A central contention in political science is that states strongly desire international status and prestige (Dafoe et al. 2014). Many scholars view the international system as a hierarchy of states seeking loftier positions, and show that this goal frequently leads to inter-state conflict (Renshon 2016). States seek these attributes for their own sake, because they can provide power and influence, and because of the material benefits associated with them, such as FDI or trade concessions (Tomz 2012). But while significant attention has been given to the importance of status in international relations, we know far less about how status is determined and how states attempt to change their status. The majority of work in this area identifies war as the primary means by which states do so, but as Renshon (2016, p. 545) concludes, “it seems likely that states might use multiple strategies to alter the beliefs of their status community as to where they ‘stand.”’Footnote 1 Understanding the ways in which states seek to shift other states’ perceptions of their rank and power is thus crucial for moving forward the debate on how status considerations shape state behavior short of war. Comprehension of these dynamics is particularly important in the current era in which states interact more than ever in political, economic, and social domains. What strategies do these states employ to influence other states’ perceptions of their strengths and capabilities?

We focus on a particular signal through which we argue that states convey increased status: self-sufficiency. While states provide assistance to other states for many reasons—including desires to stimulate economic growth and development—we show that such aid can also have the perverse effect of demonstrating the recipient state’s weakness. While donor status confers “superiority and power” on the donor, recipient status signals “inferiority and powerlessness” (Kuusik 2006, 57). This offers an explanation for why states often reject such assistance: leaders wishing to improve their international prestige reject aid to indicate their capacity for self-sufficiency. Our findings suggest that, because highly effective governments often have little need for foreign aid, many less competent governments attempt to fool international actors into believing that they also do not require aid. Our argument comports with previous research showing that governments seek out ways to send costly signals to international audiences, often to avoid becoming stigmatized (Hyde 2011; Adler-Nissen 2014). Since it is difficult to observe a government’s level of development and quality of governance, observers must rely on imperfect indicators (Heffetz and Frank 2008; Pinker et al. 2008). We claim that aid constitutes one such indicator; thus, incompetent governments that need aid may turn it down to reap the benefits associated with a better international reputation.

For example, in 2011 India’s government attempted to terminate its primary source of foreign aid from Britain, as foreign minister Nirumpana Rao worried about the “negative publicity of Indian poverty promoted by [the aid]” (Gilligan 2012).Footnote 2 Similarly, Pakistan’s Punjab cabinet voted in favor of declining foreign aid in 2011 (Pakistan Today 2011), Kenya’s President Uhuru Kenyatta encouraged African nations to shun foreign aid in 2015 (BBC News 2015), and Botswana stopped accepting much of the aid it was offered in the 1970s (Lancaster 1999, 244). For these countries, rejecting aid was a price worth paying to signal self-sufficiency.

Yet states clearly do not always decline aid offers; we argue that they do so only when international observers would find it plausible that they do not require it, and when they value status highly.Footnote 3 For instance, a government that possesses considerable resources may seem to govern quite effectively.Footnote 4 In such a case, international actors experience difficulty discerning whether the government is indeed self-sufficient, and rely more heavily on the rejection of aid as a signal. By contrast, a government possessing few resources and experiencing high levels of corruption and mismanagement is not likely to fool observers by refusing aid, and thus should not find it worthwhile to do so.

By formalizing and testing this argument, this article contributes both to the scholarship on status, as well as the literature on aid. The quality of governance in recipient countries is well-known to condition how effectively aid can be used to support development (Bräutigam and Knack 2004). We illustrate that political motivations influence even the decision to accept assistance. Our theory is related to, but distinct from, the literature on the “fiscal contract,” which grounds the notion of state legitimacy in the relationship between tax revenue and public services (Bates and Lien 1985; Levi 1998). Scholars operating in this framework express concern that if donors rather than taxpayers visibly fund public services, citizens may lose confidence in their governments (Brass 2016; Ahmed 2012; Sacks 2012).Footnote 5 Furthermore, citizens may hold governments more accountable when governments derive their budgets from tax revenues rather than aid (Paler 2013), though see De La Cuesta et al. (2017) for an alternative argument. Aid-funded projects could also entrench individuals’ beliefs that government-funded projects are highly corrupt by comparison (Milner et al. 2013), though opinions of aid-funded projects may depend on whether country systems are trusted or bypassed (Knack 2013).

