The Model Facade: Selling Jobs to the Public, Press and Policymakers

History shows models are used to obscure costly, ineffective programs

“The models are used in ways that systematically exaggerate the public benefits of proposed government projects, thus biasing government decision making in the direction of excessive government spending and expansion into areas that should be left to the private sector.” – Edwin Mills, author of “The Misuse of Regional Economic Models.”

Legislation up for consideration (Senate Bills 242-244), colloquially known as “Good Jobs for Michigan” or “New MEGA,” contains a mandate that the state employ “an industry-recognized regional economic model cost-benefit analysis [that] reveals [that] … [the program] … will result in an overall positive fiscal impact to the state.” This means nearly nothing as state-bought analyses can and have failed to accurately predict and measure impacts and are at times steeped in secrecy or used mischievously.

Stay Engaged

Receive our weekly emails!

The “industry-recognized regional economic model” is almost invariably (should the proposal become law) going to be “REMI,” which stands for Regional Economic Models, Inc. I say this with some confidence because the state has for years wielded this computer model in its service directly and also indirectly through the consultants it hires. The model was employed to estimate the jobs and revenue impacts of the state’s failed Michigan Economic Growth Authority projects (which are maintained online here), the state’s film subsidy program and the current and secretive Michigan Business Development Program analyses.

Like the current proposal, MEGA law mandated a cost-benefit analysis to ensure that each project would be a net plus for Michigan. We now know from hindsight that the vast majority of these impact analyses were profoundly inaccurate. This was a function of one deal after another not living up to expectations. The model’s output is only as good as the accuracy of the modeler's input assumptions, or what the modeler tells the computer about a project. In other words, and in a supreme twist of irony, the models can’t accurately predict the future because the model’s operators can’t accurately predict the future.

Consider just two of many examples: and Kmart. The REMI analysis performed by the state reported that Webvan Group, Inc., would employ 900 people directly at its Livonia location by 2004 and a total of 1,200 in the state by 2015 while increasing tax revenue to the Michigan treasury by $27.3 million. This, if only the state would offer up millions in incentives through the MEGA program. The state did give Webvan these tax breaks but the jobs never materialized. Just over a year after being declared a MEGA winner, the company was no more.

Kmart was given its second MEGA deal less than 17 months before it declared bankruptcy. The company later moved its headquarters to a different state. The REMI model reported that the Kmart deal would create 753 jobs in Michigan by 2013 and provide a net tax gain to the Michigan treasury of $27.2 million. But this never happened.

From 2005 to the time it closed in 2011, only 2.3 percent of MEGA’s projects met their job creation expectations. The most generous accounting reports that MEGA was responsible for about 19 percent of all direct jobs it had said it would create. Five empirical studies have been done on the overall program and four found a zero-to-negative impact from the program.

This suggests that a mandate to perform such analyses ultimately counted for little-to-nothing save for the public relations surrounding alleged jobs that would be created. Previous administrations reported the REMI jobs output with a verve and passion and many news agencies simply reported the claims without a hint of skepticism. They should not have done so.

The state mandate for MEGA and the Good Jobs for Michigan proposal does not mandate an opportunity cost-benefit analysis. The REMI analyses are not required to investigate alternative uses of precious resources, either from spending on a different basket of public goods or an across-the-board tax cut. What if, instead of handing out goodies to one company, the REMI analysis projected the economic impact of a tax cut for all? The Anderson Economic Group once made such a calculation for the MEGA program and found that an across-the-board tax cut would have created more jobs on net balance than MEGA.

Regional models can also be used irresponsibly. For instance, the Michigan Economic Development Corporation once hired Michigan State University consultants to estimate the impact of their new film incentive subsidy program after its first year. The consultants likewise employed REMI to make their estimations. They failed to note in their report, however, that their estimations did not include any of the costs associated with the program. That’s not how the real world works. Everything has a cost. Yet this study was released and pointed to as evidence of the film subsidy program’s success.

The state’s corporate welfare machine, the MEDC, has also used REMI models to forecast the impact of its subsidies to corporations under the Michigan Business Development Program. This replaced the now defunct MEGA. The MEDC claims that the development program forecasts a return on investment of more than $8 to $1. It refuses, however, to make public the input assumptions it fed into the REMI model for review. In other words, MEDC officials are just expecting everyone to take their word that their programs are successful. It is naive to do so when you consider the MEDC’s own incentive to engage in self-aggrandizing puffery.

I’m hardly the only person to notice the use (or misuse) of these types of models as a justification for an otherwise unwarranted intrusion in the marketplace. Edwin Mills, professor emeritus at Northwestern University, wrote a classic takedown of such models titled, “The Misuse of Regional Economic Models.”

He examined the use and misuse of REMI by governments and their consultants. He noted that deploying models like REMI can be effective because “it is a complex computer model that lay people cannot understand or evaluate, and it has important scientific merits. Thus, the frequent government claim that the best scientific model available shows that ‘x’ thousand jobs will be created by the project helps to carry the day.” Carrying the day is precisely what state jobs officials are trying to do. Using REMI or another regional model provides economic air cover for corporate handouts.

John Crompton, a distinguished professor at Texas A&M University, published a 2006 paper titled, “Economic Impact Studies: Instruments for Political Shenanigans?” and argued that economic impact studies “are commissioned to legitimize a political position rather than to search for economic truth. Often, this results in the use of mischievous procedures that produce large numbers that study sponsors seek to support a predetermined position.”

A mandate that forces bureaucrats to employ a particular type of model to estimate the benefits of particular policies should serve as a warning to lawmakers, not a signal that all is well. Past model use (or abuse) by state officials has shown that these models are used to help obscure the fact that such programs are costly and ineffective.

Related Articles:

Return of the Mega Subsidy

MEGA 2 Hearing Demonstrates Desperation

New Corporate Welfare Proposal is Unfair

‘But For’ Can’t Be Proved: Corporate Welfare is a Waste

Legislators Who Promote Transparency Should Start by Disclosing Corporate Welfare Deals

Secretive Bidder Wins Pure Michigan Contract Over Transparent One