: Expected credit losses. Retrieved December 30, 2021, from https://www.ifrs.org/content/dam/ifrs/project/fi-impairment/ exposuredraft-2013/published-documents/ed-expected-credit-losses.pdf IASB. (2020, March 27). Ifrs 9 and covid-19. accounting for expected credit losses applying ifrs 9 financial instrument in the light of current uncertainty resulting from the covid19 pandemic. Retrieved December 30, 2021, from https://cdn.ifrs.org/content/dam/ifrs/ supporting-implementation/ifrs-9/ifrs-9-ecland-coronavirus.pdf IMF. (2020a). Global financial stability report: Preemting a legacy of vulnerabilities. Retrieved December 30, 2021, from https://www.imf.org/en/Publications/GFSR/Issues/2021/04/06/ global-financial-stability-report-april-2021 IMF. (2020b). World economic outlook: A long and difficult ascent. Retrieved December 30, 2021, from https://www.imf.org/en/Publications/WEO/Issues/2020/09/30/ worldeconomic-outlook-october-2020 IMF. (2021). World economic outlook: Recovery during a pandemic. Retrieved December 30, 2021, from https://www.imf.org/en/Publications/GFSR/Issues/2021/04/06/ globalfinancial-stability-report-april-2021

56

Bibliography

ITG. (2015, December 11). Iasb transition resource group for impairment of finanacial instruments: Meeting summery. Retrieved December 30, 2021, from https://www.ifrs.org/ content/dam/ifrs/meetings/2015/december/itg/itgmeeting-summary-11-december-final. pdf Kinross, P., Suetens, C., Dias, J. G., Alexakis, L.,Wijermans, A., Colzani, E., Monnet, D. L., et al. (2020). Rapidly increasing cumulative incidence of coronavirus disease (covid-19) in the european union/european economic area and the united kingdom, 1 january to 15 march 2020. Eurosurveillance, 25(11), 200–285. Kirschenheiter, M., & Melumad, N. D. (2002). Can „big bath“ and earnings smoothing coexist as equilibrium financial reporting strategies? Journal of Accounting Research, 40(3), 761–796. Knock, E. S., Whittles, L. K., Lees, J. A., Perez-Guzman, P. N., Verity, R., FitzJohn, R. G., Gaythorpe, K. A., Imai, N., Hinsley, W., Okell, L. C., et al. (2021). Key epidemiological drivers and impact of interventions in the 2020 sars-cov-2 epidemic in england. Science Translational Medicine. Laeven, L., & Majnoni, G. (2003). Loan loss provisioning and economic slowdowns: Too much, too late? Journal of Financial Intermediation, 12(2), 178–197. Liang, J. J., & Dong, Z. (2018). How the 2007 global financial crisis changed the financial disclosure behavior: The case of us equity reits. Property Management. Mark Rhys, A. S. (2016). The implementation of ifrs 9 impairment requirements by banks (type). Global Public Policy Committee. Retrieved December 30, 2021, from https://www2.deloitte.com/bd/en/pages/financial-services/articles/2016-gppc-theimplementation-of-ifrs9-impairment-requirements-bybanks.html Novotny-Farkas, Z. (2016). The interaction of the ifrs 9 expected loss approach with supervisory rules and implications for financial stability. Accounting in Europe, 13(2), 197–227. Payne, J. L., & Robb, S.W. (2000). Earnings management: The effect of ex ante earnings expectations. Journal of Accounting, Auditing & Finance, 15(4), 371–392. PWC. (2017). Ifrs 9, financial instruments understanding the basics. Retrieved December 30, 2021, from https://www.pwc.com/gx/en/audit-ervices/ifrs/publications/ifrs-9/ifrs-9understanding-the-basics.pdf Rahman, M. M., Moniruzzaman, M., & Sharif, M. J. (2013). Techniques, motives and controls of earnings management. International Journal of Information Technology and Business Management, 11(1), 22–34. Rajan, R. G. (1994). Why bank credit policies fluctuate: A theory and some evidence. The Quarterly Journal of Economics, 109(2), 399–441. Ryan, S. G. (2007). Financial instruments and institutions: Accounting and disclosure rules. John Wiley & Sons. Scott, W. R., & O’Brien, P. C. (2003). Financial accounting theory (Vol. 3). Prentice Hall Toronto. Skinner, D. J., & Sloan, R. G. (2002). Earnings surprises, growth expectations, and stock returns or don’t let an earnings torpedo sink your portfolio. Review of Accounting Studies, 7 (2), 289–312. Skrabka, J. (2021). The moratorium on loan repayments during the covid-19 pandemic in europe: A comparative analysis of loan moratoria in selected European countries. Juridical Tribune-Tribuna Juridica, 291–301.

Bibliography

57

Sun, L., & Rath, S. (2010). Earnings management research: A review of contemporary research methods. Global Review of Accounting and Finance, 1(1), 121–135. Trueman, B., & Titman, S. (1988). An explanation for accounting income smoothing. Journal of Accounting Research, 127–139. WHO. (2020). Who director-general’s opening remarks at the media briefing on covid19-11 march 2020. Retrieved December 30, 2021, from https://www.who.int/directorgeneral/speeches/detail/who-director-general-s-openingremarks-at-the-media-briefingon-covid-19---12-november-2021