Skip to main content

The Influence of Crime, Disasters, and Climate on FHA Mortgage Behavior

HUD.GOV HUDUser.gov

Keywords: Mortgage Delinquency, Demographics, Natural Disasters, Climate Change, Crime, Federal Housing Administration

 
Trending
HUD USER Home > PD&R Edge Home > Trending
 

The Influence of Crime, Disasters, and Climate on FHA Mortgage Behavior

Stephanie Hawke, Economist, Housing Finance Analysis Division, Office of Policy Development and Research
Wenzhen Lin, Economist, Housing Finance Analysis Division, Office of Policy Development and Research

Front view of a row of townhomes and a street in the foreground.Understanding the local context is crucial in predicting and explaining borrowers' behavior. Photo credit: iStock.com/Kirpal Kooner

Introduction

Most research on mortgage delinquency and default focuses on micro- and macroeconomic determinants. This body of work, although valuable, overlooks a critical influence on individual behavior and decisionmaking: the local context. Our forthcoming study focuses on context and investigates the effects of crime rates, natural disasters, and rising temperatures on loan decisionmaking for Federal Housing Administration (FHA)-backed mortgages.

Extending the strategic default theory, we argue that borrowers facing acute local stressors make rational calculations regarding their mortgage payment obligations. If paying their mortgage is not in their financial best interest, these borrowers may choose delinquency and eventually default.

We have merged six publicly available datasets alongside FHA loan performance data to create a comprehensive, multilevel panel. This paper examines the mortgage termination risks associated with 90-day delinquency for single-family, fixed-rate FHA mortgages originating between 2011 and 2019. Our findings affirm the importance of borrower characteristics, mortgage details, and macroeconomic indicators while also supporting local-level variables as robust predictors of an individual's mortgage payment behavior.

Theoretical Framework

Decisions about mortgage behavior occur within a local context. We argue that individuals' perceptions of their locality — their neighborhood, community, or county — affect their perceived home valuation. Substantial neighborhood stressors may alter people's perceptions of their immediate surroundings, leading them to believe that the value of their homes is diminishing. This perspective shift causes them to feel alarm and doubt about their housing investment — the most significant investment for many Americans. When individuals believe that their home's value is declining, they reconsider continuing to invest in their homes — in this case, through mortgage payments. For some borrowers, this leads to a choice to eschew their mortgage obligations (figure 1).

Figure 1. Overarching Theory

A flowchart that reads, from left to right, "Independent Variables", "Perceived Home Value Decreases over Time", "(rational) Reassessment of Financial Investment in Home", "Increased Incentive for Mortgage Delinquency.


We test three local variables that we hypothesize will significantly influence the likelihood of mortgage delinquency. First, we link crime rates to increased delinquency risk. Borrowers perceive crime as a sign of neighborhood deterioration. Rising crime rates prompt homeowners to grow concerned about the safety and viability of their housing investment, leading them to decide to default on mortgage payments to safeguard their remaining assets in the face of perceived threats.

Second, we hypothesize that natural disasters affect borrowers' perception of their home value. A borrower who owns a home in an area affected by natural disasters may worry that the disaster depresses all home values in the affected area, even if the borrower's home was not directly affected. Borrowers in affected areas may discontinue payments because of this perceived home devaluation.

Finally, we anticipate a positive correlation between temperature increase and delinquency likelihood, which previous studies on the housing price index and purchase behavior support. Considering increased awareness of climate change and its economic effects, we posit that temperature increases can significantly influence individuals' perceptions of their neighborhoods. With this new perspective, residents may contemplate the prospect of sustained temperature increases, envisioning a future in which the climate becomes less hospitable. This conclusion, in turn, may lead them to question the safety of their home investment, potentially prompting a decision to cease mortgage payments.

Data and Methods Approach

Our sample construction involved creating a quarterly panel from the selection of a 1 percent subset of FHA loans from the Single Family Housing Enterprise Data Warehouse (SFHEDW) encompassing 97,504 loans and 1,588,543 originations between 2011 and 2019. The SFHEDW dataset includes comprehensive information on borrower and mortgage characteristics, including standard loan details such as credit scores, loan-to-value ratios, loan amounts, mortgage rates, debt-to-income ratios, current unpaid status, and property locations.

