Skip to main content

Imagine you are an economist. You are tasked with modelling a market. It goes without saying that you will have to model both the demand side and the supply side, right?

Not so in financial research. In trying to understand long-term returns in the stock market, and especially the equity risk premium, loads of research papers have concentrated on the demand for stocks. Consumer preferences, utility curves, risk aversion, loss aversion, specific tax regulations, what have you.

I’m not saying that they disregard supply in the secondary market. I’m saying that, fundamentally, the secondary market is not important in deciding long-term returns. Here’s the salient market as I see it: Investors demand financial returns. With the obvious exception of short-term returns, which is not a rational basis for expected return, this is not supplied by other investors. It is supplied by the companies in which they invest.

To me, the real puzzle is why this supply side tends to be neglected. The relationship between a company and its investors does not end when a capital issue is completed. Companies keep supplying something that investors demand: profits, whether realised or potential, whether distributed or retained. Higher pricing of their stock implies a lower cost of equity.

Fundamentally, the secondary market is not important in deciding long-term returns.

If you find the following point banal, you’re with me: Stock market returns can be decomposed into two factors – how companies do, and how the market prices shares in these companies. To me, that is a logically inevitable starting point. I still can’t figure out why so many clever people neglect the company bit.

There’s a particular problem with focusing on pricing. If, say, risk aversion produces higher returns going forward, it has to be because it depresses present prices. This begs a question of recursion. If yesterday’s required returns were similary high, previous prices would have had to be depressed even further. And so it goes.

One way to circumvent, if not eliminate, this problem would be to have a very long time series. Of course, you need to express stock market pricing in a way that makes sense in theory. In practice, however, that has never been a problem. Multiples like P/E have long figured as measures of stock market pricing.

So, let’s do a very simple exercise. Let’s decompose S&P 500 returns going back to 1871, which is as long-term as it gets in terms of earnings series. Over these more than 150 years, the average compound return computes to 9.1 per cent. In real terms, adjusted for US inflation, it is a still impressive 6.9 per cent, despite this year’s downturn.

Of course, stocks are priced a lot higher than in 1871. According to economist Robert Shiller, whose data I have downloaded, average P/E has risen from 11 to well above 19 (which is far from an all-time high). If this had happened in the space of one year, it would have magnified that year’s return by almost 75 per cent.

Over more than 150 years, however, it contributes only 0.36 percentage points to the compound return. More than 96 per cent of the compound return can be attributed to rising earnings.

I’m not convinced that the shape of consumers’ utility curves was material in producing this increase.