Ideas in Development

North Carolina offers incentives to corporations to relocate operations to North Carolina, and increasingly its counties have done so as well. While job creation and enlarged payrolls are welcome, the incentives state (and local) government offer are an investment that, like any investment, should produce increased government revenue, if only to cover the incentives and the increased costs of expansion – new schools and highway improvements, for example.

Since the state uses a fiscal and economic impact model to project potential impacts of a project and help in determining the optimal bid, it seemed like a good idea to “interrogate” the model and test its reliability as a means of avoiding over-bidding for projects and ensuring adequate tax revenues.

Based on research and analysis done by CFED and the NC Justice Center, it appears that some of North Carolina’s investments to attract new business are losers—in particular, a $242 million package for a Dell Inc. computer assembly plant in Forsyth County that topped the next highest bidder, Virginia, by $205 million. Although the state uses a fairly comprehensive economic impact model to analyze costs and benefits of incentive packages, the model itself requires not only a few adjustments in areas, such as its treatment of growth-induced governmental costs.  It also exhibits a serious flaw: the use of projected sales as a key factor. A second problem concerns several of the assumptions that are fed into the model. North Carolina officials frequently appear to make overly optimistic assumptions; they exaggerate projections of economic impact, which appear to justify large state incentives in the short run but ultimately are contradicted  by a variety of plausible projections run in for this study. None of a number of different methodologies for estimating the projected annual sales of the North Carolina-based facility matched the original projection of $2.3 billion annually.  In fact, one of the study’s credible projections was as low as one-fifth of the figure.  So, it appears that it is too difficult to estimate with any real accuracy a 20-year trend in sales.  In addition, the sales figure also does not accurately adjust for the large share of these revenues going to non-North Carolinian shareholders in Dell.  This element in the model, given its supreme importance, needs to be discarded and replaced.

When equally if not more plausible assumptions were used in the same Commerce Department model, the total projected 20-year impact on Gross State Product (GSP) decreased from $24 billion to between $5 billion and $8 billion, while the net change in state revenue dropped from positive $707 million to between negative $63 million and negative $72 million   Thus, these adjusted projections suggest that North Carolina may get back less in tax dollars than it provides in incentives.  The Dell project may be one such example.  There are 30-plus Job Development Investment Grant (JDIG) deals that should be analyzed with similar thoroughness.

Getting a decent return on North Carolina’s incentive monies is very important.  After all, attracting new business relocations is not the only strategy for economic development. Others include enhancing education and training, building infrastructure, lowering taxes on middle-income citizens, improving the safety net of social services, and promoting arts and cultural attractions. Incentive packages that are overly generous essentially subsidize corporate shareholders with taxpayer dollars that might otherwise be invested in alternative strategies that may, in fact, yield higher returns for the state and for taxpayers. The state’s incentives for business development should be accountable to citizens, transparent in their costs and contractual expectations, and cost-effective.  The NC model is a potentially great tool for achieving these aims, but the original rationale for each incentive used by the state and the actual assumptions underlying the model should be tested more regularly, rigorously and openly against actual results.

Needless to say the Governor’s Office and the NC Department of Commerce disagreed with our findings, claiming that we had misunderstood their model and so forth.

Now, more than six months have passed, a debate was held between CFED and Governor’s Office staff (in our opinion, it was a “draw”), and the heat and dust has now settled.  It’s time for a retrospective.

How does our report on “Getting Our Money’s Worth” look today?

1. I would convert the claims about whether the Departmental model used a discount rate to a footnote and soften the statement a tad.  We have no definitive proof and will give them the benefit of the doubt.


2. Our in-migration rate for such an area, with lots of under- and unemployed former manufacturing workers, was probably too high.  Their estimate may have been better.


3. Last minute data runs, which can be found in the appendices of the report, closed the gap some between our worst scenarios and their best.  But they also raised great doubts, regarding the importance and variability of their sales number.


4. Dell dictated the sales number to use as well, which may have been a bad idea, since it was the highest among the methods we used to estimate this figure.


5. So, the use of firm/facility revenues as the driver of the model still seems extremely unreliable and in our view, should be replaced by the substitution of the employment estimates for the independent variable that generates the rest of the NC model’s findings.


6. However, we have a concern about an application of the model, regarding its projection of employment generation. North Carolina’s count of supplier jobs as direct jobs strongly contradicts standard practice.  (Facilities like Dell create direct jobs, indirect jobs (e.g., suppliers), and induced jobs (hired workers buying candy at a convenience store).  In probably 92-plus percent of the cases, it’s just plain wrong and not done.  Yet, it might pass the test for this particular project, but it should not become a habit for their job calculations.  It inflates the numbers of jobs. (For more detail on this knotty problem, see report) 


7. The in-state proportion of construction spending also still looks dodgy (it’s too high). 


8. Many of our other charges—some of which are rather subtle and technical—still look strong.  In the heat of the argument that went back-and-forth during the report release, these were never understood by Commerce.  Some improved wording on our part might have aided the cause.  And few people understood why spending money on a firm that would have come without a subsidy should have its gross impacts lowered to a more accurate net figure.

In short, some of our projections might have been low, but their numbers still looked high.  Plus, using sales revenues rather than jobs as the main variable is a bad idea, necessitating an independent review of the model’s strengths and weaknesses.

This recommendation is not a suggestion. We regard it as an imperative if the state is to be a prudent steward of the people’s money.