The August bankruptcy at the solar panel manufacturer Solyndra has generated a predictable political kerfluffle, since the company had received $527 million in loan guarantees from the Obama Administration. The political issues raised by the case are fair game. But I’m more interested in a more general issue:
What is the true cost to taxpayers of loan guarantees?
Obviously, we learn the cost of a particular loan guarantee, like the one for the bankrupt Solyndra, ex post—it’s the amount of money the government has to payout to the creditors. But what is the right estimated cost ex ante?
Coincidentally, August was also the month that saw the CBO publish a report on the true cost to taxpayers of Federal Loan Guarantees for the Construction of Nuclear Power Plants.
One of its co-authors is my MIT Sloan colleague, Professor Deborah Lucas. The report is well done and touches on a number of important considerations. One of the key points is that the current, legislatively mandated method for calculating the budgetary cost significantly understates the cost because it ignores the full cost of the risk imposed on taxpayers. Future payouts on the guarantees are discounted at US Treasury rates, but the true cost of those future payouts should include a market risk premium. As the CBO report says:
A common view is that the government has a lower cost of capital than private financial institutions because it can borrow at Treasury rates. Treasury rates are low, however, because holders of Treasury bonds are protected against losses by taxpayers who absorb the risk of the government’s activities. … when the government provides such a guarantee, it is effectively shifting financial risk to taxpayers who, like investors in a financial institution, are averse to bearing that risk. From that perspective, market risk is a cost to taxpayers that is not included in budget estimates.
The difference is significant. As an illustrative example, let’s consider a guarantee for a loan of $8 billion extended to an A-rated company constructing a nuclear power plant. These parameters happen to roughly correspond to the case of the loan granted to Georgia Power (Southern Company) for construction of the new reactor at its Vogtle site. Let me emphasize that the CBO document does not analyze the cost of this particular loan guarantee, but only a hypothetical standardized plant being built by a hypothetical standardized company with a given rating. The comparison to Vogtle is my own. Assuming a 55% recovery rate, the CBO report calculates the legislatively mandated budget cost to be 1% of the loan, or $80 million. The true cost calculated by the CBO is 9% of the loan, or $720 million. That’s an underestimate of $640 million. In the nuclear loan guarantee program, companies like Georgia Power must pay a fee to the federal government to cover the budgetary cost of the guarantee. But according to the CBO report, and this example, the fee paid over time would only be $80 million, while the true cost of the guarantee is $720 million, so the company is receiving an unstated subsidy of $640 million.
The same fundamental mistake is being made by the current UK government’s strategy of de-risking investments in low-carbon power plants. It claims this de-risking lowers private investors’ costs. But what the government calls de-risking is actually just risk shifting. The risk that private investors used to bear is now passed along like a hot potato to UK taxpayers. The government’s calculations of the economic benefits of this de-risking captures the gains to private investors but ignores the cost borne by UK taxpayers. See this post on the UK calculations.
This post has been updated to correct the amount of the principal on the loan guarantee. The example was calculated assuming a principal of $8 billion, but the post originally read $800 million. Thanks to a reader for pointing out the error.