Reforming Economic Development Incentives.

• Author: Bartik, Timothy J.

Incentives were originally intended as assistance to encourage individual businesses to grow or retain jobs in a particular community. As incentives offers have increased unchecked, though, it's clear that the time has come for reforms based on benefits and cost. Separate myths from facts in this analysis.

In my recent book, Making Sense of Incentives, (1) argue that state and local economic development incentives need major reforms to both restrain excessive incentive costs and do better at improving residents' job opportunities.

Economic development incentives can be defined as assistance to individual businesses to encourage them to locate, expand, or retain jobs. More than 90 percent of such assistance is in the form of tax breaks or cash grants, although some incentives provide individual businesses with services such as customized job training. Three-quarters of the dollar volume of incentives is provided by state governments and one-quarter by local governments (mostly property tax abatements).

Why be concerned about incentive costs? As Exhibit 1 shows, incentive offers have tripled since 1990. Typical incentives today are equivalent to subsidizing the wages of these firms by three percent over a 20-year period.

Some recent incentive offers--for example, the State of Wisconsin's offer to the Foxconn manufacturing company, (2) or some of the offers made for Amazon's "Headquarters 2"--have exceeded current average incentive offers per job tenfold. If Foxconn-level incentive offers became standard, incentives would exceed 30 percent of state and local tax revenue. Such high costs would lead to large cuts in public services or large hikes in household taxes.

FACTS TO REPLACE MYTHS

Incentive reforms should be based on an understanding of benefits and costs. Common myths about incentives should be replaced with facts.

FACT 1

Incentives are not 100 percent effective. Economic development agencies sometimes do evaluations that implicitly assume that every incentive tips a firm's location decision. Job creation figures and tax revenue predictions are calculated as if none of the incented jobs would have occurred "but for" the incentives. This is false. Empirical research suggests that 25 percent or fewer of the incented jobs owe their existence in the local area to the incentive. (3) In other words, more than 75 percent of the new jobs would have been created locally without an added incentive. Most incentives are all cost and no benefit.

FACT 2

Multiplier effects are often overstated. Incentive impact studies often claim large "multipliers." Any direct jobs created in incented firms lead to "multiplier jobs" in suppliers to incented firms, or in retailers serving the additional workers. Mathematically, the local job multiplier is defined as the ratio of total local jobs created--the sum of direct jobs plus the supplier and retailer jobs--to the direct jobs. While direct jobs created in incented firms will yield other local jobs, studies often overestimate the multiplier effect.

Local multipliers are frequently claimed to be greater than two; that is, each direct job in the incented firm will yield at least one additional job for local suppliers and retailers. These local multiplier calculations, however, are overstated by one-third because they overlook negative effects of job creation on local costs. Local job creation will increase local land prices and wages, and these cost increases lead some other local firms to make job cutbacks. Typical local job multipliers are probably closer to 1.5. (4) In other words, for each 10 local jobs directly created, there might be additional net job creation of five jobs.

In some cases, the local multiplier may be closer to zero. If incentives are provided to firms that sell their goods and services locally, any expansion in incented firms will reduce sales and jobs in other local firms. For example, subsidies to sports stadiums or local shopping districts may induce new spending and jobs at these sites, but local...