Rather than asking whether aid harms a government’s domestic reputation, however, we ask what effect aid has on a government’s international status. In particular, we examine states’ attempts to change or defend their international images in order to provide a better understanding of the strategic behavior surrounding the receipt of foreign aid, along with aid’s broader consequences for governments and their citizens. We do so by developing a formal model and testing its predictions in the context of humanitarian responses to natural disasters. After offering several real-world examples, we present experimental evidence that rejecting aid after a natural disaster can improve a country’s international status. We then use new data to analyze cross-national variation in responses to aid offers, finding that states that value status highly—and are capable of improving their status—are more likely to reject needed aid. We also find that nearly half of the aid offers in our sample are rejected, underscoring the importance of understanding this puzzling behavior. We conclude by discussing the broader implications of our argument and possible extensions of our findings.

1 Signaling and reputation

We ask how states can achieve greater status within the international community short of going to war, arguing that rejecting assistance from other states represents a particularly powerful method of doing so. We develop our theory formally through a signaling model in order to derive empirical implications—which we test subsequently—and to focus attention on the key logics and causal mechanisms that drive this process. Here, we describe the game as a heuristic device and provide the full model in the Appendix of Supplementary Information (SI) due to space constraints.

The model features two primary actors: the incumbent government and the international community. The international community consists of leaders, investors, international institutions, non-governmental organizations, the mass public, and other actors. Our definition of the international community is intentionally expansive because its primary role in our model is to award status, and status is fundamentally a consensus concept. Renshon (2016, 519) writes, “Status is not one actor’s beliefs about one other actor. Rather, status describes many actors’ beliefs about what many other actors also believe.” Particularly in a globalized world, a country cannot improve its status only by earning other heads of states’ approval; its status will improve when these opinions are echoed in investment circles, the media, activist communities, and other networks. We leave it to others to unpack how this aggregation of beliefs occurs; for the purposes of this paper, we remain consistent with extant work in preserving a general international audience consisting of both elite and mass actors.

At the beginning of the game, a shock occurs, such as a natural disaster or an economic crisis. The incumbent chooses whether to reject or accept international assistance for help with the recovery. Incumbents come in two types: “competent” or “incompetent.” Competent governments have the resources to respond adequately to the shock and do not require foreign assistance. Incompetent governments do not possess these resources, and cannot rebuild their country without foreign aid.

However, the incumbent’s type is frequently unknown to the international community. A large body of scholarship illustrates the prevalence of this condition in international politics; states adopt behaviors ranging from inviting election monitors to repaying debts in order to persuade international observers of their commitments to democracy and cooperation (Gray 2013; Hyde 2011; Tomz 2012). This uncertainty is particularly pronounced when data quality is poor or where political and economic conditions change rapidly, making it difficult for observers to assess a country’s situation (Jerven 2013). Further, these conditions are most likely to prevail precisely when states are most likely to seek status and take advantage of changes in the global distribution of power. Indeed, the literature characterizing war as a status-seeking strategy argues that conflict erupts most often in ambiguous circumstances, when the status a state receives is inconsistent with the status it believes it deserves (Wohlforth 2009). In other words, status-motivated behavior—military or otherwise—tends to occur in the presence of such uncertainty, so we focus on these conditions here.

Incompetent governments can thus decline foreign assistance in order to try to appear competent, though doing so comes at a price. Specifically, if the incumbent refuses assistance, she forgoes the aid and must expend effort concealing her inadequate response to the shock. This deception could include activities such as issuing false reports of the relief efforts’ success, blocking affected citizens’ abilities to communicate or organize, or making cosmetic changes to provide the illusion of recovery. Finally, if either type of state accepts aid, it pays a small cost due to the bureaucratic strain of absorbing aid and the sovereignty issues associated with accepting assistance.

The international community prefers to accord high status to competent governments and low status to incompetent governments. If they mistakenly accorded high status to an incompetent government, that government may receive undue recognition and influence at the expense of deserving countries who would perform more effectively.Footnote 6 Therefore, states possess strong incentives to try to demonstrate competence to these international actors, as being held in greater esteem provides its own reward, as well as power, investment, and other benefits. However, incompetent states only do so when the expected benefits of obtaining greater status outweigh the costs of declining aid. Aid rejection not only carries the opportunity cost of losing the foregone aid but also introduces the risk that incompetent states will be caught in their attempts to seem competent. Incompetent states with fewer resources, along with those that care more about providing a speedy recovery, face steeper costs both to losing the assistance and to attempting to fool observers. After the shock occurs, the recipient decides whether to accept the aid and the international community updates its beliefs about the incumbent’s type.

In equilibrium, the competent type of government always refuses assistance because, though rejecting aid entails turning down free money, accepting aid brings more risks than rewards for this type of country. In particular, the marginal benefit of aid in helping such a state respond to disasters is small and it can be difficult to coordinate the international effort.Footnote 7 Incompetent types, on the other hand, are incentivized to mimic competent types by rejecting aid, but they do so only under certain conditions. First, the baseline probability that the government is competent must be high. Governments with more resources, such as middle-income governments, are more likely to be seen as competent from the outset, and thus are better able to fool the international community, whereas low-income governments are seen as too needy to trick observers. Further, the cost of pretending to be competent cannot be too great and the state cannot desire the aid too much, or the potential recipient simply cannot afford the charade. States that fall into this category again tend to possess fewer resources. Finally, states must value status highly enough to reject aid. We therefore expect the following:

Hypothesis 1

States that reject aid tend to experience a boost to their international reputations, unless they are caught covering up their incompetence.

Hypothesis 2

The more resources states have, the more likely they are to reject aid, even when they require the aid.

Hypothesis 3

The more states value status, the more likely they are to reject aid, even when they require the aid.

The model thus shows that aid rejection provides an avenue short of war through which states can signal competence and self-sufficiency to the international community, and thus the signal may be imitated by some states that actually need aid. Having derived testable predictions from the model, we now move to examine whether they receive empirical support.

2 Empirical analysis

To test the predictions of our model, we focus on a particular setting: natural disaster relief. Though our theory applies to assistance more broadly, we analyze disaster aid for several reasons. First, aid rejection following disasters represents particularly puzzling behavior due to their scale and frequency. Indeed, natural disasters often cause thousands of deaths, require billions of dollars for reconstruction, and dwarf individual countries’ capacities to respond. The international community thus typically offers considerable assistance to help with rebuilding efforts. In addition to rebuilding, aid is fungible and can be used for incumbency-protection or to pursue a leader’s goals more broadly (De Mesquita and Smith 2009, 2013; Bräutigam and Knack 2004; Knack 2001, 2004; Morrison 2014; Djankov et al. 2008). Yet despite the extreme need faced by victims of these disasters and the generosity of the assistance offered, governments of countries hit by these events often reject international assistance. As demonstrated in Fig. 1, out of the 66 disasters with over 300 deaths occurring between 2004 and 2012, governments rejected aid in 32 instances, which represents about half of the cases. Our study helps to answer a striking puzzle: Why would governments deny themselves the assistance they so urgently need?

Fig. 1
figure 1

Aid responses in 66 natural disasters, 2004–2012. Note: GDP data is missing for North Korea ‘07 (acceptance), Myanmar ‘08 (unsuccessful rejection), and Taiwan ’09 (unsuccessful rejection), so we are unable to quantitatively calculate the ability to respond ratio. Our qualitative research shows that all three states had low abilities to respond

This is particularly puzzling because disaster aid has few costs associated with its acceptance, unlike foreign aid given in normal times including conventional economic aid, social services aid, or aid for democracy support. In normal times, conditions are often attached to the aid, and if those conditions are enforced, they carry potential policy costs. For example, donors may require recipients to improve their human rights and democracy,Footnote 8 alter economic policies, or implement a variety of other items of importance to the donor (Crawford et al. 2001). Relatedly, donors may only be willing to fund projects that are not preferred by recipients. Such projects can carry political costs, particularly if they provide resources to the opposition (Carothers 2006), are unpopular with the government’s base of support, or use non-replaceable resources. Moreover, donors may require recipient governments to cover local costs, which are especially undesirable when the government would not undertake the project on its own (Winters and Streitfeld 2011).

A second reason to concentrate our analysis on this area is that it is useful for empirical reasons. In particular, while examining the universe of natural disasters does not eliminate problems of endogeneity, it mitigates them because the timing of these events is essentially random.Footnote 9 Thus, the precise moment that a disaster hits is not determined by strategic considerations related to reputation or aid provision. The suddenness of the disaster and the urgent requirements that follow also mean that disaster aid is rarely conditioned on a country’s past performance and comes with relatively few political strings attached to it.Footnote 10

Third, the receipt of aid following natural disasters is typically well-publicized, as severe disasters receive extensive international coverage due to their high salience. Indeed, our case studies demonstrate that international actors pay close attention to this form of aid. Further, because disasters fall out of the news cycle quickly, after which international interest wanes, short-term responses to disasters influence a state’s international status disproportionately. Immediate displays of self-sufficiency thus matter more than a steady long-term recovery plan in earning international approval, offering a government an attractive opportunity to earn improvements in status by changing its short-term behavior (Djerf-Pierre 2013; Eisensee and Strömberg 2007).

