PSYO 2770 Lecture Notes - Lecture 4: Logit, Probit, Fixed Effects Model
12 Oct 2017
School
Department
Course
Professor
Document Summary
See do file for q1 and addition info on other qs. In total, 157562 companies in the manufacturing industry. To eliminate z-score outliers, i used using chebyshev"s inequality; outliers are values that deviate from the mean by at least 3 standard deviations. The resulting data has a lower standard deviation and is normally distributed. As shown in the anova table above, flch_w has the largest coefficient. Which means firm-level cash-holding is most informative among the variables in predicting the default risk. Coefficients are the same for both heteroscedasticity adjusted regressions and the original but i see that there is higher standard deviations. This indicates that there is heteroscedasticity in the regression. With corrected standard errors, the conclusions about the relationships of the input variables with the risk of default do not change but because of larger standard deviation, the confidence interval is shorter.
