Academia.eduAcademia.edu

Figure 1 – uploaded by Peter Easton

See full PDFdownloadDownload figure
* ee Denote significantly different from 0 at the 10, 5, and 1 percent levels, respectively (two-tailed). t-tests (Wilcoxon signed-rank tests) are used for assessing means (medians). See Appendix A for variable descriptions. Short Window Returns around OTTI Warnings TABLE 1 Panel B: Cross-Sectional Determinants of the Market Reaction to OTTI Warnings RET_WARN

Table 1 * ee Denote significantly different from 0 at the 10, 5, and 1 percent levels, respectively (two-tailed). t-tests (Wilcoxon signed-rank tests) are used for assessing means (medians). See Appendix A for variable descriptions. Short Window Returns around OTTI Warnings TABLE 1 Panel B: Cross-Sectional Determinants of the Market Reaction to OTTI Warnings RET_WARN

Related Figures (10)

Panel A: Descriptive Statistics for Return Windows to have statistically significant coefficient estimates in both regressions. Investors’ ability to anticipate is not surprising because bad debt expense and earnings tend to be serially correlated. However, investors do not completely anticipate the information in earnings announcements, as evidenced by the statistically significantly positive estimate of the coefficient on RET_EARN,,, in both regressions. The 10-Q/10-K filing variables (RET FILING) tend not to have statistically significant coefficient estimates, which is not surprising given that earnings and bad debt expense would have already been disclosed in earnings announcements. Descriptive Statistics and Correlations TABLE 2
Panel A: Descriptive Statistics for Return Windows to have statistically significant coefficient estimates in both regressions. Investors’ ability to anticipate is not surprising because bad debt expense and earnings tend to be serially correlated. However, investors do not completely anticipate the information in earnings announcements, as evidenced by the statistically significantly positive estimate of the coefficient on RET_EARN,,, in both regressions. The 10-Q/10-K filing variables (RET FILING) tend not to have statistically significant coefficient estimates, which is not surprising given that earnings and bad debt expense would have already been disclosed in earnings announcements. Descriptive Statistics and Correlations TABLE 2
* ** *** Denote significantly different from 0 at the 10, 5, and 1 percent levels, respectively (two-tailed). Bold denotes significantly different from 0 at the 5 percent level or higher (two-tailed). t-tests (Wilcoxon signed-rank tests) are used for assessing means (medians). See Appendix A for variable descriptions.
* ** *** Denote significantly different from 0 at the 10, 5, and 1 percent levels, respectively (two-tailed). Bold denotes significantly different from 0 at the 5 percent level or higher (two-tailed). t-tests (Wilcoxon signed-rank tests) are used for assessing means (medians). See Appendix A for variable descriptions.
Panel A: Full Sample and Full Sample Excluding Q3 of 2008
Panel A: Full Sample and Full Sample Excluding Q3 of 2008
RET_4MONTH * wk Denote significantly different from 0 at the 10, 5, and 1 percent levels, respectively (one-tailed if sign is in the predicted direction, two-tailed otherwise). The model is estimated using ordinary least squares. Robust standard errors clustered by bank are presented in parentheses below coefficient estimates. See Appendix A for variable descriptions.
RET_4MONTH * wk Denote significantly different from 0 at the 10, 5, and 1 percent levels, respectively (one-tailed if sign is in the predicted direction, two-tailed otherwise). The model is estimated using ordinary least squares. Robust standard errors clustered by bank are presented in parentheses below coefficient estimates. See Appendix A for variable descriptions.
Regression Testing Pricing of CREDIT and NON-CREDIT OTTIs TABLE 5 Given the role of earnings in the pricing of unbifurcated OTTIs, we next examine the role of earnings in the pricing of each OTTI component. The fourth column includes earnings tercile (E) as an explanatory variable as well as interactions between EF and the two OTTI components. The estimated coefficient on CREDIT * E (NON_CREDIT * E) is 3.153 (1.451) and is statistically significant at the p < 0.01 level (not statistically significant). That is, strong earnings performance is associated with more negative pricing of the CREDIT component, whereas the role of earnings performance in the pricing of the NON CREDIT component is unclear. The estimated coefficients on the two interactions are not statistically significantly different. Results (untabulated) are similar when interacting the OTTI components with E_TOP and MBE. Overall, there is limited support for the notion that the sensitivity of OTTI pricing to earnings is stronger in the CREDIT component.
Regression Testing Pricing of CREDIT and NON-CREDIT OTTIs TABLE 5 Given the role of earnings in the pricing of unbifurcated OTTIs, we next examine the role of earnings in the pricing of each OTTI component. The fourth column includes earnings tercile (E) as an explanatory variable as well as interactions between EF and the two OTTI components. The estimated coefficient on CREDIT * E (NON_CREDIT * E) is 3.153 (1.451) and is statistically significant at the p < 0.01 level (not statistically significant). That is, strong earnings performance is associated with more negative pricing of the CREDIT component, whereas the role of earnings performance in the pricing of the NON CREDIT component is unclear. The estimated coefficients on the two interactions are not statistically significantly different. Results (untabulated) are similar when interacting the OTTI components with E_TOP and MBE. Overall, there is limited support for the notion that the sensitivity of OTTI pricing to earnings is stronger in the CREDIT component.
Related topics:
BusinessEconomicsAccountingFair Value
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved