Working Papers

Who Bears Flood Risk? Evidence from Mortgage Markets in Florida (Job Market Paper)

Revise and Resubmit at the Review of Financial Studies 

Abstract: This paper exploits strict flood insurance coverage limits and staggered flood map updates to show that mortgage lenders offload flood risk to the government through flood insurance contracts, and to under-insured households through higher down payments. Lender risk management leads delinquency rates to equalize inside and outside of flood zones. The combination of insurance requirements and credit rationing shift the composition of mortgages in flood zones towards richer and higher credit quality borrowers. In conclusion, lenders screen for flood risk when they retain residual exposures to it, and their credit rationing has distributional consequences for flood zones.


The Moral Preferences of Investors: Experimental Evidence, with J.F. Bonnefon, A. Landier, and D. Thesmar

Revise and Resubmit at the Journal of Financial Economics 

Abstract: We characterize investors’ moral preferences in a parsimonious experimental setting, where we auction stocks with various ethical features. We find strong evidence that investors seek to align their investments with their social values (“value alignment”), and find no evidence of behavior driven by the social impact of investment decisions (“impact-seeking preferences”). First, the willingness to pay for a stock is a linear function of corporate externalities, and is symmetric for positive or negative externalities. Second, whether charity transfers are contingent or independent on investors buying the auctioned stock does not affect their WTP. Our results are thus compatible with a utility model where non-pecuniary benefits of firms’ externalities only accrue through stock ownership, not through the actual impact of investment decisions. Finally, non-pecuniary preferences are linear and additive: willingness to pay for social externalities is proportional to the expected sum of charity transfers made by firms (even if some of these donations are negative).


Business as Usual: Bank Net Zero Commitments, Lending and Engagement, with D. Marques-Ibanez and E. Verner

Abstract: We use administrative credit registry data from Europe to study the impact of voluntary lender net zero commitments. We have two sets of findings. First, we find no evidence of lender divestment. Net zero banks neither reduce credit supply to the sectors they target for decarbonization nor do they increase financing for renewables projects. Second, we find no evidence of reduced financed emissions through engagement. Borrowers of net zero banks are not more likely to set decarbonization targets or reduce their verified emissions.  Our estimates rule out even moderate-sized effects. These results highlight the limits of voluntary commitments for decarbonization.


When Insurers Exit: Climate Losses, Fragile Insurers, and Mortgage Markets, with I. Sen and A.M. Tenekedijieva 

Abstract: This paper studies how homeowners insurance markets respond to growing climate losses and how this impacts mortgage market dynamics. Using Florida as a case study, we show that traditional insurers are exiting high risk areas, and new lower quality insurers are entering and filling the gap. These new insurers service the riskiest areas, are less diversified, hold less capital, and 20 percent of them become insolvent. Yet, despite their low quality, these insurers secure high financial stability ratings, not from traditional rating agencies, but from emerging rating agencies. Importantly, these ratings are high enough to meet the minimum rating requirements set by the government-sponsored enterprises (GSEs). We find that these new insurers would not meet GSE eligibility thresholds if subjected to traditional rating agencies’ methodologies. We then examine the implications of these dynamics for mortgage markets. We show that lenders respond to the decline in insurance quality by selling a large portion of exposed loans to the GSEs. We quantify the counterparty risk by examining the surge in serious delinquencies and foreclosure around the landfall of Hurricane Irma. Our results show that the GSEs bear a large share of insurance counterparty risk, which is driven by their mis-calibrated insurer eligibility requirements and lax insurance regulation.


Climate Risk and the U.S. Insurance Gap: Measurement, Drivers and Implications, with T. Scharleman,I. Sen and A.M. Tenekedijieva 

(Draft Available on Request)

It is widely believed that U.S. households are under-insured, but limited granular data on insurance has made this difficult to measure. This project develops a new methodology to construct the first U.S.-wide, long term dataset on homeowners insurance premiums and coverage at the individual level. We combine mortgage servicing, deeds, and property tax data, and then employ a novel algorithm to back out insurance payments from recurring mortgage payments made through escrow accounts. We then estimate coverage amounts from payments using data on insurance pricing functions. We validate our estimates using newly available data on insurance information for a subset of mortgage borrowers. We find that under-insurance is a significant problem, particularly for vulnerable borrowers in high climate risk states and with the lowest FICO scores. We show under-insurance is driven both by elastic borrowers reacting to rising premiums, as well as by behavioral inertia that limits updating coverage as inflation and construction costs change. We finally study the broader implications of under-insurance for mortgage and real estate markets.