The effect of accounting discretion on executive compensation contracts

This study suggests and tests an explanation for variation in the use of accounting variables in the determination of executive compensation. It is suggested that the importance of accounting variables in executive compensation packages may depend upon differing contracting costs associated with the use of accounting numbers. A source of contracting costs related to accounting-based pay plans is the degree of managerial accounting discretion, because through the use of accounting discretion managers may manipulate accounting income. It is costly for compensation committees to deal with the potential manipulation of accounting numbers. Therefore, it is hypothesized that when there is greater accounting discretion, the relationship between executive compensation changes and accounting income will be negatively affected.;To test the hypothesis, ordinary least squares and two-stage least squares were applied to a sample of changes in CEO compensation and accounting and market performance variables. The models allowed for interactions between both performance variables and the proxies for accounting discretion. Accounting discretion was measured by determining the variance of current and noncurrent accruals for each company. The analyses were conducted using a pooled time-series cross-sectional model and by examining four years independently.;The results of the empirical analysis provide some support for the hypothesized relationship between the extent of accounting discretion and the degree to which compensation changes are related to accounting performance variables. There is evidence that the presence of high-variance current accruals weakens the relationship between changes in compensation and accounting income. The results, however, are not consistent across techniques and years and raise several unanswered questions regarding the underlying relationships.;The relationship between accounting income and changes in CEO compensation appears to be affected by the variance of current accruals, as hypothesized. However, because of potentially serious methodological and measurement problems, the results are not conclusive and should be interpreted with caution. More research is needed to explore the underlying relationships and better model changes in executive compensation.