Fourth, the political economy of disaster relief represents an interesting and important topic in its own right. A burgeoning literature has shown that leaders prioritize relief for politically valuable constituencies and in areas where their actions are most visible (Healy and Malhotra 2009; Besley and Burgess 2002; Cole et al. 2012). Similarly, disasters that appear in the news are more likely to receive relief (Eisensee and Strömberg 2007). Like politicians, rebel groups can also attract supporters by delivering relief to civilians; thus, governments threatened by insurgencies should be especially inclined to provide sufficient relief themselves (Beardsley and McQuinn 2009). Scholars of international relations have also recognized the importance of natural disasters, but these works have typically focused on their diplomatic implications for conflict and cooperation between affected states and those offering assistance (Kelman 2011; Akcinaroglu et al. 2011). In contrast, our project explores the implications for an affected country’s international status based on its decision to categorically accept or decline international assistance. While the disaster diplomacy literature explains why states may reject assistance from enemies, scant attention has been paid to the categorical rejection of assistance that we observe and its implications for a country’s status.Footnote 11

We test our theoretical predictions in the natural disaster setting in multiple ways. We start by illustrating our argument through examples of two countries that experienced similar exigencies but responded differently to offers of assistance, which clarify our argument and make our theory more concrete. We then turn to systematic tests of our predictions, using both a survey experiment and cross-national analysis. The survey experiment tests our hypothesis that rejecting aid can improve a country’s international status. We then use our original data set of natural disaster responses to examine whether governments anticipate these status gains and seek them strategically when they are capable of convincing international audiences and when they desire status (hypotheses 2 and 3 respectively), finding support for our predictions.

2.1 Examples

To illustrate the model’s implications, consider the difference in responses to natural disasters by two governments in the same region that have experienced similar exigencies: Sri Lanka and India. Sri Lanka, which possesses far fewer resources than India, has responded to disasters by accepting aid, while India has declined aid to signal its global power status. While Sri Lanka and India differ in many ways and thus the variation in their responses cannot be attributed solely to the factors we identify, we present these examples for illustrative purposes and conduct a more rigorous empirical examination subsequently.

2.1.1 Sri Lanka

Sri Lanka tends to accept aid after disasters because it possesses such inadequate resources that it cannot convince the international community of its self-sufficiency by rejecting aid. For example, Sri Lanka lacked sufficient resources to respond to the 2004 earthquake and following tsunamis, which killed 31,000 people, left 4,000 missing, and destroyed 100,000 homes in the country. The government lacked a coordinated plan for addressing such a large disaster, a tsunami warning system, or even basic disaster plans. Instead, the country faced inaccessible roads, confused rescue efforts, and an unprepared health care system, leading to fears of disease outbreaks (Yamada et al. 2006).Footnote 12

Sri Lanka immediately requested aid, and received a total of $651.6 million in assistance, including $25 million in aid from India, itself hit hard by the tsunami (The Guardian 2014; Hall 2004). The government even opened some areas to foreign aid that it had closed off before the disaster struck (Farley 2008).Footnote 13 Reports conclude that Sri Lanka reacted in this manner because “the preexisting civil conflict coupled with the scale of the disaster proved to be too much for the government of Sri Lanka to handle. There simply was not enough capacity on the part of the government to deal with such a massive disaster” (Ching 2011, 6).

2.1.2 India

India constitutes one of the world’s most disaster-prone countries and possesses limited resources compared to the scale of these crises; thus, prior to 2004, it always accepted offers of aid.Footnote 14 Beginning with the 2004 Indian Ocean tsunami, however, the emerging power has rejected disaster aid to send a costly signal to the international community.Footnote 15 Though India claims that it no longer accepts aid because it no longer needs it,Footnote 16 expert analysis indicates that India does in fact require aid, suggesting that India is a pseudo-competent type of country trying to behave as a truly competent type. For example, after the 2004 tsunami, India’s relief efforts were criticized by the disaster relief community as “chaotic” and “pathetic” (Margesson 2005). Five years after the tsunami, the Indian government still had not delivered on many of its aid promises, and many citizens remained without homes (RT News 2010). India failed to respond adequately to subsequent disasters as well (Krishnan 2014) and a recent government audit noted that India’s disaster relief agency is “ineffective in its functioning in most of the core areas” (Bhaila 2013).Footnote 17

Despite later criticism of India’s disaster response effort by experts, the broader international audience appeared to be impressed by India’s behavior, particularly because its inadequacies did not become evident until after the media stopped covering the disaster and the public lost interest. For instance, immediately following the 2004 tsunami, the U.S. included India in a “core group” of states that would supply relief to the disaster’s victims, reflecting the perception that India had “‘graduated’ from its long-standing status as an aid recipient to a donor” (Bhaila 2013; Bidwal 2016). In response, many newspapers published articles with headlines such as, “Post-tsunami India’s Image Rises Globally” (The Times of India 2005).Footnote 18 Indeed, “many international commentators agreed that India’s response to the tsunami highlighted its emergence as a significant power, not only in South Asia but also in the wider global arena” (Harmer et al. 2005, 17).

