The past couple of months have been somewhat messy for the stock market. Inevitably, and particularly considering how easy and stress-free 2017 was, fear has caught quite a few investors. Google searches for “bear market” are actually at the highest levels they’ve been since 2009, and all around the web we read articles about the imminent arrival of the bear. Others, however, have called for a market bottom and predict the market to steadily rise in the foreseeable future.
Last year, on November 7th to be precise, I wrote an article claiming that the stock market was overvalued. In it, I built two statistically significant models: the first one estimates future earnings growth of the S&P500 based on the future GDP growth and the economy-wide after tax margin, while the second model estimates where the CAPE should be at any given time based on future earnings growth and the future interest environment. The conclusion was that the stock market was overvalued regardless of the approval of the Tax reform proposed by the Trump administration. Nevertheless, I stopped short of suggesting an imminent bear market and merely recommended readers to diversify or hedge their portfolios. How much easier 2018 was for those properly hedged investors?
Right now, the stock market is roughly unchanged compared to when I published my article. That is not to say that is has not moved since then. At its peak, the S&P500 price was 13% above its price when I first claimed that the market was overvalued, and has fallen 11% since.
That takes us to our current situation, where two camps have formed, in the characteristically polarized fashion of recent times. One claims the imminent arrival of the bear while the other camp claims that the market has already bottomed. However, very few investors and commentators are considering the implications of the current market situation on long-term future returns. That is what I attempt to do in this article.
Share of Equity in portfolio allocation: about the indicator
A few months ago, I came across this indicator as a predictor of future stock market returns. I ought to say that its predictive capabilities over the long term are incredible and far greater than those of any other indicator out there (except maybe Hussman's margin adjusted CAPE).
Given this, it should be of no surprise that there are quite a few articles about it out there. Nevertheless, I have yet to find an article that properly explains how the indicator is calculated, the mechanisms through which it is related to the stock market and its implication for future returns.
This indicator is usually estimated in two ways. The first one is through a survey conducted by AAII. However, this estimate fails to weight for portfolio size and, maybe more importantly, it fails to account for bank holdings that are not considered part of a portfolio (think of emergency funds, or people that don't invest their savings in anything more than a savings account).
This takes us to our second way of estimating the average equities share of American portfolios. The explanation is quite intuitive, and was first exposed by Jesse Livermore (pseudonym) in a terrific blog post. We first start with a simple generalization: an investor portfolio must be entirely made up from stocks, bonds or cash. Real life is evidently not as simple as that, but other asset classes don't take that huge of a space in the aggregate portfolio so their impact on the estimation should be insignificant.
So we only need to know three things in order to estimate the (weighted) average share of stocks in portfolios. The total amount of stocks, bonds and cash that investors are holding on aggregate.
Before explaining the basic calculation, I have to refresh to you a couple of economic concepts. First, it is important to note that the whole universe of real economic borrowers in an economy consists of four categories: households, non-financial corporations, governments (state, local and federal) and the rest of the world. Banks and other financial institutions are not real economic borrowers, but rather intermediaries. Although they do produce services, intermediation is the bulk of their operations. Therefore, the creation of net new financial assets must come from the real economic borrowers.
Secondly, when someone takes a loan, the money supply expands and a new deposit that didn't previously exist is created. Surely, someone must hold this new deposit and it affects the composition of his portfolio.
It follows then that we can calculate the total sum of bond and cash holdings in the aggregate portfolio by calculating the total liabilities of real economic borrowers.
One final assumption that we can make to simplify the calculation is that the foreign allocation of US equities and the US allocation of foreign equities balances out. Although evidently not true, as you'll later see it does not affect very much the predictive capabilities of the indicator.
Investor Allocation to Stocks (Average) = Market Value of All Stocks / (Market Value of All Stocks + Total Liabilities of All Real Economic Borrowers)
As you can see, the relationship between the indicator and forward 10 years returns from the S&P500 is as close as you get to perfect in the world of finance.
(By the Author)
(By the Author)
Why do average equities share of portfolios affect future returns?
A long time ago, when economics consisted rather of glib phrases than of real analyses, a critic of the science said, “If you want to make a first-class economist, catch a parrot and teach him to say ‘supply and demand’ in response to every question you ask him. What determines wages? Supply and demand. What determines interest? Supply and demand. What determines the distribution of wealth? Supply and demand.” In every instance the answer is right, but it explains nothing. - Irving Fisher
The parrot joke was intended as a criticism to economists of its time, and is certainly still valid in the 21st century. Fisher’s answer, however, tells us just what we need to know: it’s always supply and demand. Always. Applies to stock market prices too.
The hard part to any problem is understanding the drivers of the supply and the drivers of the demand.
With stocks, as with other financial assets, the laws of supply and demands don’t work exactly as with regular goods or services. With equities, the quantity in the supply and demand equation is not the number of shares, but rather the total value of shares (P*Q). This is because investors don’t care about how many shares they own, but rather care about the total amount of their money that they are “putting to work”. An investor that sees a 60% equity allocation in his portfolio as appropriate does not care about the amount of stocks that he owns, he only cares that stocks represent 60% of his portfolio.
So demand is not necessarily mainly driven by prices, it is mainly driven by the estimated appropriate allocation of stocks. And supply is not mainly driven by prices either, it is determined by the demand. This is because the total number of stocks is practically fixed, so changes in the demand (demand is driven by the estimated appropriate stock allocation) are the ones that determine the prices, and thus, the total value of equities (P*Q).
Under this framework, the main contributor to average stock prices is the desired proportion of investors’ portfolios of stocks. This proportion is a function of forecasted economic conditions, interest rates and psychological biases (evident in bubbles). Interest rates and economic conditions tend to move cyclically, and over the long-term, the psychological biases shouldn’t have a great impact. Because of these reasons, the equity share of portfolios should fluctuate around a mean. Therefore, when this proportion is high, over time it should correct itself and thus have a negative impact on stock prices. The opposite holds true for when this amount is too low.
Where do we stand now?
If you opened the FRED’s link to the indicator, you probably noticed that the last published data point available is from July of this year. This happens because the calculation comes from different series of the National Accounts. Given that they cover aggregate values and not merely small samples, National Accounts calculation takes several months. That gives us a lagged indicator. And a lagged indicator creates an interesting problem.
One simple way of correcting for this problem would be testing how reliable is the lagged indicator in predicting future returns from now on. When applying a 6 month lag in the indicator we still get an R-squared of .85, meaning we only lose about 5% of the model’s explaining power when we use 6 month old data.
However, since July the S&P 500 has returned -8,5%. This surely should increase future expected returns since now. To test how the how returns in the lagged period affect future expected returns using the lagged indicator, I did some simple math.
First, I totaled the forward 10-year returns indicated by our model with the lagged indicator. Think of raising the expected annualized returns to the 10th power. Then, I proceeded to subtract from it total returns in the lagged period. Lastly, I annualized again the expected total return.
As the graph below tells, after following this procedure we only lose about 2% of the original, non-lagged model predictive capabilities. This new lagged model historically explained more than 87% of the future annualized returns.
(By the Author)
When we apply the revised model to our current situation, the model tells us that we should expect annual returns of around 3.62% for the next ten years. This figure is not only way below historical norms, but also quite similar to what we can get simply investing in Treasury notes.
So, what can investors take from this model? Returns in the next decade won’t be as pretty as in the past one. At least not in the US equities market. Strategies to get around it probably include investing abroad and capitalizing opportunities to get higher returns in other markets. Bear in mind that you won’t get the accustomed 10% return, at least until market conditions change in the US, and place risk appropriately.
What this model does not tell us though, is when the next bear market will come or how bad it will be. In that sense, it is not a market-timing tool. But anyhow, investors can, and should, use this model as a tool when building their market strategies. Be it investing abroad, hedging portfolios, buying different assets classes or whatever they deem appropriate.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.