The Relative Grading Bias

What is it about?

Analyst A may rate an overvalued stock a “Buy” because it is the least overvalued amongst the portfolio of stocks she is rating, whereas Analyst B may rate an undervalued stock a “Sell” because it is the least undervalued amongst the portfolio of stocks she is rating. Controlling for three pairs of higher dimensional fixed effects, we find that this relative grading bias exists. Results from two other different research designs corroborate this finding. A long-short portfolio, long the “buys” of analysts like A and short the “sells” of analysts like B, results in a negative 8.95% annualized raw return.

Featured Image

Why is it important?

The relative grading versus absolute grading debate, which is really a debate about using a sample benchmark versus using a population benchmark, pervades many facets of human life. In political elections, the most popular person in the party may not be the most popular in the country. In sports, compensation of athletes may depend on whether they are a big fish in a small pond or a small fish in a big pond. In corporations, compensation of CEOs would depend on the peers they are benchmarked to. In portfolio performance, the alpha depends on the benchmark. In academia, a student grade in a class may depend on the average performance of the class, and faculty tenure may depend on the peer schools being used as a benchmark. Analyst recommendations should be free of this bias. The reason is that the fiduciary responsibility of analysts towards their clients demand that analysts give an overvalued stock a “Sell” recommendation and to give an undervalued stock a “Buy” recommendation. That is their job. Conflicts of interest may cloud these recommendations., and the literature (cited later) has shown that it does. But the literature till now has not shown a relative grading bias.

Perspectives