The negative relationship between the average and dispersion of stock analysts' buy/hold/sell opinions

Abstract

This paper measures the relationship between the level of stock analysts opinions and the dispersion of those opinions. The proposed theory is that analysts notice when there is less agreement and tend to lower their reported opinions in such cases. Hence, the thesis of the paper is that a negative relationship exists between the level of analysts opinions of a given stock and the dispersion of those opinions. Using a sample selection process that controls for outliers and the effect that the varying number-of-analysts per stock might introduce, this paper analyses the opinions of 245 stocks from the Standard and Poor's 500. With respect to the main thesis, all the test statistics have the correct sign; however, some statistics are not significant or are only marginally significant. One innovative random variable used in the tests involves assigning to a stock the cumulative distribution function value associated with the standard deviation of the opinions conditional on the average and number of those opinions. This procedure rescales the data so that comparisons can be made between stocks with different average opinions.