|
Article Excerpt
I. INTRODUCTION
The bank-lending channel (BLC) emphasizes the role of bank loans in the transmission of monetary policy. According to the BLC, contractionary monetary policy can decrease the loan supply of banks that face informational constraints. Additionally, some of the borrowers of these banks are bank dependent, so that a fall in loan supply decreases the expenditures of these customers. This channel complements the usual interest rate/money channel by offering an explanation for why contractionary policy has severe effects on the economy. It also implies that policy has distributional effects across banks and their customers.
Evidence of a BLC is gleaned from the cross-sectional asymmetric loan responses of constrained versus unconstrained banks. Empirical studies usually approach the identification of the BLC by dividing banks using balance sheet constraints. This approach implicitly assumes that bank balance sheet structure is independent of monetary policy. However, because bank balance sheet constraints can be influenced indirectly by the effects of policy on the behavior of bank customers, it is not clear whether these studies are measuring shifts in loan supply or loan demand.
This paper identifies a BLC utilizing data on a banking system where de jure deposit guarantees were differentially applied. From 1995 until 1999, the Polish government granted full guarantees to one group of banks, while the others had only a partial guarantee. A full guarantee could give banks an advantage in raising nonreservable, uninsured funds, allowing these banks to circumvent contractionary monetary policy by continuing loan growth. Conversely, the loans of banks with only partial guarantees should decrease due to the inability of these banks to raise alternative funds during contractionary policy. The existence and extent of a BLC are measured in this paper by testing for a differential loan growth response between these two groups of banks. Additionally, we test for the disappearance of this differential response in the post-1999 period when the full guarantee was eliminated and reserve requirements on all deposits were equalized. Unlike the use of balance sheet constraints to distinguish a BLC, the use of this double difference across banks and over time is an innovation that should mitigate most concerns about identification.
This approach identifies a BLC that may have widespread applicability and has implications for financial system development. Policies that establish differential guarantees are common in emerging financial markets. Banks that are given full guarantees are often those that dominate the banking system and are the most inefficient. Our BLC implies that central banks may have trouble controlling the credit growth of the fully guaranteed banks. Additionally, the adverse effect of policy attempts to limit credit growth will fall disproportionately on the partially guaranteed banks, which are often the most efficient banks. This distributional effect of policy could hamper financial system development.
The next section reviews the literature on the BLC and differentially applied deposit guarantees. Section III gives an overview of the Polish banking system focused on deposit guarantees, loan growth, and monetary policy. Section IV explains the data and specifies the empirical model used to test our loan growth and time deposit funding hypotheses. This section also presents the empirical results. The final section lays out some implications following from the results.
II. REVIEW OF THE LITERATURE
A. The BLC
In the BLC, contractionary monetary policy pulls away required reserves, forcing banks to reduce reservable deposit funding for their loans. Banks try substituting into large time deposits because of the lower reserve requirements and greater interest sensitivity. Due to informational frictions and the resulting risk of default associated with these uninsured deposits, depositors require a risk premium. However, some banks may face an external finance premium that is prohibitively high, preventing them from maintaining funding for their loans. Therefore, contractionary policy decreases the loan supply of these fund-constrained banks. If some banks can mitigate these informational problems by conveying to depositors that they have a lower probability of default than other banks, they can raise time deposits at a lower cost than other banks and maintain their loan growth. The differential loan response of constrained versus unconstrained banks allows a test of the BLC.
The key to empirical measurement of the BLC is to find a differential constraint between banks that allows one to reasonably argue that contractionary policy shifts the loan supply of the constrained banks. Most BLC studies argue that balance sheet strength allows some banks to mitigate informational problems, while the loans of banks with weak balance sheets respond to contractionary policy. Kishan and Opiela (K-O 2000, 2006) for the United States and Altunbus, Fazylov, and Molyneux (A-F-M 2002) for the European Union show that the loan growth of highly levered banks is more responsive to monetary policy than that of well-capitalized banks. I They argue that well-capitalized banks assure depositors of their low probability of default, giving these banks a funding advantage. Kashyap and Stein (K-S 1995, 2000) show that banks that are small and illiquid are more responsive to policy than other banks. They argue that large and relatively liquid banks can continue funding loans either by substituting from securities into loans or by raising large time deposits. The BLC in these studies is measured through those banks that are small, highly levered, and illiquid. The use of balance sheet constraints to identify loan supply shifts assumes that policy does not impact bank balance sheets by affecting the condition of bank customers. However, evidence of a balance sheet channel makes it probable that monetary policy does affect bank loan demand by affecting the net worth of bank customers (Bernanke, Gertler, and Gilchrist 1996). That is, highly levered and illiquid banks may have gotten that way through the adverse effect of policy on the customers of these banks. (2)
Recent papers have addressed this identification problem. Peek, Rosengren, and Tootle (2000) use confidential supervisory information on poor aggregate bank health to identify adverse loan supply shocks. Ashcraft (2006) shows that the loans of small banks not affiliated with a bank holding company are more responsive to monetary policy than those with an affiliation. Holod and Peek (2007) show that banks that are not publicly traded are more responsive to contractionary monetary policy than banks that are publicly traded. The classification of constrained and unconstrained banks in each of these approaches mitigates concerns about identification while distinguishing a particular type of BLC through the measured constraint.
B. Differential Deposit Guarantees
In this paper, we argue that differentially applied de jure deposit insurance guarantees distinguish a BLC and that this approach allays most concerns about identification of loan supply shifts. Banks with a large credible deposit guarantee should have a lower perceived probability of default than banks with a smaller guarantee. This lower chance of default should give banks with larger guarantees an advantage in raising uninsured funds during contractionary policy, mitigating the effect of policy on the loan growth of these banks. The implied funding advantage for banks with relatively large guarantees is prevalent in the risk-pricing literature. Hannan and Hanweck (1988) show that large banks paid lower rates on large CDs than did smaller banks in a 1985 sample of 300 U.S. banks, when controlling for bank-specific risk factors. Park and Peristiani (1998) show similar results for thrifts during the period 1987-1991. Both studies attribute this result, in part, to a too-big-to-fail policy. Flannery and Sorescu (1996) present evidence that large bank holding companies in the United States paid lower rates for subordinated notes and debentures than other banks in the mid-1980s, also arguing that the larger holding companies enjoyed higher implicit guarantees relative to smaller ones. They also compare the effects of possible differential guarantees between periods and find that in the latter part of their sample, this size advantage disappears probably due to a change in policy by the
Federal Deposit Insurance Corporation to limit coverage only to banks and not to their holding companies.
The idea that differential guarantees can distinguish a BLC is not new. The use of asset size as a means to separate loan-constrained from -unconstrained banks assumes that large implicit guarantees accompany large asset size (K-O...
|
