Discretion in reporting loan loss provisions
The impact of the reporting environment on the banking industry
Master Thesis – Final Version 22 June 2015
Chris de Kwaasteniet 10070451
MSc Accountancy & Control, variant Accountancy Amsterdam Business School
Faculty of Economics and Business University of Amsterdam
2 Statement of Originality
This document is written by student Chris de Kwaasteniet who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3 Abstract
Earnings management, capital management and information signaling within the financial sector has been tested by prior studies already. This reporting behaviour is mostly tested in current literature in combination with single factors from the reporting environment of these firms. In this study the single factors are combined to study how the whole reporting environment shapes the reporting behaviour of financial institutions. Previous literature shows that managers tend to exploit discretion in accounting rules by overstating the assets on the balance sheet. This will also lead to higher earnings in the income statement. By taking six different countries within Europe that have to report their financial statements according to IFRS, I find that the size of the auditor and the periods of intense scrutiny over the financial reporting lead to a stricter reporting environment in which managers report higher impairment losses. This means that a stricter reporting environment gives managers less room to exploit discretion in accounting rules, leading to higher reliability and comparability of financial statements. The results found in this study indicate that the stringency of the reporting environment can help to improve the reliability and comparability of financial statements.
4 Table of contents
1 Introduction ... 5
2 Theoretical background ... 8
2.1 Earnings management ... 8
2.2 IAS 39 - Financial instruments ... 9
2.3 The reporting environment ... 10
3 Hypotheses development ... 11
4 Data and method ... 13
4.1 Data collection and sample ... 14
4.2 Dependent variable ... 14 4.3 Independent variables ... 15 4.4 Control variables ... 16 4.5 Method ... 17 5 Results ... 18 5.1 Descriptive statistics ... 18 5.2 Regression assumptions ... 19 5.2.1 Assumption of multicollinearity ... 19 5.2.2 Assumption of independence ... 20
5.2.3 Assumption of linearity and homoscedasticity ... 20
5.3 Regression results ... 21
6 Discussion ... 22
6.1 Additional analyses ... 24
7 Conclusion ... 26
References ... 29
Appendix I - Descriptive statistics ... 32
Appendix II - Assumptions ... 33
Appendix III - Plots ... 34
5 1 Introduction
The practise of fair value accounting was challenged significantly during the financial crisis of 2008-2009 when a great number of companies needed to write-down the values of their assets. Fair value accounting is the practise of measuring assets and liabilities at estimates of their current value. The most important consequence of this accounting method is that the values of a company’s assets and liabilities need to be adjusted when market values change. These adjustments will impact both the balance sheet and the income statement of a company. One important aspect of fair value accounting is the impairment of assets. The impairment of an asset is the devaluation of an asset’s value, which is commonly described as a write-off or write-down. An impairment ensures that an asset is carried at no more than its recoverable amount. The impairment losses can either directly reduce the carrying amount of the asset or through the use of an allowance account, often called loan loss allowance. The amount of loss is directly recognised in profit or loss (IASB, 2011, para. 39.63).
The concept of fair value accounting does also apply to financial institutions. Most of the assets of banks are called financial assets or financial instruments and they are recorded in the financial statements according to IAS 39 – Financial Instruments: Recognition and Measurement. These assets mainly consist of outstanding short-term and long-term loans. To ensure that these loans are recorded at their current values banks need to report impairment losses when their debtors are not able to repay the outstanding loan. IAS 39 recognises these impairments following an ‘incurred loss model’. Under the incurred loss model it is assumed that the loan will be repaid until objective evidence of a ‘loss event’ is identified and the impact on the future cash flows can be reliably estimated (IASB, 2011, para. 39.59). The practise of estimating these impairment losses is especially important during and after economic recessions when more debtors fail to comply with their financial obligations.
The impairment of assets is subject to managers’ discretion due to the identification of ‘loss events’ and the calculation of the amount of impairment loss. Even though IAS 39 names a list of events that could trigger the recognition of an impairment loss, real world events can always be different (IASB, 2011, para. 39.59). Therefore, the identification of a ‘loss event’ requires the professional judgment of a manager. The amount of impairment loss depends on the difference between the carrying value of the financial asset and the discounted value of the estimated future cash flows using the effective interest rate. Determining the future cash flows requires professional judgment on for example credit risks and default rates when observable data is limited or no longer relevant (IASB, 2011, para. 39.63). Prior
6 literature on managers’ discretion shows that managers tend to exploit this discretion by reporting their assets opportunistically in the financial statements (Reidl, 2004; Boone and Raman, 2007; Vanza et al., 2011). Although guidance for the recognition of impairments for banks is provided by IAS 39, it does not provide an unambiguous standard for all the evidence that could trigger an impairment. By exploiting the discretion in the impairment guidelines, managers can have great impact on the value of a bank’s assets. According to Ramanna and Watts (2012), this discretion leads to avoidance of impairments within regular industries, which results in overstated assets. Their arguments are based on agency theory, which predicts that managers will use the unverifiability in accounting rules to manage financial reports opportunistically. Other literature suggests that discretion within the banking industry is also used for capital management, earnings management and even signaling effects (Beatty et al., 1995; Collins et al., 1995; Beaver and Engel, 1996; Ahmed et al., 1999). The main point is that the discretion of managers in determining the amount of loan loss provision can result in financial information that does not provide a faithful representation of reality.
To reduce the exploitation of discretion, Leuz et al. (2003) have shown that stricter legal enforcement can help. However, legal enforcement is only one component of the reporting environment of a firm. Another component of the reporting environment is the audit quality of the firm, which has been tested by Stokes and Webster (2010). Vanza et al. (2011) studied the impact of the global financial crisis on the level of impairments. In this study, I examine all three components of the reporting environment, and how they influence managers’ use of discretion in reporting loan loss provisions within the banking industry. This leads to the main research question of this paper:
Does the reporting environment influence a managers’ use of discretion in reporting loan loss provisions within the banking industry?
In examining managers’ use of discretion in different reporting environments, this study contributes to the literature in four ways. The global financial crisis of 2008-2009 placed pressure on the fair value accounting standards and techniques. Many economists blame fair value accounting for misrepresenting the value of firms during booms and bursts. Even though Leuz and Laux (2009, p. 29) found that fair value accounting cannot be blamed for contributing to the financial crisis, the need for more research is acknowledged. Since loan loss provisions are important for reporting the right fair values of a bank’s assets, further research is needed on the accounting standard IAS 39.
