Tuesday, January 13, 2009

Economic growth and stock returns

In the comments over at Angry Bear, Bruce Webb brings up a common argument regarding Social Security reform and personal accounts: that Social Security's projected insolvency is due to low economic growth, but if the economy grows slowly then stock returns will also be low, and so personal accounts investing in stocks can't really do anything to help Social Security or Social Security beneficiaries.

I've got some sympathy for the economic growth/stock returns argument in theory (short story: future economic growth will be lower because birth rates/labor force growth will also be lower; this can lead to capital deepening, such that the ratio of capital to labor rises; this in turn leads to lower returns on capital). Although, Kotlikoff, Smetters and Walliser make an interesting opposite argument, that rising entitlement spending due to population aging will soak up so much excess capital that stock returns are likely to rise.

Moreover, as I argue here, I think the rate of return argument pitting personal accounts against Social Security is generally wrong.

But anyway, here's some data I haven't seen around before that some might find interesting. If you do believe that economic growth and stock returns should be strongly correlated, at least over the long term, then you'd think that if you compared long-term economic growth/stock returns from a number of countries that we'd see a tight fit.

The chart below shows average annual GDP growth and average real stock returns for 16 countries in the period 1900-2000. (The equity returns are from Dimson, Marsh and Staunton; the GDP growth is from Maddison.) The regression shows there is a positive relationship on average, such that average equity returns equal 1.96 percent plus 0.92 times the country's rate of GDP growth. Short story: higher GDP growth should mean higher equity estimate.

The problem here is that the fit of the regression is pretty poor: the R-squared value is 0.1, meaning that about 10 percent of the difference in equity returns between countries is accounted for by differences in their rates of GDP growth. The rest of the differences are due to other things. So from this chart (and the discussion above) we can conclude that while the economic growth/stock returns argument has some merit, it's not the major driver of things.


Paul Lawin said...

There are certainly many factors that impact equity returns besides GDP growth. One that jumps out of this graph is war. It would be interesting to try the regression with a second independent varialbe, call it "percent of infrastructure destroyed by war". Or, maybe the correct variable would be "financial market damage due to war".

Andrew G. Biggs said...

Very good point, and I've done some regressions including a dummy variable for whether a country lost a war during the time. (It's not totally clear-cut since, say, Denmark was occupied in WW2 but not really attacked, etc.) In any case, the lost war dummy does improve the fit.

But here's the problem I found: losing a war shouldn't just affect stock returns but also GDP, since the capital stock that's destroyed both pays capital returns and produces output.

So while in generaly I'm always tempted to plug in more variables to up the R-squared, this did take me back to econometrics classes where they stressed that you need an economic rationale to include a variable, not just a statistical one.

These things can be done in greater detail, however, say by breaking GDP growth down into labor force growth and increases in GDP per capita. I can get a better fit that way. But the key point here was simply that the straight stocks/GDP relationship isn't all that straight.

Jim Glass said...

... Social Security's projected insolvency is due to low economic growth, but if the economy grows slowly then stock returns will also be low, and so personal accounts investing in stocks can't really do anything to help Social Security or Social Security beneficiaries.

What stock returns versus economic growth where?

This seems to go back to Dean Baker's argument that stock values can't consistently compound faster than GDP, and since US GDP growth is projected to fall in the future due to the decline in growth of the work force (demographic effect of the retirement of the baby boomers), stock returns must be projected to fall.

But that assumes all stock returns are earned domestically -- perhaps a fair approximation historically through, oh, the fall of the communist block ... but in the future?

As of today a quite substantial percentage (I forget the exact figure) of the profits of the S&P 500 traded on the NYSE are earned abroad. Beyond that, the ability of individuals to invest aboard directly is growing all the time.

In the early 2020s, if average US GDP growth falls to 2%, people won't be able to invest in the "China & India Growth Fund"?

Of course, payroll tax funding for SS is contstrained by domestic economic growth.

But why would world stock market returns available to investors be constraind by domestic US economic growth, as the US economy becomes an ever smaller percentage of the world economy?

Andrew G. Biggs said...

All good points. I think question is of degree: how much of domestic profits could be foreign generated? Baker, Krugman and DeLong have an interesting paper that looks at this in much greater detail. While I think they make an interesting case, one offshoot is that domestic interest rates would also fall, meaning that the trust fund would run out sooner and the long-term deficit would increase. The long-term deficit is VERY sensitive to the discount rate. So this argument, even if you accept it, goes both ways.