We’re geeking out today.
Assets are priced differently and we don’t know why. Treasuries have different return characteristics than stocks, for example, and these characteristics change over time, even if the risks do not. Economists have a fancy term for this (of course): the stochastic discount factor. But what is it? CAPM says it is β, the correlation to overall market return. There’s a theory out there that the key is the “covariance between the asset’s return and aggregate consumption”*, which is one that I hadn’t heard of before.
This article points us to this paper that offers a new explanation: the liquidity needs of broker dealers, the firms that actually conduct trades.
These firms need to hold capital because their core activity, market making, is risky. But they also lever up that capital, naturally, and sometimes by a lot. There’s a cycle to how much leverage they hold, in some years leverage is high and in others it’s low (see the first link above for a graph). This somewhat coincides with the business cycle, but not as much as you might think.
Assets that only make money in high leverage cycles are worth less to broker dealers than assets less sensitive to leverage availability. When a leverage crunch comes along, survival depends on being able to liquidate at book value. So, for example, dealers have more demand for treasuries than B grade debt or equities, for example.
This demand from broker dealers means that assets that they want are going to have a lower return: dealer demand dominates the market. Assets they don’t want need a higher return to compensate for the lower demand. The relative return of these assets are going to vary with the leverage cycle. When leverage is abundant their prices soar. When leverage is scarce, they collapse.
It’s a pretty simple explanation and quite a good one, I think. What interesting about it is that it isn’t clear to me that there is any identifiable process that drives the leverage cycle, so forward information about sudden contractions and expansions of leverage can’t be easily priced into these assets. In other words, if we can’t predict when the leverage cycle reverses this insight won’t get priced in.
So if you could predict when the leverage cycle is going to switch, you could make a LOT of money.
*I might be missing something, but from a data standpoint, this idea strikes me as… inadequate. Under the expenditure approach of GDP calculation aggregate consumption is calculated by household survey and extrapolated to the economy as a whole. It’s hardly what I’d call hard data. And that’s for countries that have reliable household surveys.