7 The main objective of the International Accounting Standards Board (IASB) is to develop a set of financial reporting standards, which are of high quality, transparent and comparable (IASB, 2010). If firms have too much discretion in the implementation of an accounting standard, it lowers the overall transparency and comparability of that standard. More research with regard to managers’ use of discretion in reporting loan loss provision can give more insight into the transparency and comparability of the standard.
The accounting standard for the valuation of assets of regular firms, IAS 36 – Impairment of Assets, has been tested frequently within the current literature (e.g. Ball et al., 2000; Leuz et al., 2003). Most of these studies exclude banks and other financial institutions from their sample because this industry uses different rules for the valuation of assets. Financial institutions have to report their financial assets according to IAS 39. This study focuses especially on the banking industry, since this industry is an important section of the entire economy and cannot be left untested. This study also focuses on several countries in Europe in order to examine the implementation of IAS 39 under different reporting environments, while maintaining the same set of accounting standards.
During and shortly after the global financial crisis followed a period of severe scrutiny to ensure impairments on all assets reflected the firm’s underlying economic position and performance (Stokes and Webster, 2010). Auditors were also under greater pressure from regulators to take a conservative position on impairment charges, according to Stokes and Webster. Due to the global financial crisis of 2008-2009 the reporting environment changed significantly for the banking industry and this stricter reporting environment may have resulted in different loan loss reporting behaviour by managers. Although the global financial crisis gives the opportunity to test changes in reporting behaviour, it has not been tested much in the existing literature. This study includes the period before, during and after the global financial crisis.
In Chapter 2 the theoretical background of this paper will be explained. I will start with an elaboration of the previous literature on loan loss provisioning. This will be followed by an explanation of IAS 39 – Financial Instruments: Recognition and Measurement and the reporting environment of financial institutions. In Chapter 3 the hypotheses will be developed based on the existing literature about loan loss provisioning and the reporting environment. This will be followed by an outline of the data selection procedure and the research method in Chapter 4 ad the empirical results of the regression model in Chapter 5. Then, a discussion in Chapter 6 will discuss the tested hypotheses. This section will also include some additional sensitivity analyses. I will finish the paper with a summary and conclusion in Chapter 7.
8 2 Theoretical background
This section will describe the current literature on loan loss provisions and the various incentives to shape these provisions in the financial statements. Then, I will elaborate on the financial accounting standard IAS 39 – Financial Instruments: Recognition and Measurement used by banks to measure their financial assets. Finally, I will discuss the existing literature on the reporting environment of firms.
2.1 Earnings management
Earnings quality and earnings management are two aspects of financial reporting that have been examined thoroughly in the current literature. An important accounting practice used for earnings management within the banking industry is the recognition of loan loss provisions. In October 1994 the General Accounting Office found that depository institutions maintained “significant amounts of unsupported loan loss reserves … not linked to loss exposure” (GAO, 1994, p. 5). This was followed by a demand from the SEC in November 1998 on SunTrust Banks to reduce the allowance for loan losses significantly (New York Times, 1998). In July 2001 the SEC issued SAB No. 102 - Selected Loan Loss Allowance Methodology and Documentation Issues, which provides guidelines on the loan loss recognition practice (SEC, 2001). The likely reason for financial institutions to recognise those large amounts of loan loss reserves was to soften the income increase during the 1990s economic boom and to build up reserves for bad times. This example shows how banks can use the loan loss provisions to manage their earnings. The reasons behind earnings management for banks are largely in accordance with the traditional reasons, which also apply to regular industries. Most studies have found reasons such as managers’ job security, compensation purposes, external financing and income smoothing (e.g. Fudenberg and Tirole, 1995; Joyce, 1996; Beatty et al., 2002, Kanagaretnam et al., 2003). Earnings management via the loan loss provisions is possible due to the discretion needed in determining the acurate amount.
Previous research streams have also tried to link loan loss provisioning to information signaling and capital management with mixed outcomes as result. Liu et al. (1997) and Wahlen (1994) find that the recognition of loan loss provisions is interpret as ‘good news’ by investors, even though the provisions are measures of expected losses that reduce current earnings. Ahmed et al. (1999) do not find support for the desire to signal information to investors, but they show that the banking industry uses provisions for capital management. The results of Beatty et al. (1995) support the capital management hypothesis, while the
9 studies of Collins et al. (1995) and Pérez et al. (2008) do not. What probably causes different results for capital management within the banking industry are the Basel regulations. The Basel-I regime removed loan loss allowances from the calculation of Tier 1 capital, thereby lowering the capital management possibilities via loan loss provisioning.
The current literature shows clear evidence for earnings management and suggests other incentives for shaping the amount of loan losses, such as information signaling and capital management. However, these incentives are influenced by accounting standards and other regulations within the reporting environment of banks, such as Basel-I.
2.2 IAS 39 – Financial Instruments
Since 2005 banks in the European Union have to report their financial statements according to IFRS. In IFRS the recognition and measurement of financial instruments is stated in IAS 39. IAS 39 applies to all types of financial instruments except for interests in subsidiaries, employers’ rights and obligations, share-based payment transactions and a few specific contracts. The standard distinguishes two stages of measurement; the initial measurement and the subsequent measurement. According to IAS 39, all financial instruments should initially be recognised at their fair value. The fair value is the amount for which an asset could be exchanged, or a liability settled, between knowledgeable, willing parties in an arm’s length transaction. The fair value can be determined by taking quoted market prices in an active market or using a valuation method that makes maximum use of market inputs. If both are not available for the financial instrument, then an entity must measure the instrument at cost less impairment, as stated by the IASB. The subsequent measurement depends on the kind of financial instrument. All the financial assets and liabilities should still be measured at fair value, except for loans and receivables, held-to-maturity investments and investments in equity instruments with no reliable fair value measurement. Those instruments should be measured at amortised cost and are subject to an annual impairment test (IASB, 2011).
Most problems will arise when the fair value of a financial instrument cannot be determined by using market information. This was shown by both the IASB and the FASB in 2008. In response to the financial crisis, the IASB changed IAS 39 and the FASB published FAS 157-3 to provide more guidelines on how to determine the fair value of a financial asset when the market for that asset is not active. The IASB also set up an ‘Expert Advisory Panel’ to provide additional guidance in the area of valuation techniques.
