Earnings management and bank characteristics

Table 2.4 presents regression results that examine whether the earnings management level is associated with bank characteristics: size (TA), performance (ROA), and ownership concentration (OWN). In Columns 1-7 of Table 2.4, we present results using EM1-EM4, earnings discretion (EMDSC), income smoothing (EMSMTH), and aggregated earnings management indexes (EMAGG), respectively.

Table 2.4 Effects of bank characteristics on cross-country differences of bank earnings management

EM1 Discretion to avoid

small losses

EM2 Discretion on Reported LLP

EM3 Smooth of OPI

relative to unmanaged OPI

EM4 Smooth of OPI through changes in

discretionary LLP

EMDSC

Average Percentage Rank of EM1 &

EM2

EMSMTH

Average Percentage Rank of EM3 &

EM4

EMAGG

Average of EMDSC

and EMSMTH

Coeff T-stat Coeff T-stat Coeff T-stat Coeff T-stat Coeff T-stat Coeff T-stat Coeff T-stat Constant 0.073^a 2.85 0.462^a 2.76 -1.039^a -8.79 0.269^a 3.60 66.094^a 6.01 52.600^a 4.94 59.347^a 7.83

OWN -0.004 -1.24 0.018 0.76 -0.011 -0.65 -0.006 -0.56 0.066 0.04 -1.465 -1.00 -0.699 -0.65 ROA -0.012^a -3.14 -0.111^a -3.35 0.051^b 2.28 0.041^a 2.95 -11.609^a -4.83 7.583^a 3.55 -2.013^c -1.69 TA -0.002^b -2.49 0.006 0.90 -0.005 -0.86 -0.004 -1.22 -0.916 -1.29 -0.708^c -1.90 -0.812^c -1.94 R2 0.1369 0.1183 0.0549 0.0793 0.2275 0.1125 0.0521 AdjR2 0.1057 0.0864 0.0208 0.0460 0.1996 0.0804 0.0178 Notes: The data sources and definitions of the variables are provided in Appendix A. This table reports results from regressions of earnings management measures on bank characteristics. All explanatory variables are median values in each country. The Model is:

ε α

α α

α + + + +

= ROA TA OWN

EM_{k 0 1 2 3} ,

where EM_k =

{

}

The models are estimated by linear OLS regressions with the White robust standard errors. The marks a, b, and c indicate significance at the 1%, 5%, and 10%

levels, respectively.

35 The results show that ownership concentration (OWN) has moderate but insignificant negative influence on most our earnings management measures. The insignificant results may be due to that most banks are highly leveraged which make the ownership concentration less varied across country, and thus we only get moderate effects for this factor. However, the negative coefficients still indicate that, in countries where banks have more concentrated ownership, insiders are less likely to exercise their private control benefits through earnings management. It is consistent with the argument by Beatty et al. (2002) that, with concentrated ownership structure, investors are more likely to participate in the management, directions, and operations of the banks. It is also consistent with the argument by Black (1992) that, with dispersed ownership, investors are less likely to monitor since the monitoring benefit is very limited.

Results for the effects of profitability (ROA) and bank size (TA) on banks’ earnings management are consistent with our predictions. In countries where banks are more profitable, the level of earnings discretion is less pronounced, but the level of income smoothing is more prevalent. Take the practice of earnings management to avoid small losses as an example. It is reasonable that bank managers do not need to engage in this type of earnings management when they are profitable. On the other hand, it is plausible that profitable banks may engage in more earnings smoothing for tax or other purposes. Our results also show that the levels of earnings management are generally less pronounced in countries where the median size of banks is larger. Since these bank characteristics are significantly related to different types of bank earnings management, we include them as control variables in the subsequent regressions that examine effects of different institutions.

36 6.2 The monitoring role of law protection and bank supervision/regulation practice

We analyze the effects of governance mechanisms on bank earnings management activities in this subsection. The main argument here is that earnings management can be curbed when the law and bank supervision practices are designed to facilitate protection on investor rights and to improve private monitoring on banks. Taken as a whole, our findings generally support the above argument.

We apply law origin (LAW) to proxy for a country’s overall legal protection on investor rights and the results are reported in Panel A, Table 2.5.¹⁷ Consistent with our predictions, the coefficients on law origin are all negative and are significant for regressions on EM1, EM4, EMDSC, and EMAGG. The results imply that, in a country with common-law origin, investor rights are better protected and thus earnings management by banks is less pronounced. The results are consistent with the findings by Ball et al., (2000), Burgstahler et al. (2006), and Leuz et al. (2003) that financial reporting quality measured by either earnings management or the timeliness of loss recognition is substantially higher in countries with stronger protection of investors or in common-law countries.

