Dong Jung Kim (Associate Professor in the Division of International Studies at Korea University)

Abstract

Recent U.S. emphasis on geoeconomic confrontation with China in the developing states entails the danger of spawning an economic version of the domino theory. Geoeconomic domino theory posits that if China’s economic influence is not stopped in a developing state, the United States might encounter more developing states incorporated into the Chinese economic orbit, and the falling economic domino can eventually reach the developed economies. Nonetheless, there are reasons to expect that the developing economies would not fall to China one after another. China’s gains in the developing regions in terms of material capacity, ability to govern key issue areas, and ideological appeal are also dubious. In conducting geoeconomic competition with China, the United States should concentrate on the developed economies—Western Europe and Northeast Asia—that have a large impact on the balance of power. For this, a geoeconomic version of the strongpoint defense strategy is needed.

Jun Young Han (Master’s Degree in Political Science and International Society at Korea University)

Abstract

Election-related violence is rampant in many developing countries that receive foreign aid. Although voters often elect representatives associated with violence in the developing world, little is known why they do that. In this article, I investigate why voters support politicians who resort to violence. I argue that the poor tend to vote for a candidate who delivers tangible local benefits through foreign aid projects even when the candidate uses violence in elections. To test this argument, I conduct a nationwide survey in the Philippines, inserting experiment about the effects of foreign aid and violence on voters’ electoral support. I find that poor voters residing outside the national capital region are more likely to support the candidate who has offered foreign aid projects to her constituency, regardless of her alleged electoral violence. This research sheds light on the mechanism linking poverty to electoral violence in less-developed countries. It also reveals unintended consequences of foreign aid for increasing the likelihood of violence.

Introduction

Electoral violence can result in many causalities and undermine the democratic institutions of developing countries.1 According to the recent data on election-related violence, 804 (30.1%) elections out of 2,667 held around the world from 1975 to 2021, had at least one aspect of violence (Besaw and Frank 2021). Among them, electoral violence has caused “thousands of casualties, population displacement, and protracted political crises, as in Kenya, Nigeria and Zimbabwe”(Gutierrez-Romero and LeBas 2020, 77). Also, it has undermined democratic institutions so severely that violence has become a routine of election in many developing countries such as Thailand and the Philippines.

Despite these costs, voters in developing countries routinely elect candidates associated with violence. In the Philippines, for instance, Melecio Yap Jr. has been a congressman of Negros Occidental’s 1st congressional district from 2016 to 2019, and mayor of Escalante city from 2007 to 2016 and from 2019 to present, even though in 2016, he was charged for the murder and frustrated murder of supporters of his political rival from 2007 to 2011.2 In the other region, Carmen Loreto-Cari, former mayor of Baybay city and 3-term congresswoman, and her son, Jose Carlos Loreto-Cari, have been swapping positions in the city and congress despite being under suspicion of masterminding for an assassination attempt against a rival politician in the 2013 election.3

What explains the electoral success of legislators who are alleged to use violence? Previous explanations primarily focus on the expected benefit from the violent candidate. They argue that the voters are more tolerant of the violent candidates with the same ethnicity or partisan affiliation, expecting that they might bring benefits if elected (Chandra 2004; Chaturvedi 2005). In a similar vein, many existing works find that the core supporters of violent politicians would not sanction him although they might blame the use of violence in the abstract (Collier and Vincente 2012; Hafner-Burton, Hyde and Jablonski 2014; Lynch 2014). In other words, voters might be willing to accept negative signals of the candidate if expected benefits from her are sufficiently large (Winters and Weitz-Shapiro 2013).

This line of research predicts that the poor voters would be more likely to endure candidate’s misconduct such as violence or corruption because “almost universally, clientelism is an exchange between politicians and their poor clients” (Weitz-Shapiro 2012, 570). However, empirical evidence is mixed. On the one hand, scholars find that, contrary to the conventional wisdom, poor voters in India do punish candidates with a record of crime or corruption (Banerjee et al. 2014), or even sometimes, in Brazil, they have a more powerful negative reaction to information about corruption than upper-class voters (Winters and Weitz-Shapiro 2013). On the other hand, however, from the vignette experiment in Kenya, other scholars find that poor voters are less likely to punish the candidate rumored to use violence in the previous election than their wealthier counterparts (Gutierrez-Romero and LeBas 2020).

I suggest that these existing works have limitations in mainly two aspects. First, they have not paid much attention to heterogeneity in the preferences of poor voters. For example, poor voters living in the area where the level of social spending is relatively high, they might be less prone to the candidate’s political mobilization with tangible benefits since their basic needs can be fulfilled by existing public goods or social services. If this is the case, they might be more willing to forgo benefit from the politician in the face of a criminal or violent background of her. Second, previous works have not considered various forms of benefits that candidates might offer, or voters might be aware of, in the real world. For example, as I would show later, foreign aid in developing countries often offers basic public goods and local social services from which the constituents can benefit. However, there is much less research that investigates the effect of these diverse forms of benefits on the probability of voting for the candidate with negative attributes.

Therefore, I aim to contribute to this literature by considering another important factor that might affect vote choice for the violent candidates in developing countries: foreign aid. Different from the clientelistic benefits, “which are targeted at the individual and linked to individual political behavior” (Weitz-Shapiro 2012, 569), foreign aid is more close to local benefits, or ‘pork’, in that politicians in developing countries frequently claim their credits in securing foreign aid projects or budget, and this leads to attracting more votes (Cruz and Schneider 2017; Ahmed 2012; Jablonski 2014; Knutsen and Kotsadam 2020). Also it is widely known that “aid can help recipient governments to stay in office by enabling them to maintain patronage and services” (Carnegie and Marinov 2017, 672). Despite this well-established electoral effect of foreign aid in less developed countries, however, there is much less research focusing on how it might affect voting behavior on a candidate with a criminal or violent backgrounds.

I try to fill this gap by arguing that the poor living in less-developed regions tend to vote for a candidate who delivers tangible local benefits through foreign aid projects even when the candidate uses violence in elections. Focusing on the role of foreign aid as a source of local public goods and basic social services provision, I predict that poor voters living in outside of the national capital region (NCR) will be more likely to vote for the violent candidate when she has a record of receiving foreign aid project in the district. In contrast, I predict that such a positive effect of foreign aid on the probability of voting for the violent candidate will not emerge among the voters living in the NCR or the voters living in the non-NCR but have higher income levels.

I test this theory using a survey experiment embedded in a nationwide online survey that explores individuals’ voting intentions for their legislators in the Philippines. This experimental approach has the advantage of testing the effect of foreign aid on the probability of voting for a violent candidate. Because the treatment is exogenously assigned, whether a respondent learned the candidate is rumored to use violence or got foreign aid project in her district is designed to be orthogonal to the individual attributes of voters, specifically their previously held views about foreign aid and violence. My treatment not only provides the respondent with a realistic amount of information on the foreign aid and the rumor of using electoral violence, but it also tests the impact of aid and violence on voting behavior. A main finding is that, while receiving foreign aid benefits and using violence harms the incumbent’s electoral fortune, the voters react differently to the trade-off between the positive and the negative attributes according to their residential area and economic status. Only poor voters living in non-NCR are more likely to reward the violence-related incumbent when they know she has a record of receiving foreign aid before. I find that these findings are robust to alternative explanations and measurement.

These findings imply that the electoral success of violent candidates likely reflects voters’ preferences for local public goods. While previous works find that the economic status of voters matter, they have little empirical consensus on when and how it matters. This study contributes to this literature by elaborating not just income level is important, but the environment where voters live is also important. This suggests that the regional capacity for providing local public goods and basic social services can be crucial in deciding whether or not poor voters would have the incentive to cast their vote for violent candidates. If there is a shortage of basic public goods and social services, poor voters’ demands for them would be high enough to ignore the misconduct of the candidate. I show this is true by illustrating that the positive effect of foreign aid treatment on the probability of voting for the candidate decreases as the capacity of providing public goods by respondents’ regional government gets greater.

Moreover, the results show that violence-related candidates might have an incentive to leverage these voters’ demands with foreign aid. Pre-existing studies have revealed that in less-developed countries, politicians keep claiming their credit in receiving foreign aid, and it helps them reelected (Cruz and Schneider 2017; Jablonski 2014). The results corroborate experimental evidence of this argument, and I further show foreign aid can mitigate the negative effect of electoral violence especially for poor voters living in non-NCR. This implies that donors and international NGOs might face a trade-off in implementing community-driven development projects. While these projects would provide basic public goods and social services in less-developed regions, at the same time, they can be exploited for electoral purposes by violence-related local politicians.

Local public goods provision, foreign aid, and voters’ preference

To understand why voters vote for a violent candidate, it is important to examine what affects voters’ preferences. Since legislators’ foremost interest lies in being re-elected, they respond to what their constituents want (Bailey 2001; Denzau and Munger 1986). Moreover, politicians do so even when constituent preferences are unorganized (Milner and Tingley 2010). So, to investigate what voters want, and what affects their preferences is a crucial step toward revealing why violence-related candidates keep being elected in developing countries.

Regarding voter preferences, previous works primarily focus on the economic status of voters because material interest is known as important predictor of voter policy preferences (Alesina and Ferrara 2005; Inglehart 2007). Research shows that poor, less-educated voters tend to prefer clientelistic and materialistic benefits over programmatic ones (Shin 2015; Nathan 2016; Kitschelt 2000). The reason why poor voters demand clientelistic benefits more than their wealthier counterparts might be because they need them more, or they are more risk-averse, or have shorter time-horizon (Stokes 2009; Keefer and Khemani 2005). Also, previous works find that poor voters are the main targets of vote buying and clientelism in developing countries (Jensen and Justesen 2014; Stokes et al. 2013). These findings predict that the poor will be more likely to vote for candidates with criminal records when they are expected to offer tangible local or individual benefits.

However, empirical evidence is mixed. Banerjee et al. (2014) finds that, regardless of their caste, poor voters are less likely to vote for criminal candidate. In the similar vein, Winters and Weitz-Shapiro (2013, 423) reveals that “respondents in the lowest classes have a powerful negative reaction to information about corruption” regardless of “candidate’s record of completing many public works projects during his term”. In more recent work, in contrast, Gutierrez-Romero and LeBas (2020) finds that respondents who have four or fewer assets from a list of 15 durable assets, do not penalize candidates for their rumoured involvement in electoral violence while respondents who have more than four durable assets do. As inconsistency in the results of previous experimental studies illustrates, the economic status of the voters alone seems not enough to explain electoral performances of the criminal or violent candidate. These inconclusive findings suggest that further research is needed to find under which conditions the poor voters become more tolerant of criminal or violent candidate.

In this paper, I aim to contribute to this debate by arguing that foreign aid and the level of basic public goods and social services in the residence can matter. Specifically, I argue that poorer voters in regions outside the national capital region (non-NCR) are more likely to vote for a candidate who has offered foreign aid projects to her constituency, regardless of her alleged electoral violence. In contrast, I argue that voters in the NCR and richer voters in non-NCR are less likely to vote for a candidate who is associated with electoral violence, regardless of her delivery of foreign aid projects. In sum, a main task of this paper is to see whether the effect of foreign aid on the probability of voting for the candidate is conditional on her use of violence, and how this conditional effect varies by voter’s characteristics. My arguments based on the two main assumptions: (1) Voter demand for basic public goods and social services is a function of the actual level of them in the region, and (2) foreign aid can be a source of basic public goods and social services in developing countries.

First, voters living in less-developed regions where local public goods are underprovided will demand them more than their counterparts living in more-developed regions where local public goods and services are provided sufficiently. In most developing countries, a large share of public goods and social services is concentrated in the NCR. In contrast, regions outside the NCR often have difficulty in providing an adequate levels of public goods and social services to citizens because local government has limited sources of tax revenue, and the development priority of central government is usually on the NCR. Indeed, in the Philippines, the level of government final consumption expenditure(GFCE)4 by regions, one of the indicators for regional capacities of providing public goods, illustrates that there is a huge gap in the capacities between the NCR and non-NCR. For example, as you can see from the Appendix G, average per capita GFCE in the non-NCR was 18.5 Philippine peso while per capita GFCE in the NCR was 82.1, greater than four times of non-NCR per capita GFCE.

Other indicators also imply that the non-NCR residents’ demand for basic public goods would be greater than that of the NCR residents. For examples, in 2018, average hospital beds per 10,000 population in non-NCR was 6.1 while it was 13.5 in the NCR. Considering the figure was 7.9 for Bangladesh, 9.0 for Cambodia, and 15.0 for Laos, non-NCR has fairly weak health care.5 Also, average student-teacher ratio in non-NCR was 26.2 while it was 21.5 in the NCR while the figure was 26.0 for Thailand, and 22.0 for Laos.6 Both average hospital beds per 10,000 population and student-teacher ratio indicate that non-NCR residents might be more likely to demand for medical and educational services from the government. However, not all the voters residing in non-NCR demand basic public goods or social services at the same level. For richer voters, they can afford quality services from the private market. Therefore, I expect that the most vulnerable type of voters to local benefits by the candidate would be the poor voters living in the non-NCR where local public goods and basic social services are underprovided.

Next, in many developing countries, foreign aid is an important source for basic public goods and social services provision. The vast majority of previous works find that foreign aid can be an important resources for politicians in recipient countries, and therefore, they leverage it to remain in the office longer (Dreher et al. 2019; Ahmed 2012; Jablonski 2014; Cruz and Schneider 2017). In line with this research, I expect that some types of foreign aid, for example, community-driven development projects, can boost local politicians’ electoral success by providing essential, basic public goods and social services needed by local constituents.

Having set the Millennium Development Goals (MDGs), there has been rising concern about how to satisfy the actual needs of the poor rural people since they and the local organizations that represent them, often lack the capacity to develop and articulate their needs.7 Also, facing critics of “big development” that criticize large and government-initiated projects perform poorly, scholars and practitioners began to promote participatory, small-scale development. In this context, community-based development approach has become an important tool that can help find, scale-up, and effectively satisfy the actual needs of the beneficiaries in the recipient countries. Community-based development “is an umbrella term for projects that actively include beneficiaries in their design and management”, and “The World Bank’s lending for such projects has risen from $3 billion in 1996 to $7 billion in 2003” (Mansuri and Rao 2004, 2). In 2020, there were 327 active community-based development projects in 90 countries for a total lending of $33 billion.8

Since this form of foreign aid is based on the small-scale and the actual needs of recipients, the projects often include providing basic public goods and social services in remote, local area of developing countries. Also, these projects usually aim to deliver and implement quality development projects, and benefit the poor populations in the region. So, foreign aid can play an important role as providing basic public goods and social services, and poor voters in need would be the main targets of the projects. At the same time, previous research finds that the voters in developing countries often credit receipt of foreign aid to local politicians, even when aid allocation is nothing to do with politicians’ capacity or performance (Cruz and Schneider 2017).

As a result, I expect that poor voters will be more tolerant of violence when the candidate brings foreign aid project in her district. However, since the demand for basic public goods and social services would be much lower in the NCR, I predict that only poor voters living in the non-NCR will be willing to embrace violent candidate for local benefits. In sum, I expect a violent candidate has an incentive to mitigate her negative impression with receipt of a foreign aid project, and poor voters with limited access to local public goods and basic social services will be willing to vote for her.

Hypothesis 1 : Poorer voters in non-NCR are more likely to vote for a candidate who has offered foreign aid projects to her constituency, regardless of her alleged electoral violence.

However, I expect that the positive effects of foreign aid on the probability of voting for the violent candidate would not emerge among the voters in NCR and richer voters in non-NCR. First, for voters in NCR, the marginal benefits brought by increased basic public goods and social services would be small compared to voters in non-NCR since it already has sufficient level of them. So, additional basic public goods and social services from the foreign aid projects would not be enough to make NCR voters be willing to vote for violent candidate. Second, richer voters in non-NCR are neither the main targets of the foreign aid projects nor demanders of basic public goods and social services. Also, they can afford more quality goods and services in the private markets. Therefore, I expect that foreign aid project would not increase the probability of voting for the violent candidate among richer voters in non-NCR.

Hypothesis 2 : Voters in NCR and richer voters in non-NCR are less likely to vote for a candidate who is associated with electoral violence, regardless of her delivery of foreign aid projects.

Research setting

To test the hypotheses, I conduct a survey in the Philippines, which is well-known for the practice and prevalence of election-related violence. From 1986 to 1998, on average, more than 100 people had been killed in each election (Patino and Velasco 2004). In the 2001, 2004, and 2007 elections, there had been a total 269, 249, and 229 election-related violence incidents, respectively, and more than one hundred people were killed before and after the elections of each year (ACE 2007). The death toll has been on a declining trend since 2007, but as one can see from the 2009 Maguindanao massacre, in which 58 people were gunned down at the order of the incumbent,9 and 43 incidents with 73 victims in the most recent election in 2019,10 the country is still experiencing violent elections.

At the same time, the Philippines “has been one of the largest recipients of foreign aid from bilateral donors such as Japan and the United States, and also from multilateral donors such as the World Bank and the Asian Development Bank” (Mitra and Hossain 2013). In 2019, the Philippines received 905 million dollars of net official development assistance (ODA), which was 0.22 % of its gross national income (GNI).11 More importantly, there are several less-developed regions in the Philippines whose local economy is highly dependent on foreign aid. For example, in 2014, the proportion of ODA disbursed to region to gross regional domestic product (GRDP) was 10.5% for the Eastern Visayas (Region VIII), 8.4% for the Autonomous Region in Muslim Mindanao (ARMM), 7.8% for the Caraga (Region XIII), and 7.7% for the Mimaropa (Region IVB).12

Also, regarding community-based development projects by the World Bank, these less-developed regions have more projects than others in terms of both the number of projects and the amount received.13 From 2014 to 2018, the Western Visayas had received 4,696 projects worth 9.5 million dollars, the Eastern Visayas had received 6,337 projects worth 9.6 million dollars, the Caraga had received 1,557 projects worth 4.3 million dollars, and the Mimaropa had received 1,690 projects worth 4.1 million dollars.14 Considering the annual GRDP of Eastern Visayas, Caraga, and Mimaropa in 2014 was 5.8, 3.1, and 4.2 million dollars respectively, the amount of community-based development projects in the impoverished regions is quiet huge. Also, the development projects have improved access to local public goods and services, and the quality of residents’ life in the regions. For example, the regions that had been selected as the beneficiaries for community-based development projects, have higher rates of school participation rates and employment rate, and better access to safe water and health services (Edillon et al. 2011). In addition, from 2014 to 2020, there has been 8% increase in access to roads, school, health centers and clean water in the municipalities received community-based development projects, compared to others who were not selected for the projects.15

From the theoretical perspective, the case of the Philippines is suitable for analysis since it has a deep-rooted political system represented by clientelism and pork-barrel politics, and also politicians often have criminal or violent background. Especially, Cruz and Schneider (2017, 396) finds that “incumbents in the Philippines undeservedly take credit for the receipt of foreign aid and thereby boost their chances of reelection”. They reveal that incumbents received development aid projects “significantly increased the frequency of their visits to the project sites and pursued a number of strategies to appear influential in the allocation of the project funds” although they could “neither affect the likelihood of selection for the project nor directly divert project funding” (Cruz and Schneider 2017, 396-397). So, for testing my hypotheses, the Philippines offers appropriate conditions since candidates’ receipt of foreign aid and rumor of using violence are common attributes that ordinary voters encounter in the election periods.

In addition, as I illustrate in the previous section, when I compare the number of hospital beds per 10,000 population, student-teacher ratio, and government final consumption expenditure per capita between NCR and non-NCR, regional imbalance in the capacity of providing public goods and social services is huge. This regional gap offers a good background for testing my hypotheses. Also, while the geographical background of most of the literature is biased toward African countries and India, electoral violence is global phenomenon with varying degrees and various forms. So, analyzing the case of the Philippines can shed light on the relatively unraveled aspects of electoral violence and its cause because election-related violence in the Philippines is usually not based on ethnicity, caste, or party affiliation, as in most African countries and India, but rather it is based on the more personalized motivation or political clan. Moreover, pervasive clientelism and pork-barrel politics in the Philippines offers a good opportunity to assess how community-based foreign aid project can help incumbent with bad attributes gain popular votes.

