Capital Goods - September 2009

Figuring a return on incentives
By Scott Mooneyham

Whenever some big-name company takes the bait — the big money the state uses to lure them — it seems to happen. The governor and secretary of commerce grin widely as they announce how North Carolina landed the trophy with tax breaks, grants and other financial incentives. To justify it all, they explain how the company’s investment will lead to X number of jobs and X amount of economic activity as the money spent on construction, wages and supplier contracts circulates through the economy. Then someone rains on their parade by questioning the math. The latest example is the $1 billion data center that Cupertino, Calif.-based Apple Inc. plans in Catawba County. A change in the tax law to get it is expected to cost the state — and save Apple — $46 million over 10 years. Local incentives will total $20.7 million in that period. Gov. Beverly Perdue and the Commerce Department crowed that the plant and all its associated activity should “create more than 3,000 jobs in the regional economy.” Then the Raleigh News & Observer published a piece questioning whether the multiplier effects and job projections mean that much.

In 2004, the same thing happened. Round Rock, Texas-based Dell Inc. announced that it would build a computer plant in the Triad after the state agreed to $242 million in incentives. Commerce announced that Dell’s coming would create 8,086 jobs in the region. But some nosy, know-it-all critics pointed out that Virginia, which had been competing for the same plant, projected job creation at about half that.

Should we ignore those figures and assume they are always inflated? Not according to the people who put them together. Stephanie McGarrah, assistant secretary for policy, research and strategic planning at Commerce, says they’re generated by impact analysis for planning — IMPLAN — economic-forecasting data and software that have been the standard since the 1990s. “I suspect it is used by most states,” she says. She acknowledges, however, that the numbers derived from the forecasting model, developed by the U.S. Forest Service and the University of Minnesota, are only as good as the numbers going into it.

An important figure is how much local purchasing the company anticipates, both during construction and once the operation is up and running. Commerce typically uses just a fraction of what the company reports it plans to do, unless it can make a strong case otherwise. But one reason Commerce’s numbers for Dell came under such scrutiny was that officials decided 95.5% of the construction money for the project would be spent in North Carolina and used that figure in their modeling. It turns out that 54% is the default percentage typically used.

A study by the liberal North Carolina Budget and Tax Center and Washington-based Corporation for Enterprise Development picked the numbers apart to conclude that the state grossly overpaid its incentives. When the report’s authors inserted what they said were more plausible numbers into the model, the 20-year economic impact of the project was no more than a third of the $24 billion estimated by Commerce officials.

But Andrew Brod, a UNC Greensboro economist who engages in some economic-development forecasting himself, says there’s nothing wrong with the models being used by Commerce or the people using them. Invariably, he says, disagreements will occur and judgment calls be made about which numbers are plugged in — and how they are plugged in. Sometimes mistakes will occur. “That’s why I try to be quite clear about the fact that my projections are a function of what I’m told.”

But what about the politicians hoping to show an anxious public facing double-digit unemployment that help is on the way, that jobs are coming to North Carolina? Are they always so clear? Maybe the problem isn’t an economic-forecasting model, the data that go into it or the numbers that come out. Perhaps it’s how that information is packaged and sold to the public.

It’s worth noting that politicians, when trumpeting the latest economic-development coup, often focus on that larger job number projected from all those expected spillover effects without mentioning the actual number of workers the plant will employ. When ribbon-cutting time comes at Apple, it might be easy to forget that the number to be employed by the computer maker isn’t 3,000. It’s 50. If anyone walks away unaware of that fact, it won’t be the fault of some economic-forecasting model.

Scott Mooneyham is the editor of The Insider,