Too expensive – despite MMT, taxes will still have to be raised substantially.

How much would a federal job guarantee cost taxpayers? The short answer, at the size the two reports project, is it would cost about the same as the Pentagon.

6 The Job Guarantee

The CBPP report estimates an annual cost of $583 billion, while the lower and upper bounds in the

Levy report are around $400 billion and $550 billion per year, respectively, in a ten-year simulation. The Levy report’s lower cost estimate, despite higher participation than that projected in the CBPP report, arises because its authors significantly underestimate their plan’s cost of capital and worker benefits.

Table 1 breaks down the components of both reports’ cost projections, and partially corrects areas in which we believe the Levy report has demonstrably underestimated its plan’s costs.

Because of different ways of incorporating the option of part-time work, the average annual-wage figures for the two reports are not directly comparable.

The CBPP report assumes an average annual salary of $32,500, approximately equivalent to fifteen dollars per hour. To account for the fact that some work would be part-time, it converts its projected 10.7 million participants into 9.7 million “full time equivalent” workers.

The Levy report assumes a fixed wage of fifteen dollars per hour and an “average work week” of thirty-two hours, thereby incorporating a slightly larger assumed share of part-time workers in the program (assuming full time would mean a forty- hour week).

Worker Benefits

The CBPP report then adds costs for worker benefits. While the authors do not describe the benefits in concrete terms, they appear to be comparable to standard benefits received by full-time American workers. They assume a cost of $10,000 per enrollee for direct benefits plus $2,500 for FICA taxes (which go toward future benefits).

According to these projections, 27.8 percent of enrollee compensation will be in the form of benefits. This number is somewhat lower than the 31.8 percent national average reported by the Bureau of Labor Statistics (BLS) in March 201811 and therefore might be a slight underestimate.

The Levy report’s estimate for cost of worker benefits, on the other hand, is not defensible. The authors assume benefits will cost 20 percent of what each worker is paid in wages, or 16.7 percent of

total compensation. The examples they provide for benefits are health care and child care (the latter of which is not part of the BLS calculation) though it is unlikely they would want the government to withhold any of the standard benefits from enrollees.

Replacing the Levy report’s figure with the BLS national average raises the cost of every job in the plan by 22 percent. Unless the authors provide a reasonable explanation for their original estimates, the estimates must be corrected before they can inform any policy discussion.

Capital Costs

Any work done by job-guarantee enrollees will require materials, machines, and know-how. The CBPP report assumes capital expenses of $11,000 per worker. The Levy report assumes capital costs equal to 25 percent of wage costs, or just over

$6,240 per worker (given their fixed wage of fifteen dollars per hour and assumed average work week of thirty-two hours).

Capital intensity varies greatly by type of work. Because both reports are short on details regarding the work assigned to program enrollees, checking the plausibility of these estimates is challenging. Both reports further confound such efforts by offering no source for their assumed capital costs.

The nonmarket structure of the proposed programs muddies the waters even further. Because the cost of labor is fully subsidized by the federal government rather than valued at a market rate, standard spending ratios such as capital to labor and capital to output cannot provide meaningful benchmarks to evaluate the authors’ estimates.12

One point of comparison is capital expenditure per worker. The Census Annual Capital Expenditure Survey reports total spending on equipment and structures across all industries in 2016 of $1.58 trillion. Dividing by 150 million, the approximate size

Table 1. Job-guarantee annual cost estimates from CBPP and Levy reports

of the U.S. workforce in 2016, yields an estimated average capital expenditure per worker of just below $11,000. This amount mirrors the CBPP report’s estimate, but the authors often emphasize infrastructure projects, which would be on the high end of capital intensiveness.

Once again, the Levy report’s assumed expenditure appears far too low. While relying on labor- intensive projects makes sense given the plan’s full subsidization of labor costs, the report’s lack of detail makes it impossible to reliably estimate how much capital would be used. Had the Levy report instead used the census estimate of average capital spending per worker, the plan’s cost would increase by another 11 percent.

Both reports neglect the cost of human capital. Even if most of the labor used in the proposed programs is unskilled, someone presumably needs to know how to do the work. This need is likely especially pronounced for work such as infrastructure and environmental cleanup. Other proposed work such as teachers’ aides and elder care would also require significant training. Beyond hoping that experts in these fields will become unemployed and willing

to accept relatively low-wage work, some industry

professionals would have to be hired at competitive salaries, further increasing the cost of both plans.

Administrative Costs

Neither report includes the cost of designing and implementing a permanent program with over ten million beneficiaries and a budget of hundreds of billions of dollars. Once again, projecting this cost with any accuracy is difficult given the unprecedented nature of the programs and the shortage of detail in the reports. However, it is yet another reason why the cost of a federal job guarantee would likely be higher than projected in either report.

Budgetary Impact

While neither report advocates ending other government entitlement programs in the wake of

a federal job guarantee, the authors are correct

that some of the cost would be offset by savings in unemployment insurance and programs aimed at poverty. Only the Levy report attempts to estimate the size of these offsets. However, as we describe

in appendix 2, the authors fail to include obvious negative budgetary impacts along with cost savings. Since we do not explicitly estimate these negative

impacts, presenting program costs without offsets is more informative than only including the offsets.