ESOPs and profit-sharing plans: do they link employee pay to company performance?
The increased incidence of performance-based pay for nonmanagers has stimulated discussion about the pros and cons of these alternative compensation systems. The traditional argument is that when individuals are rewarded as part of a group, members of the group will be tempted to free-ride on the efforts of others. Recently, however, a series of arguments have been made that pay systems involving stock ownership or profit-sharing might increase employee productivity in spite of the free-ridership problem. Brickley and Hevert  summarize several of these arguments.
Reviewing the literature on the impact of various nonmanagerial compensation systems. Conte and Svejnar  and Weitzman and Kruse  separately conclude that, when nonmanagerial compensation is linked to company performance, company performance is either unchanged or improved, but rarely diminished. These and other studies have in common the conclusion that pay systems which link individual compensation to company performance improve company performance sometimes. It is discomfiting that statistically stronger or more consistent results have not been achieved in either the negative or the affirmative.
In this paper, we argue that compensation systems which supposedly create a link between company performance and nonmanagerial pay often do so only weakly, if at all. This may explain some of the ambiguity in the results achieved up to the present time.
There has been limited empirical research to date on the impact of ESOPs and profit-sharing on employee pay. Mitchell, Lewin and Lawler  provide some evidence that profit-sharing increases total employee compensation, and Chaplinsky and Niehaus  provide evidence to the same effect for ESOPs. This is consistent with the notion that employers who offer performance-contingent pay must also offer a pay premium to compensate for the increased risk. However, there are other factors which can explain why firms with ESOPs and/or profit-sharing plans might provide greater income to either employees or shareholders aside from the increased riskiness (performance-dependency) of compensation. Tax advantages from adopting these plans provide opportunities for increased rents, and may have been the primary motive for many of the adoptions. (4) Plans may also have been adopted as "golden handcuffs," resulting in increased retention of highly compensated employees. Giving employees shares in the company can also increase their ability to extract rents. In each of these cases, nonmanagers are paid "more," but not necessarily "better."
The arguments both for and against ESOPs and profit-sharing plans are based on the assumption that these pay devices actually provide positive feedback from company performance to nonmanagerial pay. Using data from Internal Revenue Service pension plan filings combined with information on over 500 companies from the COMPUSTAT database, we are able to measure the sensitivity of employee income from these plans to several different measures of company performance. We compare the extent of performance-pay contingency imparted by these plans with that arising from other forms of pay. In most of the companies that we studied, we find that nonmanagerial compensation through nominally performance-contingent pay systems provides less of a financial incentive to the achievement of company-wide performance goals than does basic salary and wage compensation. It is therefore not surprising that empirical studies find the effectiveness of these plans in promoting company performance goals to be mixed.
In Section I of the paper, we provide an overview of some institutional features of ESOPs and profit-sharing plans. In Section II, we describe the databases which we used and the methodology that we employed in constructing our measures of pension income. Then, in Sections III and IV, we measure the extent of performance contingency of employee income from participation in ESOPs and profit-sharing plans.
I. Institutional Features of ESOPs,
Profit-Sharing Plans, and Defined
Benefit Pension Plans
Employee stock ownership plans were introduced by the Employee Retirement and Income Security Act (ERISA) of 1974, and consist of a stock bonus plan or a stock bonus plan combined with a money purchase pension plan. The basic distinctions between ESOPs and non-ESOP stock bonus plans are twofold: (i) whereas a stock bonus plan may be primarily or fully invested in employer securities, and ESOP must be primarily invested in employer securities, and (ii) ESOPs may be leveraged and may receive credit from the sponsoring company, a feature which is not permitted to any other type of qualified pension plan.
A typical leveraged ESOP transaction consists of the following basis elements: (i) an employee stock ownership trust (ESOT) borrows funds from a lending institution, and (ii) uses these funds to purchase stock of the company; (iii) the company then makes regular contributions to the plan in the amount of the scheduled principal and interest payments, which (iv) the ESOT then uses to repay the loan. The company takes a deduction for its loan payment (or, equally, for its contribution to the plan). At the time that a loan payment is made, shares are released to the trust which have a basis value equal to the amount of the loan payment being made. The current value of the shares as they are released may be more or less than their basis value. If the shares have appreciated in value since they were placed in suspense, employees capture the capital gain. If the shares have depreciated, employees suffer the loss.
