Market Valuation Varies in Lock-Step with Equity Allocation
Constant of motion reflects market microstructure
Robert Armstrong, the FT finance editor, debunks the idea that the Fed’s asset purchases have been driving stock market valuations. Specifically, he debunks the “famous bond manager,” Jeff Gundlach’s tall tale:
It’s like, if you take the capitalisation for the S&P 500, and you divide it by the Fed balance sheet, it looks a lot like a constant.
Armstrong checks the claim against the data and dispatches the idea:
That relationship is not constant, even just since last year. The two lines move in the same direction, but at varying rates. The market went up very quickly with the first huge round of asset buying, but it has kept going up just about as fast, even as the Fed’s buying has slowed.
Armstrong’s takeaway is that ‘Stock prices are not bound by the laws of matter, but by the uncertainties of psychology.’
Not so fast.
Just because Gundlach doesn’t know what’s up doesn’t mean that stock market valuations are driven by investor psychology; that there are no constants of motion, so to speak.
We’ve shown before that the proportion of their assets that investors allocate to risk assets, equities in particular, are not only strongly correlated with contemporary stock market valuation but in fact predict 1-qtr forward market valuation. This is an important structural feature that famous investment managers need to understand. What explains the diachronic variation in stock market valuation in the slow-moving rebalancing of institutional investors. When they move in to risk assets, the price of risk falls; when the move out, it rises. Put simply, how expensive risk assets are depends on how much money is chasing them. This is an empirical law — almost a mechanical fact about stock market valuation.
Here we use Shiller’s Total Return Cyclically-Adjusted Price to Earnings (TR-CAPE) ratio as our stock market valuation metric. We use the share of financial assets held in equities by US investors as our measure of Equity Allocation. The first figure displays the dramatic covariation in the two level series.
Given the strength of the cyclical components, we should be detrending the series. Here we detrend the two series by deducting the trailing 12-qtr moving average from the original series. We lag Equity Valuation by 1 qtr. The two move in lock-step. The predictive correlation is r = 0.87, meaning that Equity Allocation predicts three-quarters of the variation in quarterly shocks to stock market valuation. If this is not an empirical law, I don’t know what is.
The next figure displays the scatter plot for the same. The slope coefficient is close to unity (b = 0.96) and extremely significant (t = 29.0).
The relationship is as robust as you will find anywhere in finance. For all practical purposes, it is as good as a constant of motion.
Sound like Hyman Minsky made manifest.
Could Armstrong possibly believe that if the Fed were to start shrinking it's balance sheet stocks wouldn't react? Seems disingenuous.