It did not lead to reduction in the rate of utilization of the limits by the big firms (column (5)): the ratio of total turnover (the sum of all debts incurred during the year) to credit limit is not associated with the interaction BIG ∗ P OST . The additional credit limit thus resulted in an increase in bank credit utilization by the firms.
[TABLE 2 ABOUT HERE] Table 2 presents the impact of this increase in credit on sales and profits. The coefficient of the interaction BIG∗P OST in the sales equation, in the sample where the limit was increased, is 0.19, with a standard error of 0.11 (column (1)). By contrast, in the sample where there was no increase in limit, the interaction BIG ∗ P OST is close to zero (0.007) and insignificant (column (1), line 2), which suggests that the sales result is not driven by a failure of the identification assumption. The coefficient of the interaction BIG ∗ P OST is 0.27 in the credit regression, and 0.19 in the sales regression: thus, sales increased almost as fast as loans in response to the reform. This is an indication that there was little or no substitution of bank credit for non-bank credit as a result of the reform, and that firms are credit constrained.
Additional evidence is provided in column (2). We restrict the sample to firms which have a positive amount of borrowing from the market both before and after the reform, and thus have not completely substituted bank borrowing for market borrowing. In this sample as well, we obtain a positive and significant effect of the interaction BIG ∗ P OST , indicating that these firms must be credit constrained.
In column (3), we present the effect of the reform on profit. Because our dependent variable is the logarithm of profit, we can only estimate the impact on firms whose profits were positive. The effect is even bigger than that of sales: 0.54, with a standard error of 0.28. Here again, we see no effect of the interaction BIG ∗ P OST in the sample without a change in limit (line 2), which lends support to our identification assumption.
The large effect on profit is not sufficient to establish the presence of credit constraints: even unconstrained firms should see profits increase when they gain access to subsidized credit, because they would substitute cheaper capital for more expensive capital. However, if firms were not expanding, we should not expect to see sales (column (1)) or costs (not reported) expand as well.
The instrumental variable (IV) estimate of the effect of loans on sales and profit implied by