Employees income from and ESOP consists of two parts: (i) company contributions, and (ii) return on the plan's portfolio. The return on the company's stock affects ESOP plan returns because of the plan's required investment in employer securities. In a leveraged ESOP, the value of the company's contributions is also influenced by movements in share value on account of the suspense account mechanism described above. Share value growth also has a delayed impact in the case of leveraged plans because the value of stock released to individual accounts in future years is affected by current and past appreciation (depreciation). For this reason, the impact of share returns on employee income may be modestly underestimated in our empirical measurement. In a nonleveraged ESOP, only the return on the plan's portfolio is affected by share performance. Some nonleveraged ESOPs make contributions on a profit-sharing type formula. However, until 1986, many nonleveraged plans were of the tax credit type, which permitted the company to take a tax credit equal to the amount of its contribution to the plan. Contributions to tax credit ESOPs were small, and were related either to the company's investment (prior to 1980) or to its payroll (1980 and after), and not to any measure of company performance (see Conte and Lawrence ).
Because ESOPs are required to invest primarily in employer securities, the return on an ESOP portfolio is likely to be highly related to company performance. Exhibit 1 indicates the relative distribution of plan portfolios for several categories of pension plans sponsored by the companies in our sample (the sample is described hereinafter), and it can be seen that ESOPs, on average, are indeed much more highly invested-in employer securities than the other types of pension plans. In our ESOP sample, 72.3% of the assets were invested in employer securities, compared with 17.5% for the profit-sharing plans and 8.8% for defined benefit plans.
B. Profit-Sharing Plans
Profit-sharing plans have been in existence since the early 19th century.  While some profit-sharing plans provide cash payments to employees, the majority are deferred plans. This form of profit-sharing provides for the deferral of tax on company contributions to the plan and the accumulation of investment returns on these pretax
Exhibit 1. Distribution of Plan Portfolios Employer Other Pension Plan Securities Investments ESOP 72.3% 27.7% Profit-sharing 17.5% 82.5% Defined benefit 8.8% 91.2%
contributions as long as the contributions are held in trust for employees. Because contributions to cash profit-sharing plans do not qualify for special tax treatment, there are no administrative filings required and therefore data for these plans is limited. This study utilizes data only on deferred profit-sharing plans.
In spite of their long history, profit-sharing plans are something of a misnomer because in only the minority of such plans are contributions related to company profits. Neither laws nor regulations require that a mathematical relationship exist between company profits and contributions to the company's profit-sharing plan. Until 1986, companies were forbidden from making contributions to their profit-sharing plan in years when they had no taxable income, but the requirement of even this limited relationship between profits and profit-sharing was removed by the 1986 tax act.
One survey of profit-sharing plans (Hewitt Associates , pp. 12, 13) reported that 45.7% of the sample based their profit-sharing bonus entirely on managerial discretion, and another 9.7% relied on a formula with an additional discretionary contribution. Managerial discretion does not necessarily imply that performance is not the major criterion used for contribution decisions. This is a purely empirical issue, and we have been unable to find any empirical studies of the extent to which discretion affects the extent of pay contingency. However, the relationship may be looser under these circumstances than it would be under a mathematical formula.
The returns achieved by profit-sharing plan portfolios will generally be unrelated or weakly related to company-specific performance as well. Profit-sharing plans are permitted to invest any portion of their portfolios in employer stock, but may do so only if the investment is fiduciarily sound. This practically limits the extent of such investments. As a result, the returns achieved by profit-sharing plan portfolios will reflect general financial market conditions more than they will reflect the returns on employer stock.
C. Defined Benefit Plans
Income earned under defined benefit pension plans is generally not viewed as being dependent on company performance. However, because many defined benefit plans base benefit accruals on the value of an employee's earnings in prespecified employment years (e.g., the last three full years worked), and because wage and salary income may be substantially affected by company performance, there is the potential for some measurable performance contingency for income earned under these plans.
II. The Data
This study uses data from two sources: Standard and Poor's COMPUSTAT and the Internal Revenue Service Form 5500 administrative reports for qualified pension plans. By combining data from these two sources, we are able to obtain a comprehensive view of compensation expense for a large number of publicly traded companies.