Experimental design

I embed experiment in the survey to randomize treatment across individuals. This allows me to test the causal mechanism linking a candidate’s attributes to the perceptions of individual voters. Survey experiment was conducted between July 20 and August 20, 2021. Respondents were randomly recruited on the quota basis of age and gender from the Philippines nationwide subject pool registered at an international survey company Qualtrics. Respondents who are Filipino, residing in the Philippines, and whose age is above or equal to 18 participate in the study. In total, 1,632 Filipinos completed the survey. Respondents were asked demographic and political questions that are widely used in the literature. Appendix A presents descriptive information for respondents.

My main question was a survey experiment testing how the respondent’s intention to vote for the incumbent candidate with violent background was affected by the candidate’s receipt of foreign aid. Before receiving the main treatment, all respondents were asked to imagine a candidate in their constituency for a Member of Parliament. To equalize the candidate’s qualifications and promises, all respondents were told:

Imagine that there is a candidate for the 2022 House of Representatives elections in your constituency. This candidate has been elected member of the House before in another constituency similar to yours in the 2019 elections, and the candidate is promising to improve the economy of your community.

To reduce the possibility of socially desirable responses I did not mention explicitly the candidates’ party affiliation or ethnicity. The key manipulation is information about the incumbent’s record of receiving foreign aid projects in his district, or rumors of using electoral violence during the last election. For respondents in baseline, they were just told the above script only. For respondents in aid-only condition, adding to the baseline information, they were told about the information about the foreign aid project that the candidate’s district was selected to receive. For respondents in violence-only condition, adding to the baseline information, they were told about the information that the candidate is rumored to have ordered the murder of a rival politician to win the last election. For respondents in the last experimental condition, adding to the baseline information, they were told about both foreign aid and rumor of violence information, as described in the aid-only condition and violence-only condition. Verbatim instructions and vignettes can be found in Appendix B.

The manipulations are realistic in that the vignettes are created based on actual foreign aid projects implemented in the Philippines, and one of the most common forms of information about election-related violence that Filipino voters would encounter. For foreign aid treatment, I use KALAHI-NCDDP (Kapit Bisig Laban sa Kahirapan – National Community Driven Development Program), a $663.90 million project co-funded by the World Bank. This foreign aid project started in 2014 and ended in 2019, as an extension of another community-driven foreign aid project, KALAHI-CIDSS (Comprehensive and Integrated Delivery of Social Services). KALAHI-NCDDP intended to empower communities in the targeted areas to achieve improved access to services and to participate in more inclusive local planning, budgeting, and implementation (World Bank 2014). The largest share of the funding was spent on investment grants that supported community sub-project investments and activities. Based on the community-identified priorities, the project funded community-based public infrastructure and services such as roads, bridges, schools, daycare centers, etc.

Following the logic of Cruz and Schneider (2017), I choose KALAHI-NCDDP as foreign aid treatment because “it is explicitly designed to prevent the misappropriation of funds by national and local politicians by allocating funding based on a poverty formula and releasing funding directly to the villages”. Therefore, local politicians have much less control over both the awarding of projects and the subsequent allocation of funding (Cruz and Schneider 2017, 400). To reflect this neutrality of the program, I carefully choose wordings of ‘district was selected […] by the World Bank’ in vignette to minimize the possibility that respondents might think the candidate attracted or deserved it.16 This strategy will offer a hard test for my hypothesis since the effect of foreign aid often comes with the capture of funding and explicit credit-claiming by politicians (Cruz and Schneider 2017; Jablonski 2014). Also, I decide to choose 200,000 US dollars as the amount of money spent on the project by averaging the number of grants per subgroup project implemented in 2019.

For violence treatment, I specify that the candidate is rumored to have ordered the murder of a rival politician to win the previous election. This information delivers quite strong allegations against the candidate to the readers, compared to other more common forms of electoral violence such as coercion, threats, or intimidation. I also give additional information that the candidate has not been arrested for the alleged crime because this is what normally happens in the Philippines. So, my design offers a lower bound for investigating the effect of foreign aid on the probability of voting for a violent candidate since I use relatively weak foreign aid manipulation while using relatively strong electoral violence manipulation. As a result, in testing whether the foreign aid has a positive impact on the voting intention for violent candidate, this experimental design is quiet conservative.

Despite careful consideration in choosing the content and question wordings of the vignette, still, there can be criticism about the hypothetical nature of the experiment. One concern is related to the inaccuracy of responses on the questions regarding the hypothetical situation (Cummings et al. 1995). While there is debate on this issue, some scholars have shown that respondents can give meaningful answers to hypothetical questions if “don’t know” option is provided and if questions allow participants to make link their experiences to the hypothetical (Mitchell and Carson 1989, cited from Jensen et al. 2014). I include “don’t know” option in the survey, and I believe that the vignette captures the realistic aspects of information that ordinary Filipino voters might encounter in most elections. Another possible criticism might come from the desirability bias which the information about violence is likely to cause. However, since this bias would underestimate the effect of foreign aid on the probability of voting for the violent candidate, I believe that this is not a great problem in analyzing the results.

To isolate the effect of the candidate’s rumor of using violence and receipt of foreign aid from the respondent’s characteristics, I assign respondents randomly to one of four experimental conditions. Randomization was well implemented despite a minor imbalance. I conduct ANOVA tests to see whether or not the mean value of each variable listed in Appendix A varies across the four experimental conditions. As reported in Appendix C, the mean values do not statistically vary at the 5% significance level, except the ones for Tagalog. To address these unbalanced covariates on some factors, after demonstrating the comparison of means of four groups, I provide results from regression including variables listed in Appendix A as controls.

To estimate the effect of foreign aid on the probability of voting for the violent candidate, I rely on survey item asking respondents their voting intention for the candidate. As for the  outcome variable Vote, after the vignette, I asked respondents to indicate whether or not they will vote or not for the candidate. It is coded as 1 if respondents say “Yes”, and 0 otherwise.

Who tends to vote for a violent candidate?

As a first cut, I provide a comparison of the means of these four groups based on the dependent variable, Vote. As you can see from Figure D1 in Appendix D, receipt of foreign aid project increases the likelihood of voting for the candidate by about 18%p (changing the mean from 0.60 to 0.76), while rumor of using violence decreases the likelihood of voting for the candidate by about 48%p (changing the mean from 0.60 to 0.12). However, when I compare the mean of ‘Violence only’ group to ‘Aid and Violence’ group, I cannot reject the null hypothesis that the mean of the two groups is the same at the 5% significance level. In other words, for entire samples, I find that receipt of foreign aid has no positive effect on the probability of voting for the candidate with a rumor of using violence.

However, when I divide samples by residential area and household monthly income level, I find that the poor voters living in the non-NCR tend to vote for the violent candidate when her district was selected as a recipient of the foreign aid project. To compare the mean value of Vote by different treatment conditions and groups, I run a two-sample t-test. Results of the t-test are reported in Table D1, and Figure 1 is drawn from the table. In Figure 1, only among respondents living in non-NCR and with household monthly income below 10,001 Philippine pesos (lower-left panel), a mean value of Vote is higher in ‘Aid and Violence’ condition than in ‘Violence only’ condition (p = .02). This implies that in line with my theoretical prediction, candidate with the rumor of using violence can have electoral support from the poor voters living in non-NCR by her district being selected as receiving foreign aid project. For respondents living in NCR and living in non-NCR but with higher household income, however, there is no such positive effect of foreign aid on the probability of voting for the violent candidate.

Figure 1. The proportion of respondents who vote for the candidate by residential area and  household monthly income level

I also find that when there is no rumor of violence, poor voters reward foreign aid in general. Specifically, when I compare ‘Aid only’ group to ‘Control’ group, only respondents with household monthly income below or equal to the median, are more likely to vote for the candidate, whether or not they live in NCR or not. For poorer respondents living in NCR (upper-left panel), candidate’s district being selected as a target of the foreign aid projects, increases the likelihood of voting for her by about 28% (changing the mean from 0.58 to 0.86). Similarly, for poorer respondents living in non-NCR (lower-left panel), the likelihood increases by about 12% (changing the mean from 0.61 to 0.74). However, there is no such positive effect of foreign aid on the probability of voting for the candidate for wealthier voters, whether or not they live in NCR or not. I cannot reject the null hypothesis that the mean of Vote between ‘Control’ group and ‘Aid only’ group is same for wealthier voters at 5% significance level. This finding suggests that the previous works’ finding of electoral gain from receiving foreign aid, is largely driven by poor voters.

Next, I regress dependent variable, Vote, on covariates in Appendix A.17 Since the dependent variable is binary, I run logistic regression. First, as you can see from the Appendix E, information about receiving foreign aid project alone has positive effect on the probability of voting for the candidate while information about alleged electoral violence alone has negative effect on the probability of voting for her. To see how voters’ income level and residential area matters for their voting choice, I divide samples by their household monthly income level and residential area. Figure 2 shows the predicted probabilities of voting for the candidate by experimental conditions, residential area, and household monthly income level.

Figure 2. Predicted probability of voting for the candidate by treatment conditions, residential area, and household monthly income level

Table 1. The marginal effect of foreign aid on the predicted probabilities of voting for the candidate by residential area and household monthly income

Standard errors in parentheses. * p<0.05

The X-axis refers to whether or not respondents receive information on the candidate’s rumor of using violence, and both solid circle and triangle refer to whether or not respondents receive information on the candidate’s record of receiving foreign aid projects before. Therefore, each point on a panel refers to predicted probabilities according to four different experimental conditions. For example, when a pointer is a solid circle and located on the left of the x-axis, it refers to the predicted probabilities of baseline respondents while when a pointer is a triangle and located on the right of the x-axis, it refers to the predicted probabilities of respondents assigned to ‘Aid and Violence’ group. My main interest lies on the right of the x-axis and a gap between the two shapes because the gap refers to the marginal effect of foreign aid on the predicted probabilities of voting for her when the candidate is rumored to use violence. I report this marginal effect of foreign aid in Table 1. Full regression results for making Figure 2 and Table 1 are reported in Appendix F.

