The Department of Energy and Solyndra have been in the news lately. It’s a pretty high profile case of a government guarantee on a loan that is probably going to default. The loan guarantee comes from a Recovery Act program that is intended to help innovative energy companies obtain credit. I was disappointed (although not surprised) to find that this particular loan guarantee is not reported in USASpending.gov. In fact, only two of the loans or loan guarantees from this particular program are in USASpending.gov, totaling $612 thousand. On the other hand, Recovery.gov has nine awards for this program, totaling $3.3.billion. To make matters more confusing, one of the awards from USASpending.gov is not in Recovery.gov.
So between these two reporting systems, we have ten programs, totaling $3.44 billion. According to the Catalog of Federal Domestic Assistance(CFDA), the estimated credit subsidy for this program is $3.2 billion. So, Recovery.gov actually came pretty close to the total estimated at the beginning of the year. This comparison is similar to the methodology used in Clearspending, our report on USASpending.gov data quality (update to be released soon!). However, the reporting of this data is still a bit confusing on both websites. Recovery.gov only has an option for loans, not loan guarantees. Direct loans are when the government disburses money up front and is repaid later. But loan guarantees are only considered expenditures if the loan is defaulted on. For all nine awards in Recovery.gov, it’s hard to tell if the government has already disbursed this money (as a direct loan) or not (as a loan guarantee), and that is one of the main purposes of this site!
We’ve blogged about the unreliability and confusion around loan data before. It’s not as easy to estimate the ultimate cost to the government with loan reporting, as it is with direct grants. Loans are supposed to be reported with a total face value (what the government would pay in case of default) and a subsidy cost (an estimated percentage that the government will end up being liable for). However, only 20% of the loan records in USASpending.gov have the subsidy cost field populated. The Solyndra case is clearly just one of many where loan data is sorely under reported.