During the debate over the 2010 Basel III regulatory reform, one of the biggest concerns was that higher capital requirements would damage economic growth. Pessimists argued that forcing banks to increase their capitalization would lower long-run growth permanently and that the transitional adjustment would impose an extra drag on the recovery from the Great Recession. Unsurprisingly, the private sector saw catastrophe, while the official sector was more positive.
The Institute of International Finance’s (IIF) 2010 report is the most sensational example of the former and the Macroeconomic Assessment Group (MAG) one of the most staid cases of the latter. The IIF concluded that banks would need to increase capital levels dramatically and that this would drive lending rates up, loan volumes down and result in an annual 0.6-percentage-point hit to GDP growth in the United States, the euro area and Japan. By contrast, the MAG reported that the implied increase in capital would drive lending rates up only modestly, loan volumes down a bit, and result in a decline in growth of only 0.05 percentage point per year for five years – one-twelfth the IIF’s estimate.
Four years on we can start to take stock, and our reading of the evidence is that the optimistic view was correct. Since the crisis, capital requirements and capital levels have both gone up substantially. Yet, outside the still-fragile euro area, lending spreads have barely moved, bank interest margins have fallen and loan volumes are up. To the extent that more demanding capital regulations had any macroeconomic impact, it would appear to have been offset by accommodative monetary policy.
To begin, it is important to appreciate how much higher the new international capital standards are. Basel III is more rigorous than its predecessor in three fundamental ways: the definition of what constitutes capital is tighter, the coverage of what counts as an asset is broader, and the required ratio of the two is higher. In a previous post, we estimated that the effective increase in risk-weighted capital requirements is a factor of 10 to 20 times (albeit from a level that was abysmally close to zero).
Not only have requirements gone up, capital levels have, too. According to the Basel Committee on Banking Supervision’s quantitative impact assessments, from the end of 2009 to the end of 2013, capital ratios based on the new stringent definitions have risen significantly. The numbers in the table below are striking. Since end-2009, capital (as measured by “common equity tier 1,” the most effective loss absorber of the various capital categories) has risen by 4.5 percentage points of risk-weighted assets for the 102 largest banks in the world, and by 2.7 percentage points for the smaller banks in the Basel Committee’s sample. And, while there are surely differences across banks and regions, end-2013 capital ratios on average exceeded those that Basel III requires for 2019.
How did banks raise these capital ratios? Pessimists worried that banks would “de-leverage” and “de-risk” to reduce the denominator of the capital ratio. That is, the measured ratio of capital to risk-weighted assets would go up from a combination of balance sheet shrinkage and a shift away from assets with high risk weights. However, a recent study shows that – on average – this has not been the case. Outside of Europe, bank balance sheets have expanded, while capital levels (the numerator) have risen even further. Moreover, one third of this capital increase has come from new issuance, while the remaining two thirds has been from retained earnings.
Contrary to the predictions of the private-sector doomsayers, as banks were increasing their capital levels, they accepted a smaller return on assets, cutting their net interest margins and reducing their operating costs. The BIS provides data on banks in 15 large advanced and emerging market countries. Comparing 2013 results with the 2000-2007 average, the data show that the return on assets fell by roughly 40 basis points, interest rate spreads narrowed by an average of more than 30 basis points, and operating costs declined by nearly 75 basis points. Over this period, bank credit to the non-financial private sector rose.
To be sure, Europe is something of an exception. There, lending spreads are generally up and loan volumes down. However, the likely explanation for this pattern is two-fold. First, there is the troubled way in which European supervisors conducted stress tests and capital exercises. Instead of requiring banks to raise additional capital to offset a shortfall – as the 2009 U.S. stress test did – authorities allowed them to meet capital ratios by shedding assets. In fact, euro-area banks viewed as a whole did not raise capital. Instead, they reduced both their total assets and their risk-weighted assets. Second, as the most recent ECB stress test reveals, a number of continental European banks remain under pressure to further raise their levels of capitalization to meet Basel III standards.
These euro-area problems are consistent with the commonly held belief that banks with debt overhangs do not lend. A quick look at some country data bear out this view. The following chart plots the ratio of bank capital to risk-weighted assets in 2006 (computed using national definitions) on the horizontal axis, against the change in overall credit to GDP from 2006 to 2013 on the vertical axis. The raw correlation between these two series is 0.47. We note that there is a strong negative relationship between changes in bank capital ratios and credit growth over this time period, but that correlation probably reflects the market-driven pressure on countries with weak banking systems in 2006 to shore them up during and after the crisis.
Bank capital ratio in 2006 and the change in bank credit relative to GDP from 2006 to 2013
We draw two preliminary conclusions from this accumulated evidence on bank capital and credit. First, the predictions that higher capital requirements would drive up interest margins and reduce credit volumes are at odds with the evidence of smaller spreads and increased lending. Insofar as there was any aggregate macroeconomic impact, it appears to have been limited or inconsequential. Instead, especially in a weak economy, it is high levels of initial bank capital that lead to healthy lending.
Second, the evidence is bad news for Basel III’s countercyclical capital buffer. The idea of the buffer is that, when confronted with a credit boom, authorities should temporarily raise capital requirements both to increase financial system resilience if a bust comes and to limit the credit expansion.
However, the financial system can be made more resilient simply by setting equity capital requirements higher on a permanent, rather than cyclical, basis. In an earlier post, we argued for such a rules-based approach. Moreover, as recent research highlights, for the countercyclical buffer to be effective in limiting credit expansion, there are three preconditions: (1) capital requirements have to bind before they are raised; (2) equity has to be costly and difficult to raise in the short term, and; (3) alternatives to bank credit have to be relatively unavailable and costly. On all three counts, recent experience is discouraging. Higher capital requirements generally have not compelled banks to widen lending spreads; loan volumes do not look sensitive to changes in capital so long as banks are reasonably well capitalized; and, at the stage in the business cycle when the countercyclical buffer would be needed, banks’ business is likely to be booming and profitable, making it cheaper and easier to issue new equity.
We still need further study of this episode, including a proper statistical analysis that controls for macroeconomic conditions and policy responses. At this stage, however, it seems reasonably safe to conclude: (1) Regulators should seriously consider requiring further additions to bank capital, given that the social costs of post-crisis capital increases appear to have been limited, and; (2) the efficacy of time-varying capital requirements in moderating credit fluctuations is questionable.
This post summarizes a longer piece by one of us that appears as a CEPR Policy Insight here