Here, I again confirm the findings from the mean comparisons for the probability of voting. Only the poor voters living in the non-NCR (lower-left panel) reward the violence-related candidate when she has a record of receiving foreign aid projects (p = .04). Specifically, receipt of foreign aid increases the probability of voting for the violent candidate about by 10%p (changing the probability from 0.07 to 0.17). However, for NCR residents and richer voters living in non-NCR, foreign aid has no such positive effect on the probability of voting for the violent candidate. I cannot reject the null hypothesis that the marginal effect of foreign aid is zero when the candidate has a violence-related rumor. In sum, these results illustrate that poorer voters in non-NCR are more likely to vote for a candidate who has offered foreign aid projects to her constituency, regardless of her alleged electoral violence. In contrast, voters in NCR or richer voters in non-NCR are less likely to vote for a candidate who is associated with electoral violence, regardless of her delivery of foreign aid projects.

Robustness checks

In this section, I test observable implications from main findings, and whether the findings are robust to possible alternative explanations and other measure for one’s economic status. One of observable implications from my findings is regional capacity of providing public goods matters. I find that the poor voters residing in non-NCR are more likely to support the candidate who has offered foreign aid projects to her constituency, regardless of her alleged electoral violence because non-NCR has lower level of public goods. However, some might point that comparison between NCR and non-NCR is not appropriate since this could ignore heterogeneity among regions outside the NCR, and the findings might be mainly driven by NCR’s exceptionally high capacity of providing public goods. To deal with this concern, I focus on the regional variations among regions outside the NCR. If my prediction still holds within non-NCR, poor voters living in the region where a capacity of providing public goods is low will be more likely to support for the violent candidate when she has offered foreign aid to her constituency, compared to richer voters living in the regions with similar capacities or voters living in the region with higher capacities.

To see whether this prediction holds, I collect the data on government final consumption expenditure per capita in 2020 by regions from the Philippine Statistics Authority, and divide non-NCR samples by the level of expenditure and household monthly income. As you can see from logistic regression results in Figure H1 and Table H1 in Appendix H, poor voters residing in the region where the level of per capita government expenditure is below or equal to the median, are more likely to vote for the violent candidate when she has received foreign aid project. Neither voters living in regions with a high level of per capita government expenditure nor richer voters living in regions with a low level of per capita government expenditure show such a pattern.

Also, with same logic applied, I expect that for poor respondents living in non-NCR, the positive effect of foreign aid treatment on the probability of voting for the candidate will decrease as the regional capacity of providing public goods gets greater. However, for richer voters living in non-NCR, there would be no such variations in the effect of foreign aid treatment since they can afford necessary goods and services in the private markets, and therefore, have lower demands for public goods and social services offered by the government. To see whether this holds, I analyze how the effect of receiving foreign aid on the probability of voting for the candidate varies by GFCE per capita. Each point on Figure H2 in Appendix H represents marginal effect of foreign aid treatment on the probability of voting for the candidate by different level of per capita GFCE for non-NCR voters. Also, at the same time, each point represents actual per capita government expenditure by regions outside the NCR.

As you can see from the figure, for respondents who are living in non-NCR and with lower household income (left panel), the effect of foreign aid treatment on the probability of voting for the candidate is positive and statistically significant when they are living in the regions where the level of GFCE per capita is below 23.9 Philippine pesos, regardless of the candidate’s rumor of violence. However, for poor respondents living in CAR or Eastern Visayas, top two in terms of GFCE per capita following NCR, there is no such positive effect of foreign aid receipt by the candidate on the probability of voting for her. For richer voters living in non-NCR (right panel), marginal effect of foreign aid treatment on the probability of voting for the candidate is not distinguishable from zero at every level of per capita GFCE, whether the candidate has rumor of violence or not. This findings suggest, in line with my theoretical predictions, regional capacity of providing public goods matters for poor voters’ voting decision when they encounter possible local benefit from the foreign aid and negative attribute of the candidate.

Next, I test other alternative explanations. One possibly argues that the pattern I revealed is not because of the difference in the income level, but because of the difference in the level of political interest, or sophistication. For example, on the one hand, respondents with the lower level of interest and knowledge in politics may not know seriousness of the electoral violence, and therefore, they might be more tolerant of violent candidate. On the other hand, some previous works find that the more informed voters are more likely to be the target of the political violence by politicians (von Borzyskowski and Kuhn 2020), and victims of political violence tend to support the violent candidate more (Gutiérrez-Romero and LeBas 2020). To test these possibilities, I divide samples according to residential area, the level of political interests, and the level of political sophistication. I measure the level of political sophistication by asking respondents three political questions related to the minimum age for voter registration, the number of members in the House of Representatives, and the presidential term. Then, I run logistic regression to see whether there appears similar pattern as I found in the previous section. As you can see from Figure H3 and Figure H4 in Appendix H, I find no evidence for this alternative explanations. Neither non-NCR residents who are more interested (or less interested) in the politics nor who are politically less sophisticated (or more sophisticated) tend to vote for the violent candidate when they receive foreign aid treatment. These results suggest that the main findings are robust to other possible alternative explanations.

I also test whether the findings are consistent with another common measure of one’s economic status, level of wealth. To capture this another dimension of economic well-being of the voters, I ask respondents to indicate whether they have the goods listed as follows: laptop, smartphone, a car, a house which you have finished paying for, and a house which you are paying for. This survey questionnaire is originally from World Value Survey Wave 7 (Haerpfer et al. 2020), and I modify it to appropriately reflect the context of the Philippines. I code the respondent to have house if she answered to have either a house which she finished paying for or a house which she is paying for, and calculate the total number of goods that respondents own. Then, after dividing the samples by their assigned treatment, residential area, and the number of owned goods, I run logistic regression. As you can see from the Figure H5 in Appendix H, I find that only non-NCR residents who own less than three listed goods are more likely to voting for a candidate who has offered foreign aid projects to her constituency, regardless of her alleged electoral violence. This results confirm that the main findings are not driven by specific measure of economic status of the voters. Rather, this supports my claim that the poverty and the level of local public goods matter for the effect of the foreign aid on the violent candidate’s electoral performance.

Lastly, one of direct implications from the theory is that there would be more electoral violence-related incidents in non-NCR than in NCR since poor voters in non-NCR are more willing to vote for violent candidate in the face of development projects. To see whether this is true, I calculate the number of electoral violence-related incidents per 1,000,000 people by regions for the 2007 and 2010 elections in the Philippines, using the Electoral Contention and Violence (ECAV) dataset (Daxecker, Amicarelli, and Jung 2019). In the 2007 election, the number of incidents per 1,000,000 people was 0.29 in NCR while the average number of incidents per 1,000,000 people in other regions was 0.34. In the 2010 election, the number of incidents per 1,000,000 people in NCR decreased to 0.22 while the average number of incidents per 1,000,000 people in other regions increased to 0.44, a figure that is twice bigger than that of NCR. Although this is simply descriptive comparison of the numbers without taking account for the magnitude of each incident, these figures show electoral violence-related incidents are more pervasive in the non-NCR than NCR.

Discussion

The main finding of this research is that foreign aid can increase the probability of voting for violent candidate in the region where local public goods are scarce and voters are poor. However, it is worth discussing the another finding from this research: in general, Filipino voters do punish violent candidate. As one can see from the both Figure 1 and Figure 2, once respondents got to know that the hypothetical candidate has rumor of using electoral violence, probability of voting for her drastically decreased. This result is in line with research illustrates Kenyan voters who expect greater benefits from the candidate (i.e., co-ethnic or co-partisan) still sanctions violent candidate in general (Gutiérrez-Romero and LeBas 2020). In fact, my finding illustrates this sanctioning effect is greater than the positive effect of foreign aid. Therefore, this finding implies that the violent candidate will have difficulty in gaining votes, which seems contradictory to reality, considering a lots of criminal or violent candidates keep winning seats in many of developing countries.

Although this finding does have implications on the reality, I believe this results in part might stem from the limitations of my experimental designs, especially related to external validity. I believe social desirability bias is clearly the one which results in making the effect of electoral violence treatment overly strong. In other words, despite a room for violent candidate to gain support from the voters in real world, my experiment might have muted this possibility. As I explained, I intentionally choose the rumor of ordering the assassination as a key message in the vignette because the main purpose of the research is to test whether foreign aid can have positive effect on the probability of voting for violent candidate. Since ordering the murder of a rival politician is the strongest allegation among various forms of electoral violence, I expected this setting would offer conservative test of my hypotheses. However, there was a trade-off in this strategy. On the one hand, I find that poor voters living in non-NCR will vote for foreign aid, regardless of candidate’s alleged electoral violence, and this implies in reality, they might be more willing to vote for violent or criminal candidate if she brings local benefits into the district. On the other hand, however, my finding also suggests that most respondents do not vote for violent candidate. Unfortunately, within current experimental design, I cannot further analyze to what extent the results reflect the reality, and to what extent the social desirability bias plays a role. Therefore, I expect more elaborated future research that can investigate how the effect of foreign aid can differ by different levels of violence or crime, and that is more realistic.

Despite these limitations, there are anecdotal evidences suggesting the main findings of this research might explain electoral success of violent candidates in the Philippines. For example, Carmen Loreto-Cari had served as 3-term congresswoman at Leyte’s 5^th congressional district from 2010 to 2019 despite suspicion of masterminding for an assassination attempt against a rival politician in 2013. Leyte is one of poor provinces in the Philippines, whose economy is heavily dependent on farming and fishing. From 2014 to 2018, 5^th district of Leyte had received 367 sub-projects of Kalahi-NCDDP worth 7.2 million dollars, which are the greatest amount among other four districts within Leyte. Local impacts of Kalahi-NCDDP was substantial during the period as well. For example, in 2015, electricity was first introduced in Himokilan, one of the villages (barangays) in 5^th district of Leyte, thanks to Kalahi-NCDDP project.18 Also, in the other village of 5^th district, Casulongan, health stations, constructed with Kalahi-NCDDP funds in 2012, have been an important bases in providing essential health services to residents. Especially, they have played a significant role in response to Super Typoon Haiyan in 2013 and to Covid-19 pandemic recently.19 In 2016 election, Loreto-Cari won a landslide victory with 91.8% of popular votes, and her grandson, Cari Nicolas Cellona Cari, successfully succeeded her seat in 2019.

Conclusion

In this paper, I address determinants of voting choice for the violent candidate in terms of foreign aid, economic status, and residential area. Rather than focusing on individualistic benefits that candidate might offer or voters’ economic status alone, my framework reveals how local benefits that a candidate might have no control influences voters differently by their economic status and residential area. I show violent candidate has incentives to claim credit for foreign aid, and poor voters living in regions where public goods are underprovided are more likely to cast a vote for her. I test the theory through the use of a survey experiment with a nationally representative sample of the Philippines. I find strong evidence that voters, who are living in non-NCR and have lower household income, are more likely to vote for a violence-related incumbent politician whose district was selected as foreign aid projects before. Further, I also demonstrate that these findings are robust to possible alternative explanations and measurements.

