The subtle signals contained in managers' forecasts can tell the market not only about the forecasters' uncertainty, but also their motivations. That is the message in a study that looks at manager forecasts of earnings per share (EPS) which finds those ending in nickel estimates are less accurate and overly optimistic.

The study by Linda Smith Bamber, Kai Wai Hui and P. Eric Young, looked at 3,265 management point forecasts from 1996-2004 and found nearly half of them - 46.98 per cent - "heap" at nickel intervals (that is, an even dollar, half-dollar, quarter, dime or nickel). Yet the actual EPS ended in nickel intervals only 20.12 per cent of the time.

The frequency of heaping is close to that of analysts, who did so 50.62 per cent during the same time period, even though managers have much better information on which to base their forecasts.

"Prior research suggests heaping in analysts' forecasts is a heuristic response to uncertainty. However, uncertainty is unlikely to be the sole determinant of heaping in management forecasts. Managers have more precise private information and they are not required to issue point forecasts of earnings, both of which should reduce heaping," they say.

"Moreover, the managers' nickel forecasts are both less accurate and more optimistic than non-nickel forecasts, so the nickel forecasts are not simply benign responses to uncertainty."

The authors suggest there are strategic economic incentives to heap - when firms have proprietary information costs, and when managers want to give the market an impression of greater uncertainty by rounding estimates. The latter would likely reduce the likelihood that the market would attribute error in manager forecasts to intentional misrepresentation.

Evidence of this includes an eight per cent increase of heaping in initial manager earnings (where the average time horizon is 203 days) for every one-point increase in proprietary information costs, and a four per cent increase for revised forecasts (average time horizon 164 days).

Managers were also more likely to heap when they had greater incentives to bias their forecasts optimistically, with a one-standard deviation associated with an 11 per cent increase in the probability of heaping in initial forecasts and three per cent in revised forecasts.

Uncertainty also remained a factor, with a one-standard deviation increase in uncertainty leading to a 13 per cent increase in the probability of heaping in initial forecasts and four per cent in revised forecasts.

"These results suggest heaping in management forecasts stems not only from a benign psychological heuristic response to uncertainty, but also from managers' strategic incentives - especially efforts to protect the firm's proprietary information and managers' self-serving opportunism in response to their own economic incentives. This is the first evidence we know of that shows strategic economic incentives help explain heaping," they say.

Moreover, they identify consequences to managers' heaping: it influences analysts' subsequent forecasts. After managers issue nickel forecasts, the proportion of analysts issuing nickel forecasts increases by 13 per cent to 61 per cent. When managers issue non-nickel forecasts, analyst nickel forecasts drop by 11 per cent to 29 per cent.

"Managers' nickel forecasts spur even active analysts to issue forecasts heaped at nickel intervals, although analysts' forecast revisions partially adjust for the optimism and noise in managers' nickel forecasts," they say.

The authors add that a manager's choice between nickel and non-nickel disclosures is a finer element of the forecast specificity decision that has not been previously recognized.