Determinants of government aid to Katrina survivors: evidence from survey data.

1. Introduction

Natural catastrophes like Hurricane Katrina are (unfortunate) natural experiments for evaluating otherwise unobservable events such as the effectiveness of government disaster aid. Katrina was arguably the greatest natural disaster in modern U.S. history. The damage stretched across 90,000 square miles an area roughly the size of Great Britain--and at least 1836 people lost their lives. Thousands of Gulf Coast residents lost their livelihoods, and many were forced to permanently relocate. Economic losses are estimated at $81.2 billion (and growing), nearly double the costs associated with the next-most costly disaster, Hurricane Andrew (Johnson 2006).

Earthquakes, hurricanes, and other natural disasters have received considerable attention in the sociological and medical literature (e.g., Peacock, Morrow, and Gladwin 1997), (1) and there is a growing but diverse economic literature related to these disasters. For example, one strand of literature looks at the effect of hurricanes and tornadoes on insurance and housing markets. Fronstin and Holtmann (1994) suggest that the structural quality of homes is a substitute for insurance coverage. As a result of this moral hazard problem, newer homes sustained more damage than older homes in the 1992 Hurricane Andrew. Sadowski and Sutter (2005) argue that improvements in technology (e.g., early warning systems) make hurricanes less lethal. People are therefore more and more willing to incur the risk of living in hurricane-prone coastal regions; as a result, over time greater property damage is associated with less-lethal hurricanes. Ewing, Kruse, and Wang (2007) studied six different metropolitan areas and estimated an immediate but short-lived 0.5-2.0% decline in housing prices (or market losses ranging from $34 million to $580 million) following a tornado or hurricane. The price declines lasted only about four quarters, and the impact was consistent across tornados or hurricanes, indicating that markets help integrate and normalize losses.

The political economy of disasters and disaster relief has also garnered attention. Garrett and Sobel (2003) analyze the political aspects of the declaration process for natural disasters and the distribution of disaster aid. Using state-level data, they find that nearly half of all disaster relief is explained by political factors rather than by need. Congleton (2006), Shughart (2006), and Sobel and Leeson (2006) are examples of public choice explanations of government's failures to respond effectively to the tragic turn of events caused by Hurricane Katrina.

Natural disasters are a shock to local economies and local labor markets for two different reasons. The event itself is a negative shock that interrupts business activity (see Webb, Tierney, and Dahlhamer [2000] for an overview). But the subsequent recovery efforts provide a positive shock. For example, Webb (2007) estimates that as a result of Katrina rebuilding, the Mississippi State economy grew an additional 2.7% over a comparable period from the previous year, and he reports that job creation in Harrison County (devastated by Katrina's storm surge) accounted for almost half of the 26,878 new nonagricultural jobs created in Mississippi between February 2006 and February 2007. The effects can be long run in nature, too. In 1997, the U.S. Federal Emergency Management Agency (FEMA) launched Project Impact, which was designed to make hazard-prone communities less vulnerable to natural catastrophes. Ewing and Kruse (2002) find that in the pilot community of Wilmington, North Carolina, the project contributed to a lower natural rate of unemployment and a reduction in labor market risk (i.e., lower variance after controlling for trend, seasonal, and cyclical effects). Using a similar time-series model, Ewing, Kruse, and Thompson (2005) find that the 1999 Hurricane Bret relief and recovery activities helped lower the natural rate of unemployment by about 0.75% in the Corpus Christi, Texas, labor market.

Here we report results from a uniquely designed survey sample of Hurricane Katrina survivors along the Mississippi Gulf Coast, which we refer to as the Mississippi Gulf Coast Survey, or, for brevity, the "Gulf Coast Survey." Like Garrett and Sobel (2003), we are interested in assessing whether government disaster aid is distributed on the basis of need or on the basis of other criteria. Our approach differs from that of previous research in that we address this question directly by analyzing survivors' responses to the Gulf Coast Survey. We use logistic regression to model government aid as a function of need while controlling for ascribed and acquired characteristics like age, gender, and education. One interpretation of our regression results is that government aid is helpful in dealing with one- to two-month economic disruption and long-term rebuilding, but it could do better at short-term rebuilding and mitigating longer-term economic disruption.

We also present some pre- and post-Katrina descriptive statistics from the Gulf Coast Survey, as well as a basic economic risk analysis. Together with our regression analysis, these yield three additional findings. First, individuals who receive government aid and/or have physical disabilities prior to a disaster are among those most likely to incur severe economic hardship after the fact. Second, in our study, individuals who (after controlling for public aid, disability, etc.) are older, college educated, and/or married seem to be less at risk, perhaps because they have greater financial, economic, and/or social network capital available in a time of crisis. Third, our analysis underscores the importance of housing in the postdisaster resumption of basic economic activity. In our conclusion, we comment on some broad policy implications of these findings. It is our hope that our results can be used to better understand the timing and types of government aid that facilitates relief and recovery in the wake of natural disasters.

2. Federal and Private Aid after Katrina

An impressive amount of federal aid has been allocated to Gulf Coast states directly affected by Hurricane Katrina. More than 16,000 federal workers were deployed to the region; $88 billion in aid was allocated for relief, recovery, and rebuilding; and another $20 billion has been requested for future efforts (White House 2006). This aid has been accompanied by some successes: 80% of storm debris has been cleared in Mississippi, 169 miles of damaged levees have been restored, $5.8 billion in disaster loans have been administered by the Small Business Administration, an extension of unemployment benefits for Gulf Coast residents has occurred, and there has been provision of both short-term shelter and longer-term housing for Katrina survivors (White House 2006).

Despite the magnitude of government aid, the public and/or media perceptions of government officials and government disaster relief agencies are less sanguine. There are an abundance of media reports on failures of government officials and/or aid to adequately help the Katrina survivors. Sobel and Leeson (2006) provide interesting anecdotes. In some cases, aid was impaired because of burdensome paperwork, while in other cases, aid was provided only because guidelines or regulations were ignored; other examples include FEMA's lengthy delays in requesting military assistance and in delivering federal aid and the subsequent inadequate level of assistance. (2) Some decisions and/or actions were predicated on information that either was not timely or was simply wrong. For instance, Sobel and Leeson maintain that FEMA set up trailer parks that were largely unused or that were placed in locations in which they weren't needed and that the costs of these mostly unused assets were extravagant.

On the other hand, private and nonprofit relief efforts are generally viewed as the antitheses of public relief efforts. Shughart (2006) cites the activities of Wal-Mart, Home Depot, and FedEx as examples of the relative efficiency of private decision makers in reacting to the natural disaster. He reports that 126 Wal-Mart stores were shut down by the storm and, within two weeks, 111 of them had reopened. All three companies are given kudos for supplying aid to their communities and, as an example, FedEx delivered 440 tons for the Red Cross to the Gulf Coast (often at no charge). The performances of nonprofit relief agencies like the Red Cross are more difficult to characterize. Shughart contends that the lack of incentives hampers both nonprofit and government relief agencies; he cites news stories charging that the Red Cross was "missing in action for at least two days" after Hurricane Katrina hit. However, Sobel and Leeson (2006) take a more generous view. They argue that the government has an incentive to monopolize disaster relief aid and hinder the efforts of nonprofit competitors. Consistent with this hypothesis, the Red Cross' initial two-day delay in getting aid to areas hit by Katrina was not caused by their own failings but by orders from government officials.

3. The Mississippi Gulf Coast Survey