We began by searching the COMPUSTAT database for all companies which supplied information on wages and salaries for any of the years 1979 through 1988. Eight hundred forty-seven of the companies monitored in COMPUSTAT's "active" file and 341 of the companies on the "research" file reported wage and salary expense in at least one of these years. We pared this group down by removing all those with fewer than 500 employees. We took this step because smaller companies are likely to have one or more pension plans with fewer than 100 participants, and are not required to file a Form 5500. Four hundred sixty-six active file companies and 157 research file companies met these criteria. We further narrowed the sample by selecting for inclusion firms which had at least four consecutive observations on the variables in our regressions (described hereinafter). We excluded observations on these firms if there were differences exceeding a factor of two in the value of net total plan assets from one year to the next. For profit-sharing plans and defined benefit plans especially, swings of this magnitude indicate a significant change in the structure of the company, making any time-series inferences difficult.
Our data on ESOPs, profit-sharing plans and other pension plans come from the Form 5500 pension plan filings for the years 1979 through 1988. All qualified pension plans with 100 or more participants are required to file a Form 5500 in each year of the plan's operations. If we could not find a record for an ESOP or profit-sharing plan for a given company in our sample, we assumed that the company did not have such a plan. In order to be sure to capture total income, we also searched for defined benefit pension plans of each company in the sample, again assuming that the company did not have such a plan if it had not filed a Form 5500. This assumption appears to be highly valid, because plans which we identified in two nonconsecutive years were very rarely absent for the intervening year or years.
The Form 5500 reports the type of plan, number of plan participants, total contributions to the plan, beginning and end of year plan assets, and plan investments among other things. We primarily used these data to calculate the average participants' income from plan participation in the reporting year. Our method for calculating plan income is described below.
A. Measuring Employee Income and Income
We classified employee income into two major components: wage and salary income, and pension plan income. Wage and salary income was measured in one of two ways, depending on how it was reported in COMPUSTAT. COMPUSTAT provides data for "labor expense," (XLR) (6) as well as data on a variable XLRF, which indicates whether XLR includes pension expense or does not. When XLR did not include pension expense, we used its value as a measure of wage and salary expense. When XLR did include pension expense, we calculated the total value of pension expense from the Form 5500 data and subtracted it from XLR. The resulting variable was then used as our measure of wage and salary expense. (7)
Pension expense measures the accrued cost to the company from operating its pension plan(s) in a given year. For defined contribution plans, (8) pension expense equals the company's contributions to plans of this type. For defined benefit plans, pension expense is equal to the present value of future benefits accrued by employees during the year, also known as the plan's "normal cost" for the year. The sum of contributions to defined contribution plans plus the normal cost of the company's defined benefit plan(s) is therefore equal to the company's total pension plan expense.
Pension income measures the increase in employee wealth due to participation in the plan over a specified period of time. Employee income from participation in a defined benefit pension plan is equal to the company's normal cost. Income from participation in a defined contribution plan consists of two parts: employer contributions to the plan plus returns achieved by investment of plan assets. Assuming that capital markets are efficient, the sum of these two is the best measure of the present value of future (pretax) benefits which employees will enjoy as a result of participation in the plan in a given year.
We divided pension plans into five categories: ESOP's with a 401K feature, ESOPs without a 401K feature, profit-sharing plans with a 401K feature, profit-sharing plans without a 401K feature, and defined benefit pension plans. (A 401K feature permits employees to contribute to the plan via salary reductions on a tax deferred basis. Because of the discretionary employee contribution component, it is to be expected that the extent of performance contingency of income earned under 401K plans may be lower than the extent of performance contingency for plans to which only the company can contribute.) For every company which supplied wage and salary data in COMPUSTAT, we searched the Form 5500 database for all qualified pension plans that it sponsored in any of the years 1979-1988 in any of these five types.
These plan data were combined with data from Standard and Poor's COMPUSTAT. Employee pension income was calculated for each plan sponsored by each company in each year, and the income for all plans of a given type was summed to arrive at a measure of total company plan income by plan type. This was then divided by the number of plan participants. (9) Results are reported for all of these categories of pension plan except for the other defined contribution category, which was represented in only a small number of the sample firms.
