The Goodhart’s Law Problem with Public Loan Guarantees
As a way to encourage private investment in clean energy research, the Department of Energy has extended loan guarantees to many private companies involved in renewable energy development. While conservatives seem to believe that they are all Solyndras, in fact the portfolio is doing remarkably well. And that’s a problem. As Michael Grunwald says, the whole point of this program is to bankroll promising technology that offers high rewards, but is too risky for the private sector to invest in. If almost all the loan recipients are paying it back, that means the government is not only not investing in promising-enough technologies, but is actually crowding out private investment in the sector.
The political economy of public-private research partnerships are less promising than they seem initially. They are justified as offering lower costs than direct government sponsorship by risk-sharing, which is hypothetically true. However, it carries with it the inherent risk of Goodhardt’s Law – when a measure becomes a target, it ceases to be a good measure. The key measurement here is default rate – the government can hypothetically monitor the riskiness of its program by monitoring the default rate and making adjustments accordingly. However, the default rate in a loan guarantee program is actually the only visible metric available, and it is a natural rule of organizations that you manage to what you can measure. Even without the political pressure applied, it seems like one would naturally expect these types of programs to be managed to generate maximum return of capital rather than maximum investment in promising technology.
There are better ways to structure public investment into research. The simplest is to fund public research, which has been used plenty successfully in the past. There’s no monetary recovery, but it’s also not set up in such a way as to encourage it – and if the money is well-directed, the social benefit can far outweigh the accounting cost. There are also tax incentives for R&D, which are a bit less well-directed towards basic innovation but can be relatively cheap. There are public innovation prizes, which are likely underutilized and are a generally neat solution. But none of these face the perverse incentives of public loan guarantees. They’re a clunky policy tool that emerge from the contradictory desire to keep the government out of something while still using policy to drive it.