Longer Overtime Hours: The Effect of the Rise in Benefit Costs

Order Code 97-884 E CRS Report for Congress Received through the CRS Web Longer Overtime Hours: The Effect of the Rise in Benefit Costs Updated September 6, 2000 Linda Levine Specialist in Labor Economics Domestic Social Policy Division Congressional Research Service ˜ The Library of Congress Longer Overtime Hours: The Effect of the Rise in Benefit Costs Summary Anecdotal accounts of the increased encroachment of regularly scheduled overtime on employees’ family or social lives and the heightened use of overtime at manufacturers who have not rehired all their employees let go during the 1990-1991 recession have suggested to some observers that it may have become cheaper for firms to extend work hours than to hire additional employees. Between 1956 and 1963, manufacturing workers put in an average of 2.5 hours of overtime. In the next 29 years, overtime at factories lengthened less than 2 minutes per year on average to 3.4 hours a week. Between 1993 and 1999, in contrast, overtime hours expanded an average of almost 13 minutes per year to 4.9 hours a week. Although millions of jobs were created in nonmanufacturing industries during the 1990s, some contend that more people would have been hired had a change not occurred in the employmenthours calculation of firms. All other things being equal, a firm makes its employment-hours calculation based on the least cost combination of the number of workers and the number of hours per worker. A factor thought to influence this optimal combination is the peremployee (quasi-fixed) non-wage cost of labor: the higher these costs relative to total labor costs or the higher these costs compared to the overtime wage rate, the more likely firms are to substitute longer hours for new jobs. Per-employee costs, which do not vary with the number of hours employees work, include employer spending associated with employee hiring and firing (e.g., training and severance) and with certain benefits (e.g., health insurance). As studies of the relationship between the use of overtime hours and quasi-fixed non-wage labor costs have produced mixed results, claims about the job-creating effectiveness of increasing the Fair Labor Standard Act’s (FLSA) overtime premium should be viewed with caution. The coincidence of more overtime and little factory job growth could prompt proposals to raise the FLSA’s overtime premium or to shorten its standard workweek. These measures might not create as many jobs as expected if firms substitute capital for labor in the production process or if they pass the increase in labor costs on to consumers through higher prices which could then prompt a decrease in product demand and employment. The unemployment rate also might not fall as much as anticipated if workers whose hours are curtailed moonlight to maintain earnings or if the unemployed lack the skills needed for newly created jobs. Some could argue that because a mismatch already exists between job openings and jobseekers at this point in the business cycle, the government should take no action. Others might seek to have mandated benefit costs curbed. Workers’ compensation is more a state than federal program, and there are many factors besides state-set benefit levels and eligibility rules that affect the cost of the program to employers. Social security, because of its fairly high taxable wage ceiling, is largely a variable (per-hour) rather than a fixed (per-employee) cost, which means cutting payroll taxes would likely favor longer hours over more jobs. The low taxable wage base set by the government makes its share of the unemployment insurance (UI) program largely a fixed cost for firms, but much of UI’s financing is determined at the state level. Contents The Trend in Overtime Hours Worked . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Short-Run Explanations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Long-Run Explanations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Policy Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Raising the Overtime Premium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Initiating Work-Sharing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Restraining Benefit Costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 7 8 9 List of Tables Table 1. Average Weekly Overtime Hours of Production Workers in Manufacturing Industries, 1956-1999 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Longer Overtime Hours: The Effect of the Rise in Benefit Costs The failure of Verizon management and union negotiators to settle on new terms in expiring collective bargaining agreements led to a lengthy strike in August 2000 that renewed interest in employers’ seemingly increased use of overtime. In this instance, the concern about working longer hours chiefly was portrayed in the media as making it difficult for people to balance work and family responsibilities. In other words, it was reported that the telecommunications company regularly and with little notice required employees to put in overtime which, in turn, caused them to alter daycare arrangements or make other last-minute changes in family or social activities. Previously, some observers noted that manufacturers had employees working more overtime while little job growth meant that factory employment did not regain its pre-recession level by the end of the 1990s. Although many millions of jobs have been created in nonmanufacturing industries during