This work makes several contributions to the study of electoral violence and foreign aid. While numerous scholars have examined how clientelism or vote-buying affects the electoral success of the candidates with criminal or violent backgrounds, little work has focused on how local benefit or foreign aid affects them. I find evidence that foreign aid can provide opportunities for violent politicians to mitigate their negative characteristics and take credit for basic public goods and social services flowing into their district. Thus, the findings suggest that foreign aid can unintentionally lead to the electoral success of the violent legislators in developing countries.

In some developing country, poor voters punish candidate who promises to offer targeted goods such as bags of sugar and K500 bills while they reward candidate who promises to offer ‘community benefit’ such as schools and improved healthcare (Kao et al. 2017). In addition, some research also finds that citizens of recipients favor foreign aid projects than government programs, especially when they perceive the government as corrupt or clientelist (Milner et al. 2016). Therefore, these findings of previous research that the effect of foreign can be greater than typical tools of vote-buying used by criminal or violent politicians in developing world. In a similar vein, the findings of my research suggest that the success of violent candidates is likely to be driven by the presence of voters who are less likely to sanction violence, or who are more responsive to local benefits. In my experiment, they were poor voters living in non-NCR, who are the main beneficiaries of basic local public goods and social services that community-driven foreign aid project usually provides. In contrast to previous works focused on voters’ ethnicity or income level alone, this work implies there needs additional consideration, such as regional capacity for providing public goods, to analyze determinants of voting decision for the violent candidates in developing countries.

Overall, this research contributes to deepening our understanding of why violence-related politicians in developing countries keep gaining seats, and why voters continuously elect them. My experimental design reflects the trade-off that ordinary voters in developing countries would encounter every election, and therefore, I could find which type of voters are more willing to sacrifice non-violence for local benefits. This work also contributes to the study of electoral violence by providing a straightforward framework that future research might utilize. Since the evidence clearly shows voters are willing to vote for the violent candidate if she gives what they need, future research needs to find which types of voters demand what kinds of benefits. In my case, I start from the proposition that poor voters residing in non-NCR demand basic public goods and social services, and foreign aid can offer these local benefits. I believe future research can bear fruit by asking similar questions: for examples, which types of voters demand redistributive policies or foreign direct investment more.

Tae Gyoon Lim (Ph.D. in Political Science at Harvard University and Master of Political Science at Korea University), Sung Eun Kim (Professor of Political Science & International Relations, Korea University)

Kyu Teg Lim (Senior Researcher, Peace & Democracy Institute)

Abstract

This paper suggests that the widespread embrace of the US dollar as a key reserve currency does not adequately represent the global role of the US dollar. The paper critiques the Triffin Dilemma by revealing an inadequate relationship between the US dollar and US external deficits. The economic literature has misled us into a narrow characterization of the US dollar as a reserve currency primarily due to inattention to the US dollar’s external role as money of account. The real issue of the Triffin Dilemma is neither persistent US deficits nor the reserve role of the US dollar, but the US dollar’s exorbitant privilege outside the United States. This paper develops two key arguments to contribute to the study of international monetary power. First, the dynamic process of creating various forms of dollar debt outside of the United States has institutionalized the US dollar’s infrastructural power as money of account. Second, the US monetary power is structurally consolidated by the same process of financial globalization.

Introduction

The 2008 financial crisis, although not directly caused by the present global monetary system, stimulated a vigorous discussion among policymakers and economists on the problem of the monetary system. Generally, the present monetary system is viewed as fundamentally “unbalanced” and excessively dependent on the reserve role of the US dollar. Governor of the People’s Bank of China invoked Triffin Dilemma to point out a contradiction that, on the one hand, the US dollar cannot hold its value as a reserve currency, on the other, it is acting as a global currency, providing international liquidity to the world (Zhou 2009). He indicates that the US dollar’s international dominance is an inherent flaw of the post-Bretton Woods monetary system. Many scholars have argued that excessive dependence on the US dollar for financing world trade and reserves creates an unstable monetary system, and therefore the present system needs to develop into a multicurrency system that would be “better suited to a multipolar world economy” (Subacchi 2010: 667; Stiglitz 2010; Ly 2012).

The post-Bretton Woods system operating on the US dollar’s centrality does not appear inherently problematic and flawed. Dooley et al. (2004, 2009) argued that the post-Bretton Woods monetary system continues to be operational despite global imbalances because the economic rise of the East Asian region from the late 1990s re-established the Bretton Woods II regime wherein East Asian countries accumulated US dollar assets, and that they are “willing” to accumulate dollar assets as a secure way of managing exchange rates and domestic financial markets (2004: 307). Simply put, as long as they are willing to underwrite US deficits, Bretton Woods II is sustainable and has been reinforced by the recent financial crisis (Dooley et al 2009).

Despite the economic debate on the post-Bretton Wood monetary system, a fundamental feature of the present system is that the US dollar reserve role is shared. The US dollar’s current role as a key reserve currency is widely accepted in the economic literature and scholarship of International Political Economy (IPE). The US dollar certainly appears to be a characteristic of a reserve currency in the form of US dollar assets accumulated in foreign countries. The US dollar reserve role assumes that the US economy runs current account deficits to provide international liquidity to the world economy. Yet, the US borrows money from abroad to finance current account deficits (Bernanke 2005; Cohen 2006; Li 2008; Eichengreen 2011; Prasad 2014). The Triffin Dilemma is often reflected in the Chinese Renminbi (RMB) literature, according to which China should run current account deficits to make RMB internationalization successful (Yu 2012; Prasad 2014). The relationship between the reserve status of a national currency and a “real” economic feature of the issuing country – its current account deficits – is believed to be firmly established.

However, the Triffin Dilemma mistakenly assumed that US deficits produced only within the US economy can provide dollar reserves in the sphere of international trade. The world demand for dollar reserves is not entirely determined by the process of “real” economic activities such as in international trade (Altman 1961). The dynamic development of the Eurodollar market contributed to the increasing demand for dollar reserves in European countries during the 1960s (Strange 1976). Foreign actors (official and private) have produced dollar-denominated debts seen as US deficits outside of the United States, but they have not directly affected US current account deficits (Gavin 2004; Bordo & McCauley 2017). As shown in the next section, the original version of the Triffin Dilemma has more to do with the overall US balance of payment deficits than with US current account deficits. The Triffin Dilemma is essentially based on the US dollar’s importance as an international medium of exchange. Gold and US dollars are used as the medium of exchange in the sphere of international trade.

The Triffin Dilemma does not help us grapple with the US dollar’s role outside the United States. As will be shown in Section 3, the neglect of the US dollar’s external role has led economists to develop a distorted understanding of the exorbitant privileges of the US dollar. Without a conceptual understanding of the external operation of the US dollar, the persistent US current account deficits have been primarily viewed as a consequence of the US dollar’s reserve role. The foreign holdings of US dollars are highlighted in the Triffin Dilemma. The real issue of the Triffin Dilemma is neither deteriorating US balance of payments deficits nor the US dollar’s reserve role in a “real” world economic expansion, but the increasing role of the US dollar in the process of integrating global money and financial markets. This understanding requires an analytical reengagement in how persistent US balance of payments deficits is related to the US dollar’s external role in the process of financial globalization.

The aforementioned economic debate on the present monetary system reinforces the general perception about the US dollar as an international reserve currency. It brings back the Triffin Dilemma to our analysis. The economic debate does not guide us to a better understanding of the contemporary role of the US dollar in global finance, in particular. Therefore, different questions should be raised to reveal the US dollar’s global role beyond reserve status. For instance, contrary to the Bretton Woods II regime, why are East Asian countries forced rather than willing to accumulate dollar assets in the first place? What is the US dollar’s infrastructural role that enables the United States to maintain persistent US deficits? How do we understand the US dollar’s exorbitant privileges beyond international reserve currency? Does the United States really “borrow” from abroad to finance US deficits?

The economic debate on the Triffin Dilemma must be revisited to understand what led to an inadequate relationship between the US dollar and US deficits and answer the above questions. Revealing the limitations of the dollar’s economic analysis is crucial to reconsider the US dollar’s role beyond a reserve currency in a conceptual and empirical sense. That is, the primary feature of the US dollar’s reserve status is insufficient to understand its global role. This paper attempts to contribute to the study of international monetary power in the IPE literature, building on a critique of the Triffin Dilemma. While the economic literature does not help us to grapple fully with the role of the US dollar in global finance, the IPE literature has mainly discussed the concept of international monetary power within inter-state terms (Andrew 1994; Henning 2006; Kirshner 1995, 2006; Cohen 1998, 2006; Kirshner & Helleiner 2014).

Moreover, the IPE literature has not fully succeeded in incorporating financial globalization into the study of international monetary power. Braun et al. (2020: 4) recently pointed out that IPE literature of financial globalization has not appreciated the institutionalized processes of international money. This paper attempts to reveal the institutionalization of the US dollar in global finance.

An overarching argument developed in this paper is that the US dollar’s exorbitant privileges should be reconsidered with respect to the US dollar’s external role as money of account. This paper presents two sets of arguments to substantiate this overall argument. First, the contemporary pattern of financial globalization institutionalizes the infrastructural role of the US dollar as the money of account. In particular, foreign actors (official and private) have actively used the US dollar as money of account to create various forms of debt in the offshore market. Thus, the US dollar’s contemporary role needs to be adequately understood regarding the creation of offshore debts and offshore dollars beyond “real” world economic processes. Second, the dynamic process of issuing dollar-denominated foreign debts has extended the role of the US Federal Reserve as a world monetary authority, which has played an essential role in the making of global financial markets.

To develop these arguments and contribute to the IPE literature, sections 2 and 3 provide foundational reasons for limitations of the economic analysis on the US dollar. These two sections help develop arguments in sections 4 and 5. Section 2 aims to highlight an inadequate relationship between the US dollar and the US (external) deficits, as it explores the Triffin Dilemma debate. Section 3 further provides the inadequacy of the economic analysis of the US dollar’s exorbitant privileges. Section 4 focuses on the linkages between the process of financial globalization and the institutionalization of the US dollar. Section 5 is closely intertwined with this financial and monetary dynamic process; it characterizes US monetary power in four ways that distinguish this paper from the existing IPE literature on the role of the United States in global financial market. The final section summarizes the main arguments and provides brief implications.

