Permanent Earnings vs. Reported Earnings: Does the Average Difference Approximate Zero?

Abstract

This paper evaluates the hypothesis that the difference between reported earnings and permanent earnings approximates zero, on average. We measure a firm’s permanent earnings using its stock price, and the short term interest rate determines the permanent earnings to price relation. The hypothesis corresponds to the idea that a firm’s capitalized reported earnings minus the stock price equals some “noise” which on average approximates zero. In valuation terms, the hypothesis depends on growth and risk cancelling each other, on average; our modeling does not depend on, or imply, risk-neutrality. US data supports the hypothesis: reported earnings exceed permanent earnings in about half of all cases. However, the proportion of pluses vs. minuses can deviate materially from 50% in any year, and there is marked time-series correlation. The “zero average” holds only because we evaluate several decades of data. The permanent earnings hypothesis will not hold if the accounting approximates fair value accounting. Such accounting provides the underpinnings for Hick's concept of economic earnings, and it differs radically from traditional GAAP accounting. Per theory, economic earnings should exceed permanent earnings, on average. We consider this angle to the 50-50 proposition by examining financial firms. Earnings for such firms should to some extent tilt towards Hick’s earnings concept. The data supports the hypothesis: reported earnings now exceed the permanent earnings significantly more often than 50% of the time. Thus the benchmark permanent earnings hypothesis, the “fifty-fifty” proposition, applies only for industrial (non-financial) firms.