GJF Data Fuels Research on Bidding for Firms

March 25, 2023
By Kenneth Thomas

Two male arms arm wrestling in black and white with dollar bills all around.
Source: Gratisography / Pexels

A persistent question in incentives research is why government officials offer subsidies to attract firms when so many studies find this is not an effective economic development strategy. In a recent post, I examined papers which argued that campaign contributions are a big part of the answer. However, other answers are possible, with one being that states compete to attract companies, and that they would lose deals if they didn’t play the subsidy game.

A recent paper by UC-Berkeley economist Cailin Slattery addresses this very question. In her research, she finds that competing with subsidies for projects is not a good economic development strategy. She finds that even though, in theory, competing with subsidies might be good for the country as a whole, the gains are very low, and they are all transferred to the companies through tax breaks or other direct assistance. (In this, Slattery is critiquing an influential line of analysis among economists which says state vs. state competition for investment and jobs is not necessarily zero-sum.) She also cautions that even small mistakes in state estimates of what they will gain from a particular deal can wipe out the theoretically possible societal gain of subsidies.

Let’s take a closer look at her research.

Slattery created a database of individual subsidy deals, consisting mainly of our Subsidy Tracker data, for which the runner-up location was known. Then she created bidding rules for states and ran simulated auctions to see which location “won” the firm. She then compared her results with the actual location decisions and associated subsidies to determine whether her bidding rules gave accurate predictions. Let’s call this the competition scenario.

Next, she simulated a second scenario in which subsidies were not used and companies simply chose the most profitable location for each facility (let’s call this the subsidy ban scenario). Comparing the two scenarios yielded several interesting results which, overall, undermine the arguments for using subsidies.

First, under the subsidy ban scenario, the model predicted that 44% of the projects would have located in a different place. This suggests that subsidies can be effective at changing the location of a project.

Second, some of those site location changes would send jobs to areas of higher unemployment and lower per-capita income, which can be seen as desirable from a national perspective. By contrast, under the competition scenario, companies getting subsidies moved to less efficient locations, as measured by their profits. The gains from the former were larger than the losses from the latter, so the country as a whole was slightly better off with subsidies allowed.

Third, the competition scenario cost $40.9 billion in subsidies to attract projects, more than twice as much as the states’ estimated benefits from lower unemployment in high-unemployment areas, etc. Therefore, all the social gain was transferred to the companies.

Fourth, states were worse off as a whole under the competition scenario. However, some states did better under the subsidy ban, such as New York, whereas others did better under competition, like Alabama. This would make it hard to get agreement among the states for a subsidy ban.

Fifth, while states were worse off, companies ended up better off, by 28% from comparing Slattery’s simulations. So corporate lobbyists would oppose any curtailment of subsidies.

Allowing subsidy competition, she finds, makes society’s payoff from the investments in her database 3.2% higher. This is not a big number, and there are several factors which could wipe out the positive impact. Slattery points out that her model does not consider the cost the firm pays for a site location consultant, often 5-10% of the subsidy, or alternatively maintaining an internal corporate staff of site location experts. Governors who face re-election may overweight the benefits of the investment because they derive electoral benefit from offering subsidies, as some studies have shown.

Most important, in my view, is the possibility that economic development officials overestimate the value of the investment for the state. Slattery shows that even a small overestimate, 6%, completely eliminates any societal gain from subsidy competition. Slattery states that there is no evidence that overestimation occurs, but recent research by Timothy Bartik and Nathan Sotherland finds that most job multipliers fail to account for the negative effect of higher costs. That is, when a new employer enters a labor market, wages and prices tend to go up, as does the demand for public services. These costs reduce the typical multiplier by about 1/4, large enough to wipe out the social gain under Slattery’s analysis. John Crompton finds that economic impact studies are generally biased upwards because the reports are written by project boosters.

Other considerations may also reduce the actual benefits of “winning” a facility. According to Thomas Klier and James Rubenstein, in the automobile industry many components plants will locate as far as one day away from the assembly plant they supply. As a result, the indirect regional jobs expected by the winner of the assembly plant may not materialize or will require a separate subsidy to attract. Finally, many cost-benefit models ignore the fact that many plants would have located in a jurisdiction without the subsidy.

Slattery’s key takeaway: even if some economic gains might materialize from subsidized projects, those gains are transferred to companies or are entirely wiped out by even a small errors in a state’s cost-benefit analysis.