to the private sector with respect to the interest rate on government securities or the spread between the interest rate on private loans and the interest rate on government securities. The problem is that the part of the variation that comes from changes in the rate paid by the government is the same for all banks and therefore is indistinguishable from any other time varying effect on lending. The part that comes from the rates charged by the banks does vary by bank, but cannot possibly be independent of demand conditions in the bank and other unobserved time varying bank specific factors. One cannot therefore hope to estimate the true elasticity of lending by regressing loans on the spread.
Our strategy is to focus on a more limited question which we may hope to answer somewhat more convincingly: are banks more responsive to the central bank interest rates in slow growing environments? We start by identifying the banks that are particularly likely to be heavily invested in the “easy life.” These are banks that, for historical reasons, have most of their branches in the states that are currently growing slower than the rest. Our hypothesis is that it is these banks that have a particularly strong reason to invest heavily in government securities, since in a slow-growing environment it is harder to identify really promising clients. They also probably have more “marginal” loans, that they are willing to cut and reduce (or not increase) when the interest rates paid to government bonds increases. It is therefore these banks that should be particularly responsive to changes in the interest rate paid by the government.
Data: The outcome we focus on is the ln(Credit/Deposit Ratio), at the end of March of each year, for 25 public sector and 20 private sector banks. Two minor public sector banks were excluded due to lack of data, while the new private sector banks were excluded for reasons of comparability. The data are from the Reserve Bank of India.
Data on the net state domestic product are from the Central Statistical Office. For our measure of interest rate spread, we subtract from the SBI prime lending rate, the rate given as the weighted average of central government securities. Both interest rate measures are from the RBI. as is the price index used to deflate them36
Specification: Two measures of growth are used. To measure the state growth rate (growthit = ln(SDPit) − ln(SDPi,t−1)), we use a moving average of the real growth rates of previous three
36 Data are from the 2001 Edition of RBI “Handbook of Statistics on Indian Economy.” We use the CPI-UNME, for Urban Non-Manual Employees.