finalPADM5390SylvesandBzs-2007-PresidentialDisasterDeclarationDecisions1953.pdf

State and Local Government Review Vol. 39, No. 1 (2007): 3–15

Presidents’ decisions to approve or deny governor requests for declara-tions of major disaster entail judgments of both a political and managerial nature. Federal disaster relief “is based on the idea that federal aid is necessary to supplement state and local relief” (Garrett and Sobel 2003, 3). Declarations go out to states and only then to the counties affected by the dis aster or emergency in those states. A presidential declaration legitimizes the disaster for af- fected populations and is a major step in a state obtaining federal aid (Schneider 1995). Most declarations are for natural disasters. However, various human-caused calamities have warranted declarations, foremost among them the terrorist attacks on New York’s World Trade Center in 1993 and 2001 and the Pentagon in 2001, as well as the bombing of Oklahoma City’s Murrah Federal Offi ce Building in 1995. In August 2005, Pres. George W. Bush approved emergency decla- ration requests submitted by the governors of Louisiana, Mississippi, Alabama, and Flor- ida for Hurricane Katrina for pre-event mo- bilization and ensuing wind and fl ood devas- tation (Sylves 2006, 31). President Bush also encouraged governors of states that received evacuees who were displaced by the hurricane to seek presidential disaster declarations. This

Presidential Disaster Declaration Decisions, 1953–2003: What Infl uences Odds of Approval? Richard Sylves and Zoltán I. Búzás

situation was unusual in that states that were unaffected directly nevertheless were eligible to seek declarations that would reimburse them for the cost of receiving evacuees (Sylves 2006, 29).

Presidential authority to address domes- tic disasters won political support following World War II because there was no domes- tic equivalent to the foreign aid policy that necessitated that Americans pay taxes to fi - nance rebuilding and recovery efforts abroad (Homeland Security National Preparedness Task Force 2006, 7–9). Before 1950, Congress considered unique relief legislation for each disaster, which often resulted in inconsistency, delay, pork barreling, and administrative con- fusion (Platt 1999). By 1950, lawmakers deter- mined that it made more sense to entrust dec- laration decision making to the president as an executive responsibility (May 1985; Platt 1999). The presidential disaster declar ation authority that was enshrined in law in 1950 grants the president the discretion and fl ex- ibility to decide what constitutes a disaster or emergency. Presidents may interpret broadly or narrowly what is declarable as a major dis- aster or emergency (Waugh and Sylves 2002). Each president makes declaration decisions on a case-by-case basis (Committee on Gov- ernmental Affairs 1993).

4

Sylves and Búzás

State and Local Government Review

In 1974, federal law granted the president the authority to decide to approve governor requests for emergency declarations as well as to independently issue them for life protec- tion or safety purposes.1 These preemptive declarations usually are made for disasters that are about to transpire and for which fed- eral mobilization is judged necessary by the president. They also may apply in the im- mediate aftermath of an event when federal emergency response might make rescues, em- ergency medical help, and property protec- tion possible.2

Pres. Dwight D. Eisenhower issued the fi rst serially numbered, modern-era presiden- tial declaration of major disaster (DR#1) in May 1953 for a tornado in Georgia (Sylves 1996; 2005; Rubin and Tanali 2001). Since then, governors have requested presidential declarations at the rate of about four (3.93) a month, or almost once a week. From May 1953 to December 2001, 10 presidents as a group approved requests at a rate of 2.71 per month, or roughly one approval every two weeks (Public Entity Risk Institute 2006).

Context of Disaster Declarations

Presidential declarations of major disaster and emergencies are unusual because the author- ity to make the essential decision rests solely with the executive offi ce. Most federal laws require government offi cials who function in an entirely different capacity from that of the president to make implementation decisions (Anderson 1997). Once the president issues a declaration, federal agency and program offi cials, usually in concert with their state and local counterparts, follow elaborate, ex- tensive procedures (May and Williams 1986; Sylves 2005).

Historically, local and state governments have been delegated authority and responsi- bility for implementing disaster preparedness, response, and recovery according to their own laws and traditions (May and Williams 1986; Platt and Rubin 1999). The federal govern- ment comes to the assistance of a state gov-

ernment when the president determines that the state is overwhelmed by or incapable of addressing a disaster (Waugh 2000). Ordinar- ily, the governor of the disaster-stricken state asks the president for a declaration by con- tacting the relevant federal disaster agency, including various federal agencies and offi ces from 1950 to March 1979, the U.S. Federal Emergency Management Agency [FEMA] from April 1979 to March 2003, and the U.S. Department of Homeland Security [DHS] FEMA from March 2003 to the present. Today, the contact is the Under Secretary of Federal Emergency Management. The presi- dent, considering the governments affected by the disaster and ongoing human suffering and losses, may approve the governor’s re- quest for a declaration.

Over the years, federal disaster offi cials have attempted to establish defi nitive and quantitative requirements for disaster decla- ration eligibility. One such effort would have tied declarations to damage translated into dollars per capita and required that rigor- ous criteria be met (Settle 1990). Presidents have resisted, and Congress has vehemently opposed such measures (Settle 1990; U.S. General Accounting Offi ce 2001). Presidents do not want their range of declaration dis- cretion further circumscribed or ceded to federal disaster offi cials (Witt and Morgan 2002; Platt 1999; Daniels and Clark-Dan- iels 2000). Legislators want assurance that they may use their legitimate political infl u- ence to press for declarations directly from the president when their home states and districts experience incidents or events that they consider to be emergencies or disasters (Sylves 1996).

