Sage Journals: Discover world-class research

Abstract

Previous research shows that uncontrollable events like natural disasters hinder incumbent leaders’ chances at re-election, but also suggests that competent responses to such crises can benefit incumbents. Replicating and extending the work of other researchers, I show that disasters are informative events in retrospective voting where leaders have the opportunity to demonstrate competence and be rewarded by voters while incapable leaders face punishment for their failure to respond to such a crisis. Governors see greater electoral rewards for demonstrating competence when disaster damage is high as well as when the political situation is more difficult. In the context of a leader’s decision-making process, this reveals a complex and conditional relationship between voters and the leaders they are evaluating.

Keywords

Natural disasters retrospective voting gubernatorial elections

Introduction

Natural disasters often pose significant challenges to those in office. Unlike other challenges which can be managed over time through policy, natural disasters are exogenous and represent immediate events that can significantly disrupt the lives of citizens. Elected officials often have disaster relief options with which they can respond to these crises. The natural disaster event and the government’s response both factor in to a unique form of retrospective voting where citizens are able to judge the performance of elected officials in response to unforeseen circumstances.

I argue that voters are using natural disasters in a sophisticated and rational manner. The government’s response and the political context are key factors shaping how voters respond to severe weather events. Voters view natural disasters as informative events that reveal information about the competence of their elected officials. In this process, an exogenous event such as a weather disaster creates an issue that voters want resolved in the form of disaster relief. Elected officials who have the power to lobby for aid are judged based on their competence in overcoming political challenges to meet voters’ demands.

In replicating and expanding upon the work of Gasper and Reeves (2011), I show that the impact of storm severity is conditional on the government’s response. While most previous literature views high levels of storm damage as an impediment to an incumbent’s reelection effort, I find that increases in the severity of natural disasters actually benefit incumbent governors if they have received a disaster declaration from the president. Additionally, governors see increased vote shares when they secure aid from presidents who are not co-partisans. Thus, governors maintain agency in the process and have incentives to demonstrate competence in response to the most extreme disasters.

Natural disasters and retrospection

The current study builds upon previous studies concerning natural disasters and the ability of voters to reward or punish incumbent politicians in a rational way. For situations where the government is assumed to have a large amount of control, such as the economy, retrospective voting is rather straightforward (Fiorina, 1981; Key, 1966; Kramer, 1971; Tufte, 1978). Retrospection is trickier when the government has less control over exogenous events, such as with natural disasters.

One of the central debates in the literature concerning retrospective voting and natural disasters has been the rationality of voters (Achen and Bartels, 2004; Ashworth, 2012; Healy et al., 2010; Healy and Malhotra, 2009, 2013; Oliver and Reeves, 2015). Are voters behaving rationally when developing their retrospective evaluations? Recent studies find voters are influenced by events such as shark attacks and college sports (Achen and Bartels, 2004; Healy et al., 2010). According to this side of the debate, if voters are punishing sitting politicians for “acts of God,” then it calls into question the extent to which voters are rational when they engage in retrospective voting. Other studies point to more rational reactions from voters, finding that voters are sophisticated and rational in their response to such events (Abney and Hill, 1966; Arceneaux and Stein, 2006; Bechtel and Hainmueller, 2011; Chen, 2013; Gasper and Reeves, 2011; Gomez and Wilson, 2008; Lay, 2009; Maestas et al., 2008; Velez and Martin, 2013). According to this view, voters demonstrate rationality by reacting to how the government responds and acting as a “rational god of vengeance and reward” (Key, 1964: 568). Thus, when considering natural disasters and retrospection, both the initial event and the government’s response are key factors.

Gasper and Reeves (2011) evaluate how voters attribute blame when governors ask for a disaster declaration and are denied by the president. Their findings reveal a sophisticated electorate, which rewards governors for both attempts at obtaining, and the receipt of, disaster declarations, while also punishing presidents for denials of federal aid. Gasper and Reeves find disasters to have a negative impact on incumbent governor’s vote shares, but these effects are outweighed by the positive effects of obtaining a disaster declaration.

Related work by Healy and Malhotra (2010) addresses the rationality of voters in their response to tornado damage. More tornado damage negatively affects incumbent presidential vote share, but this negative impact is lessened when presidents issue disaster declarations. If a declaration is made, tornado damage has a positive, but not statistically significant, effect on vote share. Overall their findings indicate that voters are assessing the ways presidents react to disasters rather than only focusing on the disasters themselves. While it may appear that tornados pose exogenous and unfair challenges to leaders, voter behavior is more concerned with how those in government respond. These results, taken together, illustrate that leaders are able to maintain agency and that voter behavior might be regarded as more rational than previously thought.

