Friday, March 13, 2009

Equity Returns Lower than Economic Growth

Update: I had completely forgotten that reader AJK had provided detailed return analysis going back to 1950 on the same topic, so credit to him for this post...

In response to my post on gold vs. equities, Irrational Doomsday stated:

That's good. I think the next logical step would be how far investment returns can diverge from real underlying economic growth (ie Real GDP growth).
As requested... it's not too surprising that in recent years the real returns of the S&P 500 has been highly correlated to real economic growth.



Source: Yale

14 comments:

  1. Rob Arnott has a good article
    on the topic. If you have access to CFApubs.org there's a cleaner version there (CFA Magazine Nov/Dec 08 issue).

    While things like expanding multiples and new, fast-growing companies can cause stock price growth to outstrip economic growth for extended periods of time, these things don't last forever and eventually experience long periods of subpar growth.

    I think the starting point of this analysis can have a dramatic effect on the results as well. With a year like 1929 it creates such a huge hole for stocks to climb out of as stock values fell much more than economic growth did, peak to trough. Good post!

    ReplyDelete
  2. equities are actually underperforming since the 1960's as well.

    i have another chart i'll post later today / Monday which shows that equities typically swing in 30 year cycles between over/under performing GDP

    ReplyDelete
  3. Wow that is really, really surprising. That doesn't really jibe well with the data I was looking at:
    https://customers.reuters.com/d/graphics/USEQUITIES.pdf

    The difference in data sets was total equity outstanding (Reuters) vs S&P 500 + dividends (yours). But I wouldn't expect to S&P 500 to be underperforming that, especially if you include dividends.

    Are returns really (your graph) that far out of whack from the total market value (the reuters data)? What would cause that? Share dilution by issuance over time? If the data is correct, the actual returns you'd expect to see on a broad index wouldn't really justify owning stocks even in a normal environment.

    Really surprising though- I would have never thought the S&P 500 + dividends would be underperforming real economic growth at any time since the Great Depression.

    ReplyDelete
  4. irrational... any way you can send over that data to me? econompicdata@gmail.com

    i got my GDP data from the BEA and equity data here:

    http://www.econ.yale.edu/~shiller/data/ie_data.xls

    ReplyDelete
  5. It looks like the data was pulled from the BEA and the Fed's flow of funds, but I didn't write that article, so I don't really have the raw data.

    Looking more closely at it, maybe it's not so surprising after all- the main difference may anchored at the start time of the data. If you start at 1929 peak, it took until the 1950s for the stock market to fully recover. The Reuters guy starts his data here.

    Since your graph is cumulative returns, that gap gets pretty significant over that period of time between 1929-1952. Just eyeballing it, it looks like if you started at 1952, you'd have a more similar picture, just on the cumulative returns scale. And the cumulative returns scale highlights the variance which is muted in his graph- so the underperform/overperform cycle pops out in your graph.

    Looking it over now, I think the main reason why your chart was so surprising to me, while his chart was what I was expecting, was because the common investment advisor line that stocks outperform other investments over time- they usually tend to start their historical data collection at the low of the Great Depression market, not at its height.

    ReplyDelete
  6. Without doing any actual work, I would guess the problem is that the S&P includes actual numbers, while the GDP is composed of wishful thinking, voodoo assumptions, and outright lies.

    "GDP growth" for example, assumes the equivalent of 100% re-investment, since it includes such items as "equivalent rent".

    The GDP also increases with population, even when the population is gradually less wealthy. Was the S&P adjusted to reflect new investment opportunities too small to be included?

    If there were a way to invest in an imaginary number which is manipulated to show growth, then people would do so. I don't know that you could create a "GDP index". If you did, it would perform way under 3%.

    ReplyDelete
  7. It's probably worthwhile to note that just because the S&P line is below the RGDP line does not mean that equities are underperforming RGDP. The gap between the two on a cumulative scale is largely a function of the time period chosen, as Irrational said.

    It's really the relative slopes of the individual lines that matters for a given time period on a chart like this. So between 1980 and 2000, S&P is below RGDP, but because the slope of the S&P line is higher than the RGDP line's slope, equities were actually outperforming.

    Feel free to correct me if I'm wrong, but that just how I interpreted it.

    ReplyDelete
  8. jake,

    1. the s&p is to small index

    ReplyDelete
  9. 1. The adoption of fiat currency is predicted to lead to accentuated boom-bust cycles. The big swings seem to have started in the early 1970s...

    2. The democratization of the stock market means that fund managers are much less closely watched than in the past. The banks, fund managers and other financial professionals are taking the profits.

    3. Should the second plot's y-scale be logged to give a more accurate representation? This might make the huge swings seen recently shrink quite a bit. You often need to log-plot this data since economic growth is exponential, not linear.

    ReplyDelete
  10. it should be logged. i'll fix this later...

    ReplyDelete
  11. In addition to all other remarks, how about dilution? If management extracts value out of the company rewards itself with stock options, one would expect leakage. Also (as other mentioned) small companies growing into the S&P500 and large companies going bankrupt creates friction. I am not sure how stock buybacks affect the S&P + dividend calculation.

    ReplyDelete
  12. This is really interesting data and great points raised by all. I think the most interesting/fundamental question raised by this chart is:

    What is the relationship between GDP growth and equity growth.

    I can think of 4 potential scenarios:
    1. There is no relationship
    2. GDP will outperform equities over a long/infinite time frame
    3. Equities will outperform GDP over a long/infinite time frame
    4. Over a long/infinite time frame equity and GDP performance are similar and it is all about timing

    I would love to hear people's hypothesis. Based upon this data and the information referenced in AJK/Doomsday comments it seems like #4. If so the implication question is what drives the timing difference?

    ReplyDelete
  13. -starting point makes a difference

    -cooked GDP statistics

    -S&P doesn´t mirror the real economy

    -S&P is 500 stocks only. so if a component goes bust, it weighs harder on the index than on the real economy.

    ReplyDelete
  14. What is striking to me in those charts of GDP is that GDP has been declining in recent years, it appears that the late 1990's was the peak in GDP for the USA. Which ties in with what Economist Martin Armstrong says, the the USA as a civilization peaked in 1999.

    ReplyDelete