Banks’ regulatory buffers, liquidity networks and monetary policy transmission

Abstract

Based on a quarterly regulatory dataset for German banks from 1999 to 2004, this article analyses the effects of banks’ regulatory capital on the transmission of monetary policy in a system of liquidity networks. The dynamic panel regression results provide evidence in favour of the bank capital channel theory. Banks holding less regulatory capital and less interbank liquidity react more restrictively to a monetary tightening than their peers.

Acknowledgements

The views expressed in this article are those of the authors and do not necessarily represent those of the Deutsche Bundesbank, of the Kiel Institute for the World Economy (IfW), or of the International Monetary Fund (IMF) or IMF policy. We thank Kit Baum, Helge Berger, Claudia Buch, Kai Carstensen, Larry Christiano, Philipp Engler, Frank Heid, Heinz Herrmann, Skander Van den Heuvel, Thomas Kick, Michael Kötter, Thilo Liebig, Daniel Quinten, Pablo Rovira Kaltwasser, Andrea Schertler, Dennis Snower, Andreas Worms and the participants of the Bundesbank workshop on financial stability and the IfW staff seminar for helpful comments and suggestions. This article was written while C. Merkl was a guest researcher at the Deutsche Bundesbank. He would like to thank the Deutsche Bundesbank for hospitality and support.

Notes

¹ For a quite broad but somewhat outdated overview, see symposium on the monetary transmission mechanism (Mishkin, 1995). For European evidence, see Angeloni et al. (2003).

² For macroeconometric evidence, see Deutsche Bundesbank (2005).

³ The result was recently challenged by Baum et al. (2004a). They write that the described pattern is much weaker and thus economically potentially not as relevant when taking market volatility into account. For a theoretical explanation, see Baum et al. (2004b).

⁴ For a summary of the most recent results, see Angeloni et al. (2003).

⁵ For the bank lending channel to be operative, differences in the costs of finance depending on bank-specific criteria, however, are necessary in the theoretical model developed by Ehrmann et al. (2003), which was used for a number of empirical studies.

⁶ Their owners (cities, municipalities, rural districts for savings banks and states plus the local savings banks for the head institutions) guaranteed all liabilities. Thus, in the past, all institutes in the savings bank sector enjoyed the status of a de facto AAA rating (although most of them were not officially rated) and there was no default risk for noninsured liabilities.

⁷ Measured as time periods when banks had more customer deposits than customer loans.

⁸ Loans typically have a longer time to maturity than deposits. Maturity transformation is regarded as one of the main functions of a bank (e.g. Freixas and Rochet, 2004).

⁹ According to Basel I, 8% of the loan volume has to be held as capital (there are exceptions for government and other specific loans). A violation of the minimum capital requirement may have serious consequences, such as being taken under the control of the domestic supervisors or even being closed down.

¹⁰ Some other studies use ‘excess capital’ synonymously for ‘capital buffer’.

¹¹ Engler et al. (2007) interpret this result in the context of the traditional bank lending channel, since they cannot find any evidence that Austrian savings banks and credit cooperatives perform significant maturity transformation.

¹² The change in loans is an approximation for newly issued loans.

¹³ In the baseline regression, we omit the bank lending channel option. In a robustness check (Merkl and Stolz, 2006), we assume that the amount of inflowing deposits is driven by the stance of monetary policy to incorporate the features of the bank lending channel into the model. As a consequence, monetary policy can be transmitted via the head institution of the liquidity network of banks.

¹⁴ For instance, we calculate the ‘long-term’ coefficient λ using the following formula . The other ‘long-term’ coefficients are calculated accordingly.

¹⁵ Please note that Capital Buffer and Asset Buffer are corrected for the mean and, hence, take on negative values for banks with low buffers (see Appendix for details).

¹⁶ In Merkl and Stolz (2006) the estimation results from a fixed effects specification are shown and deliver the same qualitative outcome.

¹⁷ Without previously normalizing excess capital to zero, the sample split for credit cooperatives and savings banks delivers similar results.

¹⁸ To check for robustness, we validated all results obtained from the dummy approach by using an equivalent sample split for poorly and highly capitalized banks, which is somewhat less restrictive with respect to the imposed dynamics (for instance, not assigning the same estimated coefficients for the lagged dependent variable for both types of banks). Since this approach leads to the same conclusions, we do not show the estimated coefficients.

¹⁹ Using the same definition as in the dummy approach.

²⁰ See, e.g. Frühwirth-Schnatter and Kaufmann (2006) for Austria.

²¹ The fixed effects results are only shown in Merkl and Stolz (2006). If we use a pooled regression, we obtain estimated coefficients that are in between the other two coefficients.

²² Interest-bearing assets and liabilities with time-to-maturity shorter than 1 year, 1–2 years, 2–3 years, 3–4 years, 4–5 years and above 5 years.

²³ The Amendment to the Basel Accord to Incorporate Market Risks, Basle Committee, January 1996. Since these maturity classes are more detailed, we had to use averages for the available maturity classes.