Panels B of Table 2.5 presents the effects of bank regulatory supervision practices on earnings management, with control for law origin that proxy for protection of investor rights.¹⁸ The direct supervision power index (SUPPWR) and the overall capital stringency index (CAR) are applied to investigate the effects of direct government supervision/regulation. The private monitoring index (PRIIDX) is used to explore the effects of policies that improve private-sector monitoring. For direct government supervision, the results show that most coefficients of

17 Previous studies suggest that law origin (LAW) proxies for a country’s investor protection and law enforcement very well (La Porta et al., 1998; Leuz et al., 2003; Burgstahler et al., 2006).

18 The regression results are from 61 sample countries because 26 of our sample countries lack some data on supervision practice. The 26 countries are: Algeria, Bangladesh, China, Colombia, Dominican Republic, Ecuador, Egypt, Finland, France, Honduras, Hong Kong, India, Iran, Ireland, Israel, Kenya, Kuwait, Mexico, Paraguay, Qatar, Sri Lanka, Sweden, Taiwan, Tanzania, Thailand, and Vietnam.

37 capital regulation stringency (CAR) and direct supervision power (SUPPWR) are insignificant.

Although the results for EM2 suggest that banks in countries with stringent capital requirement (CAR) are less likely to do earnings discretion, the results for EM3 and EM4 suggest that banks in countries with stringent capital requirement (CAR) and stronger direct supervision power (SUPPWR) are more likely to execute earnings smoothing. Overall, our results for direct government supervision/regulation can not provide statistical evidence on whether direct government supervision limits or encourage bank behavior of earnings management. As for the supervision policies to improve private-sector monitoring, our results show that most coefficients of private monitoring index (PRIIDX) are negative and are significant for regressions on EM3, EM4, EMSMTH, and EMAGG. The results support that policies enhancing private-sector monitoring effectively limit earnings management activities and that the effects are stronger for income smoothing activities than earnings discretion activities.

Table 2.5 Effects of institutions on cross-country differences of bank earnings management

Panel A: Effects of Law Origin (# of countries =87) EM1 Panel B: Effects of law origin and bank regulation/ supervision (# of countries =61)

EM1

Panel C: Effects of legal protection (LAW, CPIX and ACCT) and sub-index of PRIIDX (LBKRATE, BKDISCL, and DEPOINSUR) (# of countries =30) Notes: The data sources and definitions of the variables are provided in Appendix A. This table reports results from regressions of earnings management measures on bank characteristics, investor protection and bank regulation variables to show the effects of monitoring mechanisms on the behaviors of earnings management by banks. Models for each Panel are:

Panel A: EM_k =α₀+α₁ROA+α₂TA+α₃OWN+α₄LAW+ε,

The models are estimated by linear OLS regressions with the White robust standard errors. The marks a, b, and c indicate significance at the 1%, 5%, and 10%

levels, respectively.

40 6.3 Underlying factors for monitoring effects of law protection and private-sector monitoring

In this section, we examine what underlying forces for law protection and private-sector monitoring have stronger effects on limiting bank earnings management. For law protection, we further include two legal protection variables, the corruption perception index (CPIX) and the rating on accounting standards (ACCT).¹⁹ We also substitute private-sector monitoring index (PRIIDX) with its sub-indexes, international rating on large banks (LBKRATE), bank accounting disclosure (BKDISCL), and official depositor insurance (DEPOINSUR).²⁰

We run the regressions with the 30 countries,²¹ which have all of the above institutional data, and report the results in Panel C of Table 2.5. The coefficients of perceived corruption index (CPIX) show that lower perception of government corruption (higher CPIX) is significantly associated with lower level of earnings management, EM2, EM4, EMDSC and EMAGG. It suggests that, in countries with lower level of government corruption, bank managers are less likely to bribe the government officials for their dishonest behavior and thus outsiders’ rights are better protected. Consistent with the effect of law origin, the corruption index (CPIX) also has stronger effects on limiting bank behavior of earnings discretion than income smoothing.

However, the results do not support that higher accounting standards (ACCT) have effects on curbing banks’ earnings management activities.

The private-sector monitoring forces, LBKRATE, BKDISCL and DEPOINSUR, indicate effects of stronger monitoring by international rating agencies, more disclosure requirement

19 We do not include anti-director rights (ANTI). It might be inappropriate to include this variable as many of our sample banks are privately-held rather than publicly-traded and anti-director rights (ANTI) measures the shareholders’ voting rights against the management, which are usually applied in public-traded firms.