the rebound from the 1990-1991 recession, these individuals believe that even more people would be hired if a change had not occurred in the employment-hours calculation of firms. An explanation that has been offered for employers’ greater reliance on overtime hours is the increase in employee benefit costs compared to overtime costs. About 6 decades ago, Congress set the overtime premium that firms must pay employees for working more than 40 hours a week in the Fair Labor Standards Act (FLSA) at 1½ times a covered worker’s straight-time wage rate. This report examines whether employers did, in fact, rely more on overtime in the 1990s than in earlier decades. It then explores why overtime hours might fluctuate in the short-run and why they might have increased in the long-run. The report concludes with an analysis of whether raising the FLSA’s overtime premium, implementing work-sharing, or curbing the growth of legally required employee benefits would prompt firms to create more jobs. The Trend in Overtime Hours Worked Statistics on overtime hours are regularly collected for only production workers in manufacturing industries who account for a small share of total employment. In 1999, for example, production workers in manufacturing industries made up just 12% of all employees on private nonfarm payrolls and just 14% of all production or nonsupervisory employees on private nonfarm payrolls.1 While one should therefore be cautious about extrapolating the overtime versus employment experience of factory 1 U.S. Bureau of Labor Statistics. Employment and Earnings, March 2000. CRS-2 workers to other sectors of the economy, the previously mentioned labor-management dispute at a major telecommunications company suggests that the following analysis may be more broadly applicable. Table 1. Average Weekly Overtime Hours of Production Workers in Manufacturing Industries, 1956-1999 Year Overtime hours Year Overtime hours Year Overtime hours 1956 2.8 1971 2.9 1986 3.4 1957 2.3 1972 3.5 1987 3.7 1958 2.0 1973 3.8 1988 3.9 1959 2.7 1974 3.3 1989 3.8 1960 2.5 1975 2.6 1990 3.6 1961 2.4 1976 3.1 1991 3.6 1962 2.8 1977 3.5 1992 3.8 1963 2.8 1978 3.6 1993 4.1 1964 3.1 1979 3.3 1994 4.7 1965 3.6 1980 2.8 1995 4.4 1966 3.9 1981 2.8 1996 4.5 1967 3.4 1982 2.3 1997 4.8 1968 3.6 1983 3.0 1998 4.6 1969 3.6 1984 3.4 1999 4.6 1970 3.0 1985 3.3 Source: U.S. Bureau of Labor Statistics. The amount of overtime hours worked in manufacturing industries has been on a rising trajectory. As shown in Table 1, overtime among factory workers was less than 3 hours per week from 1956 through 1963. It then rose to and generally remained in the 3-hour range from 1964 to 1992. Overtime did not exceed 4 hours per week until 1993, and has remained in the 4-hour range since then. Employers’ increased reliance on overtime becomes clearer when the same data are looked at in another way. Factory workers put in an average of 2.5 hours of overtime in the first 8 years (1956-1963) of the period under consideration. During the next 29 years (1964-1992), overtime lengthened by just 0.9 hours — or 0.3 hours, less than 2 minutes, per year on average — to 3.4 hours per week. Overtime among manufacturing workers then soared by 1.5 hours per week in the last 7 years of the 1990s — or .21 hours, almost 13 minutes, per year on average — to 4.9 hours. CRS-3 Short-Run Explanations Not unexpectedly, the fluctuations in employers’ use of overtime coincide with the ups and downs of the business cycle. During a recession (e.g., 1957-1958, 19601961, 1969-1970, 1973-1975, 1981-1982 and 1990-1991), employers faced with declining demand for their products cut back on the use of labor — both in terms of the number of employees and the number of (standard and overtime) hours. For some period after a recession ends and product demand is on the rise, firms work their current employees longer hours rather than invest in new plant and equipment or in human capital. Until they feel fairly certain that good economic conditions will continue, businesses do not want to incur the quasi-fixed labor costs associated with recalling laid-off employees or hiring new employees.2 Quasi-fixed non-wage labor costs are one-time expenses which firms incur when they hire or fire workers (e.g., initial training costs as well as severance and supplemental unemployment payments) and ongoing expenses which firms incur while an employee is on the payroll that vary less than proportionately with the number of hours the employee works (e.g., health insurance premiums and pension contributions). The distinction between quasi-fixed and variable labor costs, then, is that the former “are not fully adjustable when employers alter hours of work through overtime or shortened shift lengths.”3 (The concept of quasi-fixed non-wage employment costs will be addressed more fully later in the report.) In addition to the business cycle explanation for the short-run variability of overtime hours, firms might use overtime to avoid the reduced output that could otherwise result from employee absences, equipment problems or strikes. They might also have their employees work longer hours to fill unexpected rush orders placed by customers or to meet recurring seasonal spikes in demand (e.g., Christmas and summer holiday periods). The impetus to use overtime for these reasons would seem to arise intermittently and at varying times in different industries. Unless these factors have become increasingly prevalent over the years and increasingly pervasive