Policy makers and commentators have connected India’s rejection of disaster relief to its “ambitions to assert itself as a world power” (Sengupta 2005). For instance, “foreign diplomats say they are in little doubt that one of New Delhi’s key aims in the wake of the tsunami has been to project India as an emerging great power that gives out rather than receives humanitarian aid” (Luce 2005). The director of a private Indian research group said, “We are playing politics with aid, using aid to make a statement” (Sengupta 2005) and The Telegraph claimed that India’s aid refusals build “on a quiet but assertive diplomatic aid policy that has coincided with its growing economic clout. It’s a policy that has seen India change from a country that happily accepted foreign aid to tide it over natural disasters just a decade ago to a nation that routinely rejects bilateral assistance to handle such crises” (Kasturi 2013).

Similarly, a U.S. official stated, “Frankly, we feel [aid rejection] is a misplaced desire to demonstrate India’s self-importance in the region” (Rajghatta 2005) and that “the policy is also about driving home a point to countries that have traditionally led the world’s economic system” (Kasturi 2013). A retired Indian diplomat concurred, adding, “There’s a certain sense of self-confidence that we can manage it and a desire to signal that you are capable of managing things on your own” (Sengupta 2005). Indeed, the start of India’s aid refusal coincided with similar efforts, such as increasing military spending and seeking a permanent seat on the United Nations Security Council. Indian President Mukherjee stated that India was “eminently suited” for the position and had “affirmed its willingness and capacity to shoulder the responsibilities” (Indo-Asian News Service 2012). Backed by high economic growth, which has provided it with the ability to send a credible signal, India appears to have thus deployed an aid rejection strategy to bolster its status.

3 Survey experiment

We now investigate our model’s predictions more systematically, beginning with our first hypothesis: rejecting aid improves a government’s international status unless it is caught concealing incompetence.Footnote 19 Because we are interested in how learning that a country refused aid influences international opinions, we use a survey experiment in which we manipulate this feature of a natural disaster response. In particular, we asked subjects about the 2005 earthquake in India; while all subjects read about the earthquake, only those in the treatment group learned that the government rejected international assistance.Footnote 20 This evidence supports our claims that rejecting aid can be an effective strategy, and we leave tests of how that strategy is used for the cross-national analysis.

As described previously, status results from the consensus of many groups in the international community, and one of those is the mass public. Since the U.S. is the sole superpower, the opinions of its citizens are particularly relevant. Many countries have undertaken costly initiatives to improve their images in the eyes of the U.S. public, underscoring the high value placed on influencing this audience’s views. Notable examples include China’s and India’s cultural tours and centers that have been established in the U.S. and elsewhere, through which the countries seek to exercise soft power. Indeed, India has sought to “improv[e] India’s image in American minds” in order to “give legitimacy and credibility to India’s leadership role in the world” and such efforts have “now become an active element of India’s diplomacy” (Blarel 2012, 32).Footnote 21 While we the U.S. mass public is not the only source of international status, it thus constitutes an important one.

Both our treatment and control texts were borrowed from news reports, so individuals were aware that these events actually occurred. All respondents learned about the damage caused by the 2005 earthquake in India. Then, treatment respondents were informed, “India did not ask for outside assistance. The government announced that it did not need international aid to recover.”Footnote 22

This language best approximates how international observers learn of an aid rejection. The majority of the news articles we reviewed covering dozens of instances of aid rejection use this diplomatic phrasing to describe the event. Potential recipients rarely want to offend the countries offering assistance, so they simply announce that assistance is unnecessary. Since offers of assistance are nearly ubiquitous after disasters, this is tantamount to rejecting aid. If anything, we believe the diplomatic language would weaken the signal of aid rejection, making us less likely to find a treatment effect. We compare our treatment group to a control group, which does not learn anything about whether a country rejected or accepted assistance. While it is possible that some individuals in our control group may also believe that India rejected aid, this would bias against finding any results and therefore again presents a harder test of our claim.

Our dependent variable is the government’s reputation or status. The main outcome measure we use is: “Think about how much international status (respect, prestige) you believe India likely has compared with all other countries in the world. Please tell me the number you think best represents India’s status relative to the other countries in the world from 1 (the absolute lowest status country in the world) to 100 (the absolute highest status country in the world).” This question is modeled on the most widespread measure of group status in the psychology literature.Footnote 23 To test the mechanism—that rejecting aid increases confidence in a country’s self-sufficiency—we asked how confident respondents were that India could handle the disaster using its own resources.

We conducted our experiment through Amazon Mechanical Turk (N = 756) on May 4–9, 2016. While not a representative sample, we have no theoretical reason to expect that our participants would assess information about a foreign country’s behavior differently than would average Americans.Footnote 24 Further, MTurk has been shown to replicate studies that rely on nationally representative samples, partly because MTurk participants tend to reply similarly to those of other samples.Footnote 25 Randomization was conducted in Qualtrics through simple random assignment and all of our balance tests, presented in the SI, succeeded. Our manipulation checks indicate that individuals followed and understood the treatment texts: 98% could identify the type of natural disaster that struck India, and 95% of our treatment group knew that India did not accept assistance.

Table 1 presents the effects of our treatment on our main outcomes. We observe a statistically significant treatment effect on our primary status indicator. On average, learning of India’s decision to reject aid in 2005 improved an individual’s assessment of India’s status by just over three points on a 100 point scale (about one-sixth of a standard deviation). For comparison, a prominent psychology study using this measure finds a five point difference between individuals that did and did not identify with the group whose status they rated (Pettit and Lount 2010). In other words, the magnitude of our result is substantively meaningful in the individual psychology literature, with aid rejection producing nearly as much of an effect as being an in-group member.Footnote 26 Countries earn a modest but meaningful status bump when they reject aid, providing the incentive to use this tactic to project images of self-sufficiency.Footnote 27

Table 1 Treatment effects of aid rejection

We also find strong support for the mechanism implied by our theory. The direct effect of our rejection treatment on individuals’ confidence in India’s ability to handle the disaster is about .7 points on a 1–7 scale, and the effect is statistically significant. In the SI, we further model this relationship through mediation analysis, which uses a potential outcomes framework to estimate the effect of rejection on status that is transmitted through confidence in India’s capabilities (Imai et al. 2010). The results suggest that fully two-thirds of our total effect may be accounted for by this mechanism. However, we interpret these results with caution as they require minimal correlation between the background causes of confidence and status.Footnote 28

In sum, the findings from our survey experiment support our claims that rejecting aid signals self-sufficiency to international observers whose opinions contribute to a country’s international status. Our result is substantively meaningful in the psychological literature on status. While we leave it to future research to demonstrate similar effects on the opinions of other populations or elites, this evidence illustrates that rejecting aid is a plausible way for countries to improve their reputations.

4 Cross-country analysis

Our theory also predicts that self-sufficient governments tend to reject assistance following natural disasters because they often have no need for it and wish to avoid the costs associated with accepting aid. Further, since a reputation for self-sufficiency brings a host of benefits, we expect needy governments that place a high value on status to attempt to signal self-sufficiency by similarly rejecting aid. Doing so comes with a cost, though, as the government (1) does not receive the needed aid and (2) risks the detection of its true type by the international community. Because states with fewer resources face steeper costs on both counts, these governments should accept international aid more frequently than governments with more resources, as should states that view higher status as an important goal.

To test these hypotheses systematically, we constructed an original data set comprising 66 natural disasters that occurred between 2004 and 2012.Footnote 29 Our unit of analysis is the natural disaster because this represents the exogenous shock that allows countries to either accept or reject aid. These data come from the EM-DAT International Disaster Database, which records information about the occurrence and effects of natural disasters, compiling information from UN agencies, NGOs, insurance companies, and research institutions (Guha-Sapir, D., Below, R., Hoyois, P., Accessed 2014. EM-DAT: international disaster database. http://www.emdat.be). We subset our data to the disasters considered most exogenous such as earthquakes, hurricanes, and floods.Footnote 30 Because our interest lies in natural disasters that receive international attention, we select all disasters in which the death toll was at least 300.Footnote 31

For each disaster, we coded (1) whether the affected country chose to categorically reject international assistance, (2) if it rejected international assistance, whether it successfully persuaded the international community that its response, unaided, was sufficient, (3) the extent of the government’s resources to respond to the disaster without assistance, and (4) the value the state places on status. We coded our aid rejection variable by drawing upon open source information from news articles and aid agencies.Footnote 32

For each case, we assigned the disaster a binary indicator of whether aid was declined: a disaster received a 1 if the government outright rejected international assistance or tried to refuse international assistance before accepting it.Footnote 33 We selected this definition to capture all cases in which governments attempted to pursue a strategy of rejecting aid. Since we are interested in the government’s intent to decline aid, rather than the outcome of the refusal, we include cases as instances of rejection in which the government tried but failed to convince the international community that it could respond adequately to the disaster on its own and resorted to accepting aid.Footnote 34 We do not count as instances of rejection cases in which a state declined aid from a single country, as this type of behavior typically arises from interstate hostilities rather than the dynamics we identify. Further, rejecting aid in this manner often carries diplomatic costs, unlike blanket rejections which typically do not, as evidenced in our case study research. If states were to reject just some offers of aid while accepting others, this would not send the self-sufficiency signal essential to our theory, which is why we code only categorical aid rejection.

