In the last post we saw that Jiménez et al. (2012) showed that banks react to a monetary policy expansion by increasing credit supply, and that is particularly strong for ‘weaker’ banks. This is what it is known as the ‘lending channel’ of monetary policy. But something that has worried many economists is the fact that the increase in credit supply might go for riskier borrowers. In other words, what is known as the ‘risk-taking channel’ of monetary policy.
Jiménez et al. (2014) study this issue. They use again the Spanish credit register and estimate an econometric model with two stages: in the first stage, they estimate the probability of granting a loan (given that the firm has a loan application with the bank), while in the second stage they study the quantity of the loan. It is very important to study credit supply with this two-stage procedure (if one has loan applications data, obviously) because there is a sample selection bias if one only studies the quantity of the loan: one would only be observing firms for which the bank decided to give them a loan, which can be different from those that were not granted the loan.
They use fixed effects to exhaust the variation of the dependent variables so that they can identify the channel. In particular, the channel is a change in the composition of credit supply after a decrease in the policy rate: weaker banks increasing credit supply more to riskier firms. Crucially, this is a bank-firm-time channel: bank because of the differences in balance-sheet strength, firm because of the differences in riskiness of the borrowers, and time because of the monetary policy shocks. This is very important because there are other effects from decreasing the policy rate: we already know about the ‘lending channel’ (weaker banks expanding credit supply more), and there are potentially changes in the composition of credit demand (i.e., changes in the risk of firms demanding credit). The way they control for the first effect is by using bank-time fixed effects: in other words, controlling for the average probability of granting a loan and the credit given for each bank at each time. For the second effect, they rely, as in the previous paper, on firm-time fixed effects to control for credit demand.
Their results show that a decrease of 1 percentage point in the overnight rate is associated to a 8.2% higher probability of granting a loan to a risky firm and 16.1% more credit to that risky firm for weaker banks. This can be seen in the last column of Table II. Notice that when they use bank-time and firm-time fixed effects they cannot study other interesting results: for instance, does the average bank lend more to risky firms when the interest rate goes down? Since this is a firm-time issue, firm-time fixed effects capture it. Nevertheless, they report in the same table the same specification introducing fixed effects and interactions one step at the time. We can see that, in Column 4, the average bank tends to grant fewer loans to risky firms when the interest rate decreases, but they give more credit to risky firms if they are granted the loan as compared to safer firms. Taking both stages into account, the total impact of a lower policy rate on credit supply to risky firms is positive for the average bank (bottom part of Column 4, right before the fixed effects).
This paper has strong policy implications, even more so now that several central banks are in charge of both monetary policy and financial stability. The results of Jiménez et al. (2014) show that lower short-term interest rates induce banks to take on higher risk through lending, and that this is even worse for lowly-capitalised banks. Put it differently, they show that monetary policy actions have clear financial stability consequences.
– Jiménez, G., S. Ongena, J.-L. Peydró, and J. Saurina, 2014, “Hazardous Times for Monetary Policy: What do Twenty-Three Million Bank Loans Say about the Effects of Monetary Policy on Credit Risk-Taking? ,” Econometrica 82(2): 463-505.