The notion that public-sector wages are lower than private-sector wages is an oft-repeated claim, but one that has trouble standing up to reality. In addition, the comparisons can get complicated.
In an essay in the National Affairs journal, scholar Daniel DiSalvo provides evidence that public-sector workers in lower-wage jobs are paid more than their private-sector counterparts. For example, nationwide data from the Bureau of Labor Statistics show that government office clerks average $27,000 a year in salary, versus $23,000 in the private sector. Janitors’ wages show similar trends: $23,000 in the public sector compared to $20,000 in the private sector. (Public school and local government janitors earn even more, coming in at more than $28,000 on average nationwide.)
However, the public-sector-makes-less claim stands up better when limited to higher-paid employees. Accountants and tax preparers, for example, earn $74,000 a year in the private sector, versus just under $60,000 for local government accountants. On the other hand, national average salaries for state government employees overall are higher than the private sector average: $48,000 vs. $45,000.
Government employees come off even better when fringe benefits are included. On average, state and local public-sector employees earn $6 per hour more in wages than private-sector employees. When benefits are added, this rises to a $12 government employment premium over private-sector counterparts. Those figures are based on national statistics; Michigan comparisons may differ, in part because public-sector compensation levels have risen here over the past 10 years, while private-sector compensation has fallen.
Finally, the comparisons do not take into account the ability of many public-sector employees to begin collecting retirement benefits at age 55 or even earlier. Sometimes the benefits of government retirees even exceed their compensation as active employees.
The one-size-fits-all narrative of private-sector wages being greater than public-sector wages is not borne out in the data.