The Effects of Sources of Earnings Forecasts and Degree of Source Expertise on Subjects' Estimates of Earnings per Share: A Field Experiment

Abstract

Motivated by previous research in accounting, finance, and psychology, this dissertation reports the results of a field experiment to determine whether two particular variables affected subjects' estimates of earnings per share. The variable used were the source of an earnings forecast and the degree of expertise of the source of the earnings forecasts. Subjects were professional employees of bank trust departments whose jobs normally entail analyzing firms and estimating the firms' future earnings per share. The research utilized Analysis of Variance (ANOVA) for statistical tests of the dependent variables. The sources of earning forecasts used in this study were financial analysts and company officials (management). Earnings forecasts issued by financial analysts and management are widely disseminated in the financial press. Abundant theoretical and empirical literature exists that supports the notion that investors use financial analyst and management earnings forecasts when making investment decisions. The question of whether investors are influenced more by financial analyst or management earnings forecasts has not been answered. This dissertation provides empirical evidence about which source of earnings forecast was more influential in one particular research setting. The second experimental variable examined in this study was a degree of expertise of the sources of the earnings forecasts. A stream of social psychology literature has dealt with the variable source credibility, of which source expertise is a component. That literature, in a variety of experimental settings, consistently has reported that source expertise is a significant variable. However, the sources used in the psychology literature ranged from non expert to expert. Both financial analysts and company officials are considered to be expert sources of earnings forecasts. Thus, this study investigated whether differences in the degree of expertise of expert sources was a significant variable. For the experimental setting used in this study, the source variable was found to be a significant variable while the degree of expertise of the source was not significant. Thus, subjects' earning per share estimates reflected a preference for one earnings forecast source (company officials) over another source (professional financial analysts). On the other hand, subjects' earnings per share estimates were not affected by the degree of expertise of the source.