Return On Equity Versus Return On Assets

The Economist’s Buttonwood has a fascinating item based on a recent speech by (PDF) Andrew Haldane. It’s about accounting. Really. But it’s much less boring than it sounds.

His point is that there are two different ways a bank can fund itself. One is that it can raise equity. The other is that it can employ debt. The choice between equity finance and debt finance isn’t important in terms of what you can do. But it does have important corporate governance implications. That’s because the way the financial system works in practice, people who lend money to banks (including both formally insured depositors and formally non-insured bondholders) get bailed out when banks fail. People who held equity in banks, even ones that got bailed out, actually did end up losing money. So it’s equity holder who have some kind of incentive to focus on whether or not executives are actually doing a reasonable job. But as Buttonwood observes “investors are not long-term holders any more. The average holding period for US and UK bank shares fell from 3 years in 1998 to around three months by 2008.” In other words, nobody’s minding the store. What happens instead, Haldane argues, is that bank executive lever up to produce apparent gains in return on equity just by obtaining non-equity finance:

Imagine that in 1990 bank CEO pay had been indexed to bank ROE. By 2007, CEO compensation would have reached $26 million. That is precisely in line with their actual payouts. If you believed ROE were a reliable performance metric, US bank CEOs would have had a watertight defence back in 2007. Instead, of course, we had ludicrously leveraged banks that were too big to fail and brought the economy down with them. […] Imagine if the CEOs of the seven largest US banks had in 1989 agreed to index their salaries not to ROE, but to ROA. By 2007, their compensation would not have grown tenfold. Instead, it would have risen from $2.8 million to $3.4 million. Rather than rising to 500 times median household income, it would have fallen to around 68 times.

Remember the point here is to look at bank performance before the crash. It turns out that if you judged executives pre-crash based on ROE, they would look like geniuses. But had we been judging them by ROA instead their performance would have been wildly less impressive. With the benefit of hindsight, we now know that the ROE-induced view that banks were well-managed was completely mistaken. Naturally, the response is that we still use it as the main metric.