© Copyright 2002 by the Wyoming Department of Employment, Research & Planning
The Relationship between Benefit Costs, Benefits Offered, and Industry
by: Carola Cowan, Economist and Tom Gallagher, Manager
"Steep rises in costs for essential benefits have historically tended to crowd out increases in direct wage and salary compensation."
There is a broad-based need for information on the compensation of workers. Research & Planning (R&P) participates in the State/Federal Occupational Employment Statistics (OES) program to produce state and local estimates of wage rates by occupation.1 However, there is presently no similar program to collect information on employer-provided benefits that would produce locally relevant estimates of benefit costs to employers or provide workers with information about the availability of benefit packages. To fill this information gap, R&P began developing a mail survey technique in 1999 to estimate the cost of benefits to employers and the number and type of benefits offered to workers. While there are other uses of benefit information produced, a full accounting of costs and consumption remains a central consideration. The analysis of data from Wyoming’s mail survey indicates a positive correlation between costs and availability -- per employee benefit costs increase as the benefit package expands. However, there is a need for further analysis.
Nationally, employee benefits information is collected directly by personal interview on a quarterly basis by the Department of Labor, Bureau of Labor Statistics (BLS). The collection of information is conducted through interview and the collection of personnel records at the employer’s place of business. Only large firms are contacted and detailed information is collected on actual benefits paid to a sample of employees. As a function of these collections, estimates of the Employment Cost Index (ECI) are reported on a quarterly basis. Benefits information developed through this process is extremely expensive and is of little relevance for most employers. On the other hand, the information allows for the calculation of the cost of each benefit and therefore permits estimates of the cost of the entire benefits package typically offered to and received by employees. These benefit cost estimates, in addition to estimates for direct wage and salary compensation, permit the tracking (or indexing) of the total cost of labor over time. The budget request for the BLS’ Compensation and Working Conditions activity in Fiscal Year 2002 was $74.1 million. While yielding an accounting of cost information for each benefit feature across a variety of packages, this data collection strategy does not provide information at either the local or state level.
At the national level, benefit costs for September 2001 increased 5.1 percent from September 2000.2 In contrast, wages and salaries increased by only 3.6 percent during the same period. The Employment Cost Index (ECI), according to the BLS, is an indicator of labor cost pressures.3 Given the scope of the increase in benefit costs, the increased importance of tracking benefits and their cost by employers, employees, and policy makers is understandable. Due to rising benefit costs, employers may change their compensation packages to reflect these increases depending upon whether or not the costs are associated with an essential benefit affecting all workers or a benefit component associated only with a segment of workers. The BLS reports that the increase in benefit costs in private industry was largely due to increases in employers’ costs for health insurance. Increases in benefit costs for State and Local Government workers were attributed to increases in employers’ costs of health insurance and retirement benefits.4 Steep rises in costs for essential benefits have historically tended to crowd out increases in direct wage and salary compensation.
