It Is Hard to See the Future with Egg on Your Face

Last week a colleague mailed me an article about the degree to which some Michigan economists had been in error in their forecasts about the state economy.  I wasn’t surprised by this as humanity is a complex beast and human actions are fraught with surprises: war, government tax hikes, bankruptcy, and natural disasters — you name it.

What does surprise me is the continued faith that people and government have in long-term, hyper-detailed forecasts of economists and consultants. Regardless of the modern crystal ball they use, such as sophisticated computer software, it is difficult to work around at least two inescapable facts: if you put garbage in you get garbage out; and the many underlying assumptions that are fed into models begin changing the moment the modeler presses the “go” button on whatever computer software they are using.

Let’s use the Michigan Economic Growth Authority program as an example. For each MEGA deal the state has a forecast of the economic impact done by top economists. The model that is used is called REMI and the output the state gets from it include expected spin-off jobs over time, revenue to the state treasury from new jobs, etc. Some of the forecasts are as long as 20 years into the future. My colleague Michael Hicks and I looked at years of data from MEGA and concluded that many assumptions fed into the REMI model appeared to be wildly off the mark.

From our study:

Based on state documents compiled since 1995, we estimate that 127 of

MEGA’s agreements should have produced fully employed facilities through 2004. Of these 127, about 56, or 44 percent, have claimed credits under the program.

The ability to claim credits does not mean, however, that a firm has achieved

the job goals that it projected in its MEGA agreement. For instance, for most MEGA deals, a company must create an initial 75 “qualified new jobs,” usually within a year of commencing operations a particular site. If it does reach this minimum job threshold, it can begin claiming its tax credits, even if it never achieves the job count it originally projected.

 

Many of the 56 cases in which tax credits were claimed did not fully meet

original projections. In fact, only 10 of the 56 can be shown to have created the

number of direct jobs originally projected within the expected time frame — although three of the 10 (Kmart, for instance) have had setbacks following their initial success in meeting the targets.

 

Several of the companies behind those successes stumbled after our study was published as have companies who did receive tax credits for creating some jobs, but ultimately eliminated them. Plastech, which went belly up this year, seems to be a good case in point.

So, how is it that economists who can claim to forecast jobs growth up to 20 years into the future can’t see that in less than 18 months the company whose jobs impact they were hired to model will be bankrupt, as was the case with K mart? The short answer is that they cannot reliably do so.

 

It is difficult to see the future with egg on your face. Rather than continue to try the state should wipe away the yoke of government development programs and their many untenable predictions about some forthcoming economic gift to this or that region or state.