Diligence, Objectivity, Quality, and Accuracy

Abstract

Purpose. We provide a general framework of behavior under asymmetric information and develop indices of diligence, objectivity, and quality by an analyst and analyst firm about a studied firm, and relate them to the accuracy of its forecasts. We test the associations of these indices with time.

Methodology. The test of Public Information versus Non-Public Information Models provides the
index of diligence, which equals one minus the p-value of the Hausman Specification Test of Ordinary Least Squares (OLS) versus Two Stage Least Squares (2SLS). The test of Objectivity versus Non-Objectivity Models provides the index of objectivity, which equals the p-value of the Wald Test of zero coefficients versus non-zero coefficients in 2SLS regression of earnings forecast residual. The exponent of the negative of the standard deviation of the residuals of the analyst forecast regression equation provides the index of analytical quality. Each index asymptotically equals the Bayesian posterior probability, by the analyst and analyst firm about the studied firm, of the relevant behavior.

Findings. We find that ex post accuracy is a statistically and economically significant increasing function of the product of the indices of diligence, objectivity, and quality by the analyst and analyst firm about the studied firm, which asymptotically equals the Bayesian posterior joint probability of diligence, objectivity, and quality. We find that diligence, objectivity, quality, and accuracy did not improve with time.

Originality. There has been no previous work done on the systematic and objective characterization and joint analysis of diligence, objectivity, and quality of analyst forecasts by an analyst and analyst firm for a studied firm, and their relation with accuracy, and our paper puts together the frontiers of various disciplines.

Keywords: Asymmetric Information; Analyst Forecasts; Management Guidance; Earnings; Diligence; Objectivity; Quality; Accuracy; Hausman Specification Test; Wald Test; Bayesian; Big Data in Finance.

JEL Classification: G12; G14; G24; C23; C26; M41.