Federal Contract Workers Living in Poverty?

In his recent op-ed in the New York Times, M.I.T. economist Paul Osterman makes some sensible points about the troubling social implications of low-wage jobs. Although I disagree with some of his policy prescriptions, it’s difficult to deny the struggles that low-income households face on a regular basis. High rates of poverty can have a profound cultural impact, and I think this is an issue that social conservatives should take much more seriously.

Unfortunately, to make his point Osterman cites some research conducted by the Economic Policy Institute (EPI) that suggests “20 percent of federal contract employees earned less than the poverty level for a family of four, as opposed to 8 percent of traditional federal workers.”

This seemed like a pretty dubious finding, so I looked into the research methodology employed by EPI and was pretty shocked at what I uncovered.

Since the federal government doesn’t maintain any data on the number of federal contractors or their average wages, EPI collected information on the cost of each individual government contract from a database maintained by the General Services Administration’s (GSA). The GSA database also provides a record of which industry the contract targeted.

EPI compared the cost of each contract against the Bureau of Labor Statistics’ (BLS) Domestic Employment Requirements Matrix, which provides information on the number of jobs that we’d expect to see generated -- both directly and indirectly -- across all industries based on a $1,000,000 input in a particular industry.

The concept here is pretty simple. If the BLS matrix showed that a $1,000,000 input in the construction industry should be expected to generate three jobs, EPI assumed that a $100,000,000 investment in federal construction contracts would generate 300 federal contract workers.

After figuring out the total number of federal contractor workers in each industry, EPI multiplied this number by the proportion of workers in that industry who earned below the poverty wage level. This gave EPI a figure of approximately 20 percent of federal contract workers living below the poverty line.

As EPI notes, “[t]his methodology assumes that the same proportion of contract workers earn poverty level wages as workers in the private sector.” (A poverty level wage is defined as a wage rate that would place a full-time worker supporting a family of four below the poverty line.)

Even if we accept all of EPI’s other assumptions, the idea of industry-level wage equivalence between federal contract workers and other private sector workers is completely indefensible. A huge number of federal contract workers are subject to laws like the McNamara-O'Hara Service Contract Act (SCA) and the Davis-Bacon Act (DBA), which require contractors on federally-funded projects to pay their workers at least the prevailing wage in the industry. It’s almost nonsensical to assume these workers represent a random cross-section of workers in the industry. In fact, they are legally entitled to a wage that is at or above the industry average.

Moreover, if we can assert without evidence that federal contract workers are paid the same as private sector workers in a given industry, isn’t it just as fair to assert that federal contract workers are paid the same as federal employees in that industry area? The researchers who put this study together may believe that federal contract workers are paid less than similarly employed federal workers, but making this assumption undermines the entire analysis. EPI is simply asserting what it's attempting to prove.

There are two major lessons that I think need to be drawn from this example.

First, not all “research” is good research. It bothers me how often we hear that there is research “on both sides” of an issue. A blanket statement like this is an easy way to avoid checking the methodological rigor of different studies. This isn’t to say that a left-leaning or right-leaning think tank can’t do good work. It simply means that we need to understand the research design being employed so we can decide for ourselves whether the findings are valid.

Second, we shouldn’t simply trust research findings cited by an op-ed columnist to make a rhetorical point, particularly when those findings are produced by a research institute that is ideologically inclined to generate a particular result. Again, we should look at the study methodology and think through some of the assumptions.

I suspect that, when confronted with the assumptions inherent in this analysis, most people would find the results of EPI’s study questionable. But it’s hard to know that unless you’re nerdy enough or bored enough to do some real digging.

Update: A new study from the Project on Government Oversight (POGO) suggests that "the federal government approves service contract billing rates—deemed fair and reasonable—that pay contractors 1.83 times more than the government pays federal employees in total compensation, and more than 2 times the total compensation paid in the private sector for comparable services." I haven't combed through the methodology of this study yet, but a cursory examination reveals that it's far more sophisticated. At the very least, this study suggests that the key assumption in EPI's analysis is highly questionable based on other research findings.