Tuesday, August 4, 2009 1 comments

A rough estimate of long-term stock market returns

Reader MrParkerBohn wrote a thought-provoking comment touching on expected future returns for the stock market. I thought it was a topic that everyone should think about so let me reproduce the discussion in this post.

Before discussing this, note that we are just trying to come up with a rough estimate of the future. The actual numbers depend on what actually transpires in the future; what stock market you are looking at; tax rates, inflation, etc; and so on. One should just treat the estimates as applying to developed country (say USA or Canada). One should not put too much faith in the actual numbers. I wouldn't bet my life on my estimates and neither should you ;) Where these estimates are useful is to make you think about what is a realistic outcome for your portfolio.

Being a macro guy, I should also note that none of what I say incorporates any macro outlook! For instance, I believe there is a high chance of P/E contraction over the next decade so the early returns will be poor. I would be surprised if a passive investor posts anywhere near these returns in the next decade (active investors who are good stockpickers or can time properly can obviously beat anything.) However, the p/e valuation changes should cancel out over the 30 years (or whatever) that we are looking at.

MrParkerBohn estimates the very-long-term returns for the stock market as follows:

MrParkerBohn:

The way I think about it is that the market can have basically 4 sources of return

1) Real Growth in after-inflation profits
2) Distributions to shareholders (dividends & buybacks)
3) Change in Earnings Multiples (ie P/E ratio fluctuation)
4) Inflation

For a long-term investor, only points 1 & 2 really help. #3 (P/E) is not predictable, and should even out in the long-run, anyway. Gains from #4 (inflation) are not real.

So any investor looking for a market-wide return of 10% needs these 4 numbers to add up to 10% in some way.

Here's my (slightly pessimistic) projection.
1) 2% per year in Real Growth in after-inflation profits
2) 3% per year in Distributions to shareholders (dividends & buybacks)
3) 0% per year Change in Earnings Multiples (ie P/E ratio fluctuation)
4) 3% per year Inflation

This would equal a 5% real return, or a doubling every 14 years.


I think his estimate is reasonable. My estimate is more optimistic and here is mine:


I usually go with the following formula, which is conceptually similar to MrParkerBohn's:

Total Return (nominal) = Dividend Yield + Earnings growth + Valuation change

Current dividend yield on S&P 500 = 2.7% (WSJ Market Data)
Long term EPS growth has been around 6.7% (refer to page 2)
Valuation change = 0


Total Return (nominal)
= 2.7% + 6.7% + 0%
=9.4%
~=6.4% real (if we assume inflation is 3%)

I agree with MrParkerBohn's thinking that we shouldn't bank on a p/e expansion, especially given how the market is not cheap right now, so I have set that to zero.

Using the dividend yield method is somewhat controversial since some argue that dividends may be misleading--but I don't believe they are wildly off. One can use earnings yield or some other measure as well and I'll bet the answers will be close.

In real terms, money will double roughly 11.1 years with these numbers. This isn't as optimistic as my 'double in 10 years' case but they are slightly better than MrParkerBohn's estimate.


For reference, late last year (in Oct 2008) Jeremy Grantham, who is a very good macro asset allocation strategist, developed a 7 year forecast predicting 7.7% for US large cap, 12.2% for US high-quality, 10.8% for International large cap excluding Japan, and 13.2% for emerging markets. These numbers are real return forecasts but they include what GMO calls "value-added return" (basically it assumes some skill by the investor in adding value--passive investors should ignore the value-added component.) The market valuations have changed in the last 9 months so I don't know how current these estimates are. In any case, this gives us a reference from a superinvestor (for the next 7 years only.) Grantham's estimates are much higher than either of our numbers.

Tags:

1 Response to A rough estimate of long-term stock market returns

MrParkerBohn
August 6, 2009 at 1:47 AM

Your estimate looks pretty reasonable, and is probably closer to historical data trends than mine.  I would bet that many people would even think you are being pessimistic.

I set my economic growth estimate below historical trends for one reason:  selection bias.

When estimating the growth of the US market, people look at historical data on the US market.  But let's consider that over the last 140 years or so, the US rose from being a war-ravaged frontier with a population of 40 million to the world's only super-power with a population of 300 million.

So it seems to me that looking at US historical data to forecast US future returns is a profound selection bias, where we are looking only at the most successful country, and drawing a line to the sky.

I think this is a much bigger bias than most people realize.

What I would prefer to reference is returns data on all established markets over, say, the last 100-150 years.  This would include markets in countries that were seen as stable at the time, such as late 1800's Russia and pre-WWI Germany.

I don't have this data, so the best I can do is set my estimate a bit below historical trends.  Only time will tell, but it will be an interesting journey.

Post a Comment