盈餘管理對銀行債務資金成本之影響－－信用評等觀點

The Effect of Earnings Management on the Bank Cost of Debt-- The Credit Rating Approach

本文以2002 至2007 年全球的商業銀行為對象，透過信評機構對盈餘管理的評價，探討盈餘管理是否影響銀行債務資金成本，使用跨國資料可進一步探討訊息不對稱的程度是否影響盈餘管理與評等的關係。實證結果指出，當信評機構發現銀行從事盈餘管理時，會給予較差之評等等級，使得債務資金成本增加。當以盈餘平滑為盈餘管理的代理變數時，發現在高所得國家或財務報表係由前五大會計師簽證之銀行，其盈餘平滑對信用評等沒有影響，而一國的債權人保護程度較佳，也可以有效地降低訊息不對稱的效果。然而，當一銀行是位於較低度發展地區，由於存在較嚴重的訊息不對稱的現象，使加重盈餘平滑對信用評等的不利效果。當以盈餘操縱當作盈餘管理的代理變數時，發現中所得國家顯著地減緩了可裁量的呆帳費用對信用評等的不利影響，可解釋為較低度發展國家藉以傳遞有利的內部訊息給外部投資人，而一國的債權人保護程度較高也可以減緩盈餘操縱的不利效果。

The purpose of this paper tries to understand how rating agencies react to the behaviors of banks’ earnings management. We use bank data from 85 countries during 2002-2007. For credit ratings, we use S&P’s long-term issuer ratings and convert into 6 numerical ratings. The earnings management indicators include earnings smoothing and earnings manipulation. We use the correlation coefficient between the changes in loan loss provisions and earnings before provisions to measure earnings smoothing and discretionary loan loss provisions to measure earnings manipulation. The use of cross-country data further enables researching how asymmetric information influences the impact of earnings management at the country level. We posit the asymmetric information is systematically related to a country’s governance. That is, we hypothesize when a bank is in a high-income country, in an environment characterized by strong creditor protection, and when its financial statements are audited by Big-Five auditors can mitigate the information asymmetry effect. In such an environment, even when a bank manages its earnings, rating agencies view reported earnings as trustworthy, reducing the adverse impact of earnings management on ratings. Alternatively, in an environment with poor governance, rating agencies realize there is severe information asymmetry and do not trust the reported earnings, aggravating the adverse impact of earnings management on ratings. The study of firms’ earnings management has spurred a great deal of attention. Previous literatures focused on whether firms manage earnings to affect the cost of capital, the effect is chiefly on the cost of equity. For example, Sloan (1996) and Collins and Hribar (2000) have documented that firms with large accruals have subsequent negative abnormal returns, implying the price of firms with high accruals is overvalued. The explanation is that investors can not understand the relationship between accruals and future earnings and cash flows. Besides, there is compelling evidence suggesting that firms manage earnings around initial and seasoned public equity offerings, implying firms want to get more capital by increasing abnormal accrual items, but is usually associated with worse subsequent stock price performance (Teoh, Wong and Rao, 1998; Teoh, Welch and Wong, 1998a, 1998b; Rangan, 1998). This study complements the literature by studying the impacts of earnings management on credit rating and thus on the cost of debt. Because debt is a major source of capital for firms, it is more important to understand the factors that affect the cost of debt. Thus, to the extent that earnings management is an important determinant of credit ratings, it can have a significant effect on firms’ external financing costs. This study mitigates error in measuring managerial discretion by focusing on a single accrual and a single industry i.e., commercial bank. Focusing on a single accrual facilitates a better separation into its nondiscretionary and discretionary components. Previous studies show there is indeed earnings management phenomenon of banks. For examples, Shen and Chih (2005) find that banks in more than two-thirds of the 48 countries sampled are found to have avoided earnings losses and earnings decreases. Other studies also find banks use loan loss provisions to smooth earnings (Greenawalt and Sinkey, 1988; Ma, 1988; Hasan and Hunter, 1994). Another strand of studies find a positive relation between equity return and loan loss provisions (Beaver et al., 1989; Elliott, Hanna and Shaw, 1991; Griffin and Wallach, 1991; Johnson, 1989), representing equity market regards unexpected discretionary loan loss provisions as banks’ positive signal about future earnings and cash flow, that is, banks use accruals to signal firms’ better performance in the future. The question how rating agencies evaluate banks earnings management is important because if earnings management can affect ratings, it can also affect the cost of debt. Our conclusions show the better governance effect gains support for most conditions, though results differ slightly when the earnings management is proxied by either earnings smoothing or earnings manipulation. First, without considering governance variables, this study finds both proxies for earnings management, that is, earnings smoothing and earnings manipulation, display “adverse” effects on credit ratings. This suggests rating agencies downgrade the credit ratings when earnings management exists, implying earnings management increases the borrowing costs. Next, when governance variables are added and earnings smoothing is used as a proxy for earnings management, the better governance effect is strongly supported. When banks are in high-income countries, in countries with high creditor protection and audited by Big-Five auditors, information asymmetries associated with earnings management are clearly reduced because the coefficients of interaction terms between earnings management and these three governance variables are significantly positive. Additionally, earnings smoothing aggravates its adverse effect on credit ratings that occurs for banks located in middle-income and Eastern Europe and East Asian countries because they are poor credit quality regions. Therefore, the better governance effect gains support. The result, however, does not support our better governance effect for banks from the Latin region because the earnings smoothing has a positive effect on credit ratings. Finally, when governance variables are added and earnings manipulation is used as a proxy for earnings management, the adverse effect of earnings management on ratings changes little in high-income countries, but is significantly mitigated for those from middle-income countries. Because these results differ from those using earnings smoothing as a proxy for earnings management, we interpret the results as follows. In high-income countries, raters interpret increases in discretionary loan loss provisions as suggesting increases in bad loans; while in middle-income countries, the increase in discretionary loan loss provisions is used primarily to signal favorable private information to investors. Thus, an accrual, that is, discretionary loan loss provisions, upgrades the ratings in middle-income countries. This result demonstrates rating agencies assign different weights to earnings management indicators depending on country governance. Further, the negative effect of earnings manipulation on credit ratings is mitigated in countries with better creditor protection. The Big- Five auditors, however, do not affect the influences of earnings manipulation on ratings.