Presidents, assisted by their staffs and top disaster agency offi cials, must judge each gov- ernor’s request for a declaration based on need. Both managerial and political factors may enter into the president’s judgment. Ini- tial damage assessments, imminent dis aster threat (e.g., a hurricane that is about to make landfall), and news media coverage of an event may make it obvious to the president that a

5Vol. 39, No. 1, 2007

Presidential Disaster Declaration Decisions, 1953–2003

governor’s request deserves approval (Sylves 1996; Birkland 1997). However, there are many instances in which presidents and their advisers are unconvinced of the worthiness of the request.

Although presidents regularly turn down governor requests for declarations of major disaster or emergency, they sometimes ap- prove such requests when damage in the state is relatively light and recovery from the event may be possible without federal assistance (Sylves 1996; Miskel 2006). A major Senate task force report (Bipartisan Task Force on Funding Disaster Relief 1995) concluded that presidents tended to issue too many major di- saster declarations for events that state and lo- cal governments could have easily recovered from on their own. The report averred that overgenerous declaration issuance was creat- ing a situation of “moral hazard” in which fed- eral largess was discouraging state and local governments from investing in even routine disaster mitigation and preparedness because their leaders expected their jurisdictions to be “bailed out” by a presidential declaration of major disaster (see also Platt and Rubin 1999).

Many factors contribute to presidential decisions regarding requests for disaster dec- larations, including national media coverage of local calamities and subsequent calls for federal intervention; unclear criteria under which states may become eligible for a dec- laration; the tendency of governors and other state and local authorities to seek federal aid for both major and minor disasters and emer- gencies; and the political pressures and con- siderations that presidents face when they are petitioned for help, which sometimes compels them to make immediate decisions regarding approvals for requests for disaster declara- tions (Sylves 1996; Schneider 1995; Birkland 1997). Thus, presidents make their determi- nations regarding disaster requests within a complex political environment (Waugh and Hy 1991; Wamsley, Schroeder, and Lane 1996; Wamsley and Schroeder 1996; Platt 1999).

Governors and Disaster Declarations

Governors play a key role in the presidential declaration process. When a disaster strikes, local authorities and individuals often request help from state government as well as from private relief organizations. If they believe the situation warrants, governors may make a request for a state disaster or emergency declaration, typically through an executive order or proclamation. The order usually de- scribes the nature of the emergency, where it occurred, and the authority under which the governor makes the declaration (Beauchesne 1998).

All governors, including the mayor of Wash ington, D.C., and the heads of U.S. Trust or Commonwealth Territories, have the authority to request a presidential declar ation of major disaster or emergency. Although governors may bypass the federal disaster agency and petition the president directly for a declaration, since 1950 the standard pro- cess of requesting a major disaster declara- tion usu ally begins when a governor submits an offi c ial request to the president through the appropriate federal disaster agency. Un- der pro visions of the Stafford Act of 1988 (Title VI, 42 U.S.C. 5195a) and preceding law, governors submit their requests to their respective FEMA regional director (from 1953 until FEMA was created in early 1979, various federal emergency offi ces and agen- cies handled governor requests through their regional offi ces). A process similar to the one set forth in the Stafford Act of 1988 existed from 1953 to 1988, albeit with variations (National Acad emy of Public Administra- tion 1993; Platt 1999). In March 2003, when FEMA was folded into DHS, the governor request process largely was maintained as before, although the FEMA director had to report to the DHS secretary rather than di- rectly to the president (DHS 2006).

Governors usually fi rst consult their state emergency managers before they request a presidential declaration. The governor may authorize a state-level preliminary damage

6

Sylves and Búzás

State and Local Government Review

assessment if state offi cials are not already conducting one with local authorities. Some- times, if the disaster appears to be beyond the capacity of a state or locality, the gover- nor may ask the federal disaster agency to join state and local personnel in conducting a preliminary damage assessment. Local gov- ernments sometimes press state government and the governor to ask the president for a declaration of major disaster.

Ordinarily, a governor requests a presiden- tial declaration of major disaster because he or she concludes that an event has overwhelmed state emergency response capability or the state does not have the means to recover from the event without federal help. Such a request usually summarizes signifi cant aspects of the event; presents statistics relative to damage and losses; outlines the contributions made by federal, state, local, and private agencies; highlights unmet needs for which the gover- nor seeks federal assistance; and suggests a recommended course of action for the presi- dent. The president may turn down requests for declarations from governors for calamities that their states could easily recover from us- ing state and local resources or for problems that clearly are ineligible for a presidential declaration.

Previous Studies of Presidential Disaster Declarations

In their study titled “The Political Economy of FEMA Disaster Payments,” Garrett and Sobel (2003) examined disaster declaration data from 1991 to 1999 and found that presi- dential and congressional infl uences affect the rate of disaster declaration issuance and the allocation of FEMA disaster expenditures across states; states that are politically impor- tant to the president receive more declara- tions; and expenditures are higher in states that have congressional representation on FEMA oversight committees. Garrett and Sobel (2003, i, 2) claim that their models “predict that nearly half of all disaster relief is motivated politically rather than by need”

and that the “potential exists for political in- fl uence to impact the process at two distinct stages: whether or not a disaster is declared, and then how much money is allocated for the disaster.” Furthermore, “the public choice model predicts that those states politically important to the president are likely to have more disasters declared” (Garrett and Sobel 2003, 8).

Only looking at the rate of declaration is- suance, as Garrett and Sobel (2003) do, is not suffi cient to measure political bias or mo- tives because president-rejected governor re- quests for declarations are part of the politi- cal picture. In this study, the unit of analysis is fundamentally the governor request, not declarations alone. That is, the key point of political infl uence is whether the president decides to approve a governor’s request for a declaration.