These two studies make important points about the effects of disaster severity and government response. Generally, more damage results in lower vote shares for the incumbent, suggesting that voters are punishing incumbents for storm damage. The second point is that incumbents can mitigate these negative effects with disaster declarations that deliver aid to those affected. Healy and Malhotra find that there is a positive, albeit statistically insignificant, effect for damage when a disaster is declared, while Gasper and Reeves show that “[w]hile severe weather damage is detrimental to incumbent election hopes, presidential disaster declarations always more than make up the cost” (p. 351). However, one conclusion that can be drawn from Gasper and Reeves is that the positive effect of a declaration would diminish as damage increases. The positive effect of receiving a declaration would steadily erode away due to the increasingly negative effects from more severe disasters. Thus, governors should prefer a less damaging event when they are granted a declaration rather than a very severe event.

This study builds upon previous literature to show that voters consider a range of factors when confronted with natural disasters. The amount of damage caused by a disaster is important, but an incumbent’s response and the political context also provides valuable information to voters (Bechtel and Mannino, 2016; Remmer, 2014). More specifically, voters want to know if a governor can obtain a disaster declaration, which will bring material relief to the situation. Since the disaster declaration process requires governors to ask for a declaration and a president to grant it, voters also evaluate a governor’s political abilities. If the two actors are co-partisans, voters would expect the two to be able to work together, making the ability to cooperate a weak signal for competence. If, however, the two are from different parties, voters are able to learn more about the ability of a governor to persuade and lobby the president. Natural disaster severity, the government’s response, and partisanship influence how voters respond at the ballot box to exogenous weather events.

Hypotheses

Building on previous work by Healy and Malhotra (2010) and Gasper and Reeves (2011), I argue that natural disaster damage does not always work against incumbents seeking reelection. The disaster event and a governor’s response are closely linked, and demonstrating competence by obtaining a disaster declaration is likely to result in greater electoral rewards as the amount of damage increases.

Response hypothesis: If a governor receives a disaster declaration, her vote share will increase as disaster damage increases, but if no declaration is received, her vote share will decrease as damage increases.

Rather than always being to a governor’s detriment, the effect of damage severity is conditional. When no disaster declaration is made, increasingly severe storms further decrease an incumbent governor’s vote share. When a declaration is made, the increased severity actually benefits the governor, resulting in increases in vote share as the amount of damage rises.

Political challenge hypothesis: When a disaster declaration is obtained, voters will reward governors who are not co-partisans of the president more than governors who are.

When governors and presidents are not from the same party, I expect the coefficient associated with the interaction term to be larger than it is for co-partisan pairs. When the governor and the president are co-partisans, voters appreciate the disaster declaration; however, obtaining a declaration is not as informative about a governor’s competence level as obtaining a declaration from a president of another party. In such circumstances it is not clear that the receipt of a disaster declaration is an act of “political skill” by the governor (Abney and Hill, 1966). On the other hand, when the governor and president are not from the same party, it presents an opportunity for the governor to demonstrate a greater amount of competence. Voters prefer assurances that a governor can deliver disaster aid even if the governor and president are not co-partisans. Successfully navigating a more challenging political landscape reveals a higher level of political competence, which should be rewarded by the voters.

Data and methods

In order to assess the preceding hypotheses, I use data from Gasper and Reeves (2011), which cover elections from 1970 to 2006. The dataset only includes incumbent gubernatorial elections and weather events that occurred in the six months prior to the election. With the exception of disaster declarations, no other variables from the Gasper and Reeves dataset are transformed or changed from how they appear in their replication dataset.¹

The incumbent governor’s share of the two-party vote at the county level serves as the dependent variable. Natural disaster damage, a key independent variable, is measured in logged dollars per county resident and serves as a measure of the destructive impact of the natural disaster. Damage is measured at the county level, meaning that damage is assigned to the whole county even if only a small geographic portion of the county was actually affected.

Another important explanatory variable is whether or not a disaster declaration was made. For the purposes of this study, it is a dichotomous variable.² Either a governor was successful in obtaining one from the president, or no disaster declaration was made. If the president denied a request for a declaration, it is treated the same as if no declaration was made.³ Governors have practical reasons to seek federal aid and have the added incentive of wishing to appear competent in the minds of voters by bringing federal relief dollars back to their state. Failing to do so would likely be seen as a sign of incompetence by voters.

The model uses an interaction term to evaluate the Response Hypothesis. Damage is interacted with a dichotomous indicator for a disaster declaration with the expectation that the coefficient will be positive. If the coefficient is larger in magnitude than the negative coefficient associated with weather damage, it would indicate an increasing vote share as severity increases and provide support for the Response Hypothesis.