across manufacturing industries, they have not likely contributed much to overtime’s longterm rise in the nation’s factories. Long-Run Explanations It has been suggested that, since the 1980s, manufacturers have tried to keep inventories below historical levels and to instead meet fluctuations in demand through just-in-time production schedules. The fairly recent initiation of just-in-time production might thus be contributing to some of the escalation in manufacturers’ 2 Hart, Robert A. Working Time and Employment. Winchester, Mass., Allen & Unwin, Inc., 1987. (Hereafter cited as Hart, Working Time and Employment) 3 Golden, Lonnie M. Unions, Nonwage Labor Costs, and the Character of Labor Market Adjustment, 1929-1987. Quarterly Review of Economics and Finance, v. 32, no. 2, summer 1992. p. 53. (Hereafter cited as Golden, Unions, Nonwage Labor Costs, and the Character of Labor Market Adjustment) CRS-4 reliance on overtime in the 1990s.4 This practice is not likely to have affected the work schedules of employees in nonmanufacturing industries because their goods and services could by-and-large never be stockpiled. However, the belief that to remain competitive businesses must be able to suit their customers’ schedules (e.g., making cable television repair calls in the evenings and on weekends) appears to have become increasingly widespread in recent years. All other things being equal, a firm makes its employment-hours decision based on the least cost combination of the number of workers and the number of hours per worker. This optimal combination is achieved when, for a given increase in output, the marginal (i.e., additional) cost of hiring one more employee equals the marginal cost of working current employees one more hour. A factor widely regarded as influencing the optimal combination is the per-employee (i.e., quasi-fixed) non-wage cost of labor: increases in the share of total labor costs that is more affected by changes in employment levels than by changes in hours worked create an incentive for cost-minimizing firms to lengthen the workweek rather than expand employment; and higher quasi-fixed non-wage labor costs compared to the overtime wage rate raises the likelihood that employers will substitute overtime for additional hiring.5 Non-wage employment costs that largely are independent of the number of hours worked include employers’ expenditures associated with employee turnover (e.g., the cost of recruiting, processing, and training new employees or terminating current employees). Turnover costs typically are not included in empirical analyses because of the paucity of comprehensive data on all its elements (e.g., on employee orientation or on litigation). For example, it has been asserted that firms have become increasingly reluctant to hire new employees because new laws or regulations have enabled more groups of workers to sue them or to bring complaints against them before government agencies concerning why the individuals were not hired or were terminated.6 Non-wage employment costs that largely are independent of the number of hours worked also include employers’ expenditures associated with certain employee benefits. Although there is no standard definition of benefits, the following supplements to wages often are regarded as such: paid rest and lunch periods, washup time, travel time, set-up time, etc.; pay for time not worked (e.g., for vacations, holidays, personal illness, and family leave); health, retirement, and life insurance; legally required payments (e.g., unemployment and workers’ compensation and social security benefits); and other (e.g., child care and employee discounts as well as meal and housing allowances).7 Not all of these benefits are quasi-fixed (per- 4 From 1979 to 1989 (cyclical peaks), overtime averaged 3.2 hours per factory worker. From 1989 to 1999, overtime increased an average of 90 minutes to 4.7 hours. 5 Ehrenberg, Ronald. Fringe Benefits and Overtime Behavior. Lexington, Mass., D.C. Heath and Company, 1971. 6 See, for example, Gallaway, Lowell, and Richard Vedder. Labor Laws: Then and Now. Journal of Labor Research, v. XVII, no. 2, Spring 1996. 7 Hart, Robert A. The Economics of Non-Wage Labour Costs. London, George Allen & (continued...) CRS-5 employee) costs. Some benefits have both fixed and variable (per-hour) elements,8 which complicates their treatment for researchers and for policymakers. According to the U.S. Chamber of Commerce and the U.S. Bureau of Labor Statistics, employer costs for employee benefits increased from less than 3% of total compensation (wages and salaries plus supplements) in 1929 to almost 30% today. For the employer cost of employee benefits to be an explanatory factor in overtime’s long-term growth, it is not sufficient for it merely to have increased. This cost must also have risen more sharply than the marginal cost of overtime. In other words, the ratio of the marginal cost of benefits to the marginal cost of overtime would have to be higher today than in the past to have induced firms to utilize their existing workforces more intensively rather than hiring from the external labor market. In a number of economic analyses, researchers found a strong positive relationship between the use of overtime hours and per-employee benefit costs relative to the overtime wage rate. These studies suggest that employment might rise by, at most, between 0.3% and 4.0%; however, “these estimates are larger than the actual employment gains that would result since many of the assumptions upon which they are based prove to be invalid.”9 (The following are among the unrealistic assumptions, some of which will be elaborated on later in this report: hours and workers are perfect substitutes in the production process; straight-time hourly wages and fringe benefits are constant; employees will not moonlight; and labor is homogenous.) Evidence of statistical problems with the studies also calls into question the confidence that should be placed in these results. An analysis which attempted to overcome some of these limitations produced mixed results. An increase in the overtime premium would decrease overtime hours at nonmanufacturing establishments but have no effect at manufacturing establishments; in the case of manufacturers, an increase in quasi-fixed benefit costs was found to decrease overtime.10 One explanation offered for the statistically insignificant relationship between the overtime wage and hours in manufacturing is that the production process of manufacturers might make employment less substitutable for hours than is true in nonmanufacturing industries. 7 (...continued) Unwin, Ltd., 1984; and U.S. Chamber of Commerce. Employee Benefits. U.S. Chamber of Commerce, published annually. 8 For example, for employees with earnings below the unemployment insurance or social security programs’ maximum taxable wage ceilings, the employer’s tax is a variable labor cost (i.e., additional tax must be paid if more hours are worked and earnings rise); for employees with earnings above the ceiling, the employer’s tax is a fixed cost (i.e., the amount of the tax does not change regardless of the number of hours these employees work). Owen, John D. Reduced Working Hours: Cure for Unemployment or Economic Burden? Baltimore, MD, Johns Hopkins University Press, 1989. 9 Ehrenberg, Ronald G., and Paul L. Schumann. Longer Hours or More Jobs? Ithaca, N.Y., Cornell University, 1982. p. 15. 10 Ibid. CRS-6 Depending on the specification model used, the researchers estimated maximum employment gains from raising the overtime premium to double time of 0.5% to 1.5% in manufacturing and 0.8% to 1.8% in nonmanufacturing firms. The economists cautioned that the failure to find a statistically significant negative relationship between a change in the overtime wage rate and overtime hours in manufacturing should reduce confidence in the belief that an increase in the overtime premium would lead to a substantial reduction in the use of overtime hours in United States manufacturing.11 Another study, which did not separately analyze manufacturers and nonmanufacturers, similarly failed to find that the FLSA has spurred job growth by curbing the use of overtime.12 As previously noted, the above-described analyses assumed that the hourly wage remains constant when the overtime premium is changed. However, assuming that employees and employers have agreed upon a combination of total weekly hours and weekly earnings, a hike in the overtime premium could leave both weekly hours and earnings unchanged if employers are able to sufficiently reduce the straight-time wage rate. In other words, under this hypothesis, raising the overtime premium would not create additional jobs.13 The results of this analysis suggest that hourly wage adjustments occur, but not to the extent the employment effect is completely offset. Nonetheless, decreases in the straight-time hourly wage of workers earning above the federal or state minimum wage could limit the job-creating effectiveness of raising the overtime premium.14 Policy Implications A variety of implications for job creation could be drawn from the forgoing theoretical and empirical discussion. They range from government changing the costs of overtime and of benefits, to initiating work-sharing, to Congress’ taking no action to spur job growth in an economy with the lowest unemployment rate in decades. 11 Ibid., p. 23. 12 Golden, Unions, Nonwage Labor Costs, and the Character of Labor Market Adjustment. 13 Trejo, Stephen J. The Effects of Overtime Pay Regulation on Worker Compensation. American Economic Review, v. 81, no. 4, September 1991. 14 Exemptions are narrowly defined under the FLSA. Examples of groups of employees exempt from the Act’s overtime premium include: executives, managers, and professionals; outside salespersons; employees in certain computer-related occupations and of certain seasonal amusement or recreational establishments; agricultural workers, motion picture theater employees, and casual babysitters/persons employed as companions to the elderly or infirm; and some commissioned employees of retail or service establishments. CRS-7 Raising the Overtime Premium Some observers might recommend that the FLSA be amended to raise the premium for working more than 40 hours in a week from one-and-a-half times a covered, nonexempt employee’s straight-time wage rate in order to discourage employers from regularly scheduling overtime and thereby, to encourage them to spread to or share with unemployed workers the available hours of employment. Proposals to increase the overtime premium have been made in the past,15 with their proponents arguing that its deterrent effect has been eroded since its impositions some 6 decades ago by the growth in employer costs for employee benefits.16 How well an increase in the overtime premium might succeed at creating more jobs or at lowering the unemployment rate is questionable. Not all of the now regularly scheduled overtime hours might