Are capital buffers pro-cyclical?

Abstract

We estimate the relationship between the Spanish business cycle and the capital buffers held by Spanish commercial and savings banks in an incomplete panel of institutions covering the period 1986–2000, which comprises a complete cycle. After controlling for other potential determinants of the surplus capital we find a robustly significant negative relationship between the position in the cycle and capital buffers. From a quantitative standpoint, an increase of 1 percentage point in GDP growth might reduce capital buffers by 17%. This relationship is, moreover, asymmetric, being closer during upturns.

Introduction

Efforts within the Basel Committee on Banking Supervision to update the 1988 Basel Accord are close to finalising a new accord on banks' capital adequacy, which is already known as “Basel II.” There remain, however, some outstanding issues and details. Among these, that of “pro-cyclicality” has, perhaps, prompted most debate in the literature.¹

As is well known, one of the primary aims of the new accord is to link capital requirements more closely to risks. Accordingly, the former will increase whenever the latter increase. Yet risks are not observable and have to be estimated. In this regard, it is interesting to note that the standard methods to estimate risks tend to provide higher estimates during economic downturns as reflected, for instance, in the increase in the number of downgradings made by rating agencies or in the widening of credit quality spreads.

Therefore, under Basel II capital requirements are likely to increase in recessions. Yet if capital requirements increase in a recession—when building reserves from decreasing profits is difficult or raising fresh capital is likely to be extremely costly—banks would have to reduce their loans and the subsequent credit squeeze would add to the downturn. This would make the recession deeper, thus setting in motion an undesirable vicious circle that might ultimately have an adverse effect on the stability of the banking system.² The implications of this link between financial stability and macroeconomic stability in terms of the soundness of credit institutions merit being taken into account in the final design of Basel II.

But more importantly, it is unclear whether risks actually increase during downturns. As a matter of fact, they may even be thought to grow during upturns, when the imbalances that (unfortunately) will cause the next recession are building up and some lenders are acting as if the economic boom will last for ever.³ It is not obvious how, in practise, a genuine increase in risk and an increase in, so to speak, the rate of materialisation of those risks can be disentangled.⁴ Against this background, calling for some flexibility in the final design of Basel II so as to accommodate the fact that risks cannot be estimated without error would seem a sensible option.

It has been also argued, however, that if internal risk management models take properly into account the way default probabilities change throughout the business cycle, the effects on credit—and therefore, on output—should not be overstated. Moreover, the Basel Committee has recently proposed new flatter risk-weight curves, which also mitigate the problem.

While arguments highlighting or minimising the actual relevance of the pro-cyclicality problem have proliferated, the related empirical evidence is scant. Admittedly, the empirical literature on the impact of capital requirements on bank behaviour is extensive, though mainly confined to the US case.⁵ Papers have dealt with issues such as whether the introduction of minimum capital requirements leads banks to hold higher capital; the impact of capital requirements on risk-taking, competitiveness, and a level playing field; and whether capital requirements create credit crunches affecting the real economy. Nevertheless, as far as we know, these papers have not analysed the cyclical behaviour of capital requirements, perhaps because the current Basel capital accord ties capital requirements less closely to banks' capital risk.

Against this background, the aim of this paper is to provide some fresh empirical evidence which may prove useful in the debate about the pro-cyclicality of the new capital accord. In particular, we have noticed that most arguments in this debate relate to the cyclicality of the capital requirements, thus ignoring the fact that only a few banks hold just the required capital, while most keep capital buffers which, in some cases, are quite significant.⁶ The reasons why a bank may want to hold a capital buffer are manifold. As will be explained in detail later on, some of the fundamentals that determine the behaviour of these buffers may themselves have a cyclical behaviour. Yet after controlling for the effects of these fundamentals, the residual direct relationship between excess capital and the business cycle may be an important piece of information.

Thus, let us imagine first that banks tend to increase their capital buffers during upturns beyond what fundamentals indicate. This building up of excess capital when the economy grows above potential might be seen as evidence in favour of those who think that credit institutions either internalise the negative externalities of pro-cyclical requirements on macroeconomic stability mentioned above or properly take into account the possibility that risks may actually increase during booms. In other words, banks would be making use of capital buffers to offset—at least partially—the negative effects of pro-cyclical requirements.

By contrast, a negative (direct) relationship between capital buffers and the cyclical position might offer support to those who believe that, particularly during upswings, some institutions (or the models they use) tend to underestimate actual risks because they fail to properly characterise the cyclical nature of output. Accordingly, it would seem sensible to look for a way to accommodate this pro-cyclicality issue in the design of Basel II without it being to the detriment of the main goal of the new capital accord. So the question is: how do the capital buffers (excess of current capital over the minimum capital requirements) maintained by banks vary over the business cycle?

In this paper, we address this question empirically for Spanish commercial and savings banks. Using standard econometric panel data techniques, we build an incomplete panel of Spanish institutions from 1986 to 2000—thus covering a complete business cycle—and estimate an equation for the behaviour of capital buffers that includes an indicator of the business cycle.

Admittedly, focusing on a single country might limit the generality of the conclusions emerging from our analysis. While data availability (in particular regarding individual capital requirements) prevents us from a more general approach, it is worth mentioning that the business cycle seems—at least during the period considered—to have been relatively pronounced in Spain, which might render this an interesting case study. Moreover, Spanish banks are highly competitive and efficient, which reduces the probability of idiosyncratic factors biasing the results (see, for instance, ECB, 1999).

Our findings are fairly robust and quite unequivocal. After controlling for other potential determinants of the surplus capital, which could themselves have a cyclical behaviour (cost of capital, risk profile of the institution, size, etc.) we find a (robustly significant) negative relationship between the business cycle and the capital buffers that Spanish institutions held throughout the period analysed. From a quantitative standpoint, an increase of 1 percentage point in GDP growth might reduce capital buffers by 17%. This relationship is, moreover, asymmetric, being closer during upturns than during downturns.

Accordingly, our results offer support to those who think that it would be worth taking into account the so-called pro-cyclicality problem in the final design of Basel II. Interestingly, the existence of capital buffers might itself offer a way out as, under Pillar 2 of the new accord, regulators could monitor and, when deemed necessary, influence the behaviour of the excess capital of those institutions that are thought to be behaving in an excessively lax manner during economic upswings.

It should be noted, finally, that our results relate to the behaviour of banks under Basel I. Of course, it cannot be totally ruled out that Basel II may cause a structural change. It is worth noting, however, that Basel II is going to change a quantitative requirement which, for most banks, is not strictly binding. Moreover, on average, capital requirements are not going to be increased. It is therefore an open question whether such a change has necessarily to affect the behaviour of capital buffers. In any event, it goes without saying that the policy implications of our results should be viewed with due caution.

The rest of the paper is structured as follows. The theoretical framework, as well as the empirical equation we estimate, is introduced in Section 2, while the data set is described in Section 3. The fourth section shows the main results of the basic econometric analysis, and some extensions are considered in Section 5, providing more information on the pro-cyclicality of capital buffers. Finally, the conclusions of the paper are drawn in the last section.