How Many Jobs Were Stimulated?

Tom Evslin on Grading ARRA

There’s going to be a lot of controversy and confusion nationwide on October 30 when stimulus reports from states are made public on recovery.gov. The reported numbers for “jobs created or retained”  will be particularly contentious since they will be much lower than the job forecasts by state which used to be posted on recovery.gov (I can’t find them in the new and excellent redo of the site) and the total reported by the states won’t come close to the total jobs forecast to be created by the stimulus bill (aka American Recovery and Reinvestment Act or ARRA).

Actually, the numbers SHOULDN’T match but that’s tough to explain. Nevertheless, in the interest of transparency, I’ll try an explanation here.

The original forecasts were made using an econometric model. In English, that means that a bunch of economists decided how a computer should make an estimate of how many direct, indirect, and induced jobs will be created or retained by each dollar of government spending. You’ll find all of the details of the methodology at http://www.recovery.gov/Documents/Jobs_Report_Final.pdf.

Direct jobs are the people working directly on a project; think of the crew that you see fixing a bridge. Indirect jobs are the jobs created or retained at suppliers to the project; think of the people who made the steel that goes into the bridge. Induced jobs are the jobs created when people spend the money they earned or received as benefits; think of the people working in the mall where the road crew spends the money they wouldn’t have if they didn’t have jobs.

The economists go on to estimate that one job can be created for one year by $92,136 of federal government spending, $145,351 of tax cuts, or $116,603 of state fiscal relief (which presumably saves us from either having to cut services or raise taxes at the state level). There is no mention in this paper on current or future job loss caused by the increase in federal debt or the need to pay it back (I’m not opining, just reporting).

OK, now you understand where the projections came from. There’s roughly $787 billion allocated by ARRA so the economists just have to divide this into spending, tax cuts, and state tax relief to figure out how many years of jobs will be created or retained. Their answer is 6.8 million job-years through the end of 2012. They estimated 8,000 jobs for Vermont with this methodology. That’s a big number since we lost about 10,000 jobs total in the recession.

Let’s look at what the states are going to report:

1.       We report only the jobs created by the money that flowed to or through the state government – not the tax cuts, not the direct grants to companies, cities, and counties and other recipients. That’s “only” $246 billion out of $787 billion total.

2.       We report (this time) only on money spent through September 30, 2009. “If you look at the Recovery Act as a two-year marathon, we're at the nine-mile mark,”  says Vice President Joe Biden.

3.       We report only the direct jobs created by that money – the people building the bridge but not the ones who made the steel or sold goods to the crew. In most cases we report prime contractors but not subs. 80% of the money that came into Vermont through mid-September was for Medicaid ($105 million). None of this created or retained DIRECT jobs so no jobs are reported for this, although we would certainly have been in a pickle without this money and have a huge hole to fill when it’s gone.

4.       We’re not really reporting job-years but rather an estimate of the number of full time equivalents that were hired for the period (usually one quarter this time).

The truth is that we’ll never know how many jobs were saved by ARRA; the economy is simply too complex to give us a simple answer. A woman from the Federal Reserve Bank of Boston called our office the other day and asked, “If you didn’t have the ARRA Medicaid money, would you have raised taxes, cut benefits, or cut somewhere else?” We answered “probably all of the above” but we’ll never know in what proportions those three things might have happened or what the effects of them might have been.

We’ll report a job number as required by law. When we report it, we’ll give even more detail on how we calculated it. You now know why the state numbers won’t match the federal estimates.

Even though you should look at both the federal job estimates and the state reports with several grains of salt, the right questions to ask are, I think, “Was our money spent well?” and “What are we going to do in a year when most stimulus dollars will have been spent?”

Comments

[G.Cross]

Excellent explanation which includes the whole picture, thanks Tom. It is also a good lesson in the many issues related to economic forecasting. Rolling dice would probably give one results almost as accurate as those determined by complex mathematical equations.

[Milton Eaton]

If a government subcontract is cancelled, and then a new contract is opened for the same purpose, are these jobs new, created jobs? Any bets how they are counted?

It is called government accounting.

[David Usher]

From the outset 8,000 jobs 'created or saved' was a political, feel-good statement that could never be proven to have occurred even with the best tracking.

Anyone who believed that Vermont number was obtainable or measurable must have a box of pixie dust handy on his/her desk to sprinkle on whatever results, as Tom describes them, are published.

[bob zeliff]

Ok I'll ask. Was the money in Vermont well spent?

Do you think the tax breaks created jobs? How many in Vermont?