Many companies sponsor more than one plan of a given type in order to achieve a differentiation in the treatment of various groups of employees or to allow for the utilization of various plan features in differing combinations. For example, 47 companies in our sample in 1988 had more than one ESOP and 71 had more than one profit-sharing plan. The breakdown of number of plans by plan type is given in Exhibit 2. In cases where a company sponsored more than one defined contribution plan of a given type, we summed the contribution, return, asset and other information for all plans of a given type. If a company
Exhibit 2. Frequencies of Plan Occurrence by Plan Type Number of Profit-Sharing Defined Benefit Plans ESOPs Plans Plans 1 229 294 283 2 33 49 51 3 4 9 27 4 3 5 1 5 1 2 4 6 1 -- 12 7 1 1 5 8 -- -- 4 9 -- -- 1 10 -- -- 6 11 -- -- 2 12 -- -- 1 13 -- -- 1 15 -- -- 1 17 -- -- 1 19 -- -- 1 20 -- -- 1 24 -- -- 1 26 -- -- 1 38 -- -- 1 Note: Each cell indicates the number of companies with the stated number of plans of the given type. For example, 229 companies in the sample had only 1 ESOP, 33 companies had 2, etc.
had more than one defined benefit pension plan, we summed the value of normal cost under these plans, and this sum measured both employer cost and employee income under defined benefit plans.
This information provides a complete picture of employees' income from participation in pension plans of the companies in our sample. Exhibit 3 contains a summary of the average annual real income per employee for all of the income components included in the analysis. The average ESOP income for plans without a 401K feature was $631, compared with $1,702 for non-401K profit-sharing plans. The larger average size of the profit-sharing plans reflects the relative maturity of these plans compared with ESOPs. ESOPs with a 401K feature generated an average of $1,953, about the same as profit-sharing plans with a 401K feature. However, most ESOPs did not have a 401K feature, while the majority of profit-sharing plans did.
Exhibit 3. Average Real Income and Net Assets for ESOPs and Profit-Sharing Plans, Per Employee Average Average Type of Pension Plan Real Income Net Assets ESOPs with 401K feature $2,439 $11,337 ESOPs without 401K feature $687 $3,401 Profit-sharing with 401K feature $2,312 $9,505 Profit-sharing with 401K feature $1,873 $8,510
III. Do ESOPs and Profit-Sharing Plans
Directly Relate Nonmanagerial
Compensation to Company
Even without formalized incentive pay systems, employees in most companies benefit from good company performance by means of pay raises and increased availability of overtime work. As a result, basic wage and salary compensation is partially contingent on company performance. This carriers over to pension pay as well, because the rate of benefit accrual under defined benefit plans typically depends on wage and salary earnings. To an extent, therefore, nonmanagerial compensation has always been contingent on company performance. The proper way to ask the question is therefore: do ESOPs and profit-sharing plans relate nonmanagerial compensation to company performance more than it already is related via the traditional compensation system?
We employed three measures of company performance in our calculations: total shareholders returns, market-adjusted shareholder returns, and accounting earnings. To measure the extent of pay contingency, we estimated regressions of the following form:
[I.sub.t] measures an income component in year t, [P.sub.t] measures company performance in year t, and [P.sub.t-1] measure company performance in the prior year. This allows for the possibility that company performance has both a contemporaneous and a lagged impact on income. The regressions were estimated on one company's data at a time, and we therefore estimated as many regressions as there were companies in the sample with enough observations on the relevant income sources.
The regression specification in Equation (1) implies that there are two components to employee pay: the first component, estimated by the intercept, is not responsive to firm performance. It represents the fixed portion of pay. The second component, equal to the sume of the coefficients on the firm performance variable and its lag, reflects the performance-contingent portion of employee pay.
We measured employee income from six sources: salaries and wages, defined benefit plans, ESOPs without a 401K feature, ESOPs with a 401K feature, profit-sharing plans without a 401K feature, and profit-sharing plans with a 401K feature. We measured company performance three ways: total real shareholder returns, total excess real shareholder returns, and total real company earnings. Total real shareholder returns were calculated as follows: we extracted the percentage total return from COMPUSTAT and multiplied that percentage by the value of outstanding common stock at the end of the previous year, also extracted from COMPUSTAT. We then deflated the resulting number by the CPI. Excess real shareholder returns were calculated similarly, except that the percentage return was reduced by the percentage return of the S&P 500 in the same year, also taken from COMPUSTAT. Both the income and performance measures were adjusted for inflation and divided by the number of recipients. (We divided salary and wage income by the number of employees in the company and plan income by the number of plan participants.)