Methods and Data Limitations
Given state and local interest in benefits information and the high cost of personal interviews with employers to obtain this information, R&P decided to join the ranks of state research offices conducting mail surveys to collect benefits information.5 R&P uses the universe file of employers from the Unemployment Insurance (UI) tax files as the sample frame to collect data on 25 different types of benefits. This set of benefits is presented as a check list separately for full- and part-time employees and is grouped into four major categories: paid leave, insurance, retirement plans, and miscellaneous benefits. In addition to benefits, information is requested regarding four groups of compensation costs.6
Unlike the ECI, which depends upon expensive personal visits by highly trained staff to request specific information on each worker’s benefit package and associated costs, R&P’s mail survey requests the employer’s voluntary participation in providing information on the characteristics of the benefit package made available to employees grouped by full- and part-time status. We also request information on total expenditures for these benefits. As a consequence of this lower cost collection strategy, we cannot directly estimate the cost of each type of benefit in the benefit package nor directly track the evolution of specific costs over time. At this point in the development of State sponsored benefit surveys, we cannot distinguish between benefit costs attributable to changes in the cost of each component of the benefit package, and costs attributable to the changing composition in the types of benefits offered and used by employees. At the same time, it is evident that benefit costs per employee are associated with the scope of benefit packages. To explore the issue of the relationship between benefit package offerings and the costs of benefits per employee, we developed an Index of Benefits Availability (IBA), and compared it to the benefits spending information employers reported to us through mail survey. The IBA represents a single number summarizing the scope and variety of benefits offered to employees.7
In order to determine whether or not there is a relationship between the number of benefits offered and the cost of benefits per employee, we calculated the IBA for each employer. We then divided each employer’s benefit costs (excluding legally required benefits) by the number of employees in the establishment to determine the average cost of benefits per employee, and compared this cost to the IBA. A problem with this technique is that we do not know if all employees in the company actually elect or are eligible to receive the benefits offered. This could lead to a relatively low cost of benefits per employee. Also, for example, the cost of employee discounts may be low compared to health insurance. The burden on the employer for benefits varies depending on the percentage the employer pays versus the percentage the employee pays. Nevertheless, when added together to form an index, each benefit has the same weight. Therefore, if employers offer many low cost benefits, their IBA would be high, but the cost of benefits per employee would be low.
As with any survey, there is non-sampling error. For example, we asked employers for the cost of paid leave. However, we do not know with certainty that it was properly allocated or reported in the appropriate cost category since paid leave is usually considered part of salaries and wages. This is also the case for some of the miscellaneous benefits such as a Christmas bonus. A personal visit to individual employers to verify benefit costs would reduce non-sampling error.8
Another limitation is that the questions pertaining to benefits offered refer to the current year while the questions about the cost of benefits refer to the previous year. Using two different reference periods could reduce the level of correlation if a company offered more benefits in the current year than the previous year. In the future we could attempt to match companies with the same characteristics over two years to determine the effect of the time lag.
Once the average cost of benefits per employee was determined, we calculated the correlation with miscellaneous benefits between the IBA (a measure of the number of benefits) and the calculated cost of benefits per employee (see Table 1). We found a positive correlation (r = .72) that was statistically significant at the all-industry level, describing the relationship between benefit costs and the number of benefits offered.
Next, we tested if the correlation held true for the individual industries. We found that the positive correlation was statistically significant for five of the ten industries: Mining, Manufacturing, Retail Trade, Services, and Government.
We suspect the industries that did not have statistically significant correlations may provide a greater number of less expensive benefits. We considered less expensive benefits to be those grouped under miscellaneous benefits in our survey (i.e., wellness programs, child day care, educational assistance, profit sharing plans, employee discounts, tool allowances, uniforms, and Christmas bonuses). These benefits were grouped under miscellaneous benefits because they did not fit in the three major groups: paid leave, insurance, and retirement benefits. However, we need improved information on the cost of each type of benefit offered to make better determinations about the relationship between total cost and benefits offered.9 This may call for a different type of information collection to supplement what we are already collecting (e.g., a statewide survey requesting that employers provide a ranking of the costs of each benefit so that we can empirically identify high cost and low cost benefits).
Table 2 shows the percentage of companies providing selected benefits to their full- and part-time employees by industry. Certain benefits were more common in those industries that had no significant correlation (between the IBA and the cost of benefits per employee) than in those industries with significant correlations. For example, a Christmas bonus, employee discounts, and uniforms had relatively high occurrences in Agriculture, Construction, and Wholesale Trade, especially for part-time employees. As seen in Table 1, these industries do not have statistical significant correlations between benefit costs and the number of benefits offered.
