Yves here. I normally let VoxEU articles stand on their own, but this topic, of whether the bank PR that bigger banks are essential stands up to scrutiny, is near and dear to my heart.
Note that the authors point to a 1990s study that finds that a $25 billion in assets bank was the optimal size. There were a fair number of studies done then of bank size versus efficiency. I’m a bit surprised that this is the one that is most often cited, since it also came up with the biggest size threshold at which a negative cost curve kicked in (meaning the bank became more costly to run). One study found that the slightly negative cost curve started at $100 million in assets (!); more typical was somewhere between $1 and $5 billion. And remember, these studies were done in the days when banks returned checks, and check processing was believed to have strong scale economies (ie, if check processing was a bigger proportion of total costs then than now, it could arguably have increased scale economies).
Some academics were frustrated with these results. I recall reading a paper where the author argued that there were theoretical cost savings to being bigger (duh) and basically contended that the empirical data had to be wrong.
This article does make an important contribution in parsing out the impact of absolute versus relative size. And it mentions that bank executives have incentives to make banks bigger (an issue we have discussed) as opposed to safer.
However, a big frustration is that this piece treats all large banks as being of a muchness. There is a considerable difference between being a large and largely traditional bank, versus being one with large capital market operations (like Citibank, Goldman, Deutsche, SocGen, UBS, Barclays). The dealer banks are systemically risk due to the counterparty exposures and opaqueness (as in when one gets in trouble, others with a similar profile are assumed to be on the ropes too). For instance, Lehman, Goldman, and Morgan Stanley are not considered to be of “systemic size” in this study’s parameters.
Another methodological problem in looking at big bank efficiency is that a lot of supposedly off balance sheet exposures of large banks really aren’t. Examples include credit card securitizations (banks have stepped in repeatedly to shore up credit card deals gone sour), the famed structured investment vehicles, and to a degree, mortgage securitizations via putback liability. That means the asset of banks particularly active in these businesses, which of course are the biggest banks, are understated. Grossing them up would be precisely wrong but approximately more correct, and would also lower the size advantage of large banks.
By Asli Demirgüç-Kunt and Harry Huizinga. Cross posted from VoxEU
Today’s big banks are enormous. By 2008, 12 banks worldwide had liabilities exceeding $1 trillion. This column, using data on banks from 80 countries over the years 1991-2009, provides new evidence on how large banks differ in terms of their risk and return outcomes and investigates how market perceptions of bank risk are affected by bank size. It concludes that policies should reward bank managers for keeping their banks safe rather than for making them big.
In recent years, many banks have reached enormous size both in absolute terms and relative to their national economies. By 2008:
12 banks worldwide had liabilities exceeding $1 trillion, and
30 banks had a ratio of liabilities to national GDP higher than 0.5.
Large banks tend to be too big to fail, as their failure would have hugely negative repercussions for the overall economy.
Saving oversized banks, however, may ruin a country’s public finances (Gros and Micossi 2008).
Take the example of Ireland; this country provided extensive financial support to its large banks and subsequently had to seek financial assistance from the EU and the IMF in 2010. The public finance risks posed by systemically large banks suggest that such banks should be reduced in size.
Further evidence against big banks can be found from studies on banking technologies. Berger and Mester (1997) estimate the returns to scale in US banking using data from the 1990s, to find that a bank’s optimal size, consistent with lowest average costs, would be for a bank with around $25 billion in assets. Amel et al. (2004) similarly report that commercial banks in North America with assets in excess of $50 billion have higher operating costs than smaller banks. These findings together suggest that today’s large banks, with assets in some instances exceeding $ 1 trillion, are well beyond the technologically optimal scale.
The public finance risks of large banks and findings on banking cost structures together present a strong case against large banks. All the same, further evidence on how large banks perform relative to small banks is warranted to inform the debate on bank size. Additional insight is useful before one passes judgment on whether systemically large banks should be regulated or taxed out of existence.
Big banks vs. small banks: New evidence
In recent research (Demirgüç-Kunt and Huizinga 2011), we provide empirical evidence on two additional aspects of the debate on big banks vs. small banks.
First, we examine how large banks perform differently in terms of their risk and return outcomes.
For this, indices of bank risk and return based on accounting data are used.
Second, we investigate how market perceptions of bank risk, as reflected in a bank’s interest expenses, are affected by bank size.
Large banks may be perceived to be less risky on account of too-big-to-fail benefits, yielding lower funding costs for sizeable banks (see Carbó-Valverde et al. 2011 for example estimates). Alternatively, large banks are seen as more risky if they are too big to save, giving rise to higher interest rates.
These aspects of bank size are investigated for an international sample of banks from 80 countries over the years 1991-2009. These international data allow us to distinguish between a bank’s absolute size (as measured by the logarithm of its total assets) and its “systemic” size (i.e. how risky a bank is as measured by the ratio of bank liabilities to national GDP). The correlation between these proxies for a bank’s absolute and its systemic size is positive, but low at 0.1. Thus, it is meaningful to separately consider bank absolute size and systemic size.
Size matters, but is it absolute or systemic size?
The distinction between bank absolute and systemic size turns out to be important for explaining bank performance regarding bank risk and return. A bank with larger absolute size on average realizes a higher return on assets. This higher return, however, comes at a cost of higher bank riskiness. A bank’s absolute size thus implies a trade-off between bank risk and return.
The impact of systemic bank size on risk and return is very different. Systemically larger banks on average have lower returns on assets, but there is no discernible impact on bank riskiness.
Systemic size is thus a liability, as it lowers return without an offsetting reduction in risk.
In practice, expanding banks see their absolute and systemic size increase simultaneously. Banks located in smaller countries, however, see their systemic size increase relatively more, with negative implications for risk and return outcomes.
Next, we investigate how a bank’s interest expenses are affected by bank systemic size. Systemically large banks, defined as banks with a ratio of liabilities to GDP exceeding 0.1, on average are found to pay interest rates that are 40 basis points higher, suggesting a “too-big-to-save” effect. Furthermore, the interest expenses of systemically important banks are more sensitive to the bank capitalisation ratio as a proxy for bank risk. This also suggests that systemically important banks are too big to save, and that they are subject to market discipline by bank liability holders.
This new evidence of market discipline of systemically large banks contrasts with earlier evidence, mostly for the US, that absolute bank size pays off. In particular, Kane (2002) and Penas and Unal (2004) report that large bank mergers create value for bank shareholders and bond holders, respectively, as larger bank size increases too-big-to-fail subsidies. In our broader international sample, we do not find any impact of a bank’s absolute size on its interest costs, but we confirm earlier evidence that absolute size reduces market discipline by a bank’s debt holders for a sample of just US banks.
Market discipline of systemically important banks, while it exists, has been ineffective in preventing the emergence of systemically huge banks worldwide. A main reason for this may be that bank managers, rather than bank shareholders, in practice devise and implement bank growth strategies. Bank managers may well benefit from bank asset growth through higher pay and stature, even when continued bank growth is not in the interest of bank shareholders. The phenomenal growth at individual banks that we have witnessed over the last several decades may thus be a reflection of inadequate corporate governance at banks failing to align the interests of bank managers and bank shareholders.
In the absence of effective market discipline on bank systemic size, public policy in the form of regulation or taxation is required to bring down bank systemic size (see Goldstein and Véron 2011 for a discussion).
Regulation can take the form of quantitative limits on bank size, or of other regulations such as capital adequacy and liquidity requirements that are biased against systemically large banks.
Taxation can similarly take the form of, say, levies on bank liabilities that are especially geared towards systemically large banks.
In the US, the Wall Street Reform and Consumer Protection Act (or Dodd-Frank Act) passed in July 2010 prohibits bank mergers that result in a bank with total liabilities exceeding 10% of the aggregate consolidated liabilities of all financial companies, but an earlier proposal by the Obama administration to institute a levy on the liabilities of large bank failed to be enacted. In Europe, the European Commission (2010) is proposing bank levies to finance national bank resolution funds. Such levies could easily be slanted towards large banks, at the national or EU level.
Evidence that market discipline on bank systemic size is ineffective suggests that bank levies on oversised banks by themselves are not enough to reduce bank size.
Corporate governance reform in the banking sector is also needed to ensure that market discipline and taxation can be effective.
In particular, bank managers should be rewarded for keeping their banks safe rather than for making them systemically large.