Tim Bartik, Ph.D., is senior economist at the W.E. Upjohn Institute for Employment Research in Kalamazoo, Michigan. His research focuses on how broad-based prosperity can be advanced through better local labor market policies. His 1991 book "Who Benefits from State and Local Economic Development Policies?" is widely cited as an influential review of the evidence on how local policies affect economic development. His more recent work includes developing a database of U.S. economic development incentive programs, studying policies promoting local skills and their effects on prosperity, and examining how early childhood programs could promote local economic development, explored in his 2011 book "Investing in Kids." Bartik has also done extensive research with colleagues on the effects of the Kalamazoo Promise, a pioneering place-based scholarship program intended to improve the local economy.
At Site Selection's invitation, here Bartik presents an adaptation of his recent work developing a simulation model of incentives' benefits and costs.
Current trends in economic development incentives are unsustainable.
Based on my 2017 database (Bartik 2017), state and local business incentives tripled from 1990 on. The recent Foxconn incentives deal is 10 times as large per job as the average incentive deal. If Foxconn-level deals are widely imitated, incentive costs might dramatically escalate. Can such incentive trends be sustained? That seems doubtful.
Business productivity depends in part on public services — good schools and good roads. Business also depends on after-tax real wages to support consumer demand. Excessive incentives threaten public services and after-tax real wages.
More sustainable incentives can be achieved with reforms that cut costs, while simultaneously making incentives more productive. My suggested reforms:
Benefits vs. Costs
To analyze different incentive policies, earlier this year I constructed a model of local economies. This model was based on research on how business location decisions respond to incentives, how job growth affects migration and employment rates, and how growth affects state and local government finances.
The model makes realistic assumptions about how incentives affect business location decisions. Foxconn-level incentives might tip business location decisions over three-quarters of the time, but more typical incentives have been shown by research to tip less than one-quarter of business location decisions. An analysis should recognize that growth not only increases tax revenue, but also increases spending needs. Job growth attracts population growth, which in turn drives the need to hire teachers, widen roads, etc.
This model compares a baseline incentive package, with baseline policy parameters, versus alternative packages that consider different parameters. The purpose is to see how benefits and costs vary with different policies. Each alternative scenario varies only one parameter from the baseline scenario.
The baseline incentive package was set equal to the average tax incentives provided by the average state. I assumed that the jobs created by the incentives had a typical "job multiplier" of 2.5 — for each job directly created by incentives, 1.5 additional jobs are created in local suppliers and local retailers. I assume that the incentives' budgetary costs are paid for by some combination of increases in other taxes and reduction in public spending, but without severe consequences for any one budget category.
For both the baseline package, and the alternative scenarios, I analyzed the benefits of each package as its effects on the incomes of local residents. These include effects on the real earnings per capita of local residents, as well as any effects on profits of local businesses, and fiscal gains for state and local taxpayers.
The graph reports the "benefit-cost" ratio of each incentive package. This is the ratio of the total income effects for local residents and businesses, divided by incentive costs. I consider the present value of these income effects over time, with appropriate discounting of future incomes. To break even, an incentive package must have a benefit cost ratio of 1.0 or greater.
The average incentive package in the U.S. has a benefit cost ratio of 1.2 —benefits are 20 percent greater than incentive costs. Incentives pay off for local residents in higher earnings and some fiscal gains, but only modestly.
But Foxconn-level incentives, of 10 times as great, only have a benefit cost ratio of 0.3—the benefits are 70 percent less than the incentive costs. Incentives have diminishing returns because higher incentives do not proportionately increase the odds of tipping the location decision.
Incentives that are totally financed by cutting spending on K-12 education have a negative benefit-cost ratio. Cuts in K-12 spending lead to cuts in future wages.
If the incented firm has a high local job multiplier of 6 — which is supported for some high-tech firms — then the benefits of incentives are four times their costs. The job growth due to cluster effects can be large.
Finally, business services to smaller businesses have a high rate of return. Helping smaller businesses with job training, or advice on improving productivity, is cheap compared to what it does for local job creation.
Sustainable economic development strategies should be balanced. Tax incentives to large firms must be moderated to be consistent with maintaining needed public services. Customized business services are needed to help promote a healthy local small business sector.
Such a balanced strategy makes more sense than escalating the incentive wars to Foxconn levels.