Goldman Is Looking Forward To Making Less Money On Standardized Derivatives So It Can Make Much More Money On Non-Standardized DerivativesBy Matt Levine
How can you not love listening to Lloyd Blankfein? He spoke at this Merrill conference this morning and here are the slides, whatever; he is not a PowerPoint presenter, he is a philosopher. Let’s talk Philosophy of Lloyd.1
Lloyd and I share a number of passions, I assert without evidence, but a quirky one is that we both enjoy idly speculating about conservation laws in finance. In the Q&A Lloyd was asked about the clearing of derivatives and their movement to central counterparties, in which instead of banks trading opaque over-the-counter products with each other and their clients directly, taking client and fellow-bank counterparty risks, derivatives will increasingly trade in standardized cleared form with public pricing and central clearinghouse credit risk. Lloyd began by hypothesizing a “physical law of conservation of risk,” noting that “the things you do to reduce the risk of a 20-year-storm” – like reducing credit exposure to a bunch of different shaky counterparties in derivatives markets – “might make worse the risk of a 50-year storm,” like the credit exposure to one central counterparty whose failure could bring everyone down. I’m with you Lloyd: endorsed; narrow the distribution and fatten the tails etc.
The other concern about the move to clearing is that lots of people expect standardization and clearing will reduce the spreads that banks can charge on cleared things (mostly interest-rate swaps, some miscellany). Coincidentally a while back I speculated about a sort of law of conservation of abstraction, in which the abstract fantasies of our global financial system are built on the mucky underpinnings of a basement at DTCC full of damp stock certificates. That was … that was dumb, I don’t know what I was thinking.2 It’s obviously false. It’s the other way around actually: the more the inputs of the financial services industry abstract away from human activity, the more the outputs can move even higher up the abstraction chain. You see this with DTCC itself, which was seeking efficiency by dematerializing stock certificates even before a hurricane did that job for it, or in that Journal article last week about how Belgium is dematerializing its stock certificates and everyone’s all sad about it but the march of progress waits for no paper-hoarding Belgian.
You can see it especially Lloyd’s take on whether increased clearing of interest-rate swaps, etc., will help or hurt Goldman Sachs’s business. His answer was unambiguously bullish; he told a little story about how cash foreign exchange used to trade at 5-10 pip margins and now basically trades at zero, and how that helped liquidity providers like Goldman. And the reason is that it “made possible much more tailored and useful products than when liquidity was more expensive on the cash side. It made possible the creation of real hedges that are much more complex and more valuable,”3 giving the example of merger FX hedges where the acquirer swaps a future merger-related payment into the target currency but with an automatic unwind if the deal doesn’t close.
Your model here could be: the more things Goldman can trade without even thinking about it, the more value it can extract from thinking about them. If it’s expensive to trade spot FX, then you trade spot FX and charge a tiny spread. If it’s essentially free to trade spot FX, then you construct bespoke state-dependent FX derivatives that rely on free trading of spot FX, and you charge 2%. As the low-value-add products go to zero-value-add, you can spend more time and add more value with the high-value-add. If your fancy exotic derivatives only work in complete markets, the more complete markets there are, the happier you’ll be building derivatives. If you take as a given that your building blocks are super-liquid cleared low-margin things, then the addition of more abstract building blocks lets you build ever more abstract – and profitable! – structures with them.
For some reason Lloyd reminded me of this Felix Salmon post yesterday about, like, improved target-date retirement funds. Here is an innocuous-looking passage:
The idea behind these funds is that they diversify according to risk, rather than according to asset class: they take a bunch of different asset classes, disaggregate the various risks, and then work from there. For instance, if you want exposure to commodities or real estate, and buy stock in mining or housing companies, then you’re getting equity exposure at the same time.
Slow down man! That is so blasé and yet amazing. A diversification rule that is like “buy 70% stocks and 30% bonds and wait” is dumb, yes, but also easy enough for any schlub to follow at home. A diversification rule that is like “have X exposure to macroeconomic risk 1 and Y exposure to macroeconomic risk 2 and Z exposure to etc.” makes more sense in a perfect world: you don’t care about a thing called “stocks”; you care about things like real estate or inflation or the rise of China or whatever. But it’s vastly harder to do, in part because it requires you to figure out what risks you care about, in part because it requires you to math out a whole series of correlations and assumptions4 to draw lines between “instruments” and “risks,” and in part because it requires you to have easy and liquid access to all those instruments in order to get your risks right.
The fact that someone is offering a retail product that attempts to do that is really impressive and/or sort of alarming, depending on how it does between now and when you retire. And it’s possible essentially because there’s an increasing ability to trade “risks,” instead of just S&P 500 stocks, cheaply and liquidly. Sort of like what Lloyd is predicting for cleared derivatives.
There are reasons for concern, of course, if you like that sort of thing. I’m sure that the financial regulators who are pushing standardized derivatives toward clearinghouses are not really doing so because they dream of banks one day creating far more complicated second-order derivatives out of the now easy-to-trade standardized ones. If you think that simplification and clearing reduce risk, then you’ll be sad to hear that Goldman plans to use those features to facilitate complexification and new OTC businesses.5
Of course I’m with Lloyd – and Felix – here, but then I would be: I think that each move up the chain in financial-system abstraction, from “you can buy 100 shares of Apple stock” to “you can have $1,000 worth of pure abstract exposure to Chinese consumer-technology demand,” from “you can buy an interest-rate swap” to “you can buy a monster built out of other, smaller interest-rate swaps,” is just sort of cool. If the financial system is about adding value through abstraction, then more abstraction creates more value. And why shouldn’t Goldman take a cut of it?
1. Ooh but before we do, let’s talk about another thing, which is that Lloyd talked about how Goldman is reducing risk-weighted assets by basically charging every desk for capital based on the risk weights of everything they own: charging desks is a much faster way to bring down RWAs, says Lloyd, than “fiat or instruction.” This is … so obvious? Yet sort of fascinating? Like, what is a better institutional design: having a bunch of traders and risk managers make independent decisions about risk management and then making a bank aggregate them all up and hold capital against them, or having every individual trader make his or her risk decisions based in large part on the same Basel 2.5/3/whatever risk-weightings? I have a weak sense that this could end in tears. Discuss.
2. In my defense, it was in a footnote, whatever.
4. To take a simple example: how much exposure to oil prices do you get from buying an E&P company? Like, a lot, generally: higher oil prices = higher stock prices for companies who find oil. Unless they hedge. Or decide to start hedging. Or decide to stop hedging. Or their rig blows up, pushing oil prices higher while also exposing them to massive potential liability. These things are irreducibly uncertain though also susceptible to, y’know, statistics.
5. Similarly with them fancy target-date funds. If you buy a fund that is 70% stocks and 30% bonds, and stocks drop, you get it: your stocks dropped so you lost money. You didn’t know beforehand what would cause stocks to drop, or what risks you were exposed to, but then, you didn’t think you did. If you buy a fund that is 20% oil-price risk and then oil prices go up because a rig explodes but your fund goes down because its oil-price risk came from owning stock in the rig operator, you will be … confused. Confusion and systemic risk are historically correlated.