We estimated eighteen regressions per company. We then reestimated each of them using the change in employee income as the dependent variable. This approach was used by Jensen and Murphy in assessing the extent to which CEO's salaries are dependent on firm performance. We include this approach to ensure comparability of our results with theirs. This added a second group of eighteen regressions per firm.
We excluded observations from the regressions if the number of employees in the company changed dramatically from one year to the next; i.e., if employees increased by a factor of two or more, or decreased by 50% or more. In such a case, we assumed that either the data were incorrect or the company had been involved in a sale, merger or acquisition, and as a result was probably not a comparable entity over the two-year period.
We added the regression coefficients [[beta].sub.1] and [[beta].sub.2] and report quartiles of the sums in Exhibits 4 and 5. Exhibit 4 shows that most of the income components are not highly responsive to changes in market return. The component that is most responsive is salary and wage compensation. At the median, a one-dollar increase in shareholder wealth per employee increases salary and wage compensation per employee by 18.8 cents. At the seventy-five percentile, a one-dollar increase in market value increases salaries and wages by 68.1 cents. This is much greater than the 75 percentile response of ESOP or profit-sharing income, which is 2.7 cents for non-401K ESOPs and 4.4 cents for non-401K profit-sharing plans. There were not enough ESOPs with 401K features int he sample to arrive at reliable estimates of the distribution of regression coefficients. However, for 401K profit-sharing plans, the evidence is that income responsiveness to changes in shareholder wealth is somewhat more muted than is the responsiveness of the non-401K profit-sharing plans.
There are two reasons why the profit-sharing plans in the sample exhibit a greater responsiveness to the performance of employer securities than the ESOPs. First, real income per employee from the non-401K profit-sharing plans is almost three times as large as income from the non-401K ESOPs (see Exhibit 3). Therefore, even though they were much less heavily invested-in-employer securities on a proportionate basis, the profit-sharing plans held only about 25% less in dollar value of employer securities per employee. Secondly, returns on employer securities are systematically related to market returns.
The results from using Jensen and Murphy's approach are similar that the responsiveness of salaries and wages is much greater than that of ESOP or profit-sharing income. The order of magnitude for all of the responses is also preserved under the Jensen/Murphy specification, and we therefore do not report these results here. When estimating Equation (1) using excess market return, the results were extremely close to those reported in Exhibit 4, and we therefore do not report them either.
Exhibit 4 provides our estimaes of the responsiveness of income components to accounting earnings. The responsiveness of salary and wage income to accounting income is considerably lower than it is to share returns. However, income per participant in the top quartile of profit-sharing plans is relatively highly responsive to accounting earnings; i this quartile, a one-dollar-per-employee increase in accounting earnings raises a participant's income from the profit-sharing play by at least 21.1 cents. This is consistent with the institutional features of these plans discussed above, and particularly with the fact that in a minority of profit-sharing plans there is a rigid relationship between earnings and contributions to the plan. There has been no survey of formulas for contributions to ESOPs, but it is not likely that many of the nonleveraged plans make contributions on the basis of a
[TABULAR DATA OMITTED]
[TABULAR DATA OMITTED]
non-401K ESOP income rises by 6.5 cents per additional dollar of accounting earnings.
These results contrast with the common view that ESOPs and profit-sharing plans substantially increase the extent of pay contingency for nonmanagers in companies which sponsor such plans. In fact, these plans appear to provide considerably less pay contingency than basic salary and wage compensation when share returns are used as the measure of company performance. When the company performance meausre is accounting earnings, profit-sharing plans do provide substantial feedback, but ESOPs do not provide greater feedback than salaries and wages. As expected, defined benefit plans provide little in the way of performance-based income irrespective of the measure of performance.
