Do Higher Interest Rates on Loans and Deposits and Advertising Spending Cuts Forecast Bank Failures? Evidence from Russia

Abstract

This study builds a probabilistic model of Russian bank defaults. Microdata from the monthly financial and regulatory statements that Russian banks submit to the Bank of Russia are analysed, covering the period from July 2010 to December 2017. A model incorporating a standard set of reliable predictors of bank defaults is augmented by three novel predictors: the excess of deposit and loan rates over the respective cross-section averages, and the ratio of spending on advertising to the bank's assets. These predictors are statistically significant in logit regressions that forecast bank defaults and improve the forecasting power of the model, although relatively moderately. The too-big-to-fail premise is not supported by the data.