He sees it (earlier, preliminary, no-cost version here) as a public choice problem.
If there is a high probability that political leadership will be induced to pursue policies that maximize the profitability of private entities at the expense of taxpayers, then purely public options create lower social losses. If there is a high probability that leaders will pursue a populist agenda of lowering prices or borrowing costs, then catastrophic risk insurance can lead to lower social losses than either complete laissez-faire of a pure public option.
Thanks to reader John Alcorn for forwarding me the paper.
To me, the size of the market failure, if any, in housing finance seems awfully small relative to the amount of corruption and populism that government intervention introduces. Nirvana would ensue if political leadership could be induced to be bound by something like the Constitution, so that they are less prone to intervene when they are likely to do more harm than good. But there's not much chance of that, is there?