Benchmarking Benefits Methodology Sheet

Michigan’s public-sector workers are generally understood to have a more expensive benefits package than private workers. But getting a precise figure on the average level of benefits offered by the two sectors involves obtaining governmental statistical information from a number of sources, ensuring that definitions match, making needed assumptions, and then performing alternative calculations to check for robustness.

Much of the work was done for an article in 2009[1] that estimated the difference at $5.7 billion. Since publishing, interested parties have requested a more in-depth look at the methodology. This paper provides that information.

Assembling the data and mapping the calculations

No single source exists for a detailed breakdown of the benefits offered to public- and private-sector members on a state-by-state basis. Because of this, a series of assumptions needed to be made in order to estimate savings or costs from benchmarking public-sector benefits. The author proceeded on two principles: be as comprehensive as possible and be as conservative as warranted.

Michigan’s private-sector benefits information ultimately comes from the Bureau of Labor Statistics’ National Compensation Survey. The survey provides a breakdown on the costs per hour for a private-sector employee in the “East North Central” region, which includes Michigan, Illinois, Indiana, Ohio and Wisconsin.

The benefits information it covers includes the average cost per hour worked for the following types of benefits:

• Paid leave
• Vacation
• Holiday
• Sick leave
• Other paid leave
• Supplemental pay
• Overtime and premium
• Shift differentials
• Nonproduction bonuses
• Insurance plans
• Life
• Health (Which includes medical, dental, vision, prescription drug and other health insurance coverage.[2])
• Short-term disability
• Long-term disability
• Retirement and savings
• Defined benefit
• Defined contribution
• Legally required benefits
• Social Security
• Medicare
• Federal unemployment insurance
• State unemployment insurance
• Workers’ compensation

The NCS also collects data on access to “nontraditional” benefits such as child care assistance, flexible workplaces, long-term care insurance, employee assistance programs, subsidized commuting and stock options.[3] These, however, are not quantified nor are they included in its breakdown of the East North Central region.

A frequent question after publishing is whether bonuses were accounted for as a benefit by the private-sector measure. It was believed that private-sector recipients received bonuses, counted as salary by the survey, which mitigated or eliminated benefit disparities with the public sector.

This belief was mistaken. The BLS includes “nonproduction bonuses” as a benefit. These were payments to employees that were not directly tied to an individual or group’s production.[4] These types of bonuses are also available to some state and local government workers around the country as well.[5]

This survey would ideally be the source for Michigan’s average public-sector compensation along with private sector information. However, only a national average of state and local government employees’ benefits is available from this survey, so it was not used to compare Michigan public compensation. The national, state and local government employee compensation data was explored as an estimate for Michigan’s state and local government employee compensation. However, as shown below, more state-specific measures appear to be a better proxy for the average level of benefits provided to Michigan’s state and local government employees.

The state Civil Service Commission’s annual workforce report was used to estimate the benefits received by Michigan’s public sector. The report includes detailed information on the state’s 50,000 employee workforce — around 13 percent of all the state and local government employees in Michigan. The data serves as a proxy for the benefits obtained by all state and local government employees in the state.

Specifically, benefits levels are found in table 2-3 of the document.[6] It includes the following breakdown of benefits:

• Legally required payments
• FICA
• Unemployment compensation
• Workers compensation
• Group insurance
• State health plan
• Health maintenance organization
• Dental
• Vision
• Life
• Long-term disability
• Flexible benefits allowance
• Other cash payments
• Terminal annual and sick leave
• Terminal longevity payments
• Professional development
• Uniforms and dry cleaning
• Pension and deferred items
• Retirement or pension plans
• FICA for early-out retirees
• Defined contribution pension plan
• Banked leave time for employer contribution

The figures are published as a percentage of payroll.[7] However, the figures are summarized in a dollar value by multiplying this percentage of payroll by an imputed average annual salary for a state classified employee.[8]

Other data received from the Civil Service Commission after publishing the original article indicate that average wages and benefits may be underestimated, though by less than 10 percent. The average wages in the imputed workforce report were $53,495 and benefits of $31,107. By calculating an average based on total payments for wages and benefits divided by full-time equivalent employees, average wages are $58,830 and benefits are $34,209. The information from the workforce report was used because the workforce report figures are published accounts of state employee compensation.

As indicated, the categories compare well with the BLS breakdown. There are a couple of differences, however. Terminal annual and sick leave payments are not the same as payment for vacation, holiday or sick leave as in the NCS survey. Terminal annual and sick leave payments are cash payments for unused vacation and sick leave when an employee retires. As such, this would understate public-sector benefits.[9]

Employer-paid professional development fees and uniform and dry cleaning benefits offered to some state employees are also not included in the NCS survey. The state’s workforce report indicates that this equates to 0.24 percent of base salary, less than a percent of total benefits.[10]

Thus, the data available from the NCS and the state’s workforce report cover much of the same information and better alternatives are unavailable.

The data is categorically comparable, but the figures are reported differently. NCS information is available as a cost-per hour while the workforce report is presented as an annual estimate. The state annual estimates were divided by 2,088, the figure used by the state as the average number of hours in a full-time work year in its workforce report, in order to make the annual figure an hourly figure.

With the average differences in benefits in hand, it is only a matter of finding out how many employees to apply the figure to and to annualize the difference.