translate into new jobs, particularly full-time jobs, because of the reactions of employers and employees to the policy change: ! ! ! If firms substitute new employees for overtime hours, compensation costs per hour worked would rise because of the previously discussed quasi-fixed non-wage employment costs. As a consequence, firms might reduce their demand for workers and increase their use of now relatively less expensive capital.Firms might pass the increase in their labor costs on to consumers through higher prices. If consumers responded by cutting back on purchases, firms might then curtail output and employment. Raising the overtime premium would increase the amount of money that firms could save if they did not comply with the federal law. The rate of noncompliance might rise as a result, which would lower the amount of job creation associated with the policy change.If the premium increase prompted firms to cut back their use of overtime, the earnings of employees who formerly worked the extra hours would fall. These workers might decide to moonlight (i.e., take second jobs) to maintain their prior earnings level, which would reduce the number of new jobs potentially available to unemployed individuals. Employers, in order to minimize the fixed labor costs associated with substituting new employees for overtime hours, might create parttime jobs because they typically offer fewer benefits than full-time 15 Representative Conyers introduced bills in the 95th - 99th Congresses (H.R. 11784 in the 95th Congress, H.R. 1784 in the 96th - 98th Congresses, and H.R. 2933 in the 99th Congress); Representative Dent introduced H.R. 10130 in the 94th Congress; Representative Murphy introduced H.R. 5808 in the 98th Congress; and Representative Blackwell introduced H.R. 3267 in the 103rd Congress, which would have among many other things increased the overtime premium to double-time. 16 U.S. Congress. House. Committee on Education and Labor. Subcommittee on Labor Standards. To Revise the Overtime Compensation Requirements of the Fair Labor Standards Act of 1938. Hearings on H.R. 1784, 96th Cong., 1st Sess., October 23-25, 1979. Washington, U.S. Government Printing Office, 1980. CRS-8 jobs. And, the moonlighting workers mentioned above, might fill some of these part-time positions. Moreover, the very heterogeneity of labor could limit the potential employment gain or unemployment reduction associated with an increase in the overtime premium: ! The skills of those who work overtime and the skills of the unemployed would have to be similar in order for the latter to adequately function as substitutes for the former. If there were a skills mismatch between the two groups, it would reduce the employment effect of the policy change.17 A comparison of the characteristics (i.e., education, age, and occupation) of those unemployed and those working longer than 40 hours per week in March 1992 found the two groups to be quite different.18 This last point is likely to be especially relevant in the current economic climate, when the unemployment rate has been at its lowest level in three decades. Because the nation has experienced considerable job growth since the 1990-1991 recession, firms that have wanted to expand output and employment in this extremely tight labor market have had to hire workers further back in the unemployment queue. As these new hires likely are less productive (i.e., contribute less additional output) than those already holding jobs, they raise the firm’s cost of production. The cost of labor will rise even further if, in response to skill imbalances or “shortages,” firms compete for qualified workers by bidding up compensation. Instead of the unemployment rate falling much below its current level, wages and prices could rise — possibly prompting the Federal Reserve Board to raise interest rates in order to slow the rate of economic and employment growth. Consequently, other observers might argue that the government should take no action to promote additional job growth despite firms’ increased reliance on overtime work. Initiating Work-Sharing Another means that some have advocated to spread employment across more members of the labor force is lowering the number of weekly hours per employee at a single firm. This shortening of per-employee work time commonly is referred to as work-sharing. In the United States, work-sharing might take the form of reducing the 17 An analysis by Statistics Canada estimated that the mismatch between job qualifications and unemployment workers’ qualifications would reduce the job-creating effect of limiting overtime hours from 169,000 to 93,000 private sector jobs in November 1995. The geographic mismatch of jobs created and qualified unemployed workers was found to further diminish the potential employment effect of converting overtime hours to full-time jobs. Employment Policies: Converting Overtime to Jobs Not Feasible Solution, Statistics Canada Analyst Says. Daily Labor Report, January 23, 1998. 