To test our hypothesis that the low-cost, miscellaneous benefits adversely affect the correlation between benefits and costs, we calculated an IBA excluding miscellaneous benefits and ran the correlation again (see Table 3). We found that Agriculture, Construction, and Transportation, Communications, & Public Utilities (TCPU) did not have significant correlations between benefits and costs (see Table 1), while Wholesale Trade and Finance, Insurance, & Real Estate (FIRE) did have significant correlations when miscellaneous benefits were excluded. Table 3 confirms that Wholesale Trade and FIRE provided many miscellaneous benefits. TCPU also provided many miscellaneous benefits, but its correlation was still not significant. That may be the case because there are other low-cost benefits or perhaps employers in TCPU require their employees to share a larger percentage of the costs related to benefits. The results for Agriculture and Construction may reflect the seasonal nature of those industries. Even though many employers in Agriculture and Construction offer benefits, some employees may never qualify for them due to waiting periods. Construction employees also tend to be younger and may not opt to receive health insurance or pay into a retirement plan, especially if they have to pay part of the cost themselves.10 Therefore, construction companies may offer many benefits but still have relatively low costs if their employees do not qualify or elect not to participate in the benefit programs.
We suspect that if the costs of legally required benefits go up, employers may decrease their spending on other benefits. For example, employers may reduce the number of benefits they offer to pay for the increase in legally required benefits. This reduction would lead to a lower IBA and a decrease in costs for voluntary benefits. Another possibility is that employees will be required to pay a larger share of the costs for benefits themselves. This would not change the IBA, but it would reduce costs for voluntary benefits, while increasing the costs for legally required benefits. Future research will compare this year’s data with last year’s to see if our assumption is correct.
The distinction between mandatory and voluntary benefits, which relies upon statutory criteria, needs additional research. Some benefits may be defined normatively mandatory such as health care. However, time did not permit us to conduct a comprehensive review of the literature that needs to be completed as we begin addressing more extensive questions raised by this initial analysis.
Overall, the data collected show that as the number of benefits offered by an employer increases so does the total cost of benefits per employee. This relationship was significant for seven of the ten major industries when miscellaneous benefits were excluded from the analysis. Construction, TCPU, and Agriculture are exceptions to that rule. The difference may be due to the demographic characteristics of the workforce in those industries, which is a topic under study.
Future issues of Wyoming Labor Force Trends will feature survey updates. If you would like a copy of the survey results when they become available, please call (307) 473-3804 or visit our web site at <http://LMI.state.wy.us/>.
1Wyoming Department of Employment, Research & Planning, Wyoming Wage Survey: 2000, January 2002.
2Bureau of Labor Statistics, “Employment Cost Index - September 2001,” News Release, October 25, 2001, Table B, p. 2.
3John W. Ruser, “The Employment Cost Index: What is it?,” Monthly Labor Review, September 2001, < http://www.bls.gov/opub/mlr/2001/09/contents.htm > (March 1, 2002).
4Bureau of Labor Statistics, p. 2.
5This effort was undertaken jointly with the Research Offices in Nebraska Workforce Development, Department of Labor, Labor Market Information Center and the South Dakota Department of Labor, Labor Market Information Center. Other benefit surveys are conducted by research offices in Oklahoma, Maine, and New Hampshire.
6The following are the four groups of compensation costs for which information is requested and examples of some of the items included in each group:
• Wages and salaries - including overtime
• Retirement costs - 401k, pension plans
• Legally required benefits - Social Security, Medicare, Unemployment Insurance, Workers’ Compensation
• Miscellaneous costs - uniform allowance, child care, insurance
For more information, see Wyoming Department of Employment, Research & Planning, Employee Benefits in Wyoming: 2000, June 2001.
7The Index of Benefits Availability (IBA) is the sum of the number of benefits offered by a company weighted by full- and part-time employment. For an example of how the IBA is calculated:
Number of Number of
Full-time 8 5 80% * 5 = 4
Part-time 2 3 20% * 3 = .6
8If funds become available, it would be beneficial to conduct small sample verification.
9The Employment Cost Index may be used as a proxy.
10Douglas W. Leonard, “Wyoming’s Workforce: Growing Older Faster?,” Wyoming Labor Force Trends, December 2001,
Table of Contents | Labor Market Information | Wyoming Job Network | Send Us Mail
designed by Julie Barnish.
Last modified on by Krista R. Shinkle.