10 As explained by the IASB, financial assets not measured at fair value are subject to an impairment test. To report impairments on a financial asset objective evidence is required of one or more events that occured after initial recognition and that event (or events) should have an impact on the future cash flows of the instrument that can be realibly measured. Examples of ‘loss events’ are given by the IASB and contain events such as financial difficulty of the obligated party, the breach of a contract and the probability of a borrower entering bankruptcy or reorganisation. These ‘loss events’ are all related to the uncollectibility of the financial asset. At the end of each reporting period, an entity is required to assess whether there is any objective evidence of impairments. If there is any indication that an asset may be impaired, the asset’s carrying amount and the present value of estimated cash flows discounted at the financial asset’s effective interest rate should be determined. To determining the present value of estimated cash flows banks have to determine the credit risk characteristics that are indicative of its debtors’ ability to pay. Based on these characteristics the bank decides to recognise an impairment loss and the amount of the loss. The carrying amount of the asset shall be reduced either directly or through use of an allowance account. The amount of the loss shall be recognised in profit or loss. The impairment on some financial assets can also be reversed when a new event (or events) occurs that provides objective evidence of a decrease in the impairment loss. However, this reverse will never be higher than the original amount of impairment loss recorded (IASB, 2011).
2.3 The reporting environment
The reporting environment of a firm is broadly defined as the setting in which a firm has to report its financial statements. The reporting environment consists of a wide range of determinants that individually or together influence the way a firm reports its financial information. This paper only covers the mandatory reporting environment since the sample group only consists of banks that are subject to a mandatory reporting regime. The determinants of the mandatory reporting environment of banks that have been tested in the existing literature can be categorised into three groups. The first group, country-level determinants, consist of determinants as creditor rights, rights of minority shareholders, legal enforcement, political influences and bank regulation and supervision (Fonseca and Gonzalez, 2008; La Porta et al., 2002; Leuz et al., 2003). Firm-specific determinants, the second group, are for example the financial structure of a bank, the audit quality and the risk and cost of litigation (Fonseca and Gonzalez, 2008; Kim et al., 2003; Ball et al., 2003). Finally, the third
11 group, periods of intense scrutiny, contains determinants that cause a period of intense scrutiny over a firm’s financial reporting, such as the decline in equity markets during the global financial crisis (Bartram and Bodnar, 2009; Durkin, 2009). The determinants of the reporting environment of banks can either have a positive or negative relationship with the stringency of the reporting environment. The determinants tested in this paper are the legal enforcement system of a country, the audit quality of the audit firm and the influence of the global financial crisis of 2008-2009.
3 Hypotheses development
The quality of the reported accounting information of financial institutions is influenced by various incentives of managers. Previous literature has found significant evidence for earnings management within the banking industry. A couple of these studies found that managers value their assets opportunistically (Reidl, 2004; Boone and Raman, 2007; Vanza et al., 2011). Since earnings management is mostly done through the recognition of loan loss provisions, this means that managers tend to report low amounts of loan loss provisions to overstate the assets and report higher earnings. The reporting environment can, however, significantly reduce these earnings management practices. The three determinants of the reporting environment tested in this study are assumed to have a positive effect, which means that the reporting environment will be more stringent with the presence of these determinants. Three hypotheses are developed to test the impact of the reporting environment.
The first category enfolds the firm-specific determinants. The determinants in this category are firm-level characteristics that influence the way firms report their financial statements. One of these firm-level characteristics is the quality of the audit that is performed on the firm’s financial statements. Previous literature has shown that the audit quality is positively related to the size of the auditor. Kim et al. (2003) found that larger audit firms receive more media attention and therefore bear higher costs of reputation loss from audit scandals compared to smaller audit firms. It is also more likely for a large audit firm to get sued, because firms have better expectations about the benefits flowing from a lawsuit against a large audit firm with “deeper pockets”. They concluded that larger audit firms therefore tend to be more conservative in determining the reported earnings. In the study of DeAngelo (1998) another aspect of the relationship between audit quality and the size of the audit is the dependence of smaller audit firms on single (profitable) clients. A non-big four auditor has more to gain by keeping the single client and misreporting then by being critical and
12 potentially losing the client. Finally, it is shown that big four audits provide a higher audit quality due to their higher level of expertise and higher amount of available resources (Francis et al., 2005; Ferguson et al., 2003). Because big four audit firms provide an audit service of higher quality, banks audited by these audit firms are expected to be under higher scrutiny than banks audited by a small audit firm. The managers of these banks will therefore have less room to use their discretion in reporting loan loss provisions. This will result in a more conservative accounting approach, which increases the magnitude of the loan losses. This leads to the following hypothesis:
Hypothesis 1: Banks audited by big four audit firms will report a higher amount of loan loss provisions than banks audited by non-big four audit firms.
The second category contains the country-level determinants. The reporting quality of firms is not only determined by the financial accounting standards set by the IASB and the FASB, the characteristics of a country play an important role as well. La Porta et al. (1996) show that the reporting quality of firms from various countries substantially differs as a consequence of the law system used. They showed that common law countries give the best protection to both shareholders and creditors, while code law countries give the least. The United Kingdom is a common law country. The other countries tested in this study are the Netherlands, Germany, France, Spain and Belgium, which are classified as code law countries. Other studies also tried to link characteristics such as the legal enforcement system, the amount of investor protection and securities regulations to the reporting behaviour of firms. Ball et al. (2000) and Ball et al. (2003) show that financial reporting incentives could be more important than the underlying accounting standards and these reporting incentives are among others influenced by the legal system of a country. They show that accounting income in common law countries provides greater conservatism compared to code law countries, which reduces the incentives of managers to manage earnings. Finally, Leuz et al. (2003), Leuz (2010) and Barth et al. (2012) study the legal enforcement system and the rights of minority shareholders and find that differences in legal enforcement and investor protection lead to reporting differences under the same accounting standards. In a stringent reporting environment the legal enforcement system is better and investors receive higher protection from reporting firms, which results in less use of discretion by managers of banks in determining the amount of loan loss provision. This leads to the following hypothesis:
13
Hypothesis 2: The level of legal enforcement of a specific country is positively related to the amount of loan loss provisions reported by banks of that country.
