On discretion in quant models
I was listening to an episode of the Excess Returns podcast (great show BTW, well worth a listen), titled “Why eliminating discretion from quantitative models is harder than you think”, and I was struck by one of their claims.
In the talk, one of the hosts makes the claim that discretion extends to the types of factors that a quant would throw into their models. So far, so good.
The next claim however was fascinating. He goes on to say that if one looked at the last 10 years of market history, they would stuff their models with growth factors instead of value factors, despite the fact that over the long haul value has outperformed growth as a strategy. He attributes this to recency bias.
Let’s unpack this series of claims. First, whether quants pack their models with one set of factors over the other can indeed be described as discretion. But is it really recency bias, or a question of the quant making a judgment on the kind of market regime in play at this point in the market’s history? My assumption here (as always) is that market regimes drive equity performance, switching back and forth on the basis of both endogenous shocks (Minsky moments transmitting across time and asset classes), and exogenous shocks (the COVID pandemic). Discretion may apply in how one defines regimes, but recency bias is really a cloak applied to the idea of reacting to market dynamics as they play out in real time.
Which brings me to the larger point: the claim that value has outperformed growth over the “long term” has been repeated often, in the media as well as among market participants, to varying degrees of passion. What I have rarely seen (and this is an understatement) is the role folks attribute to how markets have evolved over time, and how people have evolved in the way they think about markets too.
My take: Homo economicus did not evolve sui generis, but was willed into existence by the sheer force of persuasion employed by legions of economists and political forces (h/t Milton Friedman, Hayek, von Mises, Cochrane). The middle class at large was trained to imbibe the spirit of an ownership society - there was nothing more patriotic than to own stock. As the stampeding herds of Wall Street brokers train their clients to financialize every relationship, it’s no wonder that stock market participation is at an all time high, and when markets swoon, the economy collapses. The tail wags the dog every few years, vigorously and with abandon.
Circling back, the sociology of markets matters: when broad stock market participation was limited to a select few speculators and staid mutual funds, there was an implicit assumption that markets would keep going up, and it would (mostly) pay to buy good companies for cheap, as a rising tide would lift all boats, even cheap dinghies. However (and this is important), the assumption fails to take into account how markets adapted to the following (not a comprehensive list, just events off the top of my head):
- The commodity bubble of the 1970s
- Stagflation, rendering moot the assumption that the US economy will grow forever
- Nixon free-floating the US dollar
- The invention of options markets by reverse-engineering and standardizing warrants
- The creation of futures for stock indices (there were controversies when Chicago and NY were economically tied together via Leo Melamed’s sheer ambition)
- The rise of portfolio insurance and the ensuing Black Monday of 1987, and the creation of the volatility smile
- The rise of financial contagion due to the cross-holding of assets, ably demonstrated during the Asian crisis of 1997
- The collapse of LTCM and recognizing liquidity as a risk factor
- The dot com bubble and the collapse of traditional valuation techniques
- The housing bubble, the GFC, and the destruction of the concept of real estate as an infinitely malleable asset class
- The sovereign debt crisis and the hunt for yield
- The rise of volatility as an asset class
- The collapse of transaction fees to zero, and the growth of fractional shares purchases as a brokerage offering
Each of these events set off feedback loops, modifying investor behavior and expanding their set of
rational satisficing choices. Throw in a war or two (as had happened at the beginning of the previous pandemic), and macroeconomics slowly disintegrates into a collection of exceptions and edge cases, a grand narrative of human economic systems at best, and far less of a testable, falsiable science.
TLDR: There is no grand unified theory of the markets, no universal law mandating that value factors will beat growth over the long term. It would do well for everyone to always keep this in mind.