Monday, June 27, 2005

Is growth good for shareholders?

I love articles that turn the conventional wisdom on its head (and I held this view long prior to ristening to Freakonomics), but that said, this one surprised me. What is more, in doing looking at growth and returns, Ritter offers valuable insights into what matters in forecasting expected returns.

Jay Ritter finds that shareholder returns are negatively correlated with economic growth.
In his words:
"... does economic growth benefit stockholders? This article argues on both theoretical and empirical grounds that the answer is no. Empirically, there is a cross-sectional correlation of –0.37 for the compounded real return on equities and the compounded growth rate of real per capita GDP for 16 countries over the 1900-2002 period."

"I am not arguing that economic growth is bad. There is ample evidence that people who live in countries with higher incomes have longer life spans, lower infant mortality, etc. Real wages are higher. But although consumers and workers may benefit from economic growth, the owners of capital do not necessarily benefit."
"This article argues that limited historical data on stock returns are not a constraint, since these data are irrelevant for estimating future returns, whether in emerging markets or developed countries. This point has been made before, although possibly not as explicitly, in Fama and French (2002) and Siegel (2002), among other places. Of greater originality, this article argues that not only is the past irrelevant, but to a large extent knowledge of the future real growth rate for an economy is also irrelevant."

"I argue that only three pieces of information are needed for estimating future equity returns. The first is the current P/E ratio, although earnings must be smoothed to adjust for business cycle fluctuations. The second is the fraction of corporate profits that will be paid out to shareholders via share repurchases and dividends, rather than accruing to managers or blockholders when corporate governance problems exist. The third is the probability of catastrophic loss, i.e., the chance that “normal” profits are a biased measure of expected profits because of “default” due to hyperinflation, revolution, nuclear war, etc. This third point is the
survivorship bias issue, applied to the future."
A few other highlights:
*"I believe that the large stock price effects associated with recessions are partly due to higher risk aversion at the bottom of a recession, but also due partly to an irrational overreaction."
He argues that while some claim the high returns experienced in the US are the result of a survivorship bias, this claim is probably overstated. For this argument he points out that 1. in 1900 most people were positive on the devolped world, 2. most of the developed world economies today are the same as they were in 1900 (with a few exceptions), and 3. no convincing evidence exists to suggest that the market's risk premium (i.e. the return above the so-called risk free rate) can be expalined through a survivorship bias hypothesis. Why? because the so-called "risk free rate" may increase in response to systematic risk factors.

It is for this third point that Ritter offers one of my favorite lines from any finance paper:
* "Third, even if survivorship bias is important in explaining realized stock returns, it is not clear that survivorship bias should affect the equity risk premium, since bond and T-bill investors suffer just as much from devastation as equity holders do. In the 1950s and 1960s, when it might plausibly be argued that a nuclear war involving the U.S. and the Soviet Union had a significant probability of occurring in future decades, real rates of interest on U.S. government bonds were very low, suggesting that investors had no significant desire for immediate consumption before they were incinerated." (italics are mine)
So why are returns negatively correlated with economic performance?
* "Why is there a negative correlation between real returns and real per capita income growth? ....Siegel (1998) hypothesizes, that part of the negative correlation between real stock returns and per capita GDP growth is because high growth was impounded into prices at the start of the period."
*"optimistic investors will bid up stock prices, lowering the dividend yield."
*"One reason that GDP growth does not necessarily translate into high returns for minority stockholders is that managers may expropriate profits via sweetheart deals, tunneling, etc."
So what does matter? Ritter concludes with what many may deem the most important part of the paper:
"What...does predict future equity returns? The answer is simple: the current
earnings yield. The major adjustment that needs to be made is to smooth earnings for the effects of business cycles. Economic growth doesn’t matter. As a first approximation, the return on existing shares will equal the earnings yield on these shares, subject to the caveats that expropriation by insiders or catastrophic market meltdowns will prevent minority shareholders from receiving future earnings."

If the earnings yield is the real cost of equity capital, does this mean that the textbooks are wrong to say that the E/P ratio of a company is not its cost of equity capital? The answer is yes and no. A company can rationally be expected to earn above- or below-normal ROE for a period of time (economic profits), so in general it is incorrect to state that a firm’s cost of equity capital is its earnings yield. But for the market as a whole, above- and below-normal rates of profit growth largely cancel out, so in fact the market’s smoothed earnings yield is the expected real return on the market."

An definite I^3 paper. Well worth your time!

(for the uninitiated: I^3 = Interesting, Informative, Important)

Ritter, Jay R., "Economic Growth and Equity Returns" (November 1, 2004).
Although I question the date, as it says it was last edited on June 24, 2005.


j said...

Might seem off-topic, but it goes back to a core skew in economic measures that never ceases to make me shake my head.

The seam is that the presumption is that the average *is* the reality, but that presumes a reasonably bell-curve shaped distribution, and bell-curves are very very rare in nature/the-real-world.

So mean-per-capita-income of $N can have a rasher of different meanings -- communist gulag, plutocracy, flattened scandinavian middle-ness, etc. For example, on percapita GDP, El Salvador appears like a better economy than Lesotho, but I strongly suspect if one compared median instead of mean, they'd swap rankings. And median is just better, still not clear enough.

The point worth thinking about in all this is I think there's a flaw in choice of measures. One can always get a surprising effect if one chooses the wrong measures to try to judge the reality.

j said...

Erk...substitute Laos for Lesotho.

FinanceProfessor said...

Agreed! In fact, I just searched the paper and no mention of medians at all. I would wager a reviewer will demand it.

Thanks for comment!