The third category includes periods of intense scrutiny. A period of intense scrutiny can arise when the financial statements of firms are not in accordance with the underlying values of the assets and liabilities. This situation can evolve during periods of economic growth or economic recession, but it normally gets problematic and causes harm to investors and creditors during the latter. The global financial crisis for financial institutions started on August 9, 2007, when BPN Paribas blocked withdrawals from three hedge funds after recognising liquidity problems in the US securitisation market (BNP Paribas, 2007). However, in previous literature the years 2008 and 2009 are often appointed as the years of the crisis. According to Durkin (2009) the global financial crisis of 2008-2009 was a period of intense scrutiny. In this period, management and auditors were under strict control to ensure that all the assets reflected the underlying economic performance. Stokes and Webster (2010) also argue that auditors were potentially under increased scrutiny during this period. However, they could not find evidence of a more conservative approach taken by big four auditors towards the impairment of assets. In short, the crisis was certainly a period of intense scrutiny and has possibly resulted in a more conservative approach towards financial reporting. This gave managers of banks less room to use their discretion in determining the amount of loan loss provisions. It is also assumed that this period of intense scrutiny endured in the years after the global financial crisis. This leads to the following hypotheses:
Hypothesis 3a: The amount of loan loss provisions reported by banks is higher during the period of the global financial crisis than before the global financial crisis.
Hypothesis 3b: The amount of loan loss provisions reported by banks is higher in the period after the global financial crisis than before the global financial crisis.
4 Data and method
This section will discuss the process of data collection and the method used to conduct the research. Furthermore, it will provide an explanation of the measurements of the dependent, independent and control variables.
14 4.1 Data collection and sample
In order to examine the impact of the reporting environment on reporting behaviour, a group of countries, that have to state their financial figures according to the same accounting standards, is selected. The sample consists of financial institutions within the Netherlands, Germany, France, Spain, Belgium and the United Kingdom that report their financial statements according to IFRS. In 2005, IFRS was introduced to all the companies in countries who are part of the European Union. The sample period therefore starts in 2005. Due to the scarce availability of data for the year 2014 the sample period finishes in the year 2013. The Compustat Global Database is used to collect the information and provides an initial sample of 9,108 firm-year observations divided over 1,277 financial institutions. This study excludes the financial insitutions with Standard Industry Classification (SIC) codes ranging from 6331 to 6799, because either these firms do not consistently report loan loss provisions over the sample period or the data for these firms is missing. In addition, a firm is dropped due to the availability of data if the firm reports loan loss provisions in merely two or less firm-years. It is assumed that these firms, e.g. insurance companies or investment offices do not have financial assets which are subject to annual impairment tests. Both changes lower the sample to 125 financial insitutions (SIC codes 6000 to 6311) and 1,047 firm-year observations. With regard to the missing values in the data sample for single firm-year observations two important choices are made. First, if either the amount of loan loss provision or loan loss allowance is missing for one year, the value is adjusted to the amount of zero. Second, due to the transition from domestic accounting rules to IFRS in 2005 the balance sheet information in 2004 is limited. If the amount of total assets in 2004 is missing and therefore the lagged total assets cannot be determined, the total assets of the current year are taken. Finally, the outliers are removed from the sample. This leads to the deletion of three firms which are in severe distress, lowering the sample to 122 financial institutions with 1,020 firm-year observations.
4.2 Dependent variable
The dependent variable is the amount of impairment on financial assets reported in the financial statements. The impairment is scaled by the total amount of assets in this study. This variable is normally scaled by the lagged total loans. Due to missing data on the total amount of outstanding credit by financial institutions, the total loans will be replaced by the total amount of assets minus the total amount of property, plant and equipment (PPE). There are
15 two impairments in the financial statements, one on the balance sheet and one in the income statement. When financial institutions perceive the uncollectability of their financial assets they have to write down their assets. First, a loan loss allowance is created on the balance sheet and the amount is charged against the income statement as an expense. If the financial situation of the borrower changes, the loan loss allowance can be revised either through the recognition of another expense or through a reverse charge that lowers the amount of the allowance. In the latter situation, the loan loss allowance is reversed through the income statement resulting in a profit. Only when the financial asset becomes definitely uncollectable the financial asset is lowered by the amount of loan loss. In this study the dependent variable is the loan loss expense recognised in the income statement. This is in accordance with previous literature, summarized by Beatty and Liao (2014, p. 363).
4.3 Independent variables
The independent variables consist of the determinants of the reporting environment of financial institutions. The determinants that are tested in this study are the size of the auditor, the legal enforcement system of a country and the impact of the global financial crisis.
The size of the auditor is determined by making a distinction between big four auditors and non-big four auditors. The big four auditors are Ernst & Young, Price-WaterhouseCoopers, KMPG and Deloitte & Touche, the remainder of the audit firms is classified as non-big four auditor. This variable Auditorjt is tested by creating a binary
variable, which is equal to unity if the financial institution (j) is audited by a big four auditor at the end of the fiscal year (t) and zero otherwise.
The legal enforcement system is measured by the ‘rule of law’, which is taken from the study of Kaufmann et al. (2010). The rule of law reflects the perceptions of the extent to which agents have confidence in and abide by the rules of society, and in particular the quality of contract enforcement, property rights, the police, and the courts, as well as the likelihood of crime and violence. The rule of law is a value between -2.5 and 2.5. The relationship between the rule of law and the stringency of the reporting environment is positive, which means that a value close to 2.5 gives an indication of strong governance performance and a value close to -2.5 illustrates weak governance performance. The variable
Rulejt is equal to the rule of law value of the country in which the firm (j) operates, measured
16 The global financial crisis caused a period of intense scrutiny over the financial reporting of firms. According to previous literature, the crisis started in 2008 and ended in 2009. This period is measured by using a binary variable GFCj, which is equal to unity if the
fiscal year of the firm (j) ends in the years 2008 or 2009 and zero in the other years. It is also expected that the reporting environment is stricter during the years after the global financial crisis compared with the years before. This research, therefore, includes a binary variable
AfterGFCj, which measures the stringency of the reporting environment in the years 2010 to
2013. The variable is equal to unity if the fiscal year of the firm (j) ends after 2009 and zero otherwise.
4.4 Control variables
The first control variable in this model is ΔLoanjt, which represents the loan growth of the
firm (j) in the fiscal year (t). When the amount of outstanding loans grows within a year, one would expect the amount of loan loss provision to be higher. This variable controls for financial institutions that extend their credit within a certain fiscal year. Again, due to missing data on the total amount of outstanding credit, the change in total loans will be replaced by the change in total assets minus the PPE, which will also be scaled by total assets. It is expected that the change in total assets for financial institutions is largely caused by the change in outstanding credit.
