A new paper by me and George Erickcek on incentives was published online today in the journal Economic Development Quarterly. The online version is behind a paywall for non-academic readers, but it is quite similar to a prior working paper version.

This new paper is on MEGA, a now defunct business incentive program in Michigan that provided discretionary subsidies to new plants or plant expansions based on job creation. Here is the abstract of the paper:

“This article simulates job and fiscal impacts of the Michigan Economic Growth Authority’s tax credit program for job creation, commonly called “MEGA.” Under plausible assumptions about how such credits affect business location decisions, the net costs per job created of the MEGA program are simulated to be of modest size. The job creation impacts of MEGA are simulated to be considerably larger than devoting similar dollar resources to general business tax cuts. The simulation methodology developed here is applicable to incentives in other states.”

I think it is important to clarify what this article does and does not imply about the benefits and costs of discretionary incentives for business.

First, the article simulates that MEGA’s tax incentives have modest costs per job created because MEGA was targeted at export-based businesses that paid unusually high wages (averaging over $75,000) and had unusually high multipliers (averaging almost 4; that is, for each job created directly in a MEGA subsidized business, there were almost 3 additional jobs created in other Michigan businesses). These high wages and high multipliers led to higher job creation and earnings creation impacts for MEGA than would be true of most other state and local business incentive programs. In addition, many other state and local incentive programs are not targeted at “export-base” businesses (businesses that sell their goods and services to non-state customers), but rather also go to businesses selling to state customers. Incentives that go to businesses that sell to in-state customers may increase jobs in the assisted businesses, but will displace jobs in other state-based businesses, resulting in little net job creation impact.

The bottom-line is that we would expect many other state and local business incentive programs to have much higher state costs per job created and much higher state costs per dollar of earnings for state residents generated than is true for MEGA.

Second, the article does not consider in much detail an issue I have explored in other recent work: how do the percentage of jobs created that go to state and local residents vary with an area’s initial local unemployment rate?. In a paper forthcoming in Growth and Change, I estimate that for a local area with initially low unemployment, an increase in 1000 local jobs will after 10 years only raise local residents’ employment rates by about 200 jobs, with the rest of the jobs going to in-migrants. In contrast, if the local area’s unemployment rate is initially high, after 10 years the increase in employment of local residents is around 500 jobs for a local job increase of 1000 jobs. Therefore, we should not assume that every new job going to a local area corresponds to benefits for local residents, and the extent of such benefits depends greatly on prevailing local unemployment conditions.

The bottom-line is that even if job creation is cheap, whether local job creation provides substantial local benefits depends greatly on whether the local area is currently really in need of jobs.

Third, the article’s finding that incentives often beat general business tax cuts is probably generalizable to many state incentive programs. Incentives can be cheaper per job created because they are often focused on export-base businesses making new job creation decisions, whereas general business tax cuts include many non-export-base businesses, and many businesses not considering job creation.

If a state can keep incentives tightly targeted, such incentives can usefully substitute for general business tax cuts that cause a far larger revenue loss for state and local governments than targeted tax incentives. The political problem is that tightly targeted incentives tend to grow over time.

Fourth, none of this analysis addresses the national consequences of business incentives. From a national perspective, most of any job growth in a state due to incentives is a zero-sum game. The incentives redistribute economic activity among the states, often in a random direction with no clear policy purpose, and result in an income distribution from the general taxpayer to owners of capital, who tend to be wealthier.

From a national perspective, even if incentives work for an individual state acting alone, restricting incentives may be in the interest of the nation. The U.S. should consider policies similar to the European Union, which seeks to restrict the magnitude and type of business incentives. Most state and local business incentives in the U.S. would be illegal as unfair export subsidies if they were attempted by France or Germany. In the European Union, incentives that exceed a certain size are automatically illegal unless they advance particular policy goals such as helping designated distressed regions, small business, high technology businesses, or job skills. Similar rules in the U.S. would considerably reduce the volume of incentives, and retarget U.S. incentives on particular policy goals. All states would gain some business tax revenue, and public services would be enhanced, while also helping promote smarter economic development.