Our analytical approach involves using a logit regression to predict the probability of delinquency, ranging from 0 percent to 100 percent. Our model encompasses a diverse set of control variables, including macroeconomic indicators (such as interest rates, unemployment rates, and state of residence) and microeconomic indicators (such as individual income at origination, debt, and mortgage terms) and draws from previous research in these areas. The All-Transactions House Price Index at the three-digit ZIP Code level from the Federal Housing Finance Agency gauges the movement of single-family house prices. Unemployment rates from the U.S. Bureau of Labor Statistics provide insights into current economic conditions. Table 1 summarizes borrower and loan characteristics for mortgages originating from 2011 to 2019.

Table 1. Summary Statistics

Variable

Mean

Std. dev.

Original Income

Current Loan-to-Value Ratio

Original Debt-to-Income Ratio

Original Interest Rate

Original FICO Score

First Time Homebuyer

61,516.71

78.41

35.66

4.06

688.6

0.58

45,544.76

15.35

15.61

0.53

49.22

0.49

Black 

Hispanic

White

Other 

Missing

0.10

0.16

0.63

0.04

0.08

0.29

0.37

0.48

0.19

0.27

Male

Female

0.61

0.34

0.49

0.48

Unemployment Rate (%)

4.98

2.01

Num of Loans

Num of Observations

97,504

1,588,543


To determine whether undue geographic clustering in our independent variables might have biased our models, we geographically plotted our data (figure 2). We find that no significant geographic clustering is interfering with the model. Note that the temperature data was available for only a subset of counties across the country.

Figure 2. Geographic Distribution of Independent Variables

(From the top left going clockwise) Property Crime Cumulative Change; Violent Crime Cumulative Change; Average YOY Temperature Change (C); Disasters, by county, 2011–2019


Our analytical approach involves using a series of logit regressions to predict the probability of delinquency, ranging from 0 percent to 100 percent based on our variables of interest. First, we established a base model that incorporates and validates variables suggested by existing literature. We then conducted subsequent logit regressions, introducing the independent variables to assess their significance.

Results

We discovered robust associations between our variables and delinquency, with all models outperforming a constant-only model and demonstrating improved predictability over the base model. This finding confirms our overarching hypothesis that local factors influence individual borrowers' likelihood of serious delinquency.

Base Model

Our results underscore the significance of various factors, including borrower attributes such as original FICO score, loan characteristics such as loan-to-value ratio and debt-to-income ratio, and broader economic indicators such as the unemployment rate, in predicting 90-day delinquency. These findings corroborate existing literature and remain consistent across our variables of interest (table 2).

Table 2. Logit Regression Results 


Base

Crime

Disasters

Temperature

 

Model 1

Model 2

Model 3

Model 4

Model 5

Change of Violent Crime


0.035**


 


(0.017)


Change of Property Crime


-0.029


 


(0.018)


Lag of Disasters



0.137***






(0.031)



Lag of Fire




0.151***


 




(0.057)


Lag of Hurricane




0.228***


 




(0.044)


Temperature Increase





0.125**






(0.050)

Maximum Temperature





0.019*






(0.011)

Controls Included:

Borrower and Loan Characteristics

Y

Y

Y

Y

Y

Other Disaster Types

N

N

N

Y

N

State Fixed Effect

Y

Y

Y

Y

Y

Time Fixed Effect

Y

Y

Y

Y

Y

Pseudo R-squared

0.060

0.060

0.068

0.068

0.082

 

1,307,583

1,307,583

1,004,731

 

 

N

3

3

1

1,004,534

106,231

Notes: Standard errors are in parentheses. * p<0.10, ** p<0.05, *** p<0.01. "Lag of Disasters" is defined as the number of disasters that occurred in this area in the past year.

Our analysis reveals a significant positive relationship between rising year-over-year crime rates at the county level and mortgage delinquency. We isolated and tested violent crime and property crime, and we find that violent crime predominantly influences this model. We find no substantial evidence indicating a predictive relationship between property crime and delinquency.

Natural Disasters

When considering the lagged effect of natural disasters on delinquency rates, we find a significant, positive relationship. We then isolated and tested different types of disasters and found that wildfires and hurricanes are significantly linked with increases in delinquency at the county level. This empirical evidence underscores the substantial socioeconomic implications of these catastrophic events and underscores the need to develop proactive measures and targeted interventions aimed at mitigating the adverse effects of natural disasters on local communities.

