Imagine a federal government program that created one full time job for every $5,000 of taxpayer spending. Under this program, obligations are only made after many months of careful analysis by career officials at the Office of Management and Budget and the Department of Energy, in consultation with independent private sector experts. On top of that, the program increases domestic U.S. energy production and vastly reduces the economic damage caused by pollution. Given these benefits and the fact that jobs are sorely needed right now, you might think that politicians would be clamoring to replicate and scale up the program.

Alas, such a program exists. It’s called the Department of Energy 1705 Loan Guarantees Program, but it is arguably the most reviled in the country right now. Every day for the last month or so, news reports and news commentators have criticized it mercilessly. While members of the Obama administration may deserve criticism for bestowing one beneficiary of the program--Solyndra--with favoritism, many reporters have woefully misunderstood or miscalculated the costs and benefits of the program.

The list of errors is long: The Washington Post estimated the job creation rate for the Loan Guarantee Program to be $640,000 per job. The Dakota Voice puts it at $410,000 per job. The Boston Herald estimates $4.9 million; National Review and the Richmond Times-Dispatch put it in the millions of dollars per job; and Investor’s Business Daily says it is $23 million. This is just the tip of the iceberg of wrongheaded green loan program coverage.

Even if these authors get the math right, they have the wrong numbers in both the numerator and denominator. The DOE website deserves considerable blame for this because they list $36 billion in loans underwritten and 64,776 jobs, but neither of these values are the right numbers to use in a cost-benefit analysis. The DOE highlights them, I suspect, because it sees both of them as benefits in terms of economic value supported.

First of all, the $36 billion does not in any way reflect the costs to taxpayers; for that to be the case, every loan would have to default before making any payments and the government would have to recover nothing, despite owning the assets of the companies. The probability of that happening is zero. For some projects, the government didn’t even guarantee the full value of the loan, and some loans haven’t even closed.

OMB publishes a set of assumptions used to calculate the true costs of the program over the very long-term. For the 1705 program, OMB estimates that 12.85 percent of the value of the loans will default and that the federal government will recover 48.77 percent of the value of those defaults. The 1705 program has closed loan deals worth $16.1 billion. The math shows that the taxpayers are expected to lose $1 billion if OMB’s assumptions are correct (so far they over-estimate defaults).

With costs of $1 billion, the program generates 16,738 jobs according to DOE, translating into $60,387 per job, but those jobs numbers are also not the correct ones for this calculation. Why? They only consider direct employment at the company that receives the loan and the companies that complete the construction. They do not count the jobs created by supply-chain effects (indirect jobs) and increased spending as a result of the project (induced jobs).

Fortunately, there is one easy tool for calculating the total jobs impact for energy generation projects (i.e. direct, indirect, and induced jobs). The National Renewable Energy Lab has made available a Jobs Economic Development Impact Model (or JEDI, and yes, the NREL scientists are apparently Star Wars fans). The JEDI model uses data and assumptions from the Minnesota Implan Group—a standard “impact” model package used by hundreds of universities, consulting firms, government agencies, and businesses to analyze the costs and benefits of investments. The Implan model crunches reams of data on business and consumer purchases to come up with its multipliers.

JEDI is quite easy to use. Download the solar model excel sheet; type in the number of KW that the project expects to generate (available from the DOE website); type in the state where the project is located, the year, and voila: You get the output. (Of course, the more project specific data you can enter, the more accurate the results are likely to be).

The model relies on energy production and so is not designed to study the effects of building a manufacturing plant. Therefore, I analyzed just 14 generation projects—10 for solar and 4 for wind. These represent 71 percent of the outstanding 1705 program loans. The estimated costs to taxpayers from these loans is $715 million over the next 30 years, and they are expected to create 148,797 jobs in the next year. In other words, the 1705 program creates one job for every $4,800 of federal government spending on solar and wind generation projects.

How did such an efficient and effective program become so maligned? As I’ve said elsewhere, the president and his advisors wrongfully politicized the issue by deciding to visit green firms and take credit for their growth. The administration should also answer for apparent favoritism shown to donors affiliated with Solyndra, even though the analysis of career DOE and OMB staffers was probably unaffected.

Yet, the media also deserves criticism for lack of due diligence. In interpreting one company as a representative sample of both an industry and a set of policies, they’ve sacrificed nuance in order to exploit the rightfully anxious and angry mood of the country. In reality, trends in venture capital, jobs, and exports, show that the cleantech sector is performing quite well. But let’s not exaggerate; the numbers are still small. It will take continued federal support for clean technology to realize its full potential and reach a scale that will start to significantly reduce unemployment. We already have one great tool--loan guarantees--that should be expanded, complemented, and isolated from administration policies. The proposed Clean Energy Deployment Administration is one example of how to do this.