One more factor to consider is the partisanship of the governor and president. The disaster declaration process requires that governors make a request that presidents can either accept or deny. The partisanship of the two actors is a way to evaluate the Political Challenge Hypothesis—the relative difficulty of the political situation that a governor is trying to navigate. Partisanship data come from the Klarner’s (2013) dataset on governors and the Leip Election Atlas (Leip, n.d.).

The standard control variables used by Gasper and Reeves (2011) are also included. All models control for previous gubernatorial vote, presidential vote, county median income, and include both county and year fixed effects.

Model results

The first column in Table 1 replicates the results of Gasper and Reeves (2011). All coefficients and standard errors replicate nearly exactly the results from the original study. Gasper and Reeves use a count of disaster declarations rather than the binary measure used in this extension. Additionally, Gasper and Reeves include a variable that captures instances when the president rejects a governor’s request for a disaster declaration.

Table 1.

Effects of weather damage on gubernatorial vote share conditional on disaster declaration.

Variable	Gasper and Reeves model	Extended model
Intercept	11.211^*	11.568^*
	(5.006)	(5.047)
Logged weather damage	−0.132^***	−0.143^***
	(0.021)	(0.022)
Disaster declaration (binary)		0.340
		(0.602)
Logged weather damage * disaster declaration (binary)		0.290^***
		(0.059)
Lagged president vote	0.168^***	0.174^***
	(0.008)	(0.008)
Lagged governor vote	0.637^***	0.626^***
	(0.010)	(0.010)
Median county income	−0.075^*	−0.074^*
	(0.030)	(0.030)
Disaster declarations (total)	3.988^***
	(0.274)
Turndowns	2.650^***
	(0.244)
R ²	0.527	0.519
Adjusted R²	0.414	0.404
Numerical observation	15580	15580

Standard error are in parentheses. Fixed effects for county and year are omitted from the table.

***

p < 0.001, ^**p < 0.01, ^*p < 0.05.

The Extended Model reported in the second column of Table 1 provides support for the Response Hypothesis in that the effects of weather damage and disaster declarations are both substantively important and conditional.⁴ Weather damage is statistically significant and has a negative sign, as expected. If no disaster has been declared, a governor can therefore expect to be punished in the upcoming election. Additionally, the results reported in the second column of Table 1 demonstrate the benefits of receiving a disaster declaration.⁵ Governors can expect to be rewarded with just over one-third of a percentage point in the unlikely event that there was no damage, but a disaster is declared.

When a disaster declaration is in effect and there is storm damage, the coefficient associated with the interaction term is added to the coefficient associated with the weather damage term. According to Table 1, this effect will be positive. When there is not a damage declaration, the effect of weather damage on county vote share is approximately −0.14 percentage points. If a declaration is made, the effect is approximately 0.15.⁶ These calculations indicate that governors will see a statistically significant increase in vote share as weather damage increases, thereby giving support to the Response Hypothesis.⁷

The vote share to be gained in this situation, though, depends on the amount of storm damage. Figure 1 shows how the increase in vote share rises as the amount of storm damage rises, given a declaration. The effect ranges from less than one percentage point to a maximum of over five percentage points in the most extremely damaging events. A governor who successfully lobbies for a disaster declaration can avoid the negative ramifications of storm damage and actually benefit electorally. While previous literature shows a factor like weather damage to have a negative effect (e.g., Achen and Bartels, 2004; Cole et al., 2012; Gasper and Reeves, 2011), my model shows this effect to be conditional upon the actions of the incumbent.

Figure 1.

Marginal Effects of declaration and no declaration.

If damage occurs, it is best for a governor to obtain a disaster declaration. If a disaster is declared, the reward in vote share will vary according to the amount of damage. This reward is even greater when considering how failing to receive a declaration results in governors losing vote share in relation to the amount of damage.

How does the political situation factor into the information-gathering process? In order to evaluate the Political Challenge Hypothesis, the models in Table 2 take the partisan affiliation of the governor and president into consideration. The data are split between situations where the governor and president are from the same party versus when they are from different parties.⁸ When governors and presidents are co-partisans, the coefficient associated with the interaction term between disaster declarations and weather damage is positive but not statistically significant. However, the second model shows that when the pair are from different parties, the coefficient associated with the interaction term remains positive and is statistically significant at conventional levels.

Table 2.

Effects of weather damage on gubernatorial vote share conditional on disaster declaration, by partisanship of governor and president.

Variable	Same party	Different party
Intercept	4.404	9.957
	(5.772)	(7.193)
Logged weather damage	−0.033	−0.108^***
	(0.033)	(0.031)
Disaster declaration	−0.162	3.418^***
	(0.855)	(0.893)
Logged weather damage * disaster declaration	0.065	0.466^***
	(0.082)	(0.088)
Lagged governor vote	0.508^***	0.513^***
	(0.019)	(0.015)
Lagged presidential vote	0.298^***	0.355^***
	(0.018)	(0.015)
Median county income	−0.080	0.125^**
	(0.044)	(0.043)
R ²	0.724	0.664
Adjusted R²	0.557	0.492
Numerical observation	7084	8496

Standard errors are in parentheses. Fixed effects for county and year are omitted from table.

***

p < 0.001, ^**p < 0.01, ^*p < 0.05.

Overall, Table 2 provides support for the Political Challenge Hypothesis. Since co-partisans are expected to be able to more easily obtain disaster declarations, voters are less able to evaluate gubernatorial competence. The best way for governors to demonstrate competence to voters is to obtain a declaration from a president of the opposite party. The large difference in effects depending on the partisan affiliation of the actors is notable, considering that federal aid is being distributed in both situations. Voters are reacting to the demonstration of competence rather than just the receipt of aid.

Analysis of these data fit with a larger story about information and competence. Voters reward governors for demonstrating competence when the incumbent governor and the president are co-partisans. The effect is even larger when governors and presidents are not co-partisans. That is, when the act of obtaining a disaster declaration is more politically difficult, voters can learn more about a governor’s ability to deliver desired relief. The conditional nature of “voter gratitude,” a term used by Bechtel and Hainmueller (2011), is a clear indication that voters are reacting to more than just receiving money from the government. When voters think it takes more effort to secure aid and the governor still pulls through, voters are given a signal that their governor is competent.

The results show that voters’ reactions are dependent on the actions of the government. If there is inaction, voters may conclude that their governor is not capable of solving their problems. When a disaster is handled satisfactorily, voters reward their sitting governor. The effect is larger when the political situation is more challenging—specifically when a governor and the president are not from the same party. Natural disasters can be informative events, allowing leaders to demonstrate their abilities in especially clear terms.

Conclusion

Although previous studies point to the electoral benefits of disaster declarations for incumbents, most research also holds that damage itself has a negative impact on incumbents. All else equal, a disaster declaration for a severe storm would be worth less in terms of vote share than a disaster declaration for a rather minor disaster. However, the data show that the effects of damage are conditional on obtaining federal aid. When governors demonstrate competence by obtaining disaster declarations, storm severity actually works to the governor’s benefit electorally. Rather than a declaration mitigating the effects of a disaster, with more severe damage eroding the positive effect the most, increased severity works to an incumbent governor’s benefit. The effect is especially pronounced when a governor successfully reaches out to a president from another party in order to obtain a declaration. Overall, it emphasizes the role governors, political skill, and competence play in determining the electoral effects of a crisis rather than the size of the disaster alone.

There are several possible causal mechanisms that might explain why large amounts of damage in a declared disaster would increase incumbent vote share. One possible explanation is the role of the media. More severe storms are likely to bring more media attention and greater awareness from voters. The increase in publicity can work for or against an incumbent governor given her response to the disaster. With the attention a large disaster would bring, demonstrating competence would reach a wider audience in the electorate and actually result in increases in approval and vote share.

These findings point to interesting normative implications. If it is certain that a disaster declaration is forthcoming, the data suggest that governors would benefit most if the natural disaster is more destructive. Whereas previous studies point to low damage events as optimal, these findings suggest that governors would maximize their vote share by having a declaration and a very destructive event. The key piece, though, is making sure that a disaster is declared. The large difference in benefit and punishment based on getting a declaration should incentivize governors to work harder for a declaration, especially when severe natural disasters occur.

Supplemental Material

Appendix_A_and_B_-_Weathering_the_Storm – Supplemental material for Weathering the Storm: Conditional Effects of Natural Disasters on Retrospective Voting in Gubernatorial Elections—A Replication and Extension

Supplemental material, Appendix_A_and_B_-_Weathering_the_Storm for Weathering the Storm: Conditional Effects of Natural Disasters on Retrospective Voting in Gubernatorial Elections—A Replication and Extension by Kevin R. Stout in Research & Politics

Footnotes

Declaration of conflicting interest

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Supplemental materials

The supplemental files are available at

The replication files are available at

Notes

Carnegie Corporation of New York Grant

This publication was made possible (in part) by a grant from the Carnegie Corporation of New York. The statements made and views expressed are solely the responsibility of the author.

