If you have even the slightest interest in investing, you probably notice stock market experts everywhere, whether they appear on CNBC, write for newspapers or blogs, or even serve as your advisor. Readers of this blog know to be very skeptical of their advice. We’ve already exposed the disappointing results of using Morningstar mutual fund ratings, mutual fund managers, and Wall Street earnings forecasts.
But has it always been this way? These studies generally use performance over the last few decades to support their conclusions. The stock market and its experts have been around a lot longer than that.
Just as it’s in our blood to try predicting the markets, it seems there’s always someone checking the predictors. For our grandparents and great-grandparents, that man was Alfred Cowles III.
In a paper released in 1932, Alfred Cowles conducted one of the earliest studies evaluating the record of stock market predictors. He studied a range of experts:
- 16 financial publications making more than 7,500 stock recommendations over a 4.5 year period
- 20 institutional investors selecting stocks over a 4 year period
- The editor of the Wall Street Journal, who wrote 255 editorials on the direction of the market over a 26 year period
- 24 financial publications making more than 3,300 calls on the direction of the market over a 4.5 year period
And these were his results:
|Stock Market Expert||Average Performance
Relative to Benchmark
|16 financial publications (stock picks)||-1.43%|
|20 institutional investors (stock picks)||-1.20%|
|Wall Street Journal (market direction)||-2.75%|
|24 financial publications (market direction)||-4.00%|
Yes, every category he evaluated did worse on average than just following a benchmark. It seems predicting the markets with any accuracy was just as difficult then as it is now.
Obviously just evaluating the averages is not complete, since it’s possible that a few skilled experts were hidden by poor aggregate results. Well, Cowles considered that possibility by also looking at the distribution of performance. Unfortunately, he still found no statistical evidence of skill in the experts’ performance.
Not impressed? Does the time period of 4-5 years seem far too short to support any real conclusions? Cowles may have had the same concern, because he followed up with a paper in 1944 that evaluated experts over a longer period of time: 11 to 15 years. Same methodology, longer time period, unfortunately the same results.
Of course, this doesn’t “prove” that market predictions are always doomed to fail. But next time you hear a good story about a stock or where the market is going, make sure to have a healthy sense of skepticism.
- Cowles, Alfred, “Can Stock Forecasters Forecast?”, 1932.
- Cowles, Alfred, “Stock Market Forecasting,” 1944.