20 We do not report results with the variable external auditor (AUDIT), because, among the countries examined, only in Italy banks are not required to have an external auditor and because the results remain whether we include external auditor (AUDIT) or not.

21 The 30 countries are: Argentina, Australia, Austria, Belgium, Brazil, Canada, Chile, Denmark, Germany, Greece, Italy, Japan, Korea, Malaysia, Netherlands, New Zealand, Nigeria, Norway, Peru, Philippines, Portugal, Singapore, Spain, Switzerland, Turkey, United Arab Emirates, United Kingdom, United States, and Venezuela.

41 and higher litigation risk to bank managers for reporting misleading information, and stronger depositor incentives to monitor banks. Our results show that almost all private-monitoring forces (LBKRATE, BKDISCL and DEPOINSUR) are negatively correlated with the earnings management measures. However, only some coefficients of LBKRATE and DEPOINSUR have significance level with α smaller than 10% or 5%, suggesting that, when international rating agencies and depositors are better encouraged to monitoring banks, the level of bank earnings management is lower. Further, our results suggest while monitoring by international rating agencies has stronger effects to limit earnings discretion, monitoring by depositors constrains both earnings discretion and income smoothing. Lastly, we are unable to conclude whether mandating better reporting quality and raising litigation risks of bank managers have any effects to curb earnings management activities as these coefficients are all insignificant.

7 Conclusions

In this study, we extend previous studies to show that not only the well-established legal protection of investors, but also bank regulatory supervision can have effects on explaining bank earnings management activities around the world. An international comparison in the banking industry provides us with the opportunity to gain insight into the incentives for bank earnings management, because it offers more variation in investor protection mechanisms and bank supervision frameworks. Using earnings management measures modified from the ones developed by Leuz et al. (2003), we examine the variation of bank earnings management across countries and explore the institutional factors that affect the reporting incentives of banks internationally.

We document international differences in the degree of earnings management using a unique database consisting of 39,723 bank-year data across 87 countries. We then investigate

42 whether institutional factors help explain earnings management activities in an international setting. Consistent with the corporate governance literature, we show that better legal protection of investors lowers the extent to which banks engage in earnings management. Our findings also show that, in addition to the legal protection factors, bank regulation/supervision policies are important determinants in limiting banks’ earnings management activities. In general, banks are less likely to use earnings management to conceal their performance from outsiders in countries where higher percentage of large banks are rated by international rating agencies and where depositors are less covered by official deposit insurance. This better reporting quality under these systems can be attributed to stronger private-sector monitoring on banks from these mechanisms. However, we do not find evidence to support effects of stronger direct government supervision to improve bank financial reporting quality.

To sum up, our study represents an extension of the recent work by Leuz et al. (2003) on earnings management by industrial firms around the world to the banking industry. Our study also represents an extension of a recent study by Barth et al. (2004) on the roles of bank regulation and supervision on the development and efficiency of the banking sector to their roles on the reduction of earnings management. Our evidence confirms that bank earnings management is more pervasive in countries with fewer mechanisms on enhancing legal protection of investors. In addition to legal protection, mechanisms that may improve private-sector monitoring, such as monitoring by international rating agencies and depositors, also help explain the quality of earnings across banks around the world.

Our study contributes to the literature on the monitoring role of bank supervision mechanisms in limiting earnings management activities by banks. To our best knowledge, we are among the first to examine this issue. We extend Barth et al.’s (2004) finding to the role of bank supervisory practices that promote private-sector monitoring on improving financial

43 reporting quality. We show that banks in a regulatory environment that promotes private-sector monitoring engage in less earnings management activities. Our research also contributes to the literature regarding the role of legal protection of investors on bank financial reporting quality. Using earnings management as an indicator for financial reporting quality by banks, we find that investor protection plays an important role on curbing earnings management activities. Consistent with Burgstahler et al. (2006) and Ball et al. (2000), our results support that banks from countries with stronger legal protection of investors usually have better financial reporting quality.

44

Chapter III

The Influence of Supervision and Regulation on the Conservatism of Financial Reporting by Banks

1. Introduction

In this chapter, we investigate how the conservatism of financial reporting by banks is affected by the supervision and regulation policies of banks. We focus on the financial reporting of earnings changes and loan losses. We argue that international differences in the conservatism of financial reporting are driven by institutional factors that shape the corporate governance systems of banks around the world. By reducing the information asymmetry between fund providers and banks, conservative accounting disclosure enhances the quality of accounting information. Accordingly, we argue that conservative financial reporting should be observed in countries where bank supervision and regulation policies are likely to enhance the corporate governance mechanisms of banks. To test our argument, we consider two governance mechanisms used by bank regulators: (1) direct government supervision and regulation policies and (2) indirect government policies that encourage private-sector monitoring.

