Thursday, August 07, 2008

How the pros do it:

One illustration of the difficulty of prediction is to look at the job analysts have done in predicting the earnings of companies they are paid to follow and study. Investment expert David Dreman studied analyst forecasts in collaboration with Michael Berry of James Madison University.3 The study was subsequently updated to include data to 1996. They took analysts’ quarterly forecasts and compared them to the actual quarterly earnings for the period 1973 through 1996. The forecasts included 94,251 consensus forecasts (each consensus forecast included at least four separate analyst predictions resulting in over 500,000 individual predictions).

The analysts were able to speak with management to help guide them in their own forecasts. They were also able to change their forecasts within three months of quarter-end. These analysts are highly compensated and often educated at the nation’s top schools; their compensation is often tied to their ability to predict.

Despite all these advantages, the study found the average error rate was 44 percent. The error rates also seemed to grow larger over time. Thus despite advances in communications and technology, error rates in the last eight years of the study (from 1996) averaged 50 percent, with two of those years having error rates of 57 and 65 percent.

Dreman eliminated all earnings estimates less than ten cents per share to prevent large percentage errors from distorting the study. (The difference between 3 cents and 4 cents is a whopping 33 percent.) Even after this conservative adjustment, the error rates still averaged 23 percent. This means that, on average, if the consensus forecast called for a dollar in quarterly earnings, the analysts were off by an average of 23 cents. Dreman and Berry further broke down the data and found that the error rates were indistinguishable by industry type. Mature or budding industry, analysts were often wrong by wide margins.

It is astounding that they were so wrong so often.

Then they go on to wonder how things are going at the Fed.

No comments: