Financing Recovery After Disasters: Explaining Community Credit Market Responses to Severe Events

• Published date: 01 June 2019
• Date: 01 June 2019
• Author: Benjamin L. Collier,Volodymyr O. Babich
• DOI: http://doi.org/10.1111/jori.12221

479

©2017 The Journal of Risk and Insurance. Vol.XX, No. XX, 1–42 (2017).

DOI: 10.1111/jori.12221

Financing Recovery After Disasters: Explaining

Community Credit Market Responses to Severe

Events

Benjamin L. Collier

Volodymyr O. Babich

Abstract

Credit provides a means for uninsured households and businesses to man-

age disaster losses, but access to credit may be tenuous after severe events.

Using lender fixed effects models, we examine how natural disasters affect

the amount of credit supplied by community lenders in developing and

emerging economies. We find that disasters reduce lending. We consider

two potential causes of lending reductions: (1) disasters reduce expected

returns on loans made after the event or (2) capital constraints (lenders’ dif-

ficulty replacing equity lost during the event). Wedevelop a dynamic model

that informs our empirical identification of these causes and conclude that

capital constraints cause observed credit contractions. We also examine the

effects of insurance market development and find evidence that insurance

preserves the creditworthiness of borrowers. Our results demonstrate per-

vasive disaster-related credit supply shocks in developing and emerging

economies and identify new insurance market opportunities.

Introduction

Households and businesses often turn to credit to finance recovery following a nat-

ural catastrophe (Del Ninno, Dorosh, and Smith, 2003; Berg and Schrader, 2012). For

example, Collier et al. (2016) examine effects of Superstorm Sandy on businesses in

the New Yorkarea and find that 40 percent of negatively affected firms increased their

debt because of the event. Moreover, more of the firms negatively affected by Sandy

borrowed to finance recovery than receivedinsurance payments. Credit provides cash

in a crisis and can allow firms to replace lost assets, increasing their earnings oppor-

tunities after a severe event. The role of credit in managing disaster losses seems

particularly important in developing and emerging economies, where frequently

Benjamin L. Collier is at the Fox School of Business, Temple University, Risk, Insur-

ance, and Healthcare Management, Philadelphia, PA. Collier can be contacted via e-mail:

collier@temple.edu. Volodymyr O. Babich is at the McDonough School of Business, George-

town University,Washington, DC. United States. Wethank Florian Berg, Wouter Botzen, David

Cummins, Andrew Haughwout, Howard Kunreuther, Marc Ragin, Jerry Skees, Matthew J.

Sobel, and Mary Weissfor their help and insightful comments, and Juan Zhang for her research

assistance.

Vol. 86, No. 2, 479–520 (2019).

2The Journal of Risk and Insurance

480

insurance markets are nascent and social safety nets scarce. Yet, these disasters may

also affect the ability and willingness of financial intermediaries (FIs) to lend. Can

firms and households rely on their lenders to supply credit after a catastrophe?

We evaluate the effects of natural catastrophes on the amount of credit supplied by

community lenders. Specifically,we examine the behavior of FIs that specialize in non-

corporate credit, lending to households and micro, small, and medium enterprises

(MSMEs), in developing and emerging economies. We use a panel of 929 lenders

from 78 countries spanning 18 years. Weestimate lender-specific fixed effects models

to compare a lender’s total loan allocations in years when disasters occur to those

when they do not. Wefind that natural disasters reduce lending. The most severe dis-

asters reduce annual loan growth by 50 percent on average. We consider two potential

consequences of disasters that may explain these lending declines. First, a lender may

contract credit due to capital constraints. Capital market frictions may preventa lender

from replacing the equity that it loses in the disaster from the poor performance of

existing loans (e.g., increased loan defaults). That is, an FI may adjust lending be-

cause disasters erode its capital available for lending. Second, a lender may contract

credit due to a reduction in the expected return on lending following disasters, which

may occur due to changes in either lending risks or credit demand. Disasters may

increase lending risks; for example, reduced economic productivity may increase the

likelihood of default. Similarly, disasters may affect lending interest rates: a decrease

(increase) in demand for credit may reduce (increase) lending interest rates. Thus, an

FI may adjust lending because disasters affect its “expected returns on ex post credit

investments.”