Temperature Increase

Our analysis reveals a noteworthy correlation between annual county-level temperature increases and escalating rates of mortgage delinquency, indicating a significant positive relationship. Furthermore, we ascertain that the maximum temperature within a county serves as a significant predictor for the amplification of mortgage delinquency instances. This observation suggests that regions characterized by high temperatures that are experiencing further warming trends are particularly prone to surges in mortgage delinquency rates. These findings are particularly noteworthy for stakeholders within the real estate and financial sectors, who may need to consider factoring climatic variables into risk assessment models and strategic decisionmaking processes.

Discussion and Conclusions

This study examines how three manifestations of local context affect an individual's mortgage behavior. We argue that individuals read the local context to determine whether their investment in their home is sound. Those who believe that their investment in their home is unsafe — that is, because the local context puts that investment at risk — will decide to halt their investment and stop paying their mortgage.

We find strong evidence that variables such as crime rates, natural disasters, and rising temperatures affect individuals' mortgage decisions in various ways. Crime — particularly violent crime — is strongly associated with mortgage delinquency, although nuances in this relationship exist. Natural disasters such as hurricanes and floods increase the likelihood of mortgage default. Similarly, rising average temperatures are linked to increased mortgage defaults. These findings add to the literature on climate change and its impact on housing and mortgage markets, underlining the importance of considering the effects of climate change when evaluating loan performance.

Borrowers facing acute local stressors may choose to default if paying their mortgage is not in their best financial interest. Therefore, understanding the local context is crucial for predicting and explaining borrowers' behavior. Like many social science researchers, we find that the relationships investigated are complicated. However, we uncover robust evidence that the local context significantly impacts how borrowers interact with their mortgages. Our study highlights the need for a more nuanced understanding of borrower behavior in the face of local stressors.

Stephanie Hawke (stephanie.t.hawke@hud.gov) and Wenzhen Lin (wenzhen.lin@hud.gov) contributed equally to this article. The analysis and conclusions presented are those of the authors alone and should not be represented or interpreted as conveying an official position, policy, analysis, opinion, or endorsement of the U.S. Department of Housing and Urban Development or the U.S. Government.

Giovanni Favara, Enrique Schroth, and Philip Valta. 2012. “Strategic Default and Equity Risk Across Countries,” Journal of Finance 67:6, 2051–95; Kristopher Gerardi, Kyle Herkenhoff, Lee Ohanian, and Paul Willen. 2018. “Can’t Pay or Won’t Pay? Unemployment, Negative Equity, and Strategic Default,” Review of Financial Studies 31:3, 1098–131; Luigi Guiso, Paola Sapienza, and Luigi Zingales. 2013. “The Determinants of Attitudes toward Strategic Default on Mortgages,” Journal of Finance 68:4, 1473–515. ×

Andrew Haughwout, Sarah Sutherland, and Joseph Tracy. 2013. "Negative Equity and Housing Investment," Staff Report No. 636, Federal Reserve Bank of New York. ×

Michael Craig and Adam Hoffberg. 2023. "Natural Disasters and Mortgage Risk in the Federal Housing Administration." Presented at the annual meeting of the American Economic Association, San Antonio, Texas, 5 January 2024; Katherine Kiel and Victor Matheson. 2015. "The Effect of Natural Disasters on Housing Prices: An Examination of the Fourmile Canyon Fire." Working Paper 157, Economics Department, College of the Holy Cross; Tamara Sheldon and Crystal Zhan. 2019. "The Impact of Natural Disasters on US Home Ownership," Journal of the Association of Environmental and Resource Economists 6:6, 1169–203. ×

Li Ma and Yildiray Yildirim. 2023. "High Temperature, Climate Change and Real Estate Prices," Climate Change and Real Estate Prices, 2 June; William Doerner and Wenzhen Lin. 2022. "Applying Seasonal Adjustments to Housing Markets," Cityscape 24:3, 87–122. ×

Jonghyun Yoo and YuYu Zhou. 2024. "Thermal comfort and retail sales: A big data analysis of extreme temperature's impact on brick-and-mortar stores," Journal of Retailing and Consumer Services 77, 103699. ×

Endorsements that would have received an accept decision from TOTAL regardless of PRH status also must report PRH. ×

Between 2015 and 2022, FHA’s average annual count of purchase mortgage endorsements was 800,000, and the average annual volume of endorsements was $175 billion. See: U.S. Department of Housing and Urban Development, Federal Housing Administration. 2023. “Financial Status of the Mortgage Insurance Fund: Fiscal Year 2023, Quarter Three.” ×

 
Published Date: 11 June 2024


The contents of this article are the views of the author(s) and do not necessarily reflect the views or policies of the U.S. Department of Housing and Urban Development or the U.S. Government.