The control variable ALWt-1is calculated by dividing the total loan loss allowance by the total assets minus the total PPE. This variable includes the loan loss allowance created on the balance sheet as opposed to the dependent variable of this study, which uses the loan loss recorded in the income statement. The variable ALWt-1represents the relative amount of loan loss allowance on the balance sheet in the previous fiscal year (t-1) for the firm (j). The reason this control is included in the model is that a higher loan loss allowance in a certain fiscal year most probably lowers the amount of loan loss provision in the following year.
The last control variable Leveragejt is the total amount of liabilities divided by the total
amount of equity in fiscal year (t) for firm (j), also known as the debt-to-equity ratio. A high debt-to-equity ratio shows a firm that finances his growth mainly with debt. This increases volatility in the income statement due to interest expenses and may also higher the likelihood of default or bankruptcy. The debt-to-equity ratio of a firm also highly influences the reporting behaviour of the firm. The debtholders have greater influence on the decision making of highly leveraged firms and other studies also found that these firms tend to be more
17 conservative than low leveraged firms (Beaver and Ryan, 2005; Ball et al., 2008). This control variable is included in the model to control for the performance of the firm, as this variable predicts the likelihood of default. In addition, it controls for the highly leveraged firms that report their earnings more conservative, resulting in higher loan loss provisions.
4.5 Method
The relationship between the reporting environment and the amount of loan loss provision reported by financial institutions will be tested using an ordinary least squares (OLS) regression. A regression analysis is a statistical method to estimate the relationships among variables in a set model. It identifies the impact of various simultaneous influences upon a dependent variable. This method is used for predicting the value of a dependent variable based on the value of the independent variables. The regression equation used to test the hypotheses is as follows:
LLPjt = α0 + α1Auditorjt + α2Rulejt + α3GFCj + α4AfterGFCj + α5ΔLoanjt +
α6ALWt-1 + α7Leveragejt + εjt
The dependent variable in the regression equation is LLPjt, which represents the loan loss
provision charged to the income statement of firm j in fiscal year t. The α0 is the mean value
of the loan loss provision when all the independent variables are zero. The amount of change in the dependent variable caused by the independent variables is shown by α1, α2, etc. All the
independent variables are expected to have a positive influences on the dependent variable. Thus, the size of the audit firm, stringency of the legal enforcement system and the period of intense scrutiny during the global financial crisis increase the amount of loan loss provision charged to the income statement. The independent variables are as follows; Auditorjt
represents the size of the auditor of firm j in fiscal year t, Rulejt reflects the value of the rule of
law of the country of firm j during the fiscal year t, and GFCj and AfterGFCj show,
respectively, the periods during and after the financial crisis of 2008 and 2009. The control variables are as follows; ΔLoanjt shows the loan growth of firm j during fiscal year t, ALWt-1 covers the loan loss allowance created by firm j in the previous fiscal year t-1. Finally, the
Leveragejt variable shows the debt-to-equity ratio of firm j in fiscal year t.
The first hypothesis will be accepted when the Auditorjt has a positive coefficient. That
18 higher amounts of loan loss provisions reported. Hypothesis 2 expects the same positive coefficient with the variable Rulejt. Hypotheses 3a and 3b predict the variables GFCj and
AfterGFCj to have a positive coefficient. Accordingly, managers reporting during or after the
global financial crisis report in a more stringent reporting environment leading to higher amounts of loan loss provisions.
5 Results
5.1 Descriptive statistics
This section will cover the descriptive statistics in order to get a better understanding of the collected data. As showed in Table 5.1 in the Appendix I the dataset contains 1,020 firm-year observations with information about the amount of loan loss provision. These firm-year observations are spread over a total of 122 financial institutions. The loan loss provision is scaled by the total assets of the financial institution, which results in a low mean value. This value shows that the average amount of loan loss provision recorded is approximately 0.45% of the total assets. The variable LLPjt also contains negative values due to the reverse
impairments. The number of firms audited by a big four auditor is low. Only 18.63% of the firm-year observations are audited by a non-big four auditor, which is 190 years of the total 1,020 years. Therefore, the results of this variable should be interpret with care. The Rulejt
variable shows an average of 1.549. This shows that the sample group of countries have a high rule of law value compared to the world database, which ranges from -2.5 to 2.5. The minimum value of 0.996 has been given to Spain, while the Netherlands have received the highest value of 1.841. The rule of law values of Belgium, France and Spain over the sample period are below average, the other countries score above it. Table 5.2 shows the sample composition over the different countries with the average rule of law value per country. The group of financial institutions reporting in countries with a weak legal enforcement system is slighty lower than the group reporting in a strong legal enforcement system. Table 5.3 below gives a summary of the sample size and the mean value of the LLPjt variable per time period.
This table already gives an indication of the influence of the global financial crisis on the relative amount of loan losses reported. It can be seen that in the period during the global financial crisis the loan loss provisions recognised are noticeably higher than in the periods before and after the crisis. The period after the crisis also shows a higher mean value than before the crisis.
19 Table 5.2: Sample composition
Sorted by country
Country Number of firms Number of obs. Rule of law
Belgium 2 15 1.333 France 30 267 1.440 Germany 44 373 1.666 Netherlands 5 45 1.789 Spain 15 114 1.110 United Kingdom 26 206 1.682 Total/Average 122 1,020 1.549
Table 5.3: Sample composition Sorted by time period
Time period Number of firms Number of obs. Mean value LLPjt
Before GFC 116 340 0.00282
GFC 115 226 0.00713
After GFC 118 454 0.00448
Total/Average 1,020 0.00451
5.2 Regression assumptions
The assumptions of an OLS regression are multicollinearity, the independence of residuals, linearity and homoscedasticity. The assumptions are tested by using statistical tests and plots.
