I am confused by the paper of Kan et al. (2013 - "Pricing Model Performance and the Two-Pass Cross-Sectional Regression Methodology") which proposes the execution of hypothesis tests on R² of regressions to assess whether the coefficients are different from each other. Consequently the following setup is tested: $$H_0=R²_A = R²_B \\ H_1=R²_A != R²_B $$

The test is based on a p-value for each of the R² - this is where my confusion starts. As I understood hypothesis tests the p-value results from the lookup of the z-value from the normal distribution, whereas the z-value follows $$ z-value = \frac{x-\sigma}{\sqrt{n}} $$ For the implementation of such a test I do not have sufficient information or is it necessary to consider the standard deviation of included regression coefficients to make this calculation work?

I also thought about using the methodology to compare two mean-values, however, I still need a standard deviation of the tested models. Moreover, the paper by Kan describes the method only for a single cross section, not periodical cross-sections as described by Fama & MacBeth (1973). Do you know how to transform the calculation?

Many Thanks!


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.