Previous studies have documented that conservative financial reporting is positively related to the quality of corporate governance through curtailing the information asymmetry and improving the capital allocation efficiency of investments.²² Evidence also shows that conservatism in financial reporting varies across countries and is associated with institutional factors. For example, Ball and Shivakumar (2005) show that public firms rely more on

22 Watts (2003a, 2003b) views conservatism as an efficient contracting mechanism, because it reduces information asymmetry problems through a higher degree of verification of gain recognition than of loss recognition. Bushman et al. (2005) show that firms in countries with more conservative financial reporting respond to and withdraw capital from losing projects relatively faster than do firms in countries with less conservative reporting.

45 shareholder governance mechanisms than do their private counterparts and thus financial reporting is more conservative among public firms than among private firms. It has also been shown that firms in countries with common-law origins, with less reliance on inside networks for communication, or with less political intervention are likely to report more conservative financial statements than are their counterparts.²³ Bushman and Piotroski (2006) further examine how securities laws and political forces affect conservative reporting of income among industrial firms across 38 countries. Their evidence shows that stringent regulation and intervention by governments may influence the incentives for managers to be conservative in financial reporting.

Although factors that affect the incentives to be conservative in financial reporting among industrial firms may apply to banks, two special features should be considered in examining the conservatism of financial reporting by banks: (1) information opaqueness and (2) government intervention. Caprio and Levine (2002) and Levine (2004) argue that information opaqueness and heavy intervention by governments in the banking industry complicate corporate governance issues for banks and that these two special features weaken many traditional corporate governance mechanisms for banks.

The information opaqueness among banks leads to greater information asymmetry between insiders and outsiders and thus makes it difficult for outsiders to monitor banks. We argue that outsiders can reduce the problems of information asymmetry if they demand conservative financial reporting. Such a demand for conservative financial reporting can reduce upwardly biased reporting on income and downwardly biased reporting on loan loss provisions. Although such reporting does not eliminate the information opaqueness problem completely, conservative financial reporting at least provides more informative accounting

23 See, for example, Ball et al. (2000), Ball et al. (2003).

46 information and hence complements the role of outside monitoring.

Governments intervene in the activities of banks mainly through their regulatory and supervisory policies.²⁴ Recent cross-country studies document the dangers of powerful regulatory supervisors and show the importance of private-sector monitoring to the corporate governance of the banking industry.²⁵ Evidence also indicates that both government supervision/regulation policies and private-sector monitoring mechanisms contribute to assessments of bank conditions and to executing adequate disciplinary actions against banks.²⁶ Given this evidence, we ask here how the two governance mechanisms complement each other and how authorities supervise banks and encourage private-sector monitoring as they pursue better governance systems for banks.

There are many ways to understand how corporate governance mechanisms work. In this study, we examine how existing corporate governance mechanisms affect incentives for banks to be conservative in their financial reporting. Financial reporting is an important source of information in assessing a bank’s financial conditions. As such, for corporate governance mechanisms to function well, bank supervisors and private-sector monitors will demand conservative financial reporting to ensure that the banks provide informative accounting information. We expect that corporate governance mechanisms that function well are positively related to a bank’s incentives to be conservative in financial reporting. In other words, we test the effects of these mechanisms on bank corporate governance by examining how they influence a bank’s incentives to be conservative in financial reporting.

24 In the extreme, governments intervene in banks through ownership of banks. When a government owns a bank, the conflict of interest makes its role as a monitor ineffective. Evidence also shows that the government ownership of banks has a negative impact on the development, performance and stability of the banking system (Barth et al. 2004; La Porta et al. 2002).

25 See, for example, Barth et al. (2004) and Beck et al. (2005).

26 See, for example, Berger and Davies (1998), DeYoung et al. (2001), Flannery (1998), and Gunther and Moore (2003).

47 Using a unique and large data set of information from 1,248 publicly traded and 6,481 privately held banks across 48 countries during 1997 to 2004, we document two kinds of conservatism in financial reporting. First, banks are more conservative in reporting earnings changes. Second, banks incorporate more loan losses into their financial reports when their operating cash flows decrease or when the amount of problem loans increases. We also find that banks charge off more problem loans when their loan loss provisions increase. In general, our evidence indicates that banks are conservative in their financial reporting. In addition, public banks appear to be more conservative than are private banks in financial reporting.

When we compare reporting conservatism across countries, our results show that the

When we compare reporting conservatism across countries, our results show that the