Tuesday, August 11, 2009


Early this year, projections released by recovery.gov made me skeptical that the outcome of the American Recovery and Reinvestment Act could meet the public promises on the jobs front. The issue was not the number of jobs but the cross-state equality promised in the projections. The projections were the culmination of a paper written at the Council of Economic Advisers by Jared Bernstein and CEA Chairwoman Christina Romer.

Romer and Bernstein combined proportional projections, essentially multiplying each states labor force by a constant factor, with modeling to produce a estimates of the number of jobs likely to be “Saved or created over the next two years” in each state. Figure 1 shows the number of jobs predicted to be saved or created as a share of the state’s total labor force against the unemployment rate at the time of the estimates. The results show that Idaho, a low unemployment state was expected to gain a higher proportion of jobs than Michigan, where high unemployment should have made creating jobs cheap.

The modeling work took methods used by Moody’s Mark Zandi into account, yet the results were seemingly unaffected by unemployment. Zandi’s method incorporate “resource slack,” a technical term meant to describe that people in low employment areas will take low paying jobs. When unemployment is high there is a significant amount of slack and less money is needed to induce people to work.

Not only are the results not differentiated by unemployment, but are almost constant. At the median the program was supposed to save or create 2.27 percent worth of pre-recessionary labor force worth of jobs. Even the weakest growth state, Rhode Island, was projected to gain or save 2.08 percent of it’s labor force. The only outlier in the group was DC, expected to gain or save 60 percent more jobs as a share of labor force than the average state. The results for DC likely account for the concentration of the government as a jobs provider, a sector that would grow as the government expanded programs and created new ones, like administering ARRA.

New data from Recovery.gov seems to follow this tight concentration of results. The site now provides information on the total amount spend by ARRA (divided into loans, grants, and contracts). As figure 2 shows, the money from ARRA to date has not been evenly distributed. The three biggest recipients, California, New York and Florida, have received 25 percent of all funds. Yet they are also big states.

When we compare the amount spent so far to the projected jobs impact released earlier this year, the states appear to receive a much more equal share. Figure 3 charts the cost per job in each state with the December 2007 state unemployment rate, the metric used in the jobs projections earlier this year. The trend is weakly positive (albeit not statistically so). While the figure appears to undercut the resource slack assumption, arguing that it actually costs more per job in high unemployment states to create jobs there is still a significant amount of funds to be spent.

A more complete picture will emerge at the end of this week when tentative job creation numbers will be released. I expect the new numbers to vary from those released back in the first months of this year. Likely indicating that some states have had difficulties in creating jobs.

No comments: