Bias and Efficiency: Comparison of Analyst Forecasts and Management Forecasts

In our paper, Bias and Efficiency: A Comparison of Analyst Forecasts and Management Forecasts, we compare the forecast characteristics of analyst forecasts and management forecasts. Frequently, analysts and managers provide similar type of information to investors, namely forecasts. Since managers and analysts have different incentives and different information sets, we empirically test whether those differences are manifested in their forecast characteristics. Specifically, we compare the bias, a systematic deviation of management and analyst EPS forecasts from the actual realized EPS, and efficiency, the ability of managers and analysts to incorporate prior publicly available information in their forecasts.

When comparing management forecasts and analyst forecasts, it is important to consider the implications of the difference in incentives and information available to analysts and managers. Since prior literature documents an optimistic bias in analyst forecasts, we expect that, given management incentives and cognitive biases, management forecasts will be at least as biased as analyst forecasts. In addition, since companies’ managers are exposed to private information, we expect management forecasts to better incorporate prior available information.

We find several striking results. First, we find that prior stock returns do not predict management forecast errors while they predict analyst forecast errors. Furthermore, while we find an optimistic bias in a broad sample of both management forecasts and analyst forecasts, the optimistic bias in analyst forecasts disappears in months in which management forecasts are issued. The bias is still apparent for these firms when managers do not provide forecasts.

As the fiscal-period end nears, the ability of both managers and analysts to provide accurate forecasts increases, as much of the information is already known. Consistent with this notion, we find that the optimistic bias in management forecasts diminishes as the fiscal year end approaches. Similarly, we also find that the ability of prior stock returns to predict analyst forecast errors diminishes, although it does not disappear, as the fiscal year end nears, implying that analysts’ forecasts improve over time. Lastly, we supplement our tests by examining the ability of past accruals, earnings changes, and sales growth in explaining analyst and management forecast errors. We find that none of these variables has the ability to forecast management forecast errors. In contrast, and consistent with prior literature, we find that analyst forecast errors are predicted by both prior accruals and prior stock returns.

To conclude, this paper finds that in contrast to analyst forecasts, management forecasts are efficient with respect to incorporating available public information. In addition, while in a general sample both management forecasts and analyst forecasts are optimistically biased, the bias in analyst forecasts disappears in the presence of management forecasts. This paper contributes to prior literature by showing that while management and analyst forecasts exhibit forecast errors, the sources of these errors are different. Specifically, in periods with concurrent management forecasts and analyst forecasts, the error in management forecasts is partly a result of optimistic bias, whereas the error in analyst forecasts is partly a result of analysts’ inability to fully incorporate information.

The full paper is available for download here.