Suppose the government imposed a 50 percent sales tax on food purchased at a restaurant. Such a tax would no doubt lead people to eat out a lot less. But suppose the revenue was used to provide modest income support during unemployment. People usually cut back on eating out when unemployed. With the new tax and income support, would unemployed people be more likely to eat out? Probably not, because household finances would still be tight and eating out would be even more expensive compared with eating at home.
As former Fed governor Jeremy Stein pointed out in a speech a few years ago, capital and liquidity requirements placed on banks are similar to the hypothetical dining-out tax. Because capital and liquidity buffers are expensive, they encourage banks to move away from types of lending that come with the highest capital and liquidity requirements. Indeed, regulators often establish higher liquidity and capital requirements on certain activities with the explicit goal of discouraging that activity – for example, banks aren’t allowed to treat deposits at other banks as a liquid asset because regulators want to discourage interconnectivity between banks.
But capital and liquidity buffers also act like the additional unemployment income in the example and allow banks to weather rough patches to keep operating. Sometimes, bank regulators focus on this mitigant role of the buffers and mistakenly conclude that when they require banks to hold more liquidity or capital against a specific type of lending, the banks will do more of it.
For example, while a bank facing a run would seem likely to hold all its cash inflows to pay off departing creditors, the liquidity coverage ratio (LCR) assumes that banks instead will relend a large fraction of cash inflows to households and small businesses. In explaining that assumption, the banking agencies stated that one of the purposes of the LCR is to ensure that banks are able to sustain such lending during times of stress (see p. 61512 of the LCR rule). While the goal may be lauded, the predicted impact on bank behavior is exactly wrong. Periods of stress are when banks will struggle the hardest to meet regulatory liquidity requirements (just like the unemployed person struggling to pay her bills), and therefore will work harder to reduce activities that come with high liquidity charges. So applying a high liquidity charge on a category of lending will not only discourage it in good times, it will also discourage it in difficult times.
At times, the result that banks substitute away from a loan category with a high capital or liquidity requirement appears to be an unintended consequence. In particular, the Fed’s annual stress tests are built on projected economic downturns of unprecedented severity, and so penalize banks for loans to sectors exposed to such downturns. As shown in a recent TCH research note, to pass the tests, banks have to hold very high amounts of capital against loans to small businesses, likely because small businesses struggle during recessions. Not surprisingly, as shown in another recent TCH research note, banks subject to the tests have cut back on lending to small businesses, which seems unlikely to have been the Fed’s intention.
A related confusion seems to have set in concerning the effect of capital requirements on bank lending generally. In a series of speeches and papers, Hyun Shin, the head of research at the Bank for International Settlements, documented that bank loan growth is higher at times when banks have more capital. The Clearing House recently showed that the positive correlation Shin identified is the result of banks lending more when they have extra capital over and above regulatory requirements, probably because banks only end up with extra capital when times are good and loan demand high. However, some senior bank regulators appear to have reached the incorrect conclusion that Shin’s result implies they can encourage bank lending by raising capital requirements. That’s like observing that neighborhoods with a lot of police have a lot of crime, and concluding that cutting back on police can reduce the amount of crime.
There appears to be a risk that bank regulators may be making a similar mistake when thinking about a further tightening of capital requirements proposed last fall by former Fed Governor Daniel Tarullo. In a recent note, economists at the Office of Financial Research argued that one reason to adopt Governor Tarullo’s proposal is that, by imposing higher capital hurdle rates to pass annual stress tests, banks would be able to meet day-to-day capital requirements during recessions or financial crises more easily and so be less likely to cut back on lending. Of course, as discussed in a February TCH blog post, it would be the new, more difficult stress tests that the banks would have to pass during a recession or financial crisis. So not only would banks cut back on lending now in reaction to the more draconian stress tests, they would be even more likely than without the change to cut back on lending in a crisis.
While the regulations that have been implemented since the financial crisis have contributed to a prudent buildup in capital and liquidity, the gains from further tightening are likely to be modest at best and therefore increasingly likely to not be worth the added cost in terms of slower economic growth and reduced employment. So it is especially important at this point that regulators have a clear understanding of those costs. Specifically, if they tax an activity, in this case lending, they should not expect that there will be more of it; there will be less of it.
Disclaimer: The views expressed in this post are those of the author(s) and do not necessarily reflect the position of The Clearing House or its membership.