A Less Coherent Post Than I'd Like About The Purpose Of Banking

Earlier this week I attended a conference hosted by the Roosevelt Institute about financial reform on the theme of: will it work? how will we know?

There were a variety of very short presentations – synopses, really, of papers included in the packets handed out to attendees – on a variety of topics including data collection, how to clear derivatives contracts, the likely impact of Basel III, the future of the ratings agencies (if any), and so forth. And the overall tone of the whole event was: we’ve fixed some things, we haven’t fixed everything, we need to maintain the political will to stay vigilant, and we need to keep Congress in the hands of the Democrats.

It was all interesting and I did learn a thing or two. But what kept nagging at me was: what do the architects of financial reform think banks are for?

That’s not a sarcastic question. The classic function of a bank is to turn savings into capital. They borrow from the public in the form of insured deposits. They then deploy this capital in the form of loans of various kinds. Their job, in other words, was to evaluate and hold risk – the risk of those loans. And for taking that risk, they earned a return.

But in the world we actually live in, banks have labored to make themselves appear to be service businesses that earn fees rather than risk-taking businesses. Indeed, risk is a bad word – and remains a bad word with the financial reformers. The new job of banks, say both the banks themselves and their regulators, is to be financial intermediaries. They don’t lend money against a house as collateral; they intermediate between a mortgage borrower and an investor in a mortgage-backed security. They don’t lend money to a business; they intermediate between a company looking to borrow money and an investor in a collateralized loan security. They don’t even take deposits; rather, the intermediate between short-term corporate borrowers and investors looking for near-cash instruments. And of course they intermediate between the various participants, hedgers and speculators, in the wide variety of over-the-counter derivatives markets.

True intermediaries, though, don’t need a lot of capital. They don’t even need a whole lot of funding. The old investment banks were intermediaries of this sort, and they didn’t retain big positions. They only needed enough money to do their next underwriting and maintain an orderly market in the underwritten security thereafter. They certainly didn’t need deposits – or deposit insurance. But the banks need a whole lot of money. They have enormous balance sheets. That’s because while they are intermediaries, they are not pure intermediaries. They’re intermediaries who retain positions. Specifically, the residual risk position created by millions if not billions of imperfectly matched trades: counterparty credit risks, dynamic hedging risks, legal risk on securitizations, and so forth.

There are, of course, rules to ensure that some capital is being held against these residual risks. But the capital rules are designed to be efficient – to require more capital where there is more risk, less capital where there is less. Where the residual risk appears to be negligible, the capital required to hold against it – even if the nominal position is in the hundreds of billions – is negligible as well.

Deposit insurance was created to prevent runs on banks, so that you wouldn’t have a pro-cyclical withdrawal of capital from the system at just the point that loan losses were increasing. Fed regulation was imposed to mitigate the obvious moral hazard created by deposit insurance. But if banks are not fundamentally in the business of lending money and earning a return for risk, then what business is being underwritten by that provision of insurance?

The business being underwritten is, in a fundamental sense, the business of modeling financial risk – the enormous pile of residual risk associated with intermediating-but-not-really. But, as we saw in the most recent crisis, modeling financial risk is, well, basically impossible. That is to say: you can model the world of normal financial behavior just fine, and you can definitely improve your risk-adjusted returns if you do it well. Fundamental changes in the state of the world, though, you can’t really model at all.

Citigroup, after all, was not bankrupted (or brught to the point of needing government assistance to avoid bankruptcy) because it made bad loans. Citigroup was bankrupted because it retained the super-senior (AAA rated) portion of risk associated with CDOs of subprime mortgage loan securitizations it had underwritten. It treated this risk as essentially nonexistent – sensibly enough given that its own models and those of the independent ratings agencies said there was none.

Goldman Sachs was not bankrupted (or brought to the point of needing government assistance to avoid bankruptcy) because it made bad loans. Goldman wasn’t even bankrupted because it happily retained the super-senior (AAA rated) portion of risk associated with CDOs of subprime mortgage loan securitizations it had underwritten. They were too clever to believe the models that said this risk was good. Goldman was bankrupted because while it retained that risk, it hedged the risk with a AAA rated counterparty: AIG FP, a subsidiary of the country’s largest insurer with a century-long track record of performance. What’s the residual risk on a AAA security hedged with a AAA counterparty? Enough to bankrupt Goldman Sachs, apparently.

What we have now is a financial system that is leveraged to risk modeling to an enormous degree, and what we’re basically doing is doubling down on that bet right after it failed in the most massive fashion. We are trying to “get the risk right” and to stop banks from taking overly risky bets in the future. But you can’t get the risk “right” – not in the only sense that matters, namely, when the next crisis will hit and how a particular book will behave when it does.

It seems to me that if we’re trying to avoid similar crises in the future, what we need to incentivize is not the elimination of risk but rather the simplification of the banks’ books. We don’t need to increase the capital charges on low-rated bonds and loans – we need to increase the capital charges on high-rated bonds with embedded derivatives, on derivatives counterparty risk, and so forth. “Proprietary trading” in the sense of highly leveraged betting on small differences in the value of similar securities or baskets thereof will migrate out of banks naturally, because it won’t be economical if the capital charges for such strategies are higher, as they would be. Derivatives businesses will either shrink or will migrate to exchanges where they will clear through a clearing corporation – which, of course, doesn’t eliminate counterparty credit risk but does massively increase transparency and provide clear policy levers for trying to keep risk from getting out of hand. Securitizations will either continue (if they make economic sense once the apparently riskless top piece cannot be leveraged infinitely to generate an adequate return) or not (if they don’t). Banks will be able to do true financial intermediation (of the sort that investment banks used to do), where at the end of the transaction the book is effectively liquidated, and there is truly no more risk. But intermediation-but-not-really as a business strategy will be too expensive in terms of capital to be profitable.

If the incentives are strong to eliminate residual risk that is impossible to measure, it will be much harder for banks to make money by pushing risk out into the tails of the distribution. They’ll have to go back to making money the old fashioned way: by lending it to the real economy. Which will mean taking risk. But it’s risk that we can see. And it’s risk that produces some social benefit, by actually turning deposits into capital. Which is what the banks exist to do in the first place.