Do bank robbers stimulate economic growth? If you employ the logic used by the National Renewable Energy Laboratory (NREL) to assess the impact of the stimulus bill’s renewable-energy subsidies, the answer would be “yes.”

The NREL recently issued a report on the economic impact of the Section 1603 Treasury Grant Program funded in the stimulus bill of 2009. The bullet points claim that the subsidies created 75,000 jobs and $44 billion in economic output.

The study uses an input-output table to trace the cascade of spending generated by the subsidies. It ignores the costs and employment losses of the tax revenue needed for the subsidies and the offsetting losses suffered by the conventional energy that gets crowded out (though there isn’t too much crowding out since the renewables do not produce very much energy). It ignores the economic impact on consumers’ energy bills. And the bullets attribute all of the positive impacts to the subsidies even though the subsidies are only a fraction of the costs of the projects analyzed.

What’s the point? If you ignore all the costs of any action, then it cannot be a loser.

The NREL input-output analysis starts with the spending on labor and equipment. It then traces the spending on equipment to the spending on the inputs for that equipment. For instance, the purchase of a wind turbine would trace the flow of spending to the parts suppliers to the turbine manufacturer, then to the steel mills that supply steel to the parts fabricators, then to the mines that provide the ore to the mills, etc. At each stage, it picks up a chunk of income and some jobs.

The problem is that process ignores a similar cascade of income and job creation that is cut off when the funds for the subsidies are extracted from the private sector of the economy.

Imagine a crime lord organizing a bank robbery. First he hires some thugs, creating jobs and income for them. He also has to give the thugs some guns, so the gun manufacturers see an employment increase, as do the suppliers to the gun makers and the suppliers to those suppliers, etc. Also, don’t forget the jobs in the production process for making the ski masks the robbers wear, the jobs producing and servicing the getaway car, the jobs at the motel where the thugs lie low, etc.

In this example, we ignore the losses to the bank and its depositors, the increased security costs that crime imposes on banks, the impact of higher interest rates on borrowers (that’s one of the ways banks cover losses from robberies), and every other negative impact of bank robbing, and then conclude bank robbing is an unambiguous gain for the economy. Wheeeee!

To the credit of the NREL study’s authors, they get the joke. In footnotes and in the text, they make it very explicit that the impacts they report are not net impacts. For instance, footnote 13 reads:

The…models do not estimate the displaced energy or associated jobs, earnings, and output related to existing or planned energy generation resources (e.g., jobs lost in the operation of natural gas or coal plants due to the need for less electricity production from these plants, given increased generation from wind) or increases or decreases in jobs related to changes in electric utility revenues and consumer energy bills, among other impacts. Therefore, the estimates represent gross rather than net impacts.

And how is it funded in the first place? Of course it has to come from taxpayers (taxpayers today if current taxes cover the bill, or taxpayers tomorrow if the government borrows the money). Though the authors seem to think other government expenditure is the only alternative to subsidizing wind and solar energy, they at least understand there is a very significant cost they do not include:

Similarly, the jobs and economic impacts associated with possible alternative spending of the federal funds used to support §1603 projects were not estimated in this study. Therefore, results presented in this paper should be interpreted as gross rather than net estimates.

In short, the NREL study tells us nothing about the overall impact of the subsidies. However, fundamental economic logic can guide us. Energy sources that require subsidies produce energy whose value is less than the costs of production. So these programs actually reduce national income since they take higher-valued resources and turn them into lower-valued output. There is no need for computer programs to reach this conclusion.

Looking at only the benefit side of a cost-benefit comparison, no matter how sophisticated that one-sided view may be, tells us nothing useful. It certainly doesn’t support the ridiculous job and income claims in the NREL’s bullet points.