THE TRIFFIN DILEMMA DEBATE AND LIMITATIONS

By exploring the Triffin Dilemma economic debate, this section shows that the relationship between the US dollar and the US (external) deficits is not adequately understood. The Triffin Dilemma debate is largely grounded on the shared, narrow conceptualization of the US dollar as the international medium of exchange. While the narrow understanding of the US dollar led Robert Triffin to mischaracterize deteriorating US deficits as dangerous, others did not pay due attention to the US dollar’s external role as money of account, which produces external dollar claims misconstrued as US deficits. Therefore, persistent US deficits are assumed to be primarily produced by the US dollar’s reserve role in “real” economic processes such as world trade – the current account version of the Triffin Dilemma.

The Triffin Dilemma debate reveals different understandings of US deficits, the US dollar, and the Bretton Woods system. Robert Triffin viewed deteriorating US deficits as dangerous to the US dollar and the Bretton Woods system. Triffin (1960) stressed that there were two primary forms of international liquidity under the Bretton Woods system: gold and US dollars. Gold production was insufficient, while US dollars were supplied through US balance of payments deficits (1960: ix). In his view, the present Bretton Woods system did not provide a sufficient level of international liquidity required for financing an expanding world trade (ibid: 40, 47). Triffin saw this as dangerous as the liquidity gap, which was increasingly filled in by foreign accumulation of US dollars, deepening the US balance of payments deficits. As a result, the Bretton Woods system created the famous double Triffin Dilemma: 1) if the US economy continued to provide US dollars for expanding world trade, the supply of dollars would force the overall US balance of payments to deteriorate, undermining confidence in the US dollar; and 2) if the United States attempted to eliminate overall US balance of payments deficits by refusing to provide US dollars, the rest of the world would face a lack of international liquidity, eventually turning the world economy deflationary, as experienced in the 1930s (Triffin 1960: 9, 67).

Other economists held different views on the US deficits, the US dollar, and the Bretton Woods system. Jacques Rueff (1972) regarded growing US deficits as an exorbitant privilege of the US economy. In his view, US deficits were not real deficits that the US should pay back to foreign holders, but a double monetary phenomenon of the US dollar’s reserve role. US dollars paid to foreign creditors in the form of monetary reserves (private and official) kept returning to the United States. Thus, the US government did not have to settle its balance of payments deficits with other countries (1972: 23).

Similarly, Charles Kindleberger (1965) understood US balance of payments deficits as an outcome of the banking role of the US economy borrowing short and lending long. He noted that the US balance of payments deficits was confused by “a mistaken definition of balance-of-payments disequilibrium” (1965: 3). This misunderstanding of the national balance of payment disequilibrium came from “confusion between capital markets for transferring ‘real’ assets and money markets for accommodating national liquidity preferences” (ibid: 4). In other words, US balance of payments deficits cannot be understood as an outcome of “real” economic processes in which US dollars are directly exchanged for goods. US deficits cannot be adequately understood from the conception of the national balance of payments, which conceptualizes deficit as disequilibrium (Despres et al. 1996). Rather, US deficits represent a provision of the US economy’s banking role with a large and open capital market that European countries lacked (1996: 210). US deficits were compatible with international disequilibrium.

Furthermore, Oscar Altman (1961) argued that the US balance of payments deficits seen as dangerous and therefore to be eliminated misled Triffin to exaggerate the danger of the Bretton Woods system (162–163). Altman found it even more troubling that an adequate level of international reserves was required to finance the growth of world trade. This belief was not valid because the demand for reserves was the result of various factors, such as banks’ exchange holdings (Altman 1961), banks’ arbitrage of interest rates (Schenk 1998), and the 1958 currency convertibility (IMF 1959). During the 1960s, the offshore money market’s rapid expansion forced countries to accumulate US dollar reserves (Strange 1976). Altman stressed that Triffin’s view on reserves’ requirement was “essentially a mechanistic one based on a rough relationship of reserves to imports” (1961:164). There is no reason to believe that expanding world trade would dictate the world demand for dollar reserves.

Less discussed, however, is the limited constructive engagement on the role of the US dollar beyond its reserve currency, whose production remains only within the US economy. Indeed, the Triffin Dilemma is essentially rooted in the key feature of the US dollar as a safe international medium of exchange in “real” world economic processes. Expanding world trade requires an international medium of exchange – gold and US dollars – to finance international trade transactions across borders. It thus becomes clear why Triffin feared the Bretton Woods system was reaching a dangerous point if foreign holdings of US dollars exceeded US gold stock in the mid-1960s. Viewing this quantitative gap as a precarious point indicates that the US dollar was understood as a neutral veil in commodities’ exchange ratios.

Kindleberger’s focus on the US economy’s banking role does not allow us to conceptualize the US dollar’s external role outside the United States. The emphasis on US-based banks as liquidity providers does not have room for foreign actors abroad in producing dollar claims. Altman (1963) indicated that foreign accumulation of dollar reserves is relatively independent of international trade expansion. Still, he does not go beyond the conceptualization of the US dollar outside the United States. Rueff correctly understands US deficits as exorbitant privileges but essentially shares the US dollar reserve role with Triffin. As Kindleberger (1965) and Despres et al. (1996) recognized, US deficits cannot be adequately understood by the economic concepts of the national balance of payments. More specifically, growing US deficits cannot be understood only by the reserve role of the US dollar as a safe international medium of exchange and store of value. The reserve status of the US dollar prevents us from viewing how the US dollar’s external role operates to produce external dollar claims often misinterpreted as “US deficits” outside the United States. That is, dollar claims produced outside the United States cannot be distinguished from domestic dollar claims on US banks. They may look like US debts but are not necessarily dollar claims on the United States (Mayer 1980: 68; Gavin 2002:128). During the Bretton Woods era, deteriorating US deficits can be seen as a sign of the US dollar’s increasing role in integrating money and capital markets of the world economy.

Persistent US deficits, deeply rooted in the narrow conceptualization of the US dollar as reserve status, established the popular version of the Triffin Dilemma – US current account deficits. The US dollar reserve role in “real” world economic processes subsequently turned the original version of the Triffin Dilemma into the current account version. As discussed above, the original version of the Triffin Dilemma has more to do with the overall US balance of payments deficits and less to do with US current account deficits. Triffin worried that the competitiveness of the real US economy, the exporting sector, would be undermined by the foreign accumulation of dollars, limiting the US autonomous monetary policies during the 1950s. Witnessing gold outflows in 1958, when US interest rates fell below those available in Europe, Triffin thought that foreign accumulation of dollars undermined US autonomous economic policies, such as low-interest rates (1960: 9).

The relationship between foreign accumulation of dollars and its constraint on US autonomous monetary policy, in Triffin’s view, is a partial feature of the US economy whose money was fixed to the price of gold per ounce at 35 dollars under the Bretton Woods system. Specifically, the European economy’s recovery was primarily a result of the successful operation of the European Payments Union (EPU), which enabled EPU members to overcome bilateral trade and payments relationships from the late 1940s until 1958 (Eichengreen 1993). The EPU’s operation did not depend on significant dollar accumulation but on the US dollar’s abstract role as money of account for settling claims and debts of EPU members (Kaplan & Schleiminger 1989: 92; Amato & Fantacci 2012). Empirically, US current account deficits appeared to be systematic only after the early 1980s (Cohen 2011; Bordo & McCauley 2017). Since then, the persistent US current account deficits confirm the current account version of the Triffin Dilemma. However, in the post-Bretton Woods era, US monetary policies have been more autonomous than any other surplus countries (Cohen 2016). The argument that foreign holdings of US dollars would impose limitations on US monetary policies’ autonomy in the post-Bretton Woods era is far less convincing. Instead, the famous Fleming-Mundell model, which indicates the incompatibility of three economic policies at a time, does not apply to the United States in the post-Bretton Woods era (Winecoff 2014).

Furthermore, the current account version of the Triffin Dilemma, stemming from the deterministic relationship between the demand for dollar reserves and world trade, ignores the dynamic process of offshore dollar creation during the Bretton Woods system. An active process of creating credit and debt relations denominated in the US dollar in the Eurodollar market in the 1960s (Higonnet 1985; Schenk 1998; Burn 1999). From 1955, banks in London began to attract dollar deposits through interest rate arbitrage (Schenk 1998). For the Eurodollar market to be further developed, this initial monetary accumulation needed monetary dynamics such as the 1958 currency convertibility and European central banks’ active involvement. The 1958 currency convertibility encouraged private actors to sell foreign earnings to foreign exchange markets for better rates or deposit them in the Eurodollar market (IMF 1959:125). In these favorable conditions, private banks alone increased dramatic dollar holdings, about $1.4 billion outside the United States in 1959 (IMF 1960: 5). As the increase in private holdings of foreign exchange could stimulate massive shifts in their distribution, with far-reaching effects on the execution of monetary management policies at various central banks (BIS 1959: 60), the 1958 monetary event meant that central banks became actively involved in the Eurodollar market. From the early 1960s, European central banks began to place dollars with banks operating in the Eurodollar market (Altman 1963: 57; BIS 1964: 132). European states and European banks began to issue various debts denominated in the US dollar as money of account outside the United States (Swoboda 1968). With their active involvement in the Eurodollar market, European central banks helped to secure the practice of Eurodollar businesses (Braun et al. 2020: 15).

THE REAL ISSUE OF THE TRIFFIN DILEMMA

Distorted exorbitant privileges

The real issue of the Triffin Dilemma is, therefore, neither deteriorating US balance of payments deficits nor the US dollar’s reserve role in a real-world economic expansion. It is the US dollar’s increasing role as money of account in the process of integrating global money and financial markets that is the crux of the Triffin Dilemma. More specifically, it requires us to see how persistent US deficits are related to the US dollar’s external role in the process of financial globalization. This crucial question will be answered in sections 4 and 5.

