Bank characteristics and the credit channel of monetary policy transmission in the European Economic and Monetary Union. Is there new evidence?
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The objective of this dissertation is to explore how the observed bank characteristics (such as bank size, liquidity, capitalization) and the unobserved (random) bank characteristics generate heterogeneous response of bank loans to monetary policy shocks across the euro area countries and within the banks in each country. I use microeconomic panel data on 3,793 bank balance sheets of seventeen euro area member states. The data are drawn from BankScope on an annual basis, and they cover the period from 1999 to 2013. I also employ more harmonized macroeconomic data compared to what have been used in literature.
I test for the differential reaction of the supply of bank loans by two ways: first, by following the traditional literature by employing a dynamic reduced-form model of bank loans and, second, by presenting a novel model specification known as the Random Coefficients Specification. The difference between this model and traditional model is that the parameters of this model are random rather than deterministic and, therefore, the random coefficients explain the variation in bank loans by not only the observed bank characteristics but also by the unobserved bank characteristics that are not covered in the traditional model.
I measured the change in the monetary policy by the three-month euro interbank offered rate (EURIBOR). I used the most common bank characteristics that had been used in the literature (bank size, liquidity, and capitalization) in addition to another two factors: the return on average assets and the return on average equity. I estimated the traditional model by the ordinary least square in two ways: first, by pooling data without discriminating among country-specific characteristics, and second, by estimating the model separately for each of the four largest economies in the euro area (France, Germany, Italy, and Spain). In effort to gain greater insight of the heterogeneity within the countries, I extended the county-level analysis to include Austria, Belgium, Luxembourg, and the Netherlands. The random coefficients model is estimated for all banks in all euro area countries by applying the restricted maximum likelihood estimator.
Several conclusions are reached in this study. The empirical estimation show that the random coefficients model appears to be more accurate to examine the pass-through of the monetary policy compared to the traditional model. Although the empirical outcomes of the two model seem similar in terms of the sign and the significance of the coefficients, the magnitude of the coefficients is greater for the random coefficients model. The bigger magnitude reflects the influence of the unobserved bank characteristics that are not reflected in the traditional model coefficients. Moreover, the findings demonstrate that the contractionary monetary policy reduces directly the supply of loans of the average bank in the euro area only in lags and all bank characteristics are found to affect the supply of loans only through the interaction with the monetary policy. The evidence support the argument that banks characteristics might be potential sources of the differential response of the banks to the monetary policy in the euro area. The effect of a restrictive monetary policy is found to be more pronounced for small, illiquid, poorly-capitalized, and less-profitable banks. These banks tend to react strongly to the raise in the policy rate by reducing their loans more than large, liquid, better-capitalized, and more-profitable banks.
Although the lending channel is found to operate through all bank characteristics, liquidity emerges as the dominant device to distinguish the path-through of the monetary policy; it has the greater magnitude of its coefficients in the interaction terms with the monetary policy. Liquid banks are more likely to protect their lending activity by drawing on their liquid assets to compensate for the reduction in the deposits when monetary policy tightens.
The total effect of the monetary policy on the growth rate of loans is calculated by performing a separate regression estimation as well as pooling estimation of the traditional model and by estimating the random model. It is found that the total effect varies across the euro area banks based on their characteristics and their typology. It is also found that the total effect are heterogeneous across countries. The average total effect of the monetary policy is found to be negative for small, poorly-capitalized, illiquid, and less-profitable banks, while it is positive for more-liquid and more-profitable banks. The total effect differs for large and better-capitalized banks based on the estimated model.
Based on bank typology, it is found that loans of the cooperative banks decline when the monetary rate increases. Commercial and investment banks expand their lending activity, while the total effect for the real estate and mortgage and for the savings banks are found to differ based on the estimated model. The total effect of the monetary policy on bank loans across the euro area countries appears to be more harmonized and less sensitive to the model estimates. In the exception of German and Austrian banks that are affected negatively by the monetary policy, the other countries (France, Italy, Spain, the Netherlands, Luxembourg, and Belgium), surprisingly, respond positively to a tight monetary policy.
On the country level, the findings provide evidence that the effectiveness of the monetary policy varies within the banks in each county. The evidence of Germany, France, Italy, and Austria supports that lending channel operates through all bank characteristics, but the potency of the monetary policy seems to be more affected by bank liquidity and to a lesser extent by bank size. However, the findings fail to provide evidence of the importance of liquidity and size in Spain. Lending channel seems to work weakly in Spanish banks, and it mainly operates through capitalization. In Belgium and the Netherlands, the importance is given to bank size, while liquidity is found to affect the behavior of bank loans in Luxembourg. Besides, it is found that bank profitability, specifically when it is measured by the return on average assets, is an important element in explaining the reaction of loan supply to monetary policy stance in all counters; loans of lower-profitable banks are more sensitive to the change in the policy rate than more-profitable banks.