IV. Indirect Pay Contingency
In addition to the direct relationship of plan income to company performance measured above, it is possible that the presence of one or more of these pay systems may have an indirect effect on the company's performance reward contingency via their influence on salary and wage compensation. Employees, as owners, may be less resistant to pay cuts in bad times, and/or may be more effective than non-owners at capturing a portion of total economic rents in good times. To test this hypothesis, we formed the variable TCOMP, which measures the annual total compensation in the firm per employee, including wage and salary compensation plus deferred compensation. This variable was regressed on the same measures of company performance as above, plus interactions of these company salary compensation plus deferred compensation. This variable was regressed on the same measures of company performance as above, plus interactions of these company performance measures with dummies reflecting the presence or absence of an ESOP and a profit-sharing plan. The regression was therefore:
[Mathematical Expression Omitted]
We employ two measures for X: total shareholder return per employee (R) and accounting earnings per employee (NI). E is the ESOP dummy variable and P is the profit-sharing dummy variable. Equation (2) is therefore the same equation as Equation (1) above except that it includes the dummy interaction terms. we estimated Equation (2) in first difference form to allow for differing levels of independent variables across firms. The regression was estimated across the entire sample without imposing any fixed effects. The coefficients [[beta].sub.1] and [[beta].sub.4] therefore reflect the average responsiveness of total employee compensation to the performance variable, and the other coefficients reflect any additional (or reduced) responsiveness of total compensation which results from the presence of a contingent pay plan. The total responsiveness of employee compensation to company performance in a company which has both an ESOP and a profit-sharing plan is therefore equal to the sum of coefficients [[beta].sub.1] and [[beta].sub.6].
Our regression estimates are:
[Mathematical Expression Omitted]
[R.sup.2] = 0.14, DW = 2.2, t-statistics in parentheses.
Neither model has a great degree of explanatory power, however each model contains several variables with significant coefficients. In regression Equation (3), the ESOP and profit-sharing interaction term for period t are significant and positive, implying that profit-sharing plans and ESOPs increase the sensitivity of compensation to shareholder returns. In the absence of either type of plan, a dollar of shareholder returns lowers compensation by 1.5 cents in the same period. With an ESOP only, the response is a positive 0.7 cents. With a profit-sharing plan only, it is zero, and with both types of plan in place, the average impact of an additional doolar of return per employee is to increase employee pay by 2.2 cents.
When using accounting earnings, the result is quite different. The ESOP interaction is not significant in either the current or lagged term, but the profit-sharing dummy enters negatively and significantly in both interactions. This implies that, on average, profit-sharing plans diminish the responsiveness of total compensation to accounting earnings (actually increasing the extent of negative feedback from earnings to income). Based on evidence in Section IV above, direct income from the median profit-sharing plan is positively related to accounting earnings. It therefore appears that the presence of a profit-sharing plan serves to diminish the responsiveness of one or more of the other components of income to earnings. This is consistent with the notion that profit-sharing substitutes for other components of income. The evidence here is that profit-sharing more than fully substitutes for any wage and salary increases which would have occurred in the absence of a profit-sharing plan, and that the existence of a profit-sharing plan actually makes total employee compensation less positively contingent on accounting earnings than it would be in the absence of such a plan. ESOPs appear to be neutral in the relationship between accounting earnings and total employee pay.
The results of our analysis imply that, while employee's direct income from ESOPs and profit-sharing plans is significantly and positively conditioned by shareholder returns in a substantial fraction of companies which sponsor these plans, the extent of this direct pay contingency is typically quite small. The median responsiveness of ESOP income per employee is one cent per dollar of shareholder returns per employee, while that for profit-sharing plans is approximately the same. A substantial minority of profit-sharing plans do link employee income to accounting earnings at a higher rate: the upper quartile provides a minimum of 21 cents additional income to employees through the plan for every additional dollar of accounting earnings per employee.
The indirect effects of these contingent pay plans are, however, as important or more important than the direct effects. As is to be expected, both ESOPs and profit-sharing plans raise the sensitivity of the overall compensation package to concurrent shareholder returns, but only slightly. The net impact is 2.2 cents per dollar of earnings for the average size ESOP and 1.5 cents per dollar of earnings for the average size profit-sharing plan. In contrast, profit-sharing plans on average diminish total employee pay when accounting earnings increase.
These findings underscore the fact that researchers on the quantitative impact of pay-performance contingency must measure the actual extent of such contingency as a part of their research design. To assume the ESOPs, profit-sharing plans or other alternative compensation schemes actually link pay to performance is a risky supposition, and may account for the current ambiguous state of the empirical literature on this subject.