We also used our qualitative research to code the degree of success with which governments rejected aid. Many cases exist in which the government declined aid at first, but the international community quickly discerned that it could not respond adequately on its own. We code whether governments pursuing aid rejection strategies faced international criticism for doing so: were these governments openly criticized by other governments, international organizations, and aid agencies for poor responses and for rejecting international assistance? We coded the country as 1 if its response did not attract international criticism, and 0 if it was openly condemned. By coding this qualitatively, our measure reflects the media’s view because this is our quantity of interest; we want to know whether governments were “caught” by the international community.

Additionally, we created a measure of how capable the government was of responding to the disaster using its own resources by dividing the country’s GDP by the number of people killed in the disaster.Footnote 35 A higher ratio indicates greater capabilities. We chose this measure over a more qualitative coding to avoid a measure that is a function of the strategy the government selects. However, as a robustness check, we also use logged GDP in an alternative specification.

Further, we code whether states care about status using a measure of their logged military expenditures. Since many scholars have shown that bigger militaries accord higher status to a country, military investments signify that states value that status and are undertaking costly efforts to obtain it.Footnote 36 However, since military expenditures can proxy for non-status dynamics (such as the threats faced by a regime), we also employ an alternative measure—an indicator of whether a given state possesses or seeks a permanent seat on the United Nations Security Council—as a robustness check.Footnote 37 Since certain regime types may be less receptive to assistance, we include control variables such as GDP growth, respect for human rights, democracy, and other logged foreign aid; we also control for whether the disaster was an earthquake, since these disasters are especially salient in the news.Footnote 38

Figure 1 provides a visual representation of the relationships between our key variables. In panel (a), we plot our cases according to each government’s logged ability to respond to a given disaster. Red points indicate countries that accepted aid, while green and blue points depict those that rejected aid. The figure shows that cases in which a country possessed a fewer resources tend to be red, while there exist very few red points at high levels of capacity.Footnote 39 This supports our claim that governments that are most incapable of responding to disasters (e.g. Haiti, Bangladesh, Philippines) choose to accept assistance because they cannot succeed in fooling the international community. In panel (b), we plot our cases according to a government’s value of status. The figure shows that, in accordance with our theory, all but one instances of aid rejection lie above the 1% line, as states that place higher importance on status tend to decline aid. It is worth noting that the states that rejected aid are typically either high status or ambitious middle-income countries. For high status countries (e.g., U.S., China, and Japan), accepting aid would be embarrassing.Footnote 40 For middle-income countries with ambitious aspirations (e.g. Thailand, Turkey, India, and Iran), rejecting foreign aid is hoped to change the international community’s impression of their self-sufficiency.

Having examined our data visually, we turn to a statistical analysis.Footnote 41 Since our outcome variable is binary, we run a logistic regression, with robust standard errors clustered by country. In Model 1 of Table 2, we present basic correlations, regressing the indicator variable for aid rejection on our measure of the government’s ability to respond to the disaster and its concern for status. To account for characteristics common to particular regions, Model 2 displays the results of the same regression, but adds region and year fixed effects. To check the sensitivity of our results to modeling assumptions, Model 3 uses OLS instead of logit. Models 4 and 5 use alternative measures of our key independent variables—logged GDP represents a government’s need for assistance in model 4 and an indicator of whether a country currently possesses or has seriously sought a permanent seat on the UNSC indicates that the government values status in model 5. Model 6 then adds covariates. In particular, we control for the acceptance of other aid to address the possibility that our measure of a state’s capacity to respond captures characteristics that predispose states to accept or reject aid. We also control for democracy and human rights to mitigate concerns about the types of governments that have high military expenditures.Footnote 42 Across nearly all specifications, both the ability to respond to a disaster and the desire for status are significant predictors of whether a government accepts aid.Footnote 43

Table 2 Statistical analysis of 66 natural disasters, 2004–2012

We also perform additional robustness checks, which appear in the SI due to space constraints. While our analysis focuses on the as-if random timing of disasters, we acknowledge that disaster damage is endogeneous. To mitigate these concerns, we demonstrate that our results are not sensitive to the particular sample of disasters used.Footnote 44 First, we limit our sample to earthquakes and include those with a high magnitude on the Richter Scale in which the shaking affected a population center. Second, we analyze our original data using alternative thresholds for how many people died in the disaster to warrant the disaster’s inclusion in the sample. Our results are robust to both exercises.