Data on the number of government employees is available from the Census Bureau’s governments division. It surveys every state government and a sample of local governments to determine the March payrolls, including the number of full-time, part-time, and full-time equivalent employees.

These include members of every Michigan state or local government organization — the state, state universities, community colleges, school districts, counties, municipalities, townships and special authorities — all governmental units in the state. The 2007 report, the most recent available at the time of publishing, shows that there were 411,659 state and local government employees.

The difference in average hourly benefits was annualized by multiplying it by 2,088. The annual disparity was then multiplied by the number of full-time state and local government employees in Michigan to get $5.7 billion.

Information is available from the Census Bureau on the number of part-time employees in Michigan as well as the number of full-time equivalent employees. However, these employees are excluded from this calculation. Part-time employees may or may not be offered paid leave, retirement and insurance benefits. This is somewhat inconsistent with the private-sector averages since they included employees that may or may not be eligible for benefits beyond those legally required.

Technically, a large portion of part-time employees are offered some benefits, though not always health insurance. State classified employees, judges, state police and public school employees are placed in state pensions systems and governments set aside money into pension funds for the employees regardless of full- or part-time status.

Also, private-sector benefits are calculated as a percentage of total employees, not simply those who are full-time employees.

Including the number of full-time equivalent employees in this calculation raises the difference between public and private benefits to $6.9 billion. However, using only full-time employees in the comparison is a conservative measure.

Testing the assumptions and checking for robustness

There are a few assumptions made in the calculations described above that can influence the estimated disparity between public and private benefits, including the following:

• That the East North Central Region is an adequate approximation for Michigan.
• That the state and local benefits measure for the United States is not an appropriate approximation for Michigan state and local government employees.
• That the Michigan state classified employment benefits package is roughly equal to the benefits package received by other state and local government employees in the state.

There are also data sources that provide similar information on benefits that can be used to test the robustness of the calculation.

To test the assumption that the regional measures were fit to be used as a Michigan measure, wages for the East North Central Region listed in the NCS report were compared against average wage rates for Michigan listed in the Quarterly Census of Employment and Wages.[11] Similar wage rates imply that the East North Central region is a good benchmark for Michigan compensation.

Average weekly wages were compared to the average East North Central hourly wage multiplied by 40 to account for an average workweek. Since the QCEW information also includes vacation, sick leave and holiday pay as part of wages and salaries, the NCS paid leave information data was added back into salary information.

Discrepancies in quarterly wages between the two range from the Michigan wages being 3 percent lower to being 12.9 percent higher. The average difference from the 2004 to 2008 period showed that Michigan was 3 percent higher than the East North Central average. For the sample period (the last quarter of 2007 to the third quarter of 2008), the average difference was only 1.4 percent.

While benefit compensation is unobtainable from this source, NCS was at least a representative fit for Michigan wage compensation and this calculation bolsters the case for using NCS East North Central information to compare Michigan public and private benefits costs. If Michigan private sector benefits maintain the same sample period 1.4 percent difference as wages, then adjusting the private sector benefits upwards of 1.4 percent would decrease savings estimates by $99.1 million.

To test the assumption that Michigan’s civil service compensation figures are a more appropriate match to Michigan’s state and local compensation than the national figures, we compared the Census-reported information on payroll to the two different proxies.[12] If the figures for wages are consistent, then figures for benefits are likely consistent as well.

In order to convert the Census figures to hourly rates, the March 2008 payroll was divided by the number of full-time equivalent employees and then by the number of working days in March (21), and by the number of working hours in a day (8).

As the table below indicates, the average hourly comparison between different public sectors is different by only pennies. The national public-sector average wage rates, in contrast, are 12 to 15 percent different from the Michigan public-sector average wage rate.

A final check was performed on calculations since there is an alternative source for information on benefits, though it is less comprehensive than the NCS and civil service data.

The Bureau of Economic Analysis reports on state personal income from a number of sources, including wages, “supplements to wages and salaries” (which include retirement benefits, payments for insurance benefits), and government social insurance payments. These benefits categories are consistent with the NCS “legally required benefits,” “Retirement and savings,” and “insurance plans.” However, it would be missing supplemental pay and paid leave, which would be included in the BEA’s wages and salaries category.[13]

The BEA reports on these benefits as supplemental pay in its compensation reports. It also reports on wages and salaries, where supplemental pay is excluded. Subtracting the two categories for the private sector and for the state and local sectors gives an approximate level of benefits for the two sectors.

To get an average benefits level, the total benefits figures were divided by the number employees in the industry as reported by the BLS establishment survey.[14]

Still, taking the average compensation levels by industry verifies the magnitude of benefits discrepancies. Using the two different employment calculations, state and local employees receive benefits worth 24.3 to 50.9 percent higher than private-sector workers. However, the imputed calculations involve benefits levels that are significantly less than reported information. For instance, the costs of retirement, insurance and legally required benefits per hour for Michigan private-sector workers is estimated at only $2.76 to $3.33 per hour, depending on which employment estimate is used for the BLS. This is significant since the NCS reports these costs at nearly double that rate — $5.47 per hour.

So while these figures show that there are discrepancies between public and private sector benefits, it does not include enough benefits and the analysis used to impute the measures is likely to underestimate benefits levels for both private and public employees.