Incentives and delisting. Table 5 shows the marginal effects, calculated at the mean from regression (1), testing the influence of firm incentives and mandatory IFRS adoption on the delisting decisions of the sample companies. The results of all models are based on regressions with robust standard errors, clustered by entity, and adjusted for the non-linearity of the probit model.15 The models in the table differ by their definition of the IFRS indicator variable. Model_1 and Model_2 define this indicator variable as one in 2005 and zero otherwise. Model_3 and Model_4 define it as one for 2005 and all the years thereafter. Model_1 and Model_3 include only the variables of interest, while Model_2 and Model_4 include control variables as well. All four models include industry and exchange fixed effects. Consistent with my expectations, the coefficient of the IFRS indicator variable in Model_1 and Model_2 is positive and significant at the 5 percent level. This indicates that the probability of delisting increased by about 2% in 2005, after the introduction of the new standards. The sign of this indicator variable’s coefficient remains positive and significant at the 10 percent level in Model_3, but loses statistical significance in Model_4. For all four models, my Incentives variable has the expected negative sign and is significant at the 0.01 level. This result speaks of lower delisting probability of about 1.6% for higher incentive companies during the entire sample period. Consistent with H2, the interaction between the IFRS indicator variable and the incentive variable is 15 For a linear regression with interaction effects: ß1x1 + ß2x2 + ß12x1x2 + ßX = y, δ2[Ey|x1, x2, X]/ δx1 δx2 = δ2[Ey|x1, x2, X]/ δx1x2 = ß12. However, for a probit regression with an interaction effect y = Φ(ß1x1 + ß2x2 + ß12x1x2 + ßX), where Φ(.) is the cumulative normal distribution function, the interaction effect is given by the cross-derivative δ2Φ(.)/ δx1δx2 = ß12Φ’(.) + (ß1 + ß12x2)*(ß2 + ß12x1)* Φ”(.) and is not equal to δ2Φ(.)/ δx1x2 = ß12Φ’(.) computed by Stata (Ai and Xxxxxx 2003). statistically significant in Model_1 and Model_2 and has the expected negative sign, showing that the introduction of IFRS in 2005 in the four countries in my study increased the delisting probability for their companies, but this increase is more pronounced (by about 1%) for the lower-incentive firms. While the calculated marginal effects seem modest, they are economically significant, given that the basic delisting pro...