5.2.1 Assumption of multicollinearity
Multicollinearity means that two or more independent variables within a regression model are highly correlated. To test this assumption the correlation values of the variables are produced and stated in Table 5.4 in the Appendix II. It is shown that the correlations between the dependent variable LLPjt and the independent variables Auditorjt (r = 0.029, n = 1,020,
p = 0.178) and GFCj (r = 0.132, n = 1,020, p < 0.05) are positive. The variables Rulejt
(r = -0.030, n = 1,020, p = 0.170) and AfterGFCj (r = -0.003, n = 1,020, p = 0.465) show a
negative correlation with LLPjt. The correlation values are low, meaning that the statistical
20 applies to the control variables ΔLoanjt (r = -0.036, n = 1,020, p = 0.126), ALWt-1 (r = 0.088, n = 1,020, p < 0.05) and Leveragejt (r = -0.126, n = 1,020, p < 0.05). Looking at the
correlations between the independent variables, the highest correlation is shown between
GFCj and AfterGFCj (r = -0.478, n = 1,020, p < 0.05). However, this correlation is still at an
acceptable level and does not imply multicollinearity. The assumption of multicollinearity can also be tested by the tolerance and VIF values, which are shown by Table 5.5 in Appendix II. The tolerance is an indicator of how much of the variability of the specified independent variable is not explained by the other independent variables in the model. The VIF value is the inverse. The tolerance values range from 0.674 to 0.935 and the VIF values are far below 10. This means that the model complies with the assumption of multicollinearity.
5.2.2 Assumption of independence
The assumption of independence tests whether the residuals of the individual observations are independent. The independence of the individual observations can be tested by the Durbin-Watson test. The Durbin-Durbin-Watson statistic is a value between 0 and 4. A value close to 0 gives an indication of positive autocorrelation; a value close to 4 suggests the opposite. A Durbin-Watson value of 2 means that there is no autocorrelation. The Durbin-Durbin-Watson statistic in the model is 2.102, which gives no indication of autocorrelation. Hence, the model satisfies the assumption of independence.
5.2.3 Assumption of linearity and homoscedasticity
A standard multiple regression can only reliably estimate the relationship between the dependent and independent variables if the relationship is linear. The linearity assumption can be tested by looking at the normal probability plot and the scatterplot. The normal probability plot shows the standardized residuals against the predicted values. The assumption of linearity requires the values to be distributed around a diagonal line. In Graph 5.1 in Appendix III the pattern of the values form a S-curve and thus violate the assumption of linearity. The homoscedasticity assumption is tested by looking at the scatterplot. The values plotted in this graph should be roughly rectangularly distributed, with most of the scores concentrated in the centre. Graph 5.2 in Appendix III displays that the observations are concentrated in the middle right area of the plot and the residuals are not rectangularly shaped, indicating heteroscedasticity in the dependent variable. The violation of this assumption is considered a weakness in this study.
21 5.3 Regression results
After complying with the assumptions of an OLS regression, the hierarchal regression can be performed. The results from the regression are shown in Table 5.6. In the hierarchal regression two models are created; one model with the control variables and one model with all the variables. In this way the impact of the independent variables can be tested while the model is controlled for other effects.
The results show the influence of the reporting environment on the loan loss provision reported, while controlling for the loan growth, the loan loss allowance and the debt-to-equity ratio. The reporting environment is expected to have a positive relationship with the reported amount of loan loss provisions. This means that in a stringent reporting environment the loan loss provisions should be higher. The independent variables represent the stringency of the reporting environment and are therefore expected to have a positive influence on the dependent variable as well. In Table 5.6 the standardized coefficients column show the effects of the control variables and individual independent variables on the dependent variable.
The variable ΔLoanjt has a slightly negative relationship with the loan loss provisions,
indicating that the relative amount of loan loss provisions declines when a financial institution extends his outstanding credit. The negative relationship between these variables is remarkable, since one would expect the loan loss provisions to be higher after loan growth. It is, however, likely that the relationship is disturbed by using the amount of total assets instead of the total loans. The variable ALWt-1 shows a positive relationship. This relationship is logical as this control variable gives an indication of the lagged non-performing assets as well, meaning that a higher loan loss allowance in year t-1 indicates a poorer credit portfolio in year t-1, resulting in a higher loan loss provision in the following years. However, this effect is largely offset since a higher loan loss allowance in year t-1 lowers the need for a loan loss allowance in the subsequent year. Finally, the performance variable Leveragejt reveals a
very small, negative relation with the dependent variable. This relationship is also surprising, since a higher leverage ratio gives an indication of more risky performance, normally resulting in a lower performance and thus a higher loan loss provision. However, this coefficient indicates that there is almost no relationship between the two variables.
As expected, the independent variables all have a positive impact on the LLPjt. The
variable GFCj (t = 3.276, p < 0.001) has the strongest impact on the dependent variable. This
result shows that the loan loss provisioning was significantly higher during the global financial crisis of 2008 and 2009. The results also reveal that the size of the audit firm has an
22 impact on the dependent variable, however, this finding is less convincing. The Auditorjt
variable has a significant impact on LLPjt at the 90% confidence level (t = 1.754, p < 0.10).
The results of the other two variables Rulejt (t = 0.635, p = 0.525) and AfterGFCj (t = 1.354,
p = 0.176) do not fall into the significance level of 0.05. In conclusion, the results suggest that the independent variables GFCj and Auditorjt lead to higher amounts of loan loss provisions.
Table 5.6: Model Coefficients
Unstandard. Coefficients Standard. Coefficients
B Std. Error Beta t Sig.
Constant 0.001 0.003 0.438 0.000 Auditorjt 0.002 0.001 0.056 1.754 0.080* Rulejt 0.001 0.002 0.021 0.635 0.525 GFCj 0.004 0.001 0.175 4.900 0.000*** AfterGFCj 0.001 0.001 0.051 1.354 0.176 ΔLoanjt -0.001 0.001 -0.018 -0.556 0.578 ALWt-1 0.068 0.031 0.073 2.199 0.028** Leveragejt -0.00009 0.000 -0.150 -4.595 0.000***
* significant at 0.10 level, ** significant at 0.05 level, *** significant at 0.01 level
6 Discussion
This section will provide a thorough discussion of the regression model results. The results will be used to evaluate the hypotheses created in the beginning of the paper. The findings will also be linked to previous literature within the banking industry.
This study tried to link the stringency of the reporting environment to the use of discretion in reporting loan loss provision. The determinants of the reporting environment included in the study are the size of the auditor of the firm, the legal enforcement system of a country and the global financial crisis of 2008 and 2009. All the hypotheses expected a significant and positive relationship with the amount of loan loss provision reported by banks. While testing for the determinants of the reporting environment, the model is controlled for loan growth, lagged loan loss allowance and the leverage of the firm.
