In this study I investigate the relation between information complexity and financial analysts' use of that information. I rank by complexity six tax-law changes enacted by the Tax Reform Act of 1986, and then examine analysts' explicit forecasts of effective tax rates around those changes. I show that analysts' revisions of their forecasts of effective tax rates appear to impound the effects of the less complex tax-law changes but not the more complex changes. Furthermore, as expected, if analysts assimilate less complex (but not more complex) information, the magnitude of the errors in their forecasts of effective tax rates increases with the effects of the more complex tax-law changes, but is unrelated to the less complex changes. Taken together, these results indicate that analysts assimilate less complex information to a greater extent than they assimilate more complex information. Either analysts' abilities to incorporate specific information into their forecasts is a decreasing function of the complexity of that information, or analysts choose not to assimilate complex information because the cost would exceed the benefit. In either case, complexity reduces analysts' use of information. These results demonstrate the importance of considering information attributes, such as complexity, when investigating why analysts' forecasts fail to incorporate all public information.

Management earnings forecasts expressed as a range have become the most common form of quantitative management guidance. Traditionally, the proxy for the sign and the magnitude of the information conveyed by these forecasts—the forecast news—is calculated as the difference between a pre‐forecast earnings expectation and the midpoint of the forecasted range. We provide strong evidence that this traditional measure understates the amount of information conveyed by range forecasts. More importantly, we demonstrate that information conveyed by the upper and lower bounds of the forecasts can be used to improve the classification of forecasts as conveying good or bad news and for calculating the magnitude of that news. We rely on these findings to suggest alternative methods of classifying management range forecasts as conveying good versus bad news and to refine the calculation of forecast news to include the broader information set. Our analysis also suggests that the information conveyed by the range when the forecasts are bundled (issued concurrent with an earnings announcement) is significantly different than when forecasts are not bundled. Overall, our study documents the importance of incorporating range‐related information when assessing the sign and the magnitude of the information conveyed by management range forecasts.

This paper examines the association between the cost of equity capital and levels of annual report and timely disclosure, and investor relations activities. We estimate the cost of equity capital using the classic dividend discount model. We find that the cost of equity capital decreases in the annual report disclosure level but increases in the level of timely disclosures. The latter result is contrary to theory but is consistent with managers’ claims that greater timely disclosures may increase the cost of equity capital, possibly through increased stock price volatility. We find no association between the cost of equity capital and the level of investor relations activities. We conclude that aggregating across different disclosure types results in a loss of information. Failing to include all disclosure types in regression analyses may lead to a correlated omitted variable bias and erroneous conclusions.

Managers, investors, and researchers have a compelling interest in identifying a reliable empirical proxy for firm-specific cost of equity capital (r). In theory, deducing r is possible if the market's future cash flow forecast and current stock price are observable. Practically, deducing r is dependent on the ability to estimate the market's forecasted terminal value. We evaluate five methods of deducing firm-specific r (labeled rDIVPREM, rGLSPREM, rGORPREM, rOJNPREM, and rPEGPREM) that deal with this conundrum differently. The extent to which the estimates are associated with firm risk in a stable and meaningful manner is the basis for our assessment. We find that the rDIVPREM and rPEGPREM estimates are consistently and predictably related to risk, while the alternatives are not. Based on these results, we conclude that rDIVPREM and rPEGPREM dominate the alternatives.