Despite the much-publicized black eye to Citigroup's management, the bottom line of the Federal Reserve's stress tests is that every other large U.S. bank will be allowed to pay out more cash to its shareholders, either as increased dividends or stock buybacks. And pay out more cash they will: at least $22 billion in increased dividends (that includes all the banks subject to stress tests), plus increased buyback plans.
Those cash payouts come straight out of the banks' capital, since they reduce assets without reducing liabilities. Alternatively, the banks could have chosen to keep the cash and increase their balance sheets—that is, by lending more to companies and households. The fact that they choose to distribute the cash to shareholders indicates that they cannot find additional, profitable lending opportunities.
This puts the lie to the banks' mantra that capital requirements will constrain lending and therefore reduce growth (made most famously in the Institute of International Finance's amateurish report claiming that increased regulation would make the world's advanced economies 3 percent smaller). Capital isn't the constraint on bank lending: it's their willingness to lend.
Let's look at this a little more closely. Let's say that, instead of letting the banks increase their dividends and buybacks, the Federal Reserve increased capital requirements and said that banks had to hold onto their cash to meet those higher requirements. What would happen to bank lending? Nothing. Banks wouldn't have to reduce their balance sheets because they already have the cash; they would just be not paying it out to shareholders.
The counterargument is this: banks only want to lend if their expected rate of return exceeds their cost of capital; but higher capital requirements increase the cost of capital (because equity capital is more expensive than debt capital); therefore the set of attractive lending opportunities will shrink.
But this is a fallacy, as spelled out by Admati and Hellwig in The Banker's New Clothes. According to Modigliani-Miller, capital structure doesn't affect the overall cost of capital, so retaining cash shouldn't reduce the set of attractive lending opportunities. We all know Modigliani-Miller doesn't hold in the real world, but the main reason it doesn't hold is the tax subsidy for debt. (The second-biggest reason it doesn't hold is agency costs, which dictate that more debt is bad.) In other words, to the extent that more debt lowers the cost of capital, it's due to a distorting government intervention. The lower cost of capital due to increased leverage is a social bad, not a social good.
Bank CEOs can't have it both ways. If the best use of their cash really is returning it to shareholders, then they might as well be keeping it in their accounts at the Federal Reserve. And that way we would have safer banks, and a safer financial system.