To distinguish between these potential causes of creditcontraction, we develop a dy-

namic theoretical model of a representativeFI and examine its decisions of investment,

dividend payments, and deposit subscriptions. In this model, the returns on invest-

ment portfolios are random and the distributions of returns are Markov modulated,

according to the disaster state regime. The lender’s optimal policy depends on an in-

teraction between the disaster state and the lender’s capital state (i.e., its equity stock).

An empirical identification strategy emerges from this model’s insights. The model

predicts that changes in the expected returns cause a lender with high capital to adjust

lending more than a lender with low capital. A lender with high capital determines its

credit supply by equating the expected marginal benefits and costs of lending; how-

ever,a lender with low capital must also contend with binding capital constraints and

the greater possibility of insolvency, making its supply less elastic in the expected re-

turns on lending. In contrast, the model predicts that the investment portfolio of a low-

capital lender is more sensitive to available capital and so equity losses from the event

cause a lender with low capital to reduce lending more than a lender with high capi-

tal. The model offers several additional insights that inform our empirical robustness

tests.

Using our panel of FIs, we find that lending reductions are explained by capital con-

straints due to disaster-related losses to lenders’ existing portfolios. Lenders with low

capital ratios before a disaster lend substantially less afterward; however, lenders

with high capital ratios tended to lend at the same rate following the event. For

Financing Recovery After Disasters 3

481

low-capital lenders, large disasters reduce annual loan growth by 117 percent. Thus,

access to credit to manage disaster losses would seem to vary substantially across

households and businesses depending on the capital structure of their lenders.

We also examine the effects of insurance coverage on disaster-related credit con-

traction. Insurance may prevent loan defaults by providing payments to borrowers

and/or preserve their creditworthiness by protecting collaterelizable assets. We mea-

sure insurance coverage using the ratio of a country’s nonlife insurance premiums

relative to its average annual damages from natural disasters (using data from 1975 to

2014). Following our main results, we find evidence of capital constraints in economies

with higher insurance coverage: low-capital FIs lend less due to disasters. Our finding

that loan losses constrain the credit supply in economies with high insurance cover-

age is consistent with the common finding that even in the most developed insurance

markets, a large portion of disaster losses remain uninsured (Swiss Re Sigma, 2013).

In economies with low insurance coverage, we find that FIs also contract credit due to

a decline in the expected returns on lending: low- and high-capital FIs lend less due

to disasters. These results highlight an important role for insurance in maintaining

the creditworthiness of borrowers.

The primary contributions of this article are identifying systematic effects of natural

disasters on community credit supply and explaining their causes. Previous research

provides a series of rich case studies regarding how disasters affect credit markets,

which guide our analyses. Yet, case studies areembedded in a specific disaster, credit

market, and year, and so how their findings generalize is unclear. We trade detailed

knowledge of a specific event for a more systematic comparison across many severe

disasters, lenders, countries, and years. Our results are consistent with the findings

of those case studies and indicate that the challenges disasters create for community

credit markets are widespread in developing and emerging economies.

We recognizetwo implications of our results. First, our finding that access to credit is

often unreliable after a disaster highlights the importance to the enterprise of proac-

tively managing risk. Second, our findings identify opportunities for insurance mar-

kets and public policy to reduce the consequences of severe events, which we discuss

in our conclusions. In particular, expanding the capital available to lenders would

seem to increase access to credit for households and businesses after major disasters.

Related Research

Recent trends elevate the importance of understanding and managing the conse-

quences of natural disasters. Natural disasters have caused $1.7 trillion in economic

losses and 647,000 fatalities in the last decade (2007–2016; Swiss Re Sigma, 2017).

Moreover, losses from these events will increase due to a confluence of more volatile

weather, strained natural resources, and development and urbanization (Stern, 2008;

Kunreuther and Michel-Kerjan, 2009).

Macroeconomic recovery from a natural disaster seems to depend on an economy’s

ability to reallocate resources after the event (e.g., through credit, insurance, or ef-

fective government spending; Noy, 2009; von Peter, von Dahlen, and Saxena, 2012).