5 Alternative explanations

We consider several alternative explanations for why countries could reject international assistance after a natural disaster. First, perhaps governments sometimes lack the bureaucratic capacity to receive international assistance. They might therefore decline aid simply because they cannot absorb it. However, this explanation is not likely driving our results, and actually may work against them. Countries with limited bureaucratic capacity tend to possess fewer resources in general. Thus, if this argument held up, countries with lower capacity should be more likely to reject aid—the opposite of what we observe. Further, in coding our cases, we did not encounter evidence that governments with difficulties absorbing aid rejected the aid as a result. Instead, aid was still provided but the disbursement was slow and inefficient.Footnote 45 Finally, in our main results, we control for a government’s receipt of other aid. Thus, while governments do sometimes experience difficulties managing aid flows, these problems cannot account for our findings.

Another alternative theory is that states reject disaster assistance because they want to keep international influence out of their countries.Footnote 46 Yet here too, we find little evidence in favor of this explanation. Like the previous alternative story, if this were true, it would cut against our findings. Countries that shun foreigners tend to be small and resource poor. These kinds of countries may violate international norms and seek to hide their activities, or are unstable and do not want international intervention.Footnote 47 Thus, if this explanation held, we would expect states with fewer resources to reject aid, while we instead see that states with greater resources tend to do so.

In our qualitative research, we also encountered very few instances that would support this logic. One country that did seem to reject assistance for this reason was Myanmar, which considered refusing aid after a cyclone devastated the country in 2008 in part because it feared foreign influence. In particular, the regime worried that accepting aid could increase the risk of an invasion (Selth 2008). Due to threats and pressure from the international community, Myanmar eventually accepted limited aid, but still demanded that it control its distribution (Farley 2008). The case of Myanmar shows that some truly extreme cases remain anomalies in our data. However, such cases are rare; other closed regimes – such as Sri Lanka and North Korea—opened their doors when disasters struck.Footnote 48 Even so, we account for the possibility that states shun assistance due to worries that their human rights violations will be detected by including respect for human rights as a control variable in our main specification.

Finally, perhaps some governments reject aid to signal competence to their domestic populations. However, we find this explanation to be unpersuasive for several reasons. Theoretically, previous work has shown that citizens impacted by natural disasters demonstrate increased interest in politics and punish politicians at the polls unless they receive relief (Fair et al. 2017; Sinclair et al. 2011). Since disaster aid also presents opportunities for patronage and clientelism, and domestic audiences are likely difficult to fool because they experience the government’s performance firsthand, turning down this assistance should carry high domestic costs (Achen and Bartels 2004).

Empirically, we saw many instances in which governments were punished domestically for rejecting aid in the qualitative coding of our cases, and few where doing so provided a domestic benefit. Even in cases in which the government experienced a domestic boost from doing so—as occurred after several disasters in China—the government chose to reject aid mainly due to international considerations. Further, we ran our survey on a sample of Indian citizens, and found no evidence that the same aid rejection treatment that improved Americans’ opinions of India held any sway among Indians themselves.Footnote 49 These findings suggest an international—rather than a domestic—motivation for rejecting aid.

6 Conclusion

We show that some governments attempt to reject offers of foreign assistance in order to improve their status in the international community. Whereas accepting foreign aid signals that a government requires external resources to deal with an unforeseen and costly event, rejecting foreign aid demonstrates self-sufficiency and accrues international respect, as shown through our survey experiment. Our formal model predicts that governments are more likely to use rejection strategies when they have greater ability to respond to a disaster and when they value status more highly. The results from our original survey experiment and analysis of new data comport with our theoretical predictions.

Though we focus on the domain of disaster aid, our theory applies to many other empirical areas such as development aid, loans, peacekeeping, and military assistance, as well as other kinds of shocks, such as economic and refugee crises. These remain fruitful directions for future research. More generally, other steps that states can take to limit their receipt of international assistance short of refusal could be productively examined; for example, scholars could study graduation from aid programs (Carnegie and Samii 2019) or upward revisions of income calculations that render states ineligible for foreign aid. These seemingly irrational actions could also elicit greater status by demonstrating self-sufficiency. Further, how and when states adopt methods such as aid rejection instead of engaging in wars to improve status or due to other geopolitical considerationsFootnote 50 remains an interesting avenue for scholars to explore.

Our study also suggests several policy implications. First, our findings imply that while states often heavily publicize their efforts to assist other states in order to reap the prestige and positive image associated with these activities, a more effective way to help may be to do so quietly. If assistance is not widely known, states face a smaller hit to their status when accepting it. Second, since we have demonstrated that states so strongly seek improved status that they often take actions detrimental to their economic well-being, it is possible that offering such avenues could effectively motivate states to make concessions in exchange, perhaps in areas such as human rights or democracy. Finally, since we have likely identified one of many avenues short of war through which states can improve their status, encouraging states to pursue such strategies or making them increasingly available could potentially help to avoid costly future conflicts.