Here again, the US dollar’s economic analysis has led to a distorted understanding of the currency’s exorbitant privilege as modest economic benefits. The economic literature has not advanced our understanding of the US dollar’s contemporary role in global finance. The typical economic calculus of the benefits and the costs of international currency suggests negligible benefits accruing to the issuing country (Dobbs et al. 2009; Bergsten 2009; Hai & Yao 2010; Krugman 2013 McCauley 2015; Bernanke 2016). The domestic version of national seigniorage, the difference between the value of money and its production costs (Cohen 1971: 495), is extended to the international level. The status of international currency offers economic benefits to the issuing state and domestic actors.

There are two primary forms of economic benefits: foreign holdings of national currencies as interest-free loans to the issuing country and foreign holdings of financial assets as low capital costs for domestic actors. Paul Krugman (2013) noted that international seigniorage of the US dollar provides insignificant economic benefits to the US economy. Richard Dobbs et al. (2009: 8) showed that in a “normal” year, the revenue from the dollar’s reserve status, interest-free loans to non-residents holding US notes and coins is estimated at $10 billion. The lower capital costs for US domestic actors, due to foreign actors’ large purchases of US Treasury securities, are estimated at $90 billion (2009: 8). However, these financial benefits are offset by capital inflows that elevate the dollar’s value, causing a net financial cost of $30–60 billion in 2008 (ibid: 8–9). Robert McCauley focused on the US dollar’s economic benefits. He concluded that the US dollar’s exorbitant privileges are not unique but an exaggeration (2005: 11). More seriously, Fred Bergsten argued that US dollar dominance is increasingly costly to the US economy and no longer in favor of its national interests. The US capital inflows create financial risks such as the 2008 financial crisis. The US dollar’s reserve role enables other countries to manipulate their exchange rates to undermine US exports’ competitiveness (2009: 23–24). Thus, US dollar dominance no longer confers an exorbitant privilege on the United States.

Benjamin Cohen criticized the economic analysis of the benefits and costs of the US dollar. The empirical calculus of the benefits and costs of the dollar suffers from a limited range of data. It narrowly focuses on the cost of exchange-rate appreciation or transaction costs (2011: 25). He noted that the economic literature completely ignores the US dollar’s exorbitant privilege, which often translates into international monetary power in geopolitics. The IPE scholars have been interested in this aspect (Strange 1976, Kirshner 1995, Andrew 1994; Henning 2006; Cohen 2006). Cohen also argued that the economic analysis of the US dollar’s role for decades had distorted our understanding of its contemporary role (2011). Nonetheless, Cohen agreed that the US dollar’s economic benefits and costs are “rather less obvious” (2011: 4). He suggested that the economic benefit of running an international currency is negligible; it imposes significant financial burdens (16–20).

The US dollar and the absence of money of account

The widespread embrace of the US dollar’s contemporary role as a reserve currency prevents us from analyzing its significant role in today’s global money and financial markets. The economic literature overlooks the significance of the US dollar’s role outside the country as money of account to produce various forms of dollar debt since the inception of the Eurodollar market in the late 1950s. It is well known that the expansion of the Eurodollar market has contributed to the rapid development of global finance (Burn 1999, 2006; Frieden 1987; Helleiner 1994; Schenk 1998). The offshore money market amplified the process of globalizing the US dollar during the Bretton Woods system (He & McCauley 2010). However, the offshore dollar has been largely conceptualized as forms of the dollar, such as “stateless” dollars or private dollars (Frieden 1987; Burn 1996; Schenk 1998); that is, the US dollar has been seen as a medium of exchange rather than money of account.

Swoboda (1968) recognized the significance of abstract dollar denomination practiced by European banks operating in the Eurodollar market. He emphasized that European banks issued dollar liabilities to “bid away part of the gains from dollar denomination that previously accrued exclusively to American banks” (ibid: 14, emphasis added). He regarded the European practice as “[dollar] denomination rents.” This important recognition of the US dollar as an abstract monetary denomination by foreign banks is lost in the economic literature. It does not attract adequate attention in IPE scholarship either. Rather, the US dollar’s significant role offshore is misplaced onto the US domestic market from which only US domestic actors can benefit (Cohen 2011: 15; Eichengreen 2011). The economic analysis has reduced the US dollar’s external role into its forms, such as US currencies within the US economy.

In a fundamental sense, what constitutes a large proportion of today’s foreign exchange reserves is not US currencies such as US notes and coins, but US dollar assets such as US Treasury Securities denominated in the US dollar as money of account. In a traditional sense, the economic literature does not have adequate room for conceptualizing money as “money of account.” It restricts money’s role to the so-called “unit of account” to price goods and services rather than measure credit and debt relations (Ingham 2004). There is no consensus in the precise meaning of unit of account in the economic literature (see Chinn & Frankel 2005 vs. Ito 2011). More specifically, the economic literature recognizes the role of the monetary denomination as a “unit of account” when issuing various forms of financial assets in the first place. But then, as they place those financial assets in a functional category as a store of value (e.g., Cohen 1998; Schenk 2012), the significance of money of account, which allows the value of the financial assets to be constructed, is lost in most economic analyses. Economists significantly recognize the noticeable feature of financial assets rather than money’s abstract role, which denominates those financial assets. A few economists have recently begun to recognize the significance of money as a unit of account in the analysis of RMB internationalization in “invoicing” trade transactions (Yu 2012) or denominating financial assets in global financial markets (Park 2010).

However, it is not obvious that forms of money such as national currencies are themselves “units of account” because the economic concept of a unit of account is believed to originate from the commodity theory of money (Schumpeter 1954; Ingham 2004). Money as money of account has historically developed independently from the evolution of monetary forms such as gold and paper (Ingham 1996; Wray 1998). Money as money of account acts as a means of constructing price lists and debt contracts (Keynes 1930: 3). Therefore, the meaning of money as money of account can be seen as broader than the economic concept of a unit of account in a way that money of account denominates not only forms of money such as currencies and bank credit money but also financial assets such as Treasury Securities and corporate bonds (Ingham 2004).The monetary denomination of financial assets should be categorically recognized as money of account rather than a store of value. The significance of the US dollar as money of account should be reconsidered with respect to the offshore market where various forms of dollar debt are produced.

THE INSTITUTIONALIZATION OF THE US DOLLAR

The offshore market has institutionalized the infrastructural role of the US dollar as money of account in financial globalization. The expansion of the Eurodollar market in the post-Bretton Woods era has been based on more transactions between foreign depositors and foreign borrowers outside of the United States than those between US depositors and non-US borrowers (Borio & Disyatat 2011; Shin 2011), except for the period between 2000 and 2007 (He & McCauley 2012). Foreign actors have played an important role in the expansion of offshore transactions across borders. For instance, “only $2.3 trillion out of $8.6 trillion in dollar claims on non-banks outside of the United States were held in the US” in 2013 (McCauley et al. 2015a: 5). The expansion of external dollar claims indicates the embedded practice of using the US dollar as money of account by non-US actors abroad. In the post-2008 financial crisis, a number of scholars have stressed the importance of offshore money creation (Avdijev et al. 2015; Mehrling 2016; Murau 2018).

Foreign actors have actively practiced the US dollar as money of account to issue various forms of debt. Outside of the United States, supranational agencies, such as the World Bank, Asian Development Bank, and the Inter-American Development Bank, have often issued financial debts denominated in the US dollar (Bordo & McCauley 2017: 26). The Asian Development Bank (ADB) has recently issued a large number of global bonds denominated in the US dollar as money of account, which was cleared through the Federal Reserve Banks and listed on the Luxembourg Stock Exchange (ADB 2020). ADB has consistently practiced the US dollar in the global bond market since 1994. The global government bond market expanded from $14 trillion in 1993 to about $19 trillion in 2000. While advanced countries dominated the global market by issuing sovereign bonds denominated in their own money, developing countries in Latin America have increasingly issued sovereign debts denominated in the US dollar (Claessens et al. 2002: 8–10). Since the recent financial crisis, a surge of new sovereign dollar debt issuance in global bond markets has occurred. Many countries across the world – Ghana, Pakistan, Sri Lanka, Zambia, Ethiopia, and Angola – issued sovereign debts in the US dollar between 2014 and 2015 (McCauley 2015b: 34). Emerging economies, such as Turkey and the Philippines in particular, have issued dollar-denominated sovereign debts since 2009 (ibid: 36). From 2017, Chinese state agencies like the finance ministry began to issue dollar-denominated sovereign debts offshore in London (Ding et al. 2019: 358). In 2019, the Korean government issued dollar-denominated sovereign debt of $1.6 billion.

Non-government actors such as corporations and banks in the developing world have also practiced the US dollar to issue offshore bonds. The Asia-Pacific bond market is characterized by a monetary denomination distinction between onshore issuance entirely in domestic money and offshore issuance mostly in foreign money. Firms from Australia, Hong Kong, New Zealand, the Philippines, and Singapore often used the US dollar for issuing offshore bonds between 1994 and 2008 (Black & Munro 2009: 100). In particular, firms in China, India, Brazil, Russia, and South Africa used offshore subsidiaries to issue dollar bonds outside their domestic markets (McCauley 2015b: 36). Chinese state-owned enterprises significantly contributed to the growing offshore corporate bond market in the post-financial crisis from nearly zero in the mid-2000s to $500 billion by 2016. In short, sovereign and private actors in the developing world have actively issued various dollar debts outside their domestic markets.

The process of creating foreign debts (official and private) entails three intertwined monetary dynamics, which empower the US dollar as world money. First, it expands the infrastructural role of the US dollar as money of account practiced in issuing debts. Foreign dollar debts issued outside of the United States are often exchangeable for US domestic dollar debts such as US Treasury Securities and certificates of deposit produced within the United States (Christelow 1966: 228; Mendershon 1980: 182) because they both share the dollar denomination. The production of offshore and onshore dollar debts institutionalizes the infrastructural power of the dollar across borders. Second, financing of dollar-denominated debt contracts requires lenders to provide forms of the US dollar such as US currencies. Similarly, borrowers need to repay debts with US dollars. Thus, the ongoing process of settling dollar debt contracts reinforces lenders and borrowers to seek US dollars. Typically, the payment of offshore dollar loans and dollar bonds are made in US dollars (Mendelsohn 1980; Avdjiev et al. 2015: 15). Consequently, the initial issuance of foreign dollar debts increases the US dollar’s role as world means of payment. The institutional practice of the US dollar has become closely linked to the inherent monetary dynamics of financial globalization (Amato & Fantacci 2012).