Once the president approves a governor’s request for a declaration, FEMA actually de- termines how much money is allocated to states, counties therein, and other eligible organizations under specifi c conditions (i.e., damage assessment), laws, and rules—deci- sions that are subject to audit by a variety of government offi ces. Political discretion exer- cised by the president is likely to be evident when the president approves governor re- quests regarding low damage, marginal inci- dents that often involve relatively low federal payouts. Thus, the relationship is an inverse one. In other words, the lower federal pay- outs are for various declarations, the higher the probability that political considerations at the presidential level played a role in a president’s approval of a declaration (Miskel 2006, 134–35).

Garrett and Sobel’s (2003, 1) congressional dominance model postulates that “bureaus are very responsive to the wishes of Con- gress” and that “congressional committees with budget and oversight responsibilities see that bureaucrats implement the policy preferences of the legislators (legislators are wealth maximizers).” The results of the cur- rent study do not support Garrett and Sobel’s

7Vol. 39, No. 1, 2007

Presidential Disaster Declaration Decisions, 1953–2003

claim that FEMA seeks to infl uence disaster declarations to reward legislators who are on its appropriations and oversight committees. Because the decision to approve or reject a governor’s request for a declaration is made by the president, FEMA offi cials would have only indirect infl uence. In most cases, the FEMA director compiles information for the president about the event that is the subject of the governor’s request and recommends that the president either approve or turn down the request based on in-house criteria of deservedness. This information and the recommendation are protected by rules of executive privilege and therefore are not sub- ject to public scrutiny, nor may they be used in assessing FEMA’s role.

Nevertheless, presidents may use their discretion to reward states that are the po- litical homes of key House and Senate legis- lators and advance electoral strategies benefi - cial to themselves, their fellow party members on the Hill, and others political actors whom they judge to be important, including the re- questing governors themselves. As Garrett and Sobel (2003, 4) assert, “because the Staf- ford Act (1988) allows the president to uni- laterally declare a disaster without the ap- proval of Congress, it is possible that the president may use this power to punish or reward legislators who support or oppose his policies or just simply tarnish the image of opposing party legislators in hopes of reduc- ing their probability of reelection.” Moreover, the vague language of the Stafford Act and federal laws established since the Federal Di- saster Act of 1950 about what constitutes a disaster “means that an offi cial federal disas- ter could have occurred whenever the presi- dent said it did” (Garrett and Sobel 2003, 4). Since September 11, 2001, Congress has ac- corded the president powers that have vastly increased the presidential discretion in mat- ters of disaster, as evidenced by the enactment of the Homeland Security Act of 2002 and Pres. G. W. Bush’s issuance of a series of Home land Security Presidential Directives. Disasters and emergencies are within the spec-

trum of “incidents of national signifi cance” upon which the president may decide whether to issue a declaration (Sylves 2005).

In his study on electoral politics and pres- idential disaster declarations, Reeves (2005) draws on research by economists Garrett and Sobel as well as statistics from FEMA. He combines these data with those on whether political parties view particular states as “friends,” “enemies,” or “competitive” based on their likelihood of voting for the party’s presidential candidate. Reeves’s “analysis of 10 years (1989–99) of such decisions by for- mer presidents George H. W. Bush and Bill Clinton shows that, at least in marginal disas- ters, the size of the state (in terms of electoral votes) and whether the political parties view it as ‘competitive’ matters quite a bit” (Tarcey 2004, 1).

Reeves (2005, 5–6) maintains that disaster declarations are “an easy-to-use political tool at the president’s disposal.” When an emer- gency strikes, the governor of the affected state can request a disaster declaration, which the president may accept or deny. In the de- cade studied by Reeves (1989–99), 17 percent of the requests (generally for minor weather events) were turned down (Reeves 2005, 17). When the requests are accepted, FEMA—not the president—decides how much money to allocate. The average grant for 570 disasters (not including those in Texas) was $36.4 mil- lion. Outlays ranged from $6,301 (Montana in 1999) to $7 billion (California in 1994). During his four years in offi ce, Pres. George H. W. Bush issued an average of 39 disas- ter declarations annually (Reeves 2005, 10). Reeves’s study found that there was an average of 72 disasters per year during the seven years of the Clinton presidency (Reeves 2005, 10). Focusing on the 1992 and 1996 presidential election years, Reeves found that “President Clinton was about 60 percent more likely than President Bush to declare a disaster in a pivotal, electorally important state” (Tarcey 2004, 1).

According to Reeves (Reeves 2005, 16), large states that are friendly to the president

8

Sylves and Búzás

State and Local Government Review

appear to benefi t more than do large, un- friendly states: “Even if California’s 54 elec- toral votes are virtually assured to a candi- date, [the president] will most likely be more protective of a lead.” Yet, presidents tend to give more to “enemies” than to “friends” in small states: “The positive public-relations impact of dis aster relief may worry political opponents, perhaps forcing them to spend money in a state they had previously con- sidered friendly. On the other hand, spend- ing the money in a small friendly state may be a waste of political capital” (Reeves 2005, 19). Though there may be many reasons why small, uncompetitive states receive less money, the political ramifi cations stand out: according to Tarcey (2004, 1), “an uncom- petitive state with three electoral votes is 50 percent less likely to receive a disaster declaration than a competitive state with 20 electoral votes. . . . The best predictor of a presidential disaster declaration, bar none, is actual need. The question arises in these marginal cases when it’s unclear whether to give or not.” Marginal disasters include events like the 1994 deep freeze in Michigan, which garnered less than $6 million in aid (Tarcey 2004, 1). Tarcey (2004, 1) contends that “in these (marginal) cases, small, noncompetitive states receive less help because they are just not that important in terms of an election.” He notes that if there is a major tornado, a state will receive aid but not in the case of a marginal event such as a fl ood or windstorm that does not cause as much damage.

Reeves (2005) as well as Garrett and Sobel (2003) examine the electoral importance of states that have received presidential decla- rations and consider whether the partisan- ship of the requesting governor is related to presidential approval decisions. Downton and Pielke (2001) provide evidence that presiden- tial declarations for fl ood disasters are greater in election years when the president is run- ning for reelection. Dymon and Platt (1999) argue that in the process of presidential dec- laration approvals, the main political actors involved (i.e., presidents and governors) act

in conformity with legislative theory in the sense that declarations are issued “on the basis of political motives, political pressures, and political responsiveness more than they are issued on the basis of objective need” (Dymon and Platt 1999, 58–59).

Purpose

This study addresses what presidential dis- aster declarations are, the process of their is- suance, and the controversies that surround turndowns. The chief focus is on the presi- dent-governor decision point. The study mea- sures whether certain political variables and disaster agent–type variables are statistically correlated with the approval and turndown decisions of presidents from Eisenhower to G. W. Bush (1953–2003). FEMA data from these administrations are used to examine the factors that might have infl uenced or been the basis for the decisions of presidents to either approve or deny governor requests for declarations of major disaster.

This study is distinct from that of other researchers of declaration politics, specifi cally Garrett and Sobel (2003) and Reeves (2005), who to date have not considered turndowns or conducted longitudinal analyses encom- passing more than two or three presidential administrations. Both approved declarations and governor requests for declarations that are turned down by the president rather than approvals alone are examined over a 50-year period; measures of disaster severity are not included.3

Quantitative Analysis

The database for this analysis was compiled from offi cial FEMA disaster declaration in- formation from 1994, 1997, 2001, 2004, and 2005. (Much of this information is available on the Public Entity Risk Institute Web site at www.peripresdecusa.org.) Information about governors who were in offi ce when disaster declaration requests were made came from Wikipedia (2006). State demographic infor-

9Vol. 39, No. 1, 2007

Presidential Disaster Declaration Decisions, 1953–2003

mation was drawn from the U.S. Census Bu- reau (2006) and Demographia (2006).

The statistical model used in this study re - lies on two theoretical approaches. One per- spective assumes that political factors infl u- ence the odds of a state receiving a pres i dential disaster declaration. Dymon and Platt (1999) maintain that the pattern of presidential dec- laration approvals across the United States suggests that presidents and governors have behaved in accord with legislative theory; that is, declaration requests and the pattern of presidential approvals suggest that declara- tions are issued on the basis of political mo- tives, pressures, and responsiveness more than on objective need (Peterson 1995; Joyce and Kneedler 1999).

By contrast, according to the functional fed eralist perspective, presidential disaster declarations tend to refl ect obvious need for such a declaration (Anton 1989; Peterson 1995; Walker 1995; Joyce and Kneedler 1999). In other words, the president’s approval of governor requests for declarations may be very much based on indisputable objective need fi rst and foremost in almost every case.

To gauge the possible infl uence of legisla- tive theory, this study includes political fac- tors (such as party of the president and the requesting governor) that are likely to infl u- ence presidents as they consider governor requests for declarations of major disaster. Further, to ascertain the possible infl uence of functional federalist reasoning, type of di- saster (i.e., primary incident type) is taken into account as a surrogate measurement of need. Although not an exact assessment of objective need, this approach redresses the imbalance in the literature on presidential disaster declarations by considering type of disaster.4 The analysis considers 2,170 major disaster declaration requests from governors to presidents between 1953 and 2003.

The dependent variable is “outcome,” which refers to a president’s decision to ap- prove or turn down a governor’s request for a declaration of major disaster. The variable is coded 1 for approvals; turndowns are coded 0.

The main independent variables are described as follows.

Political Independent Variables “Party of president” refers to the political party of the president in offi ce at the time the governor’s request was approved or denied. Republican presidents are coded 0; Demo- cratic Presidents are coded 1. It is expected that Democratic presidents approve a signifi - cantly higher percentage of governor requests for declarations of major disaster than do Re- publican presidents, regardless of the political party affi liation of the governor, because of ideological differences. In general, Demo- cratic presidents are more likely to support an invigorated federal role in dis aster policy whereas Republican presidents are more likely to consider disaster management to be a state government responsibility. Related to this assumption, Republicans likely set a higher threshold for states to qualify for federal relief than do Democratic presidents.

“Reelection year” is a dichotomous vari- able, coded 1 for presidential election years when the incumbent president is running for reelection; any other year is coded 0. Regard- ing this variable, presidents are assumed to take a greater interest in disaster declarations in a reelection year. Thus, the percentage of presidential approvals of governor requests for major disaster declarations should increase signifi cantly in a reelection year.

“‘Friend’ or ‘enemy’” is a dichotomous var iable, coded 0 when the partisanship of the president and the state governor is differ- ent (“enemy”); the variable is coded 1 when it is the same (“friend”). Reeves (2005) and Garrett and Sobel (2003) posit that partisan difference between a requesting governor and the president is an important political factor in explaining whether the president approves or denies a governor’s request for a major disaster declaration. For many in the emergency management community, there is a standard assumption that holds that presi- dential decision making in matters of gover- nor requests for disaster declarations is very

10

Sylves and Búzás

State and Local Government Review

often a political and partisan-biased determi- nation of the president. If true, the odds of approval would be expected to be less when requests come from governors of the opposite party. In other words, presidents who receive requests for major disaster declarations from governors of the opposite party are likely to turn down those requests at greater rates than they are requests from governors of their own political party.

“Governors’ partisanship” refers to the po- litical party of the governor in offi ce at the time the declaration decision of the president is announced. It is rare that a requesting gov- ernor is not also the governor in offi ce at the time the president issues an approval or turn- down. This dummy variable is coded 0 for Republican governors; Democratic governors are coded 1. Independents, third-party, and ap pointed (Trust Territories only) governors are excluded from this analysis.5 Democratic governors tend to be more likely to win presi- dential declarations of major disaster than Republican governors, even when controlling for electoral votes of the states in question, and Democratic presidents are more likely to approve more declaration requests than are Republican presidents; therefore, Demo- cratic presidents and Democratic governors are likely to be mutually supportive of one an- other. “Electoral votes” refers to the number of electoral votes of each requesting state as of the most recent presidential election.

Also employed are variables for friendly and pivotal states. A state is categorized as pivotal if in the preceding presidential elec- tion, the margin of victory or loss of the presi- dent (always the president making the dec- laration decision) is equal to or below �/�5 percent.6 The pivotal state variable is coded 1 for pivotal states; all other states are coded 0. Similarly, a state is categorized as friendly if the vote margin of victory for the winning president is above 5 percent. This variable is coded 1 for friendly states; all other states are coded 0. For example, Pres. G. H. W. Bush’s de cisions on whether to approve or deny gov- ernor requests for major disaster declarations

were assumed to be infl uenced by an electoral strategy based on how those requesting states voted in the 1988 presidential election. States that requested disaster declarations during the period between President Bush’s inauguration in January 1989 until immediately before the 1992 presidential election are categorized as friendly or pivotal based on how they cast their electoral votes in the November 1988 election.

Also included is a control for “lame duck” transitional periods.7 In the 1988 election, for example, Texas was a friendly state because Pres. G. H. W. Bush had a 12.6 percent mar- gin of victory in that election. In other words, when President Bush considered Texas re- quests for major disaster declarations during the period between January 1989 and early November 1992, Texas was a friendly state in terms of his electoral strategy. Based on election results in the November 1992 elec- tions, Texas became a pivotal state because President Bush’s margin of victory in Texas was 3.5 percent; therefore, all Texas requests between the election in November 1992 up to President Bush’s departure from offi ce are considered as having come from a pivotal state. When President Clinton took offi ce, Texas remained a pivotal state (even though Clinton lost) because Texas chose President Bush in 1992 by a 3.5 percent margin.

Type Disaster Independent Variables Disasters have many manifestations and after- effects. Federal agency assignment of “pri- mary-incident” type for any single disaster request is based on what a governor describes as the nature of the event and what appears to be the lead or primary cause or force produc- ing the disaster (FEMA 2001; 2002; Sylves 1998). Certain categories encompass sets of destructive forces. This study collapsed com- parable primary-agent disaster requests into major categories, including fl ood and tornado, severe storm, hurricane or typhoon, human- caused disaster, and earthquake and volcano. Because the extremely high rates of approv- al for declaration requests for earthquake,

11Vol. 39, No. 1, 2007

Presidential Disaster Declaration Decisions, 1953–2003

volcano, and other seismic disasters would re- sult in low variation in the dependent variable, these disaster types were not included in the study. The effect of each dependent variable is assessed relative to other types of disasters.

“Flood and tornado” is a dummy variable coded 1 for all major disaster requests with the primary-incident designators fl ood, tor- nado, fl ood and tornado, and rain. These des- ignators are combined into a single set. All other disaster types are coded 0. Because most approved and denied major disaster requests are for fl oods and tornadoes, the dependent variable is expected to be dramatically affect- ed by whether the governor request is for a fl ood- or tornado-type event. The direction of the correlation (i.e., positive or negative) is not predicted before the test, however.

The “severe storm” dummy variable (cod- ed 1) represents all major disaster requests with the primary-incident designators severe storm, snow, coastal storm, freezing, severe ice storm, fi re, or beach erosion; all other disaster types are coded 0.8 Requests for declarations for disasters in the severe storm category are the next-largest pool of disaster type, but there are many more requests in this category that may or may not be worthy of a presiden- tial declaration. The assumption is that states that can recover from a disaster using their own resources will be seen as only margin- ally deserving of a presidential declaration of major disaster. Consequently, presidents are likely to judge severe storm–type requests on political grounds more than on need.

“Hurricane or typhoon” is a dummy vari- able. Major disaster requests for hurricane (Atlantic) and typhoon (Pacifi c) are coded 1; all other disaster types are coded 0. Although one might expect that declaration requests for hurricanes and typhoons would have a dispro- portionately higher rate of approval relative to other types of disasters, these events do not always produce enough damage to justify presidential approval of a governor’s request. Nevertheless, it may be diffi cult for a presi- dent to deny disaster relief to states that have very little hurricane damage when adjacent

states that have more hurricane damage win automatic approvals of their declaration re- quests. Thus, the assumption here is that it is easier for a president to justify approving requests for hurricane or typhoon disasters than it is for disaster types that produce less damage.

The dummy variable for human-caused dis asters encompasses all major requests for disaster relief for acts of terrorism, explosions, toxic substance releases, and losses resulting from fi sheries closures, for example. These disasters are coded 1; other disaster types are coded 0. Although one might expect that ter- rorism events would receive automatic ap- provals—as did the attacks in 1993 and 2001 on the World Trade Center, in 1995 on the Murrah Federal Offi ce Building in Oklahoma City, and in 2001 on the Pentagon—requests for declarations for other kinds of human- caused events may be turned down. Some explosions, toxic substance releases, and loss- es sustained from fi sheries closures do not reach a threshold of damage to warrant ap- proval from the president. Human-caused disasters receive no greater or lesser rates of approval than do other types of disasters. However, human-caused requests encompass a variety of marginal events as well as indis- putably catastrophic disasters. The assump- tion is that presidents who are considering small-scale, human-caused disaster requests will have a strong incentive to approve or reject these requests in accord with their po- litical considerations and motives.

The Model

Binary logistic regression was used in this in- vestigation to analyze the data set. That is, the dependent variable is dichotomous, and the independent variables may be of any type.9 The percentage change in odds regarding presidential approval or turndown of gover- nor requests for major disaster declarations is considered within the context of political and disaster-endemic factors.

12

Sylves and Búzás

State and Local Government Review

The results show that presidential disaster declarations are infl uenced by both political factors and type of event, corroborating both functional federalist and legislative theory explanations (see Table 1). Further, the fi nd- ings tend to confi rm the suppositions. All the variables except for friendly and pivotal state, electoral votes, human-caused event, and “friend”/“enemy” are statistically signifi cant at the .05 level.

The following equation is used in the mod- el to calculate the odds of approval of a gov- ernor’s request for a presidential disaster dec- laration for specifi c values of the independent variables: Odds of approval = �.865 + .292 (party of president) + �.522 (governors’ parti- sanship) + .052 (“friend”/“enemy”) + .681 (re- election year) + appropriate coeffi cient (type of disaster) + �.005 (electoral votes) + .145 (“friend” state) + �.005 (pivotal state).10

The variable for type of event for hurricane or typhoon has the strongest positive impact on odds of approval relative to the other dis- aster categories. Expressed in odds ratio, the odds of receiving an approval for hurricane or typhoon requests are 24.20 times more likely compared with the reference category (i.e., all other disaster types). The variable for se-

vere storm has a positive but lesser impact on odds of approval than do fl ood and tornado. Whereas the odds of approval are 8.753 times higher for fl oods and tornadoes than for the reference category, the odds of approval for severe storm are 2.277 times higher than are those for the reference category. As expected, the odds of approval for disaster declarations for human-caused events were not statisti- cally signifi cant. Moreover, the factor was negatively correlated with odds of approval. Turndowns of nonterrorist human-caused requests more than cancel out the approvals of the few terrorist case requests.

The political variables for friendly state, pivotal state, electoral votes, and “friend”/ “enemy” are not statistically signifi cant. Of the signifi cant variables, only the partisanship of the governor is negatively correlated with the dependent variable. Contrary to what was expected, Democratic governors are .593 times less likely to receive approvals than are Republican governors—a 40.7 percent lower odds of approval. On the other hand, Demo- cratic presidents are 1.339 times more likely to approve requests than are Republicans. Further, whether the request for presiden- tial declaration is made in a reelection year

Table 1. Binary Logistic Regression for Presidential Disaster Declarations, 1953–2003

Wald Percentage Model Estimates Coeffi cient Signifi cance Odds Ratio Change in Odds

Constant �0.865 Party of president 0.292* .010 1.339 33.9 Governors’ partisanship �0.522** .000 0.593 �40.7 “Friend” or “enemy” 0.052 .626 1.053 5.3 Reelection year 0.681** .000 1.977 97.7 Flood and tornado 2.169** .000 8.753 775.3 Severe storm 0.823** .000 2.277 127.7 Human caused �0.434 .284 0.648 �35.2 Hurricane or typhoon 3.187** .000 24.205 2320.5 Electoral votes �0.005 .286 0.995 �0.5 “Friend” state 0.145 .352 1.156 15.6 Pivotal state �0.005 .977 0.995 �0.5

*p < .05; **p < .01; N = 2,170; –2 log likelihood = 2253.835; Cox and Snell R2 = .150; Nagelkerke R2 = .207; percent correct prediction of the model = 70.8. Note: The dependent variable is outcome (i.e., approval or turndown).

13Vol. 39, No. 1, 2007

Presidential Disaster Declaration Decisions, 1953–2003

is very important: the results indicate that the odds of receiving an approval are 1.977 times (or 97.7 percent) higher than in non- reelection years.

Conclusion

Consistent with the fi ndings of Garrett and Sobel (2003) and Reeves (2005), the results of this study indicate that political factors do matter in presidential decisions regard- ing disaster declarations. However, different measurements and variables were employed. For example, state partisanship was not based on whether a state is considered to be friendly or pivotal as it was in Reeves’s (2005) study. Rather, the codifi cation of state partisanship to gauge whether a presidential electoral strat- egy may have infl uenced an approval or turn- down decision depended on the outcome of the previous presidential election. Moreover, this study examined a much longer period (1953–2003) than have previous studies.

In contrast with Reeves’s (2005) conclu- sions, this study fi nds that variables related to the partisanship of a requesting state do not signifi cantly affect chances of presiden tial approval or turndown decisions. There is no statis tically signifi cant evidence that odds of approval are any more or less if a governor is of the same political party as the decid- ing pres ident. However, the odds of approval are greater among Democratic than Repub- lican presidents. The four Democratic Party presidents (Kennedy, Johnson, Carter, and Clinton) tended to approve a larger percent- age of requests than did the six Republican Party presidents (Eisenhower, Nixon, Ford, Reagan, G. H. W. Bush, and G. W. Bush), even though from 1953 to 2003, Republican governors were more likely than Democratic governors to receive approvals.

