The working paper was written by economist and fellow
* * *
Here are excerpts:
Millions of properties are exposed to increasing threats from natural disasters. Yet, many at-risk homes are uninsured against the costliest disaster: flooding. We show that low home equity is an important driver of low flood insurance take-up. To isolate the causal effect of home equity on flood insurance demand, we exploit price changes over the housing boom and bust. Insurance take-up follows house price dynamics closely, with a home price elasticity around 0.3. Multiple mechanism tests suggest that mortgage default acts as implicit disaster insurance. As a result, households do not fully internalize their disaster risk.
* * *
Large disasters cause severe financial distress for many households and lead to mortgage delinquency and default./2
An analysis from CoreLogic finds that the sequence of devastating hurricanes and wildfires in 2017-2018 tripled mortgage delinquency rates in affected areas (Betten et al., 2019)./3
Ouazad and Kahn (2019) find that major hurricanes caused a 1.6 percentage point increase in the probability of home foreclosure. However, millions of flood-prone properties in the country remain uninsured for flood damage, contributing to the mortgage system's exposure to disaster risk. Identifying the causes of this flood insurance demand gap is critical for understanding how climate change will affect households and financial markets.
In this paper, we provide the first empirical evidence for a novel, incentive-based cause of low flood insurance take-up: mortgage default as implicit disaster insurance. After a flood, leveraged homeowners can default and limit their losses, effectively making their home equity their deductible if uninsured. For low-equity households, mortgage default can crowd out demand for formal insurance. Through this mechanism, home equity can have a positive causal effect on flood insurance demand by raising the cost of defaulting.
We estimate the effect of home equity on the demand for flood insurance from the National Flood Insurance Program (NFIP). The main challenge to establish such a causal relationship comes from the correlation between equity and other determinants of insurance demand, such as income and disaster risk. To overcome this issue, we use the sudden variation in home prices from the housing boom and bust in the 2000s, which drove similar changes in home equity. This housing market cycle created price variation within and across housing markets driven primarily by changing land values and independent of gradual changes in flood risk, economic fundamentals, and demographics. Therefore, this setting is ideal for isolating the effect of home equity on flood insurance demand from that of the value of the physical structure at risk and other confounding factors.
We find a large, positive relationship between home prices and flood insurance take-up during this period. For a measure of the housing boom, we use estimated structural breaks in each MSA's home price trend from Charles et al. (2018), most of which occurred during 2003 to 2005. Figure 1 provides a reduced-form depiction of our results in the raw data. The left panel shows that MSAs with larger housing booms saw greater increases in flood insurance take-up between 2002 and 2007, which roughly correspond to the beginning and the peak of the boom. The right panel, in contrast, shows that over the housing bust from 2007 to 2012, the MSAs with the largest initial booms had the lowest growth in flood insurance policies.
Our formal difference-in-differences specification exploits variation in the timing and magnitude
* * *
3/ Other studies documenting the housing finance impacts of disasters include Anderson and Weinrobe (1986), Morse (2011), Billings et al. (2019), Issler et al. (2019), and Kousky et al. (2020).
* * *
of housing booms across MSAs and tracks the dynamics of home prices and flood insurance takeup over the boom-bust cycle. The results shows that flood insurance take-up closely follows the dynamics of home prices, has no pretrends, and is robust to controlling for annual income, housing turnover, population, employment, recent floods, and risk-dependent trends. Using housing boom size and timing as instruments in an instrumental variable (IV) framework, we estimate a home price elasticity of flood insurance take-up around 0.3. We also run a series of robustness checks to verify that the effect reflects voluntary purchases made by households and to address concerns about the exclusion restriction for our instrument.
We identify two mechanisms that may drive the relationship between home equity and flood insurance demand in our results. First, homeowners with more home equity may have a higher willingness to pay for flood insurance because post-disaster mortgage defaults are more costly (henceforth the "default incentive" mechanism). Second, higher home equity combined with easier credit access during the housing boom may have given households greater liquidity to pay annual flood insurance premiums (henceforth the "liquidity" mechanism). Theoretically, we demonstrate how each mechanism can create a positive relationship between home equity and insurance take-up in a stylized model. We then empirically explore these mechanisms by testing a series of hypotheses.
If home equity increases flood insurance demand by improving access to liquidity, then we expect a negative relationship between insurance lapsation, or nonrenewal, and home prices. Over 20 percent of flood insurance policies lapse in their second year (Michel-Kerjan et al., 2012). Across insurance settings, lapsation has been shown to be driven by liquidity constraints./4
We find that the relationship between home prices and flood insurance renewal rates is flat, which does not support the liquidity mechanism.
On the other hand, if home equity increases flood insurance demand through the default incentive mechanism, then we would expect a larger effect of home equity in states where default costs are low. We show that the home price elasticity of flood insurance take-up is significantly higher in states with borrower-friendly judicial foreclosure laws. Another prediction by this mechanism is that insurance demand should be more responsive to home equity in areas with greater tail risk exposure, which would induce default. Using a new national database of property-level flood risk, we find that MSAs with more tail risk exposure also have significantly higher home price elasticities of flood insurance take-up. Finally, take-up during the housing bust declines the most for homes built at the peak of the boom, exactly the group with the least home equity at the market's nadir. All of these findings support the default incentive mechanism.
