Are Fannie Mae and Freddie Mac Doubling Down and Creating a New "Moral Hazard" Effect in Subsidizing Risky Mortgage Loans?

 I am the co-author of a controversial 2022 empirical paper that argues that Fannie Mae and Freddie Mac have implicitly subsidized home ownership in areas facing flood risk. This creates a moral hazard effect as real estate developers, mortgage lenders and home buyers conduct mutually beneficial trades and if something bad happens (a big flood) that the American tax payers are on the hook for a bailout.

As a retired old scholar, I asked Grok this morning a question. I asked Grok what have been new recent working papers;

Grok also pointed me to this paper.

Kim, You Suk, Feng Liu, and David Zhang. “The Elasticity of Home Purchases to Financing Costs: Evidence from Recent GSE Pricing Changes.” SSRN Working Paper, December 10, 2024 (last revised June 20, 2026).
SSRN: https://ssrn.com/abstract=5222560

Their paper taught me about a 2023 GSE policy change that Grok explains;

Note what Grok writes at the bottom here. The new GSE policy implicitly subsidizes low credit score borrowers. IF these individuals tend to buy homes in lower quality areas (where home prices are lower), then this new policy implicitly subsidizes more vulnerable households who buy in flood zones. So, this new policy doubles down on the mortgage finance/moral hazard effect that Amine and I studied.

What is almost funny here is that I had hoped that the news coverage of my work with Amine would lead the GSEs to start to price flood risk in the G-Fee so that riskier properties required a larger payment. Was I naive here about how the GSEs operate and learn? Clearly they were under pressure from Congress.

Have you read my old paper?

Gabriel, Stuart A., Matthew E. Kahn, and Ryan K. Vaughn. “Congressional influence as a determinant of subprime lending.” Journal of Housing Economics 28 (2015): 91-102.

I fed this Substack to Grok and here is what GROK wrote

If Ouazad and Kahn are correct that the GSEs’ refusal to price flood risk creates a moral-hazard channel in which lenders originate and then securitize riskier loans after disasters, the May 2023 LLPA redesign supplies a high-stakes test of that thesis. By cutting the upfront component of the g-fee for lower-credit-score borrowers while leaving geography and flood exposure unpriced, the new grid reduces the all-in cost of GSE financing precisely for the borrower types that empirical work shows are over-represented in inland and riverine flood zones.

Going forward, private lenders therefore face an even stronger incentive to originate conforming loans in those areas—especially for lower-score borrowers—and to sell them to Fannie Mae and Freddie Mac rather than hold them on balance sheet. The test is straightforward: relative to pre-2023 benchmarks and to higher-score borrowers, we should observe an increase in the share and volume of GSE-securitized originations on properties that carry elevated flood risk (particularly outside formal Special Flood Hazard Areas). If that pattern materializes and subsequent flood events produce larger credit losses for the enterprises, the 2023 rule will have amplified the very transfer of climate risk the 2022 paper documented.

Whether the GSEs ever internalize that risk remains an open political question. Their pricing continues to be shaped by a congressional mandate for broad credit access rather than by actuarial climate exposure, a constraint already visible in earlier work on legislative influence over subprime flows. The 2023 change shows that, under pressure, the enterprises can adjust the credit-score gradient of g-fees; it also shows that adjusting the flood-risk gradient has so far remained outside the feasible set.

Matt here again; one nuance is whether the advent of CRTs (credit risk transfer) mitigates the issue. Some Wharton professors have claimed this. Grok disagrees.