As we have argued, the increase in the ratio of bank loans to total assets can result from a decline in total assets, not just froman increase in bank loans, so we reestimate the regression using the annual growth rate of bank loans.
The F-test that all i= 0 is accepted, so we estimate using OLS regression.
We specify the annual differences in X as control variables. The estimated results for the growth rates of bank loans are shown in columns (4)–(6) of Table 3.
As shown, the estimated coefficient for Guarantee × year dummy 2009 is not statistically significant (column 4).
However, the estimated coefficients for the interactions between Guarantee ×year dummy 2009 and ln(the amount of bank loans) and the bank loans ratio in year t − 1 are positive and statistically significant at the 1% level,
which indicates similar results to when we specified the ratio of bank loans as the dependent variable.