(1) See Conte and Lawrence .
(2) We treat only deferred profit-sharing plans in this paper. According to Hewitt Associates , 78% of all profit-sharing plans are deferred, while cash-only plans represent 4%, and the remaining 18% combine elements of both.
(3) See Kruse .
(4) See conten and Svejnar  and Scholes and Wolfson  for a description of the tax advantages. Chang  coded the ESOPs in his event study sample according to the publicly given reasons for adopting the plan. He utilized four categories of reasons: employee benefit, takeover defense, wage concession, and leveraged buyout. Tax advantages are not explicitly mentioned as a reason. However, ESOPs provide a tax-advantaged mechanism to accomplish each of these goals.
(5) See Jehring .
(6) The variable XLR includes salaries, wages, pension costs, profit-sharing and incentive compensation, payroll taxes and other employee benefits. As discussed in the text below, our methodology involves subtracting pension expense, defined as the sum of employee income from participation in pension plans of all types, from XLR, and using this as a measure of salary and wage expense. Our salary and wage expense variable therefore includes salaries, wages, payroll taxes, and nonpension employee benefits. On average, therefore, our estimate of salary and wage expense is overstated by the value of payroll taxes and nonpension employee benefits. The inclusion of health expenses will not substantially affect our conclusions because health expenses per nonmanagerial employee do not vary strongly with salary. Inclusion of payroll taxes biases our estimate of wage and salary pay contingency upward by a factor equal to the sum of federal social security tax plus state unemployment compensation tax as a percentage of salary and wage expense. Federal social security tax varied over the period of analysis, and the unemployment compensation tax rate varies across states, across time and across companies in the same state at the same time. Overall, however, the sum of, payroll taxes probably average about 15% of salary and wage expense over the period of analysis, and therefore our estimated sensitivity of S&W expense may be biased upward by about this much when COMPUSTAT included pension expense in XLR. This occurred in approximately 9% of the observations in our data set.
(7) COMPUSTAT reports employee benefit expense as it is reported in the company's annual report. This may or may not include administration costs. When subtracting pension expense calculated from the Form 5500 data, plan administration costs are not included. Therefore, if the company includes administration costs in its calculation of pension expense, the measure of salary and wage expense is biased upward by the amount of the plan administration expense. However, salary and wage expense was generally not included in COMPUSTAT's XLR variable.
(8) Qualified pension plans fall into two categories: defined benefit plans and defined contribution plans. Both plan types provide for individual employee accounts. However, the method of determining retirement benefits differs substantially. Defined benefit plans provide an annual or lump sum retirement benefit based on factors specific to the plan, usually involving salary levels and years of service. Employers make contributions to the plan in relation to (their forecast of) their future payout liabilities. Irrespective of the performance of the plan portfolio, the employer is required to provide the benefit which the retiring employee is entitled to under the plan formula. The employer therefore bears the risk associated with the performance of the plan's investment portfolio. On the other hand, defined contribution plans do not promis a specific future payout. The contribution is determined by factors specific to the plan, and the payout then depends on the investment performance of the resulting plan assets. Employees therefore bear the investment risk of defined contribution plans. Profit-sharing plans are the most common form of defined contribution plans, and ESOPs fall into this category as well.
(9) When the number of plan participants varied between plans of a given type, we used the following algorithm to select a total number of participants for the given plan type. If the sum of plan participants as given in the Form 5500 was less than 1.2 times the total numbers of employees as given by COMPUSTAT, we assumed that the number of (Form 5500) participants was equal to the maximum number of participants in a plan of the given type. We used the facto 1.2 rather than 1 because participants and employees were measured at two different points in time, and we therefore allowed for a potential variance of up to 20%. If the sum of the participants exceeded 1.2 times the number of employees, we assumed that the plans covered different (mutually exclusive) employee groups, and took the sume of the participants in plans of a given type as the number of total participants in that plan type. In general, the COMPUSTAT measure of total employees corresponded closely to the Form 5500 measure. Because Form 5500 data is not audited, we screened for large time-series variances and rejected less than 5% of the observations on the basis of large employment jumps (a factor of 10 in either direction).
 J. Brickley and K. Hevert, "Direct Employee Stock Ownership: An Empirical Investigation," Financial Management (Summer 1991), pp. 70-84.