Several broad predictions may be made re garding the probability of presidential ap- proval of a governor’s request for a disaster declaration. For instance, approval is likely under the following conditions: the disaster can be included in the hurricane or typhoon

category, the requesting governor is a Repub- lican, the president is a Democrat, and it is a reelection year. On the other hand, the odds of approval are comparatively lower if there is a human-caused disaster, the governor is a Democrat, the president is a Republican, and it is not a reelection year. Nevertheless, although both political and type-of-event factors infl uence presidential decisions, it appears that presidents take the nature of a disaster into account more than they do the political aspects of a request.

Presidents’ decisions to approve or deny governor requests for declarations seem to rest on matters of law and public management; however, their decisions also involve electoral and partisan politics. Differentiating between political and type-of-event variables may help explain presidential decision making regard- ing declarations of major disaster, which not only are indicative of federal-state relations and but also ultimately have implications for grants of federal aid to state governments.

Richard Sylves is a professor of political science and international relations and a senior policy fel- low at the Center on Energy and Environmental Policy at the University of Delaware. In addition to having been an appointed member of the Na- tional Academy of Science Disasters Roundtable, he is coeditor of Disaster Management in the U.S. and Canada (Charles C. Thomas, Publisher 1996) and author of The Nuclear Oracles (Iowa State University Press 1987).

Zoltán I. Búzás is a doctoral student in the De- partment of Political Science and International Relations at the University of Delaware. His cur- rent interests involve socioconstructivist interpre- tations of trans-Atlantic relations and European and global governance.

Notes 1. When governors request major disaster declarations,

they often must provide some evidence or documen- tation of loss; that is, they must demonstrate need or

14

Sylves and Búzás

State and Local Government Review

provide a justifi cation for their request. Emergency declarations are often fast-tracked requests that do not need to document loss but merely indicate that a certain urgent state of affairs exists. Federal emer- gency declarations cap federal spending at $5 mil- lion, but the president merely need notify Congress by letter that an emergency will cost more than $5 million in order to spend more.

2. Emergency requests and declarations are omitted from this analysis for two reasons. First, many emer - gency declarations are converted in a matter of hours or days into a major disaster declaration. Sec ond, emergency declarations complicate matters in so far as some emergency declarations are issued by the president without having been fi rst requested by a governor. Emergency turndowns muddle the analy- sis because they are few in number and sometimes encompass anomalous types of incidents.

3. Because the investigators have no information re- garding the severity of the events that were turned down, they have no basis for incorporating severity into the analysis. FEMA turndown records, in which governor requests for major disaster declarations were denied, do not contain any information about the severity of each event; therefore, the severity of events that were denied presidential disaster declara- tions could not be gauged.

4. Although type of disaster is only a modest indica- tor of need, certain pooled categories of disaster type manifest relatively high rates of turndown and marginality.

5. Major disaster declaration requests and approvals from U.S. Trust and Commonwealth territories are excluded from this analysis, as are requests for emer- gency declarations, because they do not have elec- toral votes in U.S. presidential elections.

6. Pivotal or friendly states are categorized from the perspective of the president in offi ce at the time of the declaration decision.

7. The study controls for lame-duck presidential per- iods. For example, the Kennedy and Johnson ad- ministrations are grouped jointly for the 1961–64 period. Similarly, the Nixon and Ford administra- tions are considered jointly up to August 1974, when Nixon resigned the presidency, to when Ford left offi ce in January 1977.

8. Some requests and approvals for fi re disasters pertain to major urban fi res; other declarations are made for fi re events that occur at the urban-wildland interface and forest fi res that affect community areas. This study did not include fi re suppression actions as- sociated with the U.S. Forest Service and FEMA because presidents are not personally involved in the review and issuance of such actions.

9. Binary logistic regression applies maximum likeli- hood estimation after transforming the dependent variable into a logit variable. Unlike ordinary least- squares regression, however, logistic regression does not assume linearity of the relationship between the independent and dependent variables, require

normally distributed variables, or assume homosce- dasticity (that is, that the variance around the regres- sion line be the same for any value of an independent variable). Several researchers in the fi eld of disaster declarations adopt Poisson regression as their esti- mation technique because the dependent variable they test is a count variable (Reeves 2005; Garrett and Sobel 2003).

10. In order to calculate the predicted probability of approval, the appropriate values of the independent variables (for instance, for party of president, 0 for Republicans and 1 for Democrats) must be intro- duced and then multiplied by the coeffi cients. The result is expressed in log odds.

References

Anderson, James E. 1997. Policy implementation. In Public policymaking: An introduction, ed. Charles O. Jones, 213–79. 3rd ed. Boston: Houghton-Miffl in.

Anton, Thomas J. 1989. American federalism and public policy. New York: Random House.

Beauchesne, Ann M. 1998. A governor’s guide to emergency management. Washington, DC: Natural Resources Policy Studies Division, National Governor’s As- sociation Center for Best Practices.

Birkland, Thomas A. 1997. After disaster: Agenda setting, public policy, and focusing events. Washington, DC: Georgetown University Press.

Daniels, R. Steven, and Carolyn L. Clark-Daniels. 2000. Transforming government: The renewal and revita li za- tion of the Federal Emergency Management Agency. Ar- lington, VA: Pricewaterhouse Coopers Endowment for the Business of Government.

Demographia. 2006. U.S. population by state from 1900. www.demographia.com/db-state1900.htm. Accessed June 20.

Downton, Mary W., and Roger A. Pielke. 2001. Discre- tion without accountability: Politics, fl ood, damage, and climate. Natural Hazards Review 2:157–66.

Dymon, Ute J., and Rutherford H. Platt. 1999. U.S. federal disaster declarations: A geographical analysis. In Disasters and democracy: The politics of extreme natu- ral events, ed. Rutherford B. Platt. Washington, DC: Island Press.

Garrett, Thomas A., and Russell S. Sobel. 2003. The po litical economy of FEMA disaster payments. Eco- nomic Inquiry 41:496–509.