The infrastructural role of the US dollar as money of account facilitates cross-border transactions. Within a currency block, the use of a dominant currency reduces transaction costs (Cohen 1998: 72; Helleiner 2005: 7). In contemporary global money and financial markets, the US dollar as money of account has reduced transaction costs. For example, Chinese importers and exporters use the US dollar to denominate the value of goods and services for cross-border transactions. They can subsequently use Chinese currencies to settle cross-border transactions to reduce transaction costs (Yu 2012: 16). Korean corporations and banks have actively used the US dollar to denominate the value of goods and services for cross-border transactions to reduce exchange and currency risks (Eichengreen 2011: 168). The European Union members like France and Germany used the dollar to denominate about 30 percent of exports in 2008 (Fields & Vernengo 2013: 750). Park (2010) argued that the denomination of financial assets rather than trade invoicing requires extensive use of international money. In other words, the role of money as money of account becomes more important in contemporary financial markets. Avdjiev et al. characterized the dominance of the US dollar as “the global unit of account in debt contracts” (2015: 15). The institutionalized practice of the US dollar as money of account across borders allows the US dollar to facilitate cross-border transactions.

Foreign actors, however, become more dependent on US dollars as means of payment as they continue to issue their debts in the US dollar. Ironically, US dollars and US treasuries become more valuable for foreign banks and foreign states during financial crises. Indeed, the foreign recognition of US Treasury Securities as “risk”-free assets can be traced back to the end of the Bretton Woods era in the early 1970s (Garbade 2007; Sarai 2008). The Volcker Shock in the early 1980s pulled an enormous amount of foreign capital into the US Treasury and mortgage debt markets (Greider 1987: 561–563; Krippner 2011: 101; Schwartz 2014: 70). The 1997–98 Asian financial crisis led East Asian countries like South Korea to accumulate US treasuries dramatically to secure domestic financial markets (Duncan 2005: 106; Stiglitz 2010: 161). In the 2008 financial crisis, during which the value of US mortgage bonds fell, financial actors were forced to “bid aggressively for dollars to repay their dollar debts, pushing the dollar’s value in the process” (Avdjiev et al. 2015: 4, emphasis added). The “survival” of those banks and relevant actors that purchased or issued dollar securities becomes largely dependent on their access to Federal Reserve money, which is the most transferable of all dollars. But as will be discussed below, the access to Federal Reserve money depends on how the Federal Reserve accepts financial assets as collateral. Regardless of the narrow discussion of the economic benefits and costs of US currencies, the institutionalization of the US dollar as money of account enforces foreign actors to seek US dollars and US treasuries as safe assets.

US MONETARY POWER

Thus, despite the growing volume of international private capital flows, the dynamic process of creating dollar assets/debts and dollars offshore has necessarily extended the US Federal Reserve’s role as a world monetary authority. The new monetary power of the United States is linked to the monetary dynamics of financial globalization. It, therefore, needs to be understood beyond the sphere of geopolitics or inter-state relations popular in the IPE literature. The monetary power of the United States can be characterized in several ways. First, the US Federal Reserve as a world monetary authority has emerged since the early 1980s. Second, the United States does not face liquidity problems originating from the mismatch of banks’ liabilities and assets in terms of money of account. Therefore, it does not face default risk on sovereign debt. Third, the “burden” of running the US dollar as world money is often “shared” with other surplus countries. Finally, the United States plays a vital role in the development of global financial markets.

The US Federal Reserve emerged as a world monetary authority in the early 1980s when the original version of the Triffin Dilemma shifted to its current account version. The Volcker Shock turned the US economy into systemic current account deficits. It was precisely at this moment the unprecedented rise of the US Federal Reserve as a world monetary authority disciplined banks through the US money market and the Eurodollar market (Greider 1987; Kyuteg 2019). Volcker argued that the shift to monetary targeting was ‘the nature of uncertainty it created’ (FOMC 1980: 22). As Kirshner (1995) noted, hegemonic states could disrupt international money and financial markets. The monetary targeting of the US central bank provided little ground for the interest rate calculation of international banks during the Shock. It, therefore, forced banks to pay close attention to their monetary policymaking. The US monetary authority began to expand the provision of collateral and open privileged access to central bank money for any US-based foreign banks and financial institutions (Timberlake 1985; Guttentag & Herring 1993). The process of so-called financial deregulation from the 1980s effectively increased the power of US monetary authority over the valuation of dollar loans and dollar debts across borders. While the US economy fell into deep current account deficits, the United States could sell US debts effectively, meaning that other countries like Japan were forced to accumulate safe dollar assets such as US treasuries (Duncan 2005). US current account deficits known as the Triffin Dilemma have been a sign of the extended role of the US monetary power over global money and financial markets despite the relative decline of the US share in world trade. Similarly, reflecting on the recent financial crisis, an enormous amount of foreign capital flowed into US financial markets seeking safe financial assets such as US treasuries (Prasad 2014; Hager 2017).

Second, the United States can prevent a systemic crisis stemming from individual banks’ liquidity problems. The extraordinary power of the US dollar enables the United States and US domestic actors to issue their debts in the US dollar and draw foreign actors (official and private) into issuing their debts in the same dollar. Individual US banks can face a liquidity problem stemming from the maturity imbalance between assets and liabilities (Eichengreen 1996). Still, the US state does not face the same liquidity problem for two plausible reasons.

Firstly, unlike foreign banks, US banks are not likely to face double mismatches of assets and liabilities (maturity and money denomination). In contrast, East Asian banks faced double mismatches in the 1998 Asian crisis (Eichengreen et al. 2003). European banks faced the same double mismatches in the 2008 financial crisis (McGuire & von Peter 2009). US banks face only the maturity problem of assets and liabilities. Secondly, US banks experiencing the maturity problem are more likely to be supported by the US Federal Reserve. The US monetary authority can prevent the transformation of individual banks’ liquidity problems into a systemic financial crisis. Given this implicit support and no need to borrow abroad or issue debt in foreign money, US banks and financial institutions are exorbitantly privileged. The foreign states are not free of the liquidity problem stemming from the dollar exposure of their banks operating in global money markets. They should be concerned with their banks’ international operations (Advjiev et al. 2015; McCauley et al. 2015c).

Third, the “burden” of running the US dollar as world money has often been “shared” with other major economic powers in inter-state terms. Most economic literature points out that the economic benefits from the international currency status of the US dollar can be offset by its management costs, such as currency appreciation and even macroeconomic inflexibility (Bergsten 2009; Dobbs et al. 2009: 10; McCauley 2015). These economic ideas are based on the Fleming-Mundell thesis in a broad sense. However, the universal economic claim does not apply to the central position of the US in global money and financial markets (Winecoff 2014; Schwartz 2019). The Fleming-Mundell thesis is inadequately equipped to understand the monetary power of the United States.

Rather, the persistent US current account deficits provide the US government a solid justification to blame other surplus countries for manipulating exchange rates. Considering global imbalances as problematic and requiring adjustment toward “international equilibrium,” which is deeply rooted in money’s role as a medium of exchange in a “real” world economy, the US can easily translate that into politics of exchange rates. In 1985, the US forced Japan and Germany to appreciate the external values of their respective currencies – the Japanese yen and the German Deutschmark (Funabashi 1988). Henning (2006) argued that the US government effectively used the exchange-rate weapon to compel surplus countries like Germany and Japan until the creation of the euro and regional monetary corporations. Nonetheless, the exchange rate politicization continues to be “an instrument of economic influence” for the USA (Henning 2006: 138; Kirshner 2006). In 2005, Ben Bernanke, chairman of the Federal Reserve, argued that developing countries with trade surpluses in East Asia created a “global saving glut” that stimulated excessive lending in the United States (Bernanke 2005). Following this line of argument, in 2009, the US Treasury secretary explicitly targeted China’s manipulation of the exchange rate (Dooley et al. 2009: 3). As persistent US current account deficits enable the US government to pressure trade surplus economies, it can be claimed that the financial burden of managing the US dollar has been “shared.”

Finally, the US Federal Reserve’s role goes beyond an international lender of last resort. On governing the 2007–2009 financial crisis, the US monetary authority implemented unprecedented monetary policies while maintaining almost zero rates between 2008 and 2015 (Bernanke 2013; Bowman et al. 2013; Langley 2015). Existent IPE scholarship pays particular attention to the unprecedented extension of dollar swaps between the US Federal Reserve and other foreign monetary authorities (McDowell 2012; Chey 2012; Broz 2015; Helleiner 2016; Sahasrabudhe 2019). The US Federal Reserve’s role is extended to include foreign central banks, which lend dollars to domestic banks.

The international role of the Federal Reserve as lender of last resort does not reveal its active role in the making of global financial markets. The extension of dollar swaps needs to be understood in a context where the Federal Reserve acted as a market maker to prevent the collapse of global money markets originating from securities markets. After the collapse of Bear Stearns and Lehman Brothers, particularly in September 2008, short-term money markets such as the US Treasury repo market and onshore/offshore money markets froze up. Banks and financial institutions faced a “global shortage” of US dollars (McGuire & von Peter 2009). In particular, banks lacking access to the Federal Reserve were squeezed to bid for US dollars and drove LIBOR rates to unprecedented levels (Mehrling 2011: 121). To manage the internal and external liquidity problems, the Federal Reserve accepted a wide range of collateral.

The Federal Reserve created various liquidity programs and extended dollar swaps with foreign central banks. Immediately following the collapse of Bear Stearns in March 2008, the US monetary authority created the Primary Dealer Credit Facility (PDCF), which provided dollar liquidity directly to primary dealers in the US Treasury debt market. The US authority lent Treasury Securities against illiquid financial assets not accepted in the repo market (Mehrling 2011: 120) to help further. After the collapse of Lehman Brothers in September 2008, the Federal Reserve created the Commercial Paper Funding Facility (CPFF), which provided dollars directly to commercial paper issuers such as nonbank financial institutions. In establishing transactions between the Fed and nonbank financial institutions, the Federal Reserve used primary dealers as agents between itself and commercial paper issuers. Specifically, it aimed to restore the price of asset-backed commercial paper by purchasing illiquid asset-backed commercial paper directly from eligible issuers (Adrian et al. 2010: 6). Opening a discount window to commercial paper issuers was