(This is the translation of a previous entry).
In this post we are going to analyse the historical evidence between stock market returns and economic growth from a long-term perspective. As you read, you will see that some results are a bit counterintuitive. The following example shows why.
Imagine that in 1993 someone asks you what country you would like to invest your savings for the next 20 years. The gist of the game is that that “someone” knows for sure (a gift out of reach of mortals, not to mention economists) the future GDP growth rates for the next two decades and you have access to this invaluable data for your investment decision. The reason for using this data should be obvious: one should expect higher business profits and better stock valuations if the growth in revenues is high (a financial analyst would say that the future cash-flows to be discounted are higher).
Well, there is no need in having the GDP growth data at hand to know that for the period 1993-2013 China was the winner, with growth rates (in real terms) around 9%-10% – although if you want to check the data, follow this link.
It turns out that if you had followed the advice and you had bet on the Chinese stock market, the investment return in real terms would have been -3.8%, according to the Credit Suisse Global Investment Returns Yearbook 2014. Some knowledgeable investor of the Chinese market will object me that I am gaming the example, given that at the end of the 90s China had to undertake the huge reestructuring process of its 4 biggest banks (see this book for the details of the story and this a-bit-dated paper for a summary), and that that somehow provoked some general distrust among international investors (it is worth pointing out that the banks were not listed yet at the time). In any case, and being generous, if you had decided to invest from 2000 to 2013, the returns would have been around 2.6%, somehow better, but low at any rate for such high economic growth rates.
Although in the last few years several economic commentators have tried to explain the poor Chinese performance based on the “microeconomic” idiosyncrasy of Chinese people (e.g. that the Chinese investor is irrational, that the local investors are better informed than the foreign ones, that the local enterprises distort the numbers, etc.), the truth is that the Chinese case, even though counterintuitive, is not something exceptional from a historical point of view. The following graph shows the relationship between economic growth (measured as GDP per capita growth) and stock returns (measured as a traditional yield, dividends plus capital gains) in the 20th century for a group of countries:
You can see that, in general, countries with higher growth rates have obtained lower shareholder returns. Even if we are conservative at conceding the historical different experience for every country (as we should be, given that some Central European countries suffered two world wars, and so the destruction of many of their assets) and the long-time frame of the analysis, for us it is enough to say that the expected relationship “high GDP growth-high shareholder returns” is, to say the least, murky in the long-run. Not surprisingly, such (lack of) relationship has been dubbed in the literature as the “growth puzzle”.
The culprit is the q
Although we believe that a comprehensive explanation of such phenomenon is hard, we can give some clues that we think might have played some role. In a paper published in 2002, R. Grinold y K. Kronner showed through an analysis applied to the US case that the equity yield could be decomposed in the following way:
Shareholder return = Dividend yield + Earnings growth – Dilution effect + change in the valuation ratio
The dividend yield and the earnings growth are well-understood concepts for stock investors. The dilution effect is negative and measures the loss of purchasing power for shareholders mainly due to stock issuance (it is worth remembering that such negative effect was quite important up to the 80s in the US; since then, share buybacks have made that the dilution effect becomes positive). Finally, the change in the valuation ratio is, for instance, the change from buying at 15x price-earnings ratio (PER) and selling at 17x PER. Summing up, in order to understand how GDP growth affects shareholder returns at the macroeconomic level, we have to understand, first of all, how such growth affects the components of the previous equation, and then, we have to assess whether the overall effect is positive or negative – at this point, the reader may be wondering that this is a daunting task, and rightly so.
Today we will deal with the impact of GDP growth on the last component of the equation, the change in the valuation ratio. Although the relevant ratio in the previous equation is the PER, we will use for convenience the historical evidence of a different and widely used ratio, Tobin’s q – it seems that Tobin didn’t find a better letter to choose when writing his seminal paper. Tobin’s q is simply the ratio of the enterprise value at market prices to the enterprise value at replacement cost (the replacement cost measures how much would cost to buy the assets of a firm or to start up a new enterprise from scratch at current cost prices). Economic theory predicts that q should be around 1, since values different from 1 would imply arbitrage opportunities: for instance, a value higher than 1 would suggest that it would be profitable to start up a new company at current costs and then sell it in the stock market. However, the historical evidence not only shows that at the macroeconomic level (for the corporate sector as a whole) q has not been equal to 1, but that moreover it has displayed an upward trend in the last few decades. The following graph shows the evolution of q in some developed economies in the last forty years:
Beyond the fact that q has been consistently less than 1, the most startling thing has been that such increase in stock valuations has coincided with an era of lower growth in the developed economies. For example, despite of its “lost decade”, the q in Japan was higher in the late 90s than in the early 70s. Anglosaxon countries (US and UK) are perhaps extreme cases of changes in q when growth rates go down – and these cases also hint that there must have been other factors, other than low growth, that have additionally affected the evolution of q.
Said that, what is the reason for the apparent inverse relationship between stock valuations and growth rates? We think that it is due to the fact that when the volume of investment in real assets is reduced, total savings are not reduced automatically in the same amount. In the economics textbooks such problem should never happen, because investment and savings will be brought back to equilibrium through short-term interest rates (furthermore, in the long-run is usually supposed that investment is not independent from savings). However, investment and savings have always to be equal by definition. We think that such equilibrium is reached in “real economies” through savings channelled to financial products, pushing up thus the price of these assets. Firms’ investment decisions are thus responsible to a large extent for the stock valuations in the long-run – and quite independent from what people really wish to save.
In a future entry we will see why this negative relationship between stock valuations and GDP growth rates is important for several of the debates conducted by Wall Street financial analysts. For the time being, it will be interesting to track in the next few years the Chinese stock market valuations.