Earnings distribution, corporate governance and CEO pay

Abstract

We investigate the relationship between earnings differentials and the pay of CEOs of 190 British companies between 1970 and 1990. We find that (i) changes in the differential between the 90th and 50th weekly earnings percentiles for non‐manual adult male workers [90:50] explain changes in the level of real CEO salary and bonus in our sample of companies; (ii) changes in this differential also account for changes in the elasticity of CEO pay to firm size; (iii) a broader measure of earnings inequality does far worse than 90:50 at explaining changes in both the level and the firm size elasticity of CEO pay; (iv) fitting the model on data for 1970–1983 and predicting pay levels for the period starting with the widespread adoption of executive share option schemes in 1984, we find a structural break in the relationship between lower management pay differentials and the pay of the CEO. We conclude first that top executive pay prior to 1984 was a stable function of both firm size and earnings differentials lower on the administrative ladder, consistent with a hypothesis advanced by Herbert Simon in 1957; and second that the use of share options from 1984 onward represents not simply a change in the mode of top executive compensation, but a de‐linking of the pay of top executives and that of their subordinates.

Keywords:

• Executive pay
• earnings
• inequality

Notes

The ratio is conservatively stated (and also ignores certain cyclical factors) in that the figure for median annual earnings of male workers is obtained by multiplying the weekly rate by 52, hence assuming year‐round employment. The CEO figure is on an annual basis to begin with.

Simon's theory was framed as an alternative to marginal productivity theories of executive pay, and incorporates assumptions which many economists find uncongenial: a career ladder which is internal to the firm, and pay differentials between hierarchical levels based on social norms. Various efforts have been made to reconcile the structure and conclusions of the model with marginal productivity theory: Calvo and Wellicz (1979) and Rosen (1982). Both motivate differentials within the hierarchy on the basis of sorting by ability, while in another paper Calvo and Wellicz (1978) show that the same conclusion can be reached with on the basis of monitoring costs and enforcement rents. All versions of the theory depend on some relationship being constant over the relevant range in the organization in question. For Simon this was the ‘rule of proportionality’, which is a social norm. In ability sorting and monitoring cost models, the constancy emerges from a chain of causation which includes an increase in marginal product of managerial ability (or effort) as one moves up an organizational hierarchy, and a distribution of ability (or functional form mapping pay into effort) which is such that marginal productivity generates the same proportional differential between levels from the bottom of the hierarchy to the top. Which of these stories you find more parsimonious may depend on the discipline in which you are trained.

Coefficient selection was done using Schwartz Bayesian Criterion. See Table 2.

F = 2.30. With four restrictions and 3906 observations, the critical value at the 5% level is approximately 2.37.