Guru Grades

Abby Joseph Cohen, the Sunny Side (Last Updated 1/3/08)

A reader suggested that we review the stock market commentary of Abby Joseph Cohen, partner and chief U.S. investment strategist at Goldman Sachs. The public record for Ms. Cohen, available most robustly via MarketWatch and Bloomberg.com, focuses on her 12-month forward forecasts for the levels of major stock market indexes, such as the S&P 500 index. All citations we found for 1999-2008, concentrated in the period 1999-2002, are bullish. Given the quantitative nature of her forecasts, we evaluate her annual S&P 500 index forecasts made near the end of each prior year versus two benchmarks: (1) the expert averages from the annual Business Week stock market forecast surveys; and, (2) simple mechanical extrapolations of the actual historical performance of the S&P 500 index. Using data for 1999-2007 (nine years), we find that:

The following chart shows the absolute errors for Abby Joseph Cohen's annual S&P 500 index forecasts (yellow columns) along with: (1) the absolute errors for the expert averages from the same-year Business Week forecasts (blue columns); and, (2) the absolute errors for yearly mechanical extrapolations of historical S&P 500 index performance (red columns). We perform the extrapolations for the latter benchmark using the average annual change in the S&P 500 index from 1951 through the year preceding the forecast year. The chart shows that:

• For 1999-2000, Abby Joseph Cohen's forecasts were not particularly accurate, but they were more accurate than the expert averages. In 1999, her forecast was too low; in 2000, too high.
• For 2001-2002, Ms. Cohen's forecasts were especially inaccurate (dramatically overoptimistic).
• For 2003-2007, Ms. Cohen's forecasts were fairly accurate, mostly beating the expert averages. For 2003-2005 and 2007, her forecasts were a little too high. For 2006, her forecast was slightly too low.

There are slight indications that Ms. Cohen is a better forecaster than the average expert but a worse forecaster than a very simple algorithm. Across all nine forecasts, Abby Joseph Cohen's mean annual absolute forecast error is 12.6%, compared to 14.4% for the expert averages and 10.9% for the simple mechanical extrapolations. Because the sample size is so small (nine forecasts), we cannot be confident that the relative order of these mean errors will persist. There is also some indication (seven out of nine years) that, assuming no systematic improvement, Ms. Cohen's future forecasts will be too high rather than too low.

In summary, very limited evidence suggests that Abby Joseph Cohen is a little better as a stock market forecaster than her average peer, but a little worse than a simple algorithm.

See Guru Grades for a snapshot of the accuracies of various experts in forecasting the U.S. stock market, including links to evaluations of individual gurus. We do not include Abby Joseph Cohen in the snapshot because of the different nature of this evaluation.