18 Fitzgerald, Terry J. Reducing Working Hours: American Workers’ Salvation? Federal Reserve Bank of Cleveland’s Economic Commentary, September 1, 1996. CRS-9 FLSA’s 40-hour-a-week standard19 or of implementing programs to preserve (as opposed to create) jobs in periods of deficient demand.20 Work-sharing has gained more interest abroad than in the United States, where in recent decades the jobless rate has been lower and where unions have relatively fewer members.21 As in the case of raising the overtime premium, the effectiveness of work-sharing at creating jobs is affected by differences in the skill mix of workers and by the reactions of employers and employees to a cutback in work time. According to a review of studies that looked at the relationship between reduced hours and employment or output, [t]he bulk of the nonexperimental evidence suggests that work-sharing can raise employment through the substitution of bodies for hours. However, most of the evidence is generated by employer and employee interactions, not by governmental intervention in markets. Employers and employees may respond differently to state-imposed hours reductions or reductions in hours negotiated by union and management, than to changes that arise from their own initiative. In addition, most of the studies do not explore the possible effects of work-sharing on hourly pay, effects that can be crucial to its impact on employment.22 An increase in hourly pay to maintain annual earnings in response to a decrease in an employee’s standard hours at a given firm would limit work-sharing’s ability to spur job creation. So too would currently employed workers’ refusal to curtail total hours. Those employees who already are at or above their desired level of hours when work-sharing is instituted might demand little or nothing in the way of compensation for a shortened workweek. However, those employees who are working fewer hours than they would like might demand an increase in the straighttime hourly wage or might choose to moonlight. Either one of these actions would dampen the impact of work-sharing on the unemployed. Restraining Benefit Costs Rather than focusing on the hours side of the equation, other interested parties might urge that the rate of increase in benefit costs be curbed in order to promote greater job growth. Federally mandated benefit programs include workers’ compensation (WC), social security (SS), and unemployment insurance (UI). 19 See legislation in footnote 15. 20 The latter form of work-sharing has been used by some states. The unemployment insurance (UI) system favors adjustment to demand through layoffs rather than through reduced per-employee hours. “Short-time compensation programs” have tried to preserve jobs and offset the UI system’s bias against work-sharing by allowing employees whose hours were cutback to receive partial unemployment benefits. 21 Hunt, Jennifer. Hours Reduction as Work-Sharing. Brookings Papers on Economic Activity, 1:1998. 22 Freeman, Richard B. Work-Sharing to Full Employment: Serious Option or Populist Fallacy? In Freeman, Richard B., and Peter Gottschalk. Generating Jobs: How to Increase Demand for Less-Skilled Workers. N.Y., Russell Sage Foundation, 1998. CRS-10 Congress historically has had a very limited role in the WC program. It sets rules only for federal, maritime, and railroad employees. Thus, the program is largely mandated by state, not federal, law. WC costs under the state systems are affected by a number of factors, only two of which — the level of benefits and rules for proving eligibility of benefits — are directly controlled by state governments. Other important factors include the return on insurance company investments (the majority of program costs are financed through private insurance companies); employers’ injury experience; the riskiness of the covered employment; changes in medical costs, workplace safety, and in payroll size; the extent of regulation by state insurance commissioners; and the degree to which companies choose to self-insure. Because numerous factors influence the cost of WC, states would likely have to make relatively major changes in benefit levels or eligibility rules in order to have much of an impact on employer expenditures. With respect to SS, it is unlikely that Congress would want to reduce the cost to employers because that would mean less revenue to pay for retirement and health benefits for the elderly — programs expected to experience increasing financial strain in the coming years as the baby-boom generation leaves the labor force. Moreover, as previously mentioned, some benefits are a mixture of per-hour (variable) and perworker (fixed) costs. The employers’ SS contribution is largely a variable cost because employees typically have earnings below the program’s maximum taxable wage ceiling. In other words, the amount of employers’ payroll taxes vary directly with the number of hours worked by, and hence, wages paid to employees up to the program’s earnings ceiling. Therefore, “it is unlikely that reduction in firms’ payroll taxes would provide incentives to create new jobs. [Based on] the underlying theory, ... it is average hours rather than employment that may rise as a result of cuts in taxes.”23 In contrast, the federal unemployment tax rate is assessed against a fairly low earnings level. The federal share of the tax to finance the UI program thus is by-andlarge a fixed cost for employers. Economic theory suggests that changing the program’s financing structure by increasing the federally set wage ceiling (and correspondingly decreasing the tax rate) would make the program more of a variable cost to firms and therefore reduce employers’ propensity to substitute hours for jobs. However, much of UI’s financing is determined at the state level where such factors as experience rating as well as differing taxable wage bases and tax rates greatly affect the actual cost of the program to employers. 23 Hart, Working Time and Employment, p. 274.