Does the mere presence of big banks affect macroeconomic outcomes? Given that large banks can indeed be important for macroeconomic outcomes and financial stability, a number of current policy initiatives are aimed at limiting the impact of bank size: levies to finance bank-restructuring funds are often progressive in bank size; under Basel III, capital surcharges are higher for systemically important banks; in the Eurozone, the Single Supervisory Mechanism under the ECB applies in particular to banks whose total assets exceed €30 billion or 20% of their home economy’s GDP.
Some observers go further by advocating breaking up the big banks (Johnson and Kwak 2010).
Despite this plethora of policy measures and policy proposals, studies of the link between bank size and macroeconomic outcomes are surprisingly few, so our understanding of the implications of the presence of large banks remains limited. Conceptually, there are several different reasons why bank size may matter. Bailout expectations by large banks may invite imprudent risk-taking (‘too big to fail’), and close linkages between large banks and highly leveraged shadow financial institutions may destabilise the entire financial system (‘too connected to fail’). Along these lines, researchers have investigated issues of connectedness, spillovers, and exposure to common macroeconomic shocks.1 In a recent study (Bremus, Buch, Russ, and Schnitzer 2013), we ask whether bank size matters in a more basic sense – even in the absence of contagion, spillover effects, or shared responses to macroeconomic shocks. We focus on granular effects as a channel through which large banks can affect macroeconomic outcomes.
The theory of granularity in economics posits that, if some firms in an industry serve a very large share of the market, idiosyncratic shocks to the largest producers do not average out across the population of firms, but rather affect aggregate outcomes (Gabaix 2011). For US non-financial firms, Gabaix (2011) shows that, if market concentration is high, aggregate fluctuations of output growth are proportional to the product of market concentration and idiosyncratic, firm-level fluctuations. As a consequence, an increase in either concentration or in firm-level fluctuations increases aggregate fluctuations.
Given the high degree of concentration in the financial sector, we apply this concept to the banking industry in two steps:
- First, we determine whether the banking sector in theory and in practice fits the necessary conditions for granular effects to arise;
- Second, we test whether the presence of big banks as measured by a high level of market concentration is associated with a statistically significant relationship between bank-level credit growth and macroeconomic outcomes.
Our answer to both questions is ‘yes’.
Cross-country evidence on bank-size distributions
The focus on bank size in public policy debates and the media is inspired by some sensational bank failures, but also by the general observation that the banking sector in many countries is very concentrated. Figure 1 shows the median values of concentration in the banking sector for a panel of more than 80 countries.2 The graph illustrates that the share of the three largest banks’ assets in a country’s total bank assets is larger than 50% throughout the sample period for both OECD and non-OECD countries.3 Hence, banking sectors are highly concentrated with just a few banks serving a very large share of the market. Moreover, the OECD (2010) points out that merger activity during the global financial crisis has, in fact, led to even higher concentration in many countries. Other industries are highly concentrated as well, but banking is impressive even in this context. For example, the top ten manufacturing firms in Germany account for about 30% of the overall business volume in manufacturing compared to a share of roughly 50% of overall business volume in the banking sector for the ten largest banks (Monopolkommission 2012).
Figure 1. Concentration in the banking sector (three-bank concentration ratio based on assets)
Notes: This figure displays the median values of three-bank concentration ratios for 83 countries. All countries” represents the median across the full sample, while “OECD” and “Non-OECD” show median values across the OECD and non-OECD countries within the sample.
Source: Financial Structures Database, The World Bank.
Not only are banking systems highly concentrated in general, but banks’ assets relative to GDP have also increased quite sharply in OECD countries. This magnifies the effect that shocks to any one bank may have for the real economy.
The necessary condition for such granular effects to emerge from the banking sector is that the bank-size distribution is strongly skewed to the right. The largest banks in an economy have to be large enough relative to the entire market such that bank-specific fluctuations in the largest banks’ credit growth do not to average out in the aggregate. Mathematically speaking, this means that bank size must follow a fat-tailed power law.
Using data on banks’ total assets from Bankscope, we show for many countries that bank size distributions are highly skewed and indeed follow a power law with a fat right tail. Hence, the necessary condition for granular effects is fulfilled. Figure 2 plots the empirical bank size distributions for OECD and non-OECD countries, where size is measured by banks’ total assets.4 Country-specific plots look very similar. The bars indicate the frequency of banks of a given size. In order to enhance visibility, the top 5% of banks in terms of size are not plotted. The graphs show that many small banks coexist with a few extremely large ones. This, in turn, means that concentration is very high.
Figure 2. Histogram of bank-size distributions, 2009
Notes: This figure shows the empirical distribution of banks’ total assets (in billions of dollars) for 83 countries in 2009, divided into OECD and non-OECD countries. The top 5% of banks with respect to total assets are not included to enhance visibility.
Source: Bankscope, Bureau van Dijk.
Granular effects from the banking sector
Having seen that the banking sector is highly concentrated, the question arises as to whether bank-specific shocks can have a perceptible effect on macroeconomic aggregates like credit or GDP. To answer this question from a theoretical point of view requires two ingredients:
- First, the nature of heterogeneity across banks and the nature of idiosyncratic, bank-specific shocks need to be discussed.
Idiosyncratic shocks can result from product innovations, changes in the management team, or unexpectedly high default rates in specific market segments.
- Second, a model is needed which allows quantitative analysis of the impact of these idiosyncratic shocks for the macroeconomy.
Do these shocks affect the macroeconomy because banks are linked among each other, because they support too-big-to-fail subsidies, or simply because the allocation of productivity across banks is heterogeneous?
In a recent theoretical contribution, we abstract from many of these issues and take a very simplistic approach to modeling a banking firm (Bremus et al. 2013). Banks are funded by deposits from households, and they provide working capital loans to firms. The only feature that distinguishes our model from a ‘plain vanilla’ banking model is heterogeneity. Similar to recent advances in the international-trade literature (di Giovanni and Levchenko 2012), we assume that banks draw productivity from a Pareto distribution, to match the size distribution observed in the data, then we build in strategic competition in loan pricing common in other banking models. Using these very simple ingredients, we show that granular can effects arise under very feasible conditions: if concentration in the banking sector is high so that a few very large banks dominate the credit market, idiosyncratic shocks to large banks translate into fluctuations in aggregate credit growth. Given that firms have to fund at least part of their investment by bank loans, fluctuations in the loan market can be transmitted to the real economy via firms’ external funding situation.5
Our empirical results based on this framework using a linked micro-macro dataset of more than 80 countries for the period 1995-2009 confirm granular effects emerging from the banking sector. In order to get a measure of idiosyncratic, bank-level shocks, we have to purge each individual bank’s credit growth from common banking and macroeconomic factors. To do so, we follow Gabaix (2011) and take the difference between each individual bank’s credit growth and the average credit growth in the bank’s domestic market. We then compute a measure of bank-level shocks for each country by taking the weighted sum across every bank’s idiosyncratic change in credit growth, the weights being the bank’s market share in its home country. The weighted sum of bank-level credit shocks is named ‘banking granular residual’. Using fixed-effects panel estimation, we analyse whether the banking granular residual affects aggregate credit and GDP-growth in our sample.
The results show that idiosyncratic bank-level shocks have a positive and statistically significant effect on both aggregate credit and GDP growth. Thus, as banking sectors are highly concentrated, idiosyncratic shocks to large banks do not cancel out, but are rather linked to the variation in macroeconomic variables. Under a less skewed distribution of bank size, this would not be the case: bank-level shocks would not be felt in the aggregate.
Implications for regulatory policy
The presence of big banks – by itself – can drive variation in the aggregate supply of credit or output (i.e. GDP). Thus, policies which lead to increased concentration can lead to increased macroeconomic fluctuations, even in normal times. What implications can policymakers retrieve from our study? One immediate reaction may be that, indeed, drastically reducing bank sizes may be the way out. However, such policies would be an extreme intervention into market forces with serious side effects: large banks with proper incentives can be more diversified than smaller ones and thus more stable (especially large banks that are active internationally are important for the financing of international trade) and perhaps most importantly, small banks can be systemically important as well if many small banks are exposed to the same macroeconomic risks (‘too many to fail’). Hence, a reduction of bank sizes may not necessarily bring about higher financial stability.
For these reasons, we advocate a more balanced approach to policymaking:
- First, reducing the risk-taking incentives of larger banks and thus the magnitude of idiosyncratic shocks will limit granular effects.
These reduced risk-taking incentives can arise through changes in the governance structure of banks, enhancing monitoring incentives of equity owners, and also from reduced public subsidies to banks.
- Second, the higher the risk-buffers of the banks themselves, the less severe are the implications of idiosyncratic shocks for the macroeconomy.
This has (partly) been acknowledged in the new capital adequacy regulation under Basel III but we need more research to fully understand the feedback between capital requirements, market structure in banking, and macroeconomic developments.
- Third, reduced concentration in the banking sector could be a means of mitigating granular effects. Hence, competition policy has an important role to play.
In bank restructuring cases, the way that the assets of failed banks are liquidated matters. Regulators should avoid helping the big banks to get bigger. In short, analysing bank mergers only in terms of market power or loan pricing systematically underestimates their impact for the overall economy. One reason is that it overlooks the implications of mergers en totem for the size of future fluctuations in the aggregate credit supply and GDP.
Acharya, V V und S Steffen (2012), “Analyzing systemic risk of the European banking sector” in Fouque, J P und J Langsam (eds.) Handbook on systemic risk, Cambridge, forthcoming.
Adrian, T and M Brunnermeier (2011), CoVaR, NBER Working Paper No 17454, Cambridge, MA.
Amiti, M and D E Weinstein (2013), “How much do bank shocks affect investment? Evidence from matched bank-firm loan data”, NBER Working Paper No 18890, Cambridge, MA.
Ashcraft, A and D Duffie (2007), “Systemic illiquidity in the federal funds market”, The American Economic Review 97, 221-225.
Beck, T, Demirgüç-Kunt, A, and R Levine (2009), “Financial Institutions and Markets Across Countries and over Time: Data and Analysis”, World Bank Policy Research Working Paper 4943, May.
Bremus, F, C M Buch, K N Russ and M Schnitzer (2013), “Big Banks and Macroeconomic Outcomes: Theory and Cross-Country Evidence of Granularity”, NBER Working Paper No. 19093, Cambridge, MA.
Calderon, C and K Schaeck (2013), “Bank Bailouts, Competition, and the Disparate Effects for Borrower and Depositor Welfare”, Bangor Business School and The World Bank, mimeo.
Čihák, M, Demirgüç-Kunt, A, Feyen, E, and R Levine (2012), “Benchmarking Financial Development Around the World”, World Bank Policy Research Working Paper 6175.
Corbae, D and P D'Erasmo (2013), “A Quantitative Model of Banking Industry Dynamics”, Working Paper, University of Wisconsin.
Di Giovanni, J, and A A Levchenko (2012), “Country Size, International Trade, and Aggregate Fluctuations in Granular Economies”, Journal of Political Economy 120(6), 1083-1132.
Gabaix, X (2011), “The Granular Origins of Aggregate Fluctuations”, Econometrica 79, 733-772.
Hale, G (2012), “Bank relationships, business cycles, and financial crises”, Journal of International Economics 88, 312-325.
Johnson, S, and J Kwak (2010), 13 bankers: the Wall-Street takeover and the next financial meltdown, Pantheon Books, Random House, New York.
Monopolkommission (2012), “Neunzehntes Hauptgutachten der Monopolkommission 2010/2011”, June, Bonn.
OECD (2010), “Competition, Concentration and Stability in the Banking Sector”, Technical report, Policy Roundtables.
Tarashev, N, C Borio, and K Tsatsaronis (2009), “The systemic importance of financial institutions”, BIS Quarterly Review, 75, 1-87.
Tarashev, N, C Borio, and K Tsatsaronis (2010), “Attributing systemic risk to individual institutions”, BIS Working Paper 308, Basel.
1 See, in particular, Acharya and Steffen (forthcoming), Adrian and Brunnermeier (2011), Ashcraft and Duffie (2007), Corbae and D’Erasmo (2013), Hale (2012), or Tarashev et al. (2009, 2010).
2 The data is taken from the Financial Structures Database by the World Bank (Beck et al. 2009, Cihak et al. 2012). The countries included here are Algeria, Argentina, Australia, Austria, Bangladesh, Belgium, Benin, Bolivia, Brazil, Bulgaria, Cameroon, Canada, Chile, China, Colombia, Costa Rica, Croatia, Czech Republic, Denmark, Dominican Republic, Egypt, El Salvador, Estonia, Finland, France, Georgia, Germany, Ghana, Greece, Guatemala, Honduras, Hungary, India, Indonesia, Ireland, Israel, Italy, Japan, Jordan, Kenya, Korea, Kuwait, Latvia, Lithuania, Malawi, Malaysia, Mali, Mauritius, Mexico, Mozambique, Nepal, Netherlands, Nicaragua, Norway, Pakistan, Panama, Paraguay, Peru, Philippines, Poland, Portugal, Romania, Russia, Rwanda, Senegal, Slovak Republic, Slovenia, South Africa, Spain, Sri Lanka, Sudan, Sweden, Switzerland, Thailand, Tunisia, Turkey, Uganda, United Kingdom, United States, Uruguay, Venezuela, Zambia, Zimbabwe.
3 Note that these statistics are based on Bankscope data which does not include all banks in a given country and year. Due to the incomplete coverage, concentration ratios are just a proxy for the market share of the three largest banks.
4 See Bremus, Buch, Russ, Schnitzer (2013) for the bank-size distributions of individual countries.
5 Using linked firm-bank data on lending, Amiti and Weinstein (2013) find that idiosyncratic shocks to banks lead to fluctuations in aggregate credit and investment.