Looking at the influence of the size of the auditor, the results show that the impact of this variable is significant at a 10% level (t = 1.754, p < 0.10). Therefore, hypothesis 1 is
23 accepted, although the outcome is not very convincing. This finding is in accordance with the findings of Kim et al. (2003), Ferguson et al. (2003) and Francis et al. (2005) that linked the auditor’s size to the media attention and the amount of resources and found a more conservative approach towards reporting earnings at big four auditors. My finding is consistent with the idea that the size of the audit firm influences managers’ use of discretion in reporting loan loss provisions. Thus, the presence of a big four audit firm results in a stringent reporting environment, which leads to the recognition of higher amounts of loan loss provisions compared to a weak reporting environment.
Hypothesis 1: Banks audited by big four audit firms will report a higher amount of loan loss provisions than banks audited by non-big four audit firms.
Previous literature on the legal enforcement of a specific country and the reporting behaviour of firms found that the reporting behaviour is more conservative in countries with more stringent investor protection (e.g. La Porta et al., 1996, Leuz et al., 2003). The results of the rule of law variable, which represents the stringency of law enforcement per country, show a insignificant coefficient (t = 0.635, p = 0.525). This results in the rejection of hypothesis 2. My finding indicates that a higher level of legal enforcement does not imply a stringent reporting environment and, therefore, does not significantly lower the discretion used in reporting loan loss provisions. This finding contradicts previous literature. However, not many studies performed these tests on a sample group reporting under exactly the same accounting standards. Since 2005, Europe provides the unique setting of comparing different legal enforcement systems, while maintaining the same set of accounting standards. This setting might have led to different results.
Hypothesis 2: The level of legal enforcement of a specific country is positively related to the amount of loan loss provisions reported by banks of that country.
Finally, the stringency of the reporting environment is expected to be influenced by the global financial crisis, which caused a period of intense scrutiny. A distinction is made between the period during and after the crisis. The period during the global financial crisis has a significant effect on the amount of loan loss provisions reported (t = 4.900, p < 0.001). This finding is in line with previous literature of Durkin (2009) and Stokes and Webster (2010)
24 that suggested the same period of intense scrutiny over the financial reporting of firms. Hence, hypothesis 3a is accepted. I can conclude that the global financial crisis caused a stringent reporting environment for financial institutions, resulting in higher amounts of loan loss provisions.
Hypothesis 3a: The amount of loan loss provisions reported by banks is higher during the period of the global financial crisis than before the global financial crisis.
In contrast with the results of the period during the global financial crisis, the period after the crisis does not significantly effect the dependent variable (t = 1.354, p = 0.176). This suggests that the period of intense scrutiny did not endure in the years after the crisis. An analysis of a longer period after the crisis compared to a period before the crisis might lead to other results. Following the results, hypothesis 3b is rejected. This finding is consistent with the idea that the period after the global financial crisis did not cause a stringent reporting environment resulting in less use of discretion in reporting loan loss provisions.
Hypothesis 3b: The amount of loan loss provisions reported by banks is higher in the period after the global financial crisis than before the global financial crisis.
6.1 Additional analyses
This section will cover two additional analyses. These analyses will test whether the results are robust after controlling for the non-performing assets and the return on equity ratio. Both variables could have been included in the OLS regression, however, this would have lowered the sample significantly due to the scarce availability of data on both variables. I will perform the additional analyses for both variables individually, since the data on non-performing assets is much lower than the data on the return on equity.
The non-performing assets are frequently used as control variable in loan loss provision models. This variable can reflect past and forward-looking information on the performance of the loan portfolio. Banks might use this variable to calculate the loan loss provision for the fiscal year (Beatty and Liao, 2014). The non-performing assets are included in the model with three different variables. The variables ΔNPAjt and ΔNPAjt+1 represent the amount of change in non-performing assets divided by the lagged total assets for firm j during the fiscal years t and t+1 respectively. These variables contain forward-looking information
25 on the loan portfolio. The variable ΔNPAjt-1 contains past information and represents the amount of change in non-performing assets divided by the lagged total assets for firm j during fiscal year t-1. The sample of this additional analysis contains 340 firm-year observations divided over 78 firms.
The return on equity is included in the model as a performance measure. This variable is not included in the current loan loss provision models, but the stock return is sometimes taken as a substitute (Beatty and Liao, 2014). The performance of financial institutions may be linked to the amount of loan loss provisions reported in the financial statements. Financial institutions that are in financial distress will report their financial statements with other incentives than high performing firms. An example could be a low performing firm that lowers its loan loss provisions to show a positive net income. A high performing firm might do the opposite to create a ‘safety net’ for bad times. The variable ROEjt is included in the
model to control for these differences in reporting behaviour. The ROEjt represents the net
income divided by the total equity of firm j in fiscal year t. The sample of this analysis contains 95 firms with 729 firm-year observations.
The results of the additional analyses are shown in Table 6.1 and 6.2 in Appendix IV. The results in this table show that both the incorporation of the non-performing assets and the return on equity in the regression model change the results of the first regression. It can also be seen that the non-performing assets have the largest impact on the model. The coefficients of the independent variables are still positive, however, the magnitude of the impact is lower. The first hypothesis, covering the impact of the audit firm on the loan loss provisions, is no longer accepted at a significance level of 10% with the incorporation of one of the two control variables, since the Auditorjt variable now shows a p-value of 0.769 (with ΔNPAj) and 0.151
(with ROEjt). The results do not have impact on the other hypotheses. In the additional
analysis, the hypotheses 2 and 3b are still rejected, while hypothesis 3a is accepted. The GFCj
still shows a significant impact on the LLPjt in the model with non-performing assets
(t = 2.979, p < 0.05) and the model with return on equity (t = 4.716, p < 0.001).
After controlling for the non-performing assets and the return on equity of financial institutions, I can conclude that only periods of intense scrutiny over a firm’s financial statements cause a stringent reporting environment in which those firms report higher amounts of loan loss provisions compared to a weak reporting environment.
26 7 Conclusion
The financial statements are used by management, investors and debtholders, who all have different ideas about the way the accounting figures should be measured. Therefore, the IASB created the IFRS to make the accounting information relevant, reliable and comparable for all users of the financial statements. Even though IFRS provides these general guidelines, the different incentives of firms and countries have an important impact on the reported information as well. This study provides evidence on the reporting behaviour of firms by analysing the effect of the reporting environment on the recognition of loan loss provisions according to IAS 39.