These findings suggest that leveraged households do not fully internalize their environmental risk and that part of their risk is transferred to lenders instead. Lenders, in turn, rely on mortgage securitization to reduce their disaster risk exposure (Laux et al., 2017; Ouazad and Kahn, 2019; Keenan and Bradt, 2020). The government-sponsored enterprises (GSEs) that underwrite residential mortgage securitization do not price disaster risk or enforce mandatory flood insurance purchase outside of floodplains./5
As a result, the remaining risk is ultimately borne by taxpayers along with
* * *
4/ For a discussion of lapsation in broader insurance contexts, see Hambel et al. (2017) or Gottlieb and Smetters (2021).
5/ Even within floodplains, evidence is inconsistent on whether mandatory purchase requirements are well enforced (Hecker, 2002;
* * *
obligations from a host of post-disaster public transfers (see Deryugina (2017) and Billings et al. (2019)). As long as neither homeowners nor lenders bear the full cost of disasters, homes in risky areas will receive an implicit subsidy, a distortion that will grow with increasing climate risk.
This paper provides novel insights into the relationship between environmental risk and housing finance. We are the first to estimate the causal effect of home prices on disaster insurance takeup, and our estimates show an economically important causal relationship./6
Given a growing literature suggesting that climate change may already be influencing home prices, our estimates will be relevant to ongoing policy discussions around how climate change will affect financial and insurance markets./7
We also present and test a mechanism where mortgage default serves as implicit insurance, offering a new explanation for the insurance gap to complement studies on the role of adverse selection and information frictions (Gallagher, 2014; Mulder, 2019; Wagner, 2021), affordability issues (Netusil et al., 2021), and disaster aid (Billings et al., 2019; Kousky et al., 018b). Our findings suggest that, as with macroeconomic shocks, default can insure households against climate shocks (Mitman, 2016), albeit at the social cost of reducing incentives to formally insure or invest in adaptation.
This paper also relates to a larger literature on the effect of home prices and equity on household finance decisions. Our theoretical framework for understanding how home prices can influence insurance demand draws on an extensive set of studies examining the effects of leverage on homeowner incentives to default (Foote et al., 2008; Ferreira et al., 2010; Melzer, 2017; Ganong and Noel, 2020). In our setting, natural disasters often make homes uninhabitable, removing a key barrier to "strategic default." Our empirical analysis builds on the literature studying the impacts of changing home prices over the housing boom and bust on consumption and investment (Charles et al., 2018, 2019; Kaplan et al., 2020b; Mian et al., 2013). These results show that real estate finance has economically significant effects on insurance demand and homeowner disaster risk management.
Finally, our findings extend and are consistent with research on the interactions between implicit insurance from default and demand for conventional insurance. Most relevant to this study, Mahoney (2015) finds that bankruptcy acts as implicit health insurance and a higher cost of bankruptcy induces greater insurance demand. Similarly, Finkelstein et al. (2019) find that the availability of uncompensated care to uninsured patients can explain their low willingness to pay for formal health insurance. We show that default can also act as implicit disaster insurance, affecting demand for formal flood insurance and shifting environmental risk onto governments and creditors./8
The rest of the paper proceeds as follows. Section 2 provides a simple theoretical framework to motivate our empirical analysis. In Section 3, we describe our data and key features of the National Flood Insurance Program and the housing boom and bust. Section 4 explains our empirical design, Section 5 describes our results, and Section 6 concludes.
* * *
6/ Several studies have examined how insurance take-up in the NFIP is correlated with various factors (Kriesel and Landry, 2004; Kousky, 2011; Atreya et al., 2015). Typically, the analysis involves regressions that include home values as one of the covariates but not a formal treatment of unobserved confounding variables.
7/ See the related literature studying how climate and disaster risk are capitalized into home prices (Bernstein et al., 2019; Baldauf et al., 2020; Keys and Mulder, 2020; Murfin and Spiegel, 2020; Ortega and Taspinar, 2018) and how disasters affect housing markets (Gibson and Mullins, 2020; Kousky, 2010; Zivin et al., 2020).
8/ See Dobkin et al. (2018) for an example of how uninsured health costs spill over onto third parties.
* * *
We find a significant and positive relationship between home prices and flood insurance take-up over the housing boom and bust of the early 2000s. These price changes reflect a large increase in home equity for existing homeowners but little difference in their actual structural value at risk. After ruling out alternative explanations, such as new construction or mandatory purchase requirements imposed by the NFIP, our findings suggest that home equity plays a causal role in flood insurance demand. Moreover, the magnitude of the effect is comparable to other primary factors, such as premiums and flood events, in shifting flood insurance demand.