Homeland Security National Preparedness Task Force. 2006. Civil defense and homeland security: A short history of national preparedness efforts. Washington, DC: U.S. Department of Homeland Security.

Joyce, Philip G., and Amy E. Kneedler. 1999. Emergency management competencies and incentives and impli- cations for funding in the intergovernmental context. Paper presented at the 60th National Conference of the American Society for Public Administration, Denver.

15Vol. 39, No. 1, 2007

Presidential Disaster Declaration Decisions, 1953–2003

May, Peter J. 1985. Recovering from catastrophes: Federal disaster relief policy and politics. Westport, CT: Green- wood Press.

May, Peter J., and Walter Williams. 1986. Disaster policy implementation: Managing programs under shared gov- ernance. New York: Plenum Press.

Miskel, James F. 2006. Disaster response and homeland secu- rity. Westport, CT: Praeger Security International.

National Academy of Public Administration (NAPA). 1993. Coping with catastrophe: Building an emergency management system to meet people’s needs in natural and manmade disasters. Washington, DC: NAPA.

Peterson, Paul E. 1995. The price of federalism. Washing- ton, DC: Brookings Institution.

Platt, Rutherford H. 1999. Federalizing disasters: From compassion to entitlement. In Disasters and democracy: The politics of extreme natural events, ed. Rutherford B. Platt. Washington, DC: Island Press.

Platt, Rutherford H., and Claire B. Rubin. 1999. Stem- ming the losses: The quest for hazard mitigation. In Disasters and democracy: The politics of extreme natural events, ed. Rutherford B. Platt. Washington, DC: Island Press.

Public Entity Risk Institute. 2006. All about presidential disaster declarations. www.peripresdecusa.org. Ac- cessed November 6.

Reeves, Andrew. 2005. Political disaster? Presidential di- saster declarations and electoral politics. www.people. fas.harvard.edu/~reeves/papers/fema.pdf. Accessed March 7, 2007.

Rubin, Claire B., and Irmak R. Tanali 2001. Disaster time line: Selected milestone events and U.S. outcomes (1965–2001). Arlington, VA: Claire B. Rubin and Associates.

Schneider, Saundra K. 1995. Flirting with disaster: Public management in crisis situations. Armonk, NY: M.E. Sharpe.

Settle, Allen K. 1990. Disaster assistance: Securing pres- idential declarations. In Cities and disaster: North American studies in emergency management, ed. Rich- ard T. Sylves and William L. Waugh Jr. Springfi eld, IL: Charles C. Thomas, Publisher.

Sylves, Richard T. 1996. The politics and budgeting of federal emergency management. In Disaster man - agement in the U.S. and Canada, ed. Richard T. Sylves and William L. Waugh Jr. Springfi eld, IL: Charles C. Thomas, Publisher.

_________ . 1998. Disasters and coastal states: A policy analysis of presidential declarations of disaster, 1953–97. DEL-SG-17-98. Newark: Sea Grant College Pro- gram, University of Delaware.

_________ . 2006. President Bush and Hurricane Ka tri na: A presidential leadership study. Annals of the American Academy of Political and Social Science 604:26–56.

Tarcey, Brian. 2004. Flooding the ballot box: The politics of disaster. Harvard Magazine. www.harvardmagazine. com/on-line/030492.html. Accessed March 7, 2007.

U.S. Census Bureau. 2006. www.census.gov. Accessed June 20.

U.S. Congress. Senate. Committee on Governmental Af fairs. 1993. Hearing on rebuilding FEMA: Preparing for the next disaster. 103rd Cong., 1st sess.

_________ . Bipartisan Task Force on Funding Disaster Relief. 1995. Appendix II: Profi les of major federal disas- ter assistance programs. Federal disaster assistance. 104th Cong., 1st sess.

U.S. Department of Homeland Security (DHS). 2006. www.dhs.gov/dhspublic/index.jsp. Accessed June 20.

U.S. Federal Emergency Management Agency (FEMA). 2001. Federal Emergency Management Information System (FEMIS) data download. Accessed Decem- ber 12.

_________ . 2002. www.fema.gov/library/. Accessed June 20, 2006.

U.S. General Accounting Offi ce. 2001. Disaster assistance: Improvement needed in disaster declaration criteria and eligibility assurance procedures. GAO-01-837. Wash- ington, DC: GPO.

Walker, David B. 1995. The rebirth of federalism: Slouch ing toward Washington. Chatham, NJ: Chat ham House.

Wamsley, Gary L., and Aaron D. Schroeder. 1996. Es- calating in a quagmire: Changing dynamics of the Emergency Management Policy System. Public Ad- ministration Review 56:235–44.

Wamsley, Gary L., Aaron D. Schroeder, and Larry M. Lane. 1996. To politicize is not to control: The path- ologies of control in federal emergency management. American Review of Public Administration 26:263–85.

Waugh, William L., Jr. 2000. Living with hazards, dealing with disasters. Armonk, NY: M.E. Sharpe.

Waugh, William L., Jr., and Ronald John Hy. 1991. The utility of all-hazards programs. In Handbook of emergency management, ed. William L. Waugh Jr. and Ronald John Hy. Westport, CT: Greenwood Press.

Waugh, William L., Jr., and Richard T. Sylves. 2002. Organizing the war on terrorism. Public Administra- tion Review 62 (special issue): 81–89.

Wikipedia.en.wikipedia.org/wiki/List_of_United_States _Governors. Accessed June 16, 2006.

Witt, J. L., and J. Morgan. 2002. Stronger in the broken places: Nine lessons for turning crisis into triumph. New York: Times Books.