References

Abney

Hill

(1966) Natural disasters as a political variable: The effect of a hurricane on an urban election. The American Political Science Review 4: 974–981.

Achen

Bartels

(2004) Blind retrospection: Electoral responses to drought, flu, and shark attacks. Working paper. Princeton University. Available at: https://www.ethz.ch/content/dam/ethz/special-interest/gess/cis/international-relations-dam/Teaching/pwgrundlagenopenaccess/Weitere/AchenBartels.pdf

Arceneaux

Stein

(2006) Who is held responsible when disaster strikes? The attribution of responsibility for a natural disaster in an urban election. Journal of Urban Affairs 28(1): 43–53.

Ashworth

(2012) Electoral accountability: Recent theoretical and empirical work. Annual Review of Political Science 15: 183–201.

Bechtel

Hainmueller

(2011) How lasting is voter gratitude? An analysis of the short- and long-term electoral returns to beneficial policy. American Journal of Political Science 55(4): 852–868.

Bechtel

Mannino

(2016) Distributive politics, fairness, and the allocation of disaster relief. SSRN. Available at: https://ssrn.com/abstract=2943046 (accessed at 15 November 2017).

Brambor

Clark

Golder

(2006) Understanding interaction models: Improving empirical analyses. Political Analysis 14: 63–82.

Chen

(2013) Voter partisanship and the effect of distributive spending on political participation. American Journal of Political Science 57(1): 200–217.

Cole

Andrew

Werker

(2012) Do voters demand responsive governments? Evidence from Indian disaster relief. Journal of Development Economics 97(2): 167–181.

10.

Fiorina

(1981) Retrospective Voting in American National Elections. New Haven, CT: Yale University Press.

11.

Gasper

Reeves

(2011) Make it rain? Retrospection and the attentive electorate in the context of natural disasters. American Journal of Political Science 55(4): 340–355.

12.

Gomez

Wilson

(2008) Political sophistication and attributions of blame in the wake of Hurricane Katrina. Publius: The Journal of Federalism 38(4): 633–650.

13.

Healy

Malhotra

(2009) Myopic voters and natural disaster policy. American Political Science Review 103: 387–406.

14.

Healy

Malhotra

(2010) Random events, economic losses, and retrospective voting: Implications for democratic competence. International Quarterly Journal of Political Science 5(2): 193–208.

15.

Healy

Malhotra

(2013) Retrospective voting reconsidered. Annual Review of Political Science 16: 285–306.

16.

Healy

Malhotra

(2010) Irrelevant events affect voters’ evaluations of government performance. Proceedings of the National Academy of Sciences 107(29): 12804–12809.

17.

Key

(1964) Politics, Parties, and Pressure Groups (5th ed.). New York: Thomas Y. Crowell.

18.

Key

(1966) The Responsible Electorate: Rationality in Presidential Voting, 1936–1960. Cambridge, MA: Belknap Press of Harvard University Press.

19.

Klarner

(2013) Governors Dataset. IQSS Dataverse Network, Public Version. Available at: http://hdl.handle.net/1902.1/20408 (accessed at 9 September 2017).

20.

Kramer

(1971) Short-term fluctuations in U.S. voting behavior. American Political Science Review 77(1): 92–111.

21.

Lay

(2009) Race, retrospective voting, and disasters the re-election of C. Ray Nagin after Hurricane Katrina. Urban Affairs Review 44(5): 645–662.

22.

Leip

(n.d.) Dave Leip’s Atlas of U.S. Presidential Elections. Available at: http://uselectionatlas.org (accessed 9 October 2014).

23.

Maestas

Atkeson

Croom

, et al. (2008) Shifting the blame: Federalism, media, and public assignment of blame following Hurricane Katrina. Publius: The Journal of Federalism 38(4): 609–632.

24.

Oliver

Reeves

(2015) The Politics of Disaster Relief. Emerging Trends in the Social and Behavioral Sciences: An Interdisciplinary, Searchable, and Linkable Resource. 1–8. Available at: http://onlinelibrary.wiley.com/doi/10.1002/9781118900772.etrds0258/abstract (accessed 18 December 2017).

25.

Remmer

(2014) Exogenous shocks and democratic accountability: Evidence from the Caribbean. Comparative Political Studies 47(8): 1158–1185.

26.

Tufte

(1978) Political Control of the Economy. Princeton University Press.

27.

Velez

Martin

(2013) Sandy the rainmaker: The electoral impact of a super storm. PS: Political Science & Politics 46(2): 313–323.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.36 MB