We explore two potential mechanisms for this effect. First, we test whether higher home equity increased demand by improving homeowner liquidity. We find no evidence that renewal rates increased with home prices, which does not support this mechanism. By contrast, our tests suggest that home equity may have affected demand by changing the implicit insurance value of mortgage default. For leveraged households facing a large flood loss, defaulting allows them to cap their losses at the value of their home equity. Thus, an increase in home equity lowers this implicit insurance value and increases demand for flood insurance. Consistent with this mechanism, we find higher home price elasticities of flood insurance demand in states with judicial review laws, where default is less costly, and in states with higher tail risk, where implicit insurance would be more valuable.
These results have important implications for understanding the likely impact of climate change on housing markets. As disaster risk increases over time, more homeowners will face the choice between purchasing insurance or risking default after a flood. The significant elasticity between changes in home prices and flood insurance take-up, combined with continuing low take-up rates in the NFIP, suggests that many leveraged households will choose not to insure. This means that some of their losses will ultimately be borne by the broader housing finance system or the GSEs that securitize mortgages and the taxpayers that support them. Home price declines driven by a bursting "climate bubble" along the coast (Bakkensen and Barrage, 2017; Bernstein et al., 2019; Keys and Mulder, 2020) could exacerbate these dynamics by reducing insurance demand.
However, our findings do point to two promising policy interventions. First, expanding the mortgage purchase requirement to high-risk non-SFHAs may lead homeowners and lenders to better internalize their flood risk. The SFHA mortgage mandate exists in part for this reason, and our findings suggest that underinsurance due to misaligned incentives in leveraged markets is prominent outside the SFHA. Second, GSEs themselves could start pricing the risk of disaster-induced default into securitization. This would improve lenders' incentive to require borrowers to maintain flood insurance.
* * *
Anderson, D. R. and
Bakkensen, L. A. and L. Barrage (2017). Flood risk belief heterogeneity and coastal home price dynamics: Going under water? Technical report,
Baldauf, M., L. Garlappi, and
Billings, S. B.,
Callaway, B. and
Case, K. E. and
Cengiz, D., A. Dube, A. Lindner, and
Charles, K. K.,
Charles, K. K.,
Collier, B. L. and
Deryugina, T. (2017). The fiscal cost of hurricanes: Disaster aid versus social insurance.
Dobkin, C., A. Finkelstein,
Einav, L., A. Finkelstein, and
Ferreira, F. and
Ferreira, F. and
Foote, C. L.,
Gallagher, J. (2014). Learning about an infrequent event: evidence from flood insurance take-up in the united states.
Ganong, P. and
Gibson, M. and
Goodman-Bacon, A. (2021). Difference-in-differences with variation in treatment timing.
Gottlieb, D. and
Hecker, J. Z. (2002). Flood insurance: Extent of noncompliance with purchase requirements is unknown. Technical report, GAO,
Keenan, J. M. and
Keys, B. J. and
Kousky, C. (2010). Learning from extreme events: Risk perceptions after the flood. Land Economics 86 (3), 395-422.
Kousky, C. (2011). Understanding the demand for flood insurance. Natural Hazards Review 12 (2), 96-110.
Kousky, C. (2017). Disasters as learning experiences or disasters as policy opportunities? examining flood insurance purchases after hurricanes. Risk Analysis 37 (3), 517-530.
Kousky, C., M. Palim, and
Kriesel, W. and
Laux, C., G. Lenciauskaite, and A. Muermann (2017). Foreclosure and catastrophe insurance. In Allied Social Sciences Associations Annual Meeting, Topics in Risk and Economics.
Mahoney, N. (2015). Bankruptcy as implicit health insurance. American Economic Review 105 (2), 710-46.
Mayer, C. (2011). Housing bubbles: A survey. Annual Review of Economics 3 (1), 559-577.
Melzer, B. T. (2017). Mortgage debt overhang: Reduced investment by homeowners at risk of default.
Michel-Kerjan, E. O. (2010). Catastrophe economics: the national flood insurance program.
Mitman, K. (2016). Macroeconomic effects of bankruptcy and foreclosure policies. American Economic Review 106 (8), 2219-55.
Morse, A. (2011). Payday lenders: Heroes or villains?
Mulder, P. (2019). Dynamic adverse selection in flood insurance. Available at SSRN 3435324 .
Murfin, J. and
Netusil, N. R.,
OpenFEMA (2020). Fima nfip redacted policies. https://www.fema.gov/openfema-data-page/ fima-nfip-redacted-policies.
Ortega, F. and
Ouazad, A. and
Sanderson, E. and F. Windmeijer (2016). A weak instrument f-test in linear iv models with multiple endogenous variables.
Sinai, T. (2012). House price moments in boom-bust cycles. In Housing and the Financial Crisis, pp. 19-68.
Wagner, K. (2021). Adaptation and adverse selection in markets for natural disaster insurance. Available at SSRN 3467329 .
Zivin, J. S. G.,
* * *
View full report at https://www.rff.org/documents/3068/WP_21-25.pdf