Prior literature has already found evidence on earnings management within the financial industry (e.g. Beatty et al., 2002, Kanagaretnam et al., 2003). Other studies also tried to link capital management and signaling to the reporting behaviour (Wahlen, 1994; Beatty et al., 1995; Ahmed et al., 1999). Within the financial industry loan loss provisions are mostly used for earnings management, since it is the most important bank accrual. A loan loss provision is an reserve created for a loan portfolio as a result of the uncollectability of loans. Managers of financial institutions have certain discretion in determining the amount of these provisions, due to the professional judgment required in setting the credit risk and default rates of a loan portfolio. That this can go wrong has been proved by the announcement of the General Accounting Office in 1994, stating that depository institutions held “significant amounts of unsupported loan loss reserves”.
The reporting environment can have a significant impact on the discretion used by managers of financial institutions. In this context, I expect a stringent reporting environment to reduce the amount of discretion used in reporting loan loss provisions. The hypotheses are tested by using firm-specific determinants, country-specific determinants and periods of intense scrutiny over managers’ reporting behaviour. The study uses a sample of 1,020 firm-year observations divided over 122 financial institutions in six different countries in Europe. The sample period is 2005 to 2013 and only contains years after the implementation of IFRS in Europe. The results show that the size of the audit firm and the period during the global financial crisis influence the reporting environment significantly, resulting in the recognition of higher loan loss provisions. During the additional analyses the results of the audit size appeared to be weaker and insignificant. This is probably due to the smaller sample size of the additional analyses.
27 The findings of this study are in accordance with the prior studies of Kim et al. (2003), Ferguson et al. (2003) and Stokes and Webster (2010), who also find evidence for the relationship between more conservative appliance of accounting standards and the variables audit firm size and the global financial crisis. The results deviate from the studies of La Porta et al. (1996) and Leuz et al. (2003), since I do not find the level of legal enforcement to significantly influence the amount of loan loss provisions reported. The reason for this might be the difference in sampled countries between our studies. La Porta et al. (1996) and Leuz et al. (2003) studied a sample of countries around the world, where this paper focused on six different European countries with the same accounting standards. However, the overall conclusion is that the stringency of the reporting environment influences a manager’s use of discretion in reporting loan loss provisions within the banking industry. This reduction of discretion will likely improve the reliability and comparability of the financial statements.
Besides the contributions to the literature in studying a unique period during and after the global financial crisis with focus on a less examined industry, the study also has several limitations. The first limitation is the distribution of financial institutions within the sample group which are audited by big four audit firms and non-big four audit firms. Only 18.63% of the sample is audited by a non-big four audit firm, which are 24 firms of the total 122 firms. This might have an impact on the empirical results, which can also be seen in the additional analyses where the sample group was significantly changed, resulting in a weakened result for the auditor size variable. The second limitation is the two-sized impact of the financial crisis. The financial crisis can either impact the amount of loan loss provisions through the lower economic performance of loan portfolios overall or through the stringency of the reporting environment (regulators become more cautious). The model is controlled by performance indicators such as the leverage ratio and the return on equity. However, future research might also want to control for macroeconomic changes. The last limitation is the scaling of all variables by the total assets instead of total outstanding credit. The variables were scaled in this way due to the lack of available data on the latter. This might have resulted in differences, especially for financial institutions that are also active in other markets than credit facility. For these firms the differences between total assets and total outstanding credit might be high. The study performed in this paper is focused on the IAS 39 standard, which still uses the ‘incurred loss model’ for reporting loan loss provisions. Future research in the banking industry will likely focus on the new ‘expected loss’ impairment model of IFRS 9, effective on 1 January 2018 (IFRS Foundation, 2014). It will be interesting to study the impact of this new model on the manager’s use of discretion. Future research can also extend this study by
28 taking into account the specific requirements of IAS 39 (or IFRS 9). In this way it is possible to link the effect of the reporting environment to the appliance of a specific accounting standard. This information can be relevant for standard setters.
29 References
Ahmed, A. S., Takeda, C., & Thomas, S. (1999). Bank loan loss provisions: a reexamination of capital management, earnings management and signaling effects. Journal of
Accounting and Economics, 28(1), 1-25.
Ball, R., Kothari, S. P., & Robin, A. (2000). The effect of international institutional factors on properties of accounting earnings. Journal of accounting and economics, 29(1), 1-51. Ball, R., Robin, A., & Wu, J. S. (2003). Incentives versus standards: properties of accounting
income in four East Asian countries. Journal of accounting and economics, 36(1), 235-270.
Ball, R., Robin, A., & Sadka, G. (2008). Is financial reporting shaped by equity markets or by debt markets? An international study of timeliness and conservatism. Review of
Accounting Studies, 13(2-3), 168-205.
Barth, M. E., Landsman, W. R., Lang, M., & Williams, C. (2012). Are IFRS-based and US GAAP-based accounting amounts comparable?. Journal of Accounting and
Economics, 54(1), 68-93.
Bartram, S. M., & Bodnar, G. M. (2009). No place to hide: The global crisis in equity markets in 2008/2009. Journal of International Money and Finance, 28(8), 1246-1292.
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, 231-261.
Beatty, A. L., 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), 339-383.
Beaver, W. H., & Engel, E. E. (1996). Discretionary behavior with respect to allowances for loan losses and the behavior of security prices. Journal of Accounting and Economics,
22(1), 177-206.
Beaver, W. H., & Ryan, S. G. (2005). Conditional and unconditional conservatism: Concepts and modeling. Review of Accounting Studies, 10(2-3), 269-309.
BNP Paribas (2007, August 9). BNP Paribas Investment Partners temporaly
suspends the calculation of the Net Asset Value of the following funds: Parvest Dynamic ABS, BNP Paribas ABS Euribor and BNP Paribas ABS Eonia. Retrieved from http://www.bnpparibas.com
Boone, J. P., & Raman, K. K. (2007). Does implementation guidance affect opportunistic reporting and value relevance of earnings?. Journal of Accounting and Public Policy,
26(2), 160-192.
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, 263-291.
DeAngelo, L. E. (1981). Auditor size and audit quality. Journal of accounting and economics,
3(3), 183-199.
Durkin, P., (2009). ASIC warns directors on valuations, solvency. The Australian Financial
Review, 29 June, 1.
Ferguson, A., Francis, J. R., & Stokes, D. J. (2003). The effects of firm-wide and office-level industry expertise on audit pricing. The accounting review, 78(2), 429-448.
Fonseca, A. R., & Gonzalez, F. (2008). Cross-country determinants of bank income
smoothing by managing loan-loss provisions. Journal of Banking & Finance, 32(2), 217-228.
