How does product market competition shape incentive contracts?

Abstract

This paper studies the effects of product market competition on the explicit compensation packages that firms offer to their CEOs, executives and workers. We use a large sample of both traded and non-traded UK firms and exploit a quasi-natural experiment associated to an increase in competition. The sudden appreciation of the pound in 1996 implied different changes in competition for sectors with different degrees of openness. We provide difference in differences estimates and our results show that a higher level of product market competition increases the performance pay sensitivity of compensation schemes, in particular for executives.

Three different compensation measures are used as dependent variables. These are derived from the annual company statements. The first one is total compensation of the highest paid director, and contains all of the firm’s payments to the highest paid director in a particular year, including both fixed and variable compensation elements, such as stock options.¹² Although occasionally it may be the chairman, in most cases the highest paid director is the CEO.¹³ This is the only publicly available measure of top executive pay for the UK, and the one used in virtually all related studies.¹⁴ In fact the amount of information provided on each company varies, in particular many firms do not report pay to the highest paid director explicitly.

Secondly, we use a measure of average executive pay, which contains the average remuneration received by the board members. Given that individual data is not available, this measure is calculated as the ratio of total board compensation over the number of directors. These include the top executives of the firm including the CEO, but also a proportion of non-executive directors of the firm. Ideally one would like to separate these two different types of directors, as their roles are not exactly the same. However, this is not possible in our sample. In any case, even though non-executive directors do not make direct management decisions, they do influence the strategic decisions of the firm, and can be seen as agents of the shareholders, in a way similar to executive directors. Furthermore, the presence of non executive directors in the UK is quite low when compared with the US. Previous studies estimate that the proportion of non-executive directors on the board is about 40-50% for large quoted firms. However among non quoted firms, the percentage of firms with at least one non-executive director is between 33% and 47% for large firms, and 19% for small and medium sized firms (less than 50 employees). Given the predominance of small and medium sized firms in our sample, it is likely that the proportion that do not have any non-executive director represents more than three quarters of the total number of firms.¹⁵ The pay measure is the average total remuneration of all board members, so it includes the total remuneration that executive directors receive for their executive and board activities, and the remuneration associated with being a member of the board for non-executive directors.

Finally, we use average wage in the firm constructed as total wages paid over total number of employees.¹⁶ The density of information on these three compensation variables is not constant. For the variable covering the highest paid director there is an average of 2.1 observations per firm, while for the variables on average executive pay and average wages there is a mean of 3.7 and 4.1 observations per firm respectively. We exclude from the sample firms with less than 5 employees in which CEOs and directors are hardly comparable with the rest of the sample. We also drop observations where the pay variable is zero because this appears to come from mis-coding. Table 2 contains the summary statistics of the relevant variables.

The performance measure used is earnings before interests and taxes. Most of the firms in the sample are not publicly traded. This has the advantage that it is a very broad sample of firms, representative of the whole economy. It also implies that one cannot use stock market based performance measures. Much existing literature focuses on executive compensation of publicly traded companies and uses stock market returns as their measure of performance. The fact that the vast majority of our firms are not listed on the stock market implies that the only performance measure we can use is accounting based. Existing research supports that accounting profits are a relevant measure of performance when examining compensation packages (Bushman and Smith, 2001).

To allow for any non linearities (such as minimum profits to qualify for a bonus or caps) we also include a measure of profits squared in the regressions. Size is computed by the logarithm of total assets. Year dummies, firm fixed effects and a sector-specific time trend (at 3 digit SIC) are also included in all the regressions. All the monetary variables are in constant 1987 pounds.

The measures of openness are import penetration and export share of output measured at a sector level defined by the SIC classification at three digits, as a proportion of total output plus net imports and total output respectively. Since openness itself may be endogenous to changes in the exchange rate, the measures of openness are defined at a sector level as the average openness in the years before 1996 (1993 to 1995), which is kept constant for the whole sample.¹⁷

Finally, the distribution of total pay is highly skewed to the left and contains several extreme values. For this reason we eliminate as outliers observations whenever the pay variable exceeds the value of the top 99% percentile of the sample.¹⁸

Download full text paper via ziddu