 A.A. Alchian and H. Demsetz, "Production, Information Costs, and Economic Organization," American Economic Review (62, 1972), pp. 777-795.
 G.P. Baker, M.C. Jensen and K.J. Murphy, "Compensation and Incentives: Practice vs. Theory," Journal of Finance (July 1988), pp. 593-616.
 A. Blinder (ed.), Paying for Productivity: A Look at the Evidence, Washington, D.C., The Brookings Institution, 1990.
 Bureau of Labor Statistics, Employee Benefits in Medium and Large Firms, Washington, D.C., USGPO, Bullentin 2281, 1986, p. 81.
 R. Bushman and R. Indjejikian, "Accounting Income, Stock Price and Managerial Compensation," University of Chicago, August 1991.
 Chamber of Commerce of the United States, Employee Benefits 1988, Washington, D.C., 1989.
 S. Chang, "Employee Stock Ownership Plans and Shareholder Wealth," Financial Management (Spring 1990), pp. 48-58.
 S. Chaplinsky and G. Niehaus, "the tax and Distributional Effects of Leveraged ESOPs," Financial Management (Spring 1990), pp. 29-38.
 M. Conte and H. Lawrence, "Trends in ESOPs," Trends in Pensions 1992, Washington, D.C., USGPO, 1992.
 M. Conte and J. Svejnar, "The Performance Effects of Employee Ownership Plans," in Paying for Productivity: A Look at the Evidence, Alan S. Blinder (ed.), Washington, D.C., The Brookings Institution, 1990, pp. 143-181.
 Hewitt Associates, 1985 Profit-Sharing Survey, Chicago, Il, Profit-Sharing Council of America, 1985.
 Hewitt Associates, 1987 Profit-Sharing Survey (1986 Experience), Lincolnshire, Il, 1987.
 J.J. Jehring, "The Development of the Profit-Sharing Idea in the U.S.," in A New Approach to Collective Bargaining?: Progress Sharing at American Motors, J.J. Jehring, W.J. Howell, Jr., and L.R. Tripp (eds.), Madison, University of Wisconsin, Center for Productivity Motivation and Industrial Relations Research Center, 1962.
 M.C. Jensen and W. Meckling, "Rights and Production Functions: An Application to Labor-Managed Firms and Codetermination," Journal of Business (Vol. 52, No. 4), pp. 469-506.
 M.C. Jensen and K.J. Murphy, "Performance Pay and Top-Management Incentives," Journal of Political Economy (April 1990), pp. 225-264.
 D. Kruse, "Pension Substitution in the 1980's: Why the Shift Toward Defined Contribution Pension Plans," NBER Working Paper No. 3882, October 1991.
 E. Lawler, High-Involvement Management, San Francisco, Jossey-Bass, 1986.
 D. Mitchell, D. Lewin and E. Lawler, "Alternative Pay Systems, Firm Peformance, and Productivity," in Paying for Productivity: A Look at the Evidence, Alan S. Blinder (ed.), Washington, D.C., The Brookings Institution, 1990, pp. 15-88.
 O.S. Mitchell and S. Pozzebon, "Wages, Pensions and the Wage-Pension Tradeoff," Cornell University, Working Paper, August, 1987.
 M. Scholes and M. Wolfson, "Employee Stock Ownership Plans and Corporate Restructuring: Myths and Realities," Financial Management (Spring 1990), pp. 12-28.
 M.L. Weitzman and D.L. Kruse, "Profit-Sharing and Productivity," in Paying for Productivity: A Look at the Evidence, Alan S. Blinder (ed.), Washington, D.C., The Brookings Institution, 1990, pp. 95-141.
Michael A. Conte is a Professor with the Department of Economics and Finance, University of Baltimore, Baltimore, Maryland. Douglas Kruse is an Assitant Professor with the Institute of Management and Labor Relations, Rutgers University, and a Faculty Research Fellow with the NBER.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Corporate Compensation Policy Special Issue; employee stock ownership plans|
|Author:||Conte, Michael A.; Kruse, Douglas|
|Date:||Dec 22, 1991|
|Previous Article:||Management ownership and firm compensation policy: evidence from converting savings and loan associations.|
|Next Article:||The effects of board composition and direct incentives on firm performance.|