Losing my religion
No one who is even remotely interested in financial markets and investments has managed to avoid the discussion about “the bubble of everything” over the past decade. We have diced and sliced that very theme in this blog on a number of occasions. Leaving the question of the bubbliness of financial markets somewhat to the side for a moment, my view is that we do best to keep in mind that almost all policy remedies that central banks have administered over the past years, in particular the unconventional ones, have had the purpose of making us (feel) rich.
Hence, it is no wonder that one might feel asset valuation, not to mention asset allocation, to be a tricky business these days. And while bond valuation is almost purely a function of what you believe will happen with inflation and real interest rates, the equity market sometimes seems to be able to completely ignore the fundamental shackles that we economists think should tie valuations down. In previous blogs on this topic, I have looked at a well-known set of indicators that has historically been quite good at guiding long-term investment decisions; Tobin’s Q, the cyclically adjusted Shiller P/E-ratio and variants of the Buffet indicator (market capitalization divided by GDP (or a similar measure)).
Despite all measures being comfortably past the peaks from before the latest recession, most of them still have some way before reaching territories of the (admittedly) overly exuberant dotcom-era. And even so, given that high valuations have been a stated goal of policy makers, can’t they reverse-engineer our way back to normal P/E’s without that nasty drop in the numerator?
If history is anything to go by, current P/E-ratios have been followed by a total return (including dividends) of around 6% (p.a.), for five years, but there’s a confidence band of +/- 11 percentage points (p.p.) around those forecasts. It gets even worse when you look at the shorter perspective when the confidence interval is +/- 25 p.p.!
– That said, I do believe that an orderly realignment of asset valuations is exactly what central banks have in mind, yes. The problem thus far has been that every time they try, it begets an unwelcome drop in real activity which is really a no-go given the problems they are having with producing sufficient (demand-led) inflation.
Despite all these charts and calculations, we are still nowhere closer finding “an indicator that works”, in the sense that it can give us some sense of direction and – importantly – timing. Talking to a friend about it the other week, who also happens to be a stock equilibrist, she gave me an idea – why don’t you look directly at data on investor allocations instead? Her firm uses it extensively and she feels it works quite well.
Let’s begin exploring the database to see what we can come up with! I know there are all kinds of (very expensive) private datasets and surveys where investors reveal what they think about the outlook for different markets. For that exact purpose, I do think we have some useful data in our add-on databases, but I believe you are best served if we try to use what’s available in our core database.
On the one hand, it is quite easy. We only need to find the total value of all cash, bonds, and equity that investors hold. On the other hand, and come to think of it, maybe not so easy. We will need to subtract holdings by monetary financial institutions such as the FED and banks as their allocation, by and large, can be said to be endogenous. The same could be said about depository institutions like banks. To make it more straightforward, I guess it is easier to look at the five non-financial sectors in Federal Reserve’s Flow-of-Funds data: (1) Households; (2) Non-financial corporations (NFC:s); (3) Federal government; (4) State & Local government, and; (5) “Rest-of-World” (RoW). And this last “sector” creates a complication as the statistical definitions exclude US holdings of stocks and bonds denominated in foreign currencies (which we would, ideally, like to include). To some extent, of course, the same can be said about foreign holdings of US denominated stocks and bonds, which should balance out some of the issues related to that problem. However, and especially given the reserve status of the USD (and the skewed allocations that might produce), this is admittedly something to, at least, keep in mind.
Macrobond moment: Finding data in the Federal Reserve’s Flow-of-Funds data and similar vast data sets are in most cases an absolute nightmare. But thanks to the continuing adding of structure and indentations following, in this case, the FoF-release, it’s now a breeze in Macrobond.
As can be seen, this measure starts to drop some time before any recession, but there are – unfortunately – a number of false positives. What is more informative is that it is currently on very high levels with only a couple of historical precedents (the late 1960’s and the dot-com bubble in the late 1990’s). – Clearly, this will take some more work to become digestible.
Let’s start by discussing the obvious: What is the supply of assets? – Well, the supply of bonds increases in a mutually dependent process with economic growth where economic agents issue debt/bonds to fund productive investments.
Historically, this number has been anywhere between 2½% and 17½% y/y, and even though it is currently above its absolute lows in the wake of the global financial crisis (GFC), it is till low from a historical perspective. That said, an investor wanting to keep a constant portfolio allocation to equities will obviously need to see an equivalent growth in new equity or face a constant relative price increase of equity vs bonds.
Indeed, when studying the supply of new equity, it stands clear that – especially since the beginning of the 1980’s – the corporate sector has decreased rather than increased the supply of new equity (mainly through buyback and acquisitions). Hence, if investors want to keep a relatively constant allocation of their portfolio in equity, they will – almost arithmetically – contribute to the price return.
Obviously, the simple observations above put investor psychology at the forefront as total return from equity is reduced to:
- The price return from changes in investors choice of equity allocation, adjusted for;
- The price return from the net supply of bonds (minus new equities), plus;
- The dividend return.
And where ‘1’ is by far the largest contributor to total returns.
This exercise should have clearly demonstrated why it is much better to look directly at investor allocations to find our fundamental investment cues. Admittedly, the variability of this indicator is as bad over the short term as the 12m forward P/E we looked at above; the forecast interval is a stunning +/- 25 p.p. However, when we look at the variability on a 5-yr horizon – where a lot of the noise is averaged out – it shrinks to +/- 9 p.p,. and it works better than any other indicator (incl CAPE, Tobin’s Q etc.) I have tested on even longer horizons and out of sample (as you are free to try for yourself if you have Macrobond and download the documents below).
Macrobond moment: These kinds of data exercises are really gratifying when using Macrobond. I know that when the new (Q4 2019) Flow-of-funds are released (12 March) I can just hit that refresh button (even from Powerpoint) and instantly start telling the continuation of the story above.– An analyst’s life is sweet when using Macrobond!