And the importance of such a measure.... the ten year forward annualized change in the S&P 500 index vs. the starting S&P 500 to nominal GDP ratio since 1929.
The above has a remarkable 0.73 r-square.
Where are we now? Well, assuming the economy grew at the 3.2% annualized rate that has been forecasted for Q1, at current levels the S&P 500 is ~8.4% of nominal GDP (in billions $$).
Or just about at the 80 year average.
Source: BEA
I can't help but think this is somewhat distorted by including times when GDP is shrinking or the S&P is clearly in a bubble phase and mentally I'm thinking "what if we got rid of the outliers?"
ReplyDeleteMy immediate self-rebuttal is that the market doesn't really care though and doesn't give mulligans so I guess even though it includes data that's clearly during overvalued times and that affects the average it would make sense.
Fascinating
This comment has been removed by the author.
ReplyDelete(My original comment was posted after only reading half the post, so it was wrong :)
ReplyDeleteWhile we're at average valuation, we're also at a very low rate of dividends compared to the past. So even if index returns (i.e. capital gains) are average, total return can be expected to be lower than usual.
agreed, but dividends are tough to predict. they are low now, but will they increase in 5-10 years?
ReplyDeleteoverall, the analysis surprised me. both by its predictive power in the past (who knows about future), as well as relative valuation today vs. years past
That's a really great regression. I just replicated it myself. The test statistics are better than the more famous valuation measures such as Schiller's 10-year P/E or Tobin's Q!
ReplyDeleteWouldn't it be even better if you used the market value of the US stock market rather tan just the level of prices?
burt- that would be interesting to see. any idea where i can get my hands on that data?
ReplyDeletebtw- is this type of analysis new? if so, i need to throw my name on it somehow!
Jake, nice regression! although as my usual skeptical self, did you trail the GDP data by 3 months to avoid look-ahead bias? The quarterly GDP data is always released 1 quarter later.
ReplyDeletehenry bee- i matched the "actual" GDP to the 10 year forward change in the S&P 500 as of that "actual" date (i.e. the 12/31/09 data was Q4 '09 data, which was last updated March '10).
ReplyDelete"actual" because nothing is actual and the data is constantly revised.
I know of some analysts who look at the ratio of corporate profits to GDP. This is somewhat similar to your measure, and it also provides a pretty good signal.
ReplyDeleteYou can get data for total market cap from Wilshire Associates from 1980. Prior to that, you will probably have to search the academic literature.
1. Warren Buffet has suggested similar methodology, comparing stock market to GNP, so the idea isn't new. It's a good idea though, it's very hard (impossible?) for corporations as a group to grow much faster than the economy.
ReplyDelete2. Dividends are far less volatile than earnings and index prices, they tend to grow fairly smoothly.
Interesting but what about international earnings for companies in the S&P. These have become much more significant and may continue to be out-sized in the future?
ReplyDeleteanother good point, but global growth should in theory help grow the U.S. economy as well.
ReplyDeleteHey Jake: I've done some work putting this valuation measure into an actual system and running it. Results are good. If you want to discuss, I'm at rothberg at alum dot mit dot edu
ReplyDeleteburt- email sent...
ReplyDelete