The conservatism of financial reporting by country

We estimate equations (3.1) to (3.3) by country and report the results in Tables 3.4 to 3.6, respectively. The following discussions focus on the regression coefficients that indicate conservative financial reporting. We also compare the results with the existing literature.

Although the degree of conservatism varies across different conservatism measures, listing status, and countries, the results are consistent with the findings of the existing studies.

First, we compare the results of CNSV_NI (α3) from public banks in Table 3.4 with the findings on public industrial firms by Bushman and Piotroski (2006). We find conservative reporting in earnings changes for public banks in many countries, although the results from Belgium, Hong Kong, Lebanon, Morocco and Spain are counter to conservative financial reporting. If we compare the results of countries examined in this study and those in Bushman and Piotroski’s (2006) study, our results are consistent with their findings except for Belgium, Hong Kong and Spain. Second, consistent with the findings by Ball and Shivakumar (2005) for industrial firms in the U.K. and with the findings by Nichols et al. (2005) for banks in the U.S., our results show that, in most countries with English law origins, public banks are more conservative in reporting earnings changes than are private banks.

69 Table 3.4 Conservative reporting of earnings changes by country

Results for Public Banks Results for Private Banks Country

70 Results for Public Banks Results for Private Banks

Country

Singapore 0.0048 0.2017 0.2702 0.10 41 0.0066 0.1308 -0.4650 0.03 34 (0.89) (1.36) (1.36) (0.57) (0.33) (-0.90)

Switzerland 0.0037 0.0635 -0.0855 0.00 74 -0.0026 -0.2184 -0.1725 0.10 1,128 (0.32) (0.76) (-0.72) (-1.55) (-1.03) (-0.67)

Venezuela -0.0021 0.1502 0.3135 0.04 67 0.0123 -0.151 0.1685 0.03 111 (-0.31) (0.52) (0.81) (0.58) (-0.90) (0.45)

Public bank sample Private bank sample

# of countries with significant negative α3 =20 # of countries with significant negative α3 =15

# of countries with significant positive α3 =4 # of countries with significant positive α3 =6 Notes: UAE stands for United Arab Emirates. The data sources and definitions of the variables are provided in Appendix B. The table reports the results from the following regression:

it

The models are estimated by linear OLS regressions with the White robust standard errors. The robustness t-statistics are in parentheses. The marks a, b, and c indicate significance at the 1%, 5%, and 10% levels, respectively.

71 The results on conservative reporting of loan loss provisions relative to changes in operating cash flows (CNSV_LLP_CF or β3) are presented in Table 3.5. The results resemble the results on earnings changes (CNSV_NI or α3) reported in Table 3.4. However, for private banks, the variation in conservatism across countries is larger for earnings changes than for loan losses. It appears that both public and private banks tend to be conservative in reporting loan loss provisions relative to changes in operating cash flows. Moreover, public banks are usually more conservative than are private banks.

Table 3.6 reports results on conservative reporting of loan loss provisions relative to changes in problem loans (CNSV_LLP_PL or γ2) and to net charge-offs (CNSV_LLP_COF or γ4) for 14 countries. The reason for the sudden drop in the number of countries examined is due to the fact that the information on problem loans and net charge-offs is usually reported in footnotes and is unavailable for many of the banks in the countries sampled. Again, in many countries, both public and private banks are conservative in reporting loan losses relative to problem loans and to net charge-offs. However, public banks are in general more conservative than are private banks. Our results from the U.S. show that public banks recognize more loan loss provisions at the time when the amount of problem loans increases and charge off more problem loans at the time when loan loss provisions increase. In contrast, private banks recognize less loan loss provisions at the time when the amount of problem loans increases.

The results support the findings by Nichols et al. (2005) that, in the U.S., public banks are more conservative in reporting loan losses than are private banks.

72 Table 3.5 Conservative reporting of loan loss provisions to changes in cash flows by country

Results for Public Banks Results for Private Banks

Country β1 β2 β3 R² Obs. β1 β2 β3 R² Obs.

73 Results for Public Banks Results for Private Banks

Country

Public bank sample Private bank sample

# of countries with significant negativeβ3 =21 # of countries with significant negativeβ₃ =18

# of countries with significant positiveβ3=1 # of countries with significant positiveβ3=1 Notes: UAE stands for United Arab Emirates. The data sources and definitions of the variables are provided in Appendix B. The table reports the results from the following regression:

it

The models are estimated by linear OLS regressions with the White robust standard errors. The robustness t-statistics are in parentheses. The marks a, b, and c indicate significance at the 1%, 5%, and 10% levels, respectively.

Table 3.6 Conservative reporting of loan loss provisions relative to changes in problem loans and to net charge-offs by country Results for Public Banks Results for Private Banks COUNTRY

Public bank sample Private bank sample

# of countries with significant positiveγ2 =10 # of countries with significant positiveγ2 =8

# of countries with significant negativeγ2=1 # of countries with significant negativeγ2=1

# of countries with significant positiveγ4 =8 # of countries with significant positiveγ4 =7

# of countries with significant negativeγ4=1 # of countries with significant negativeγ4=1 Notes: The data sources and definitions of the variables are provided in Appendix B. The table reports the results from the following regression:

it

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. .

75 In sum, our results on conservative financial reporting by banks are consistent with the results from the existing studies on industrial firms in the U.K. and banks in the U.S. However, the estimated regression coefficients that represent the degree of conservatism vary substantially across countries. In some countries, we even find evidence counter to conservative reporting, either for different listing status or for different measures of conservatism. It would be interesting to examine whether the variation in conservative financial reporting can be explained by the country-level institutional factors that may influence financial reporting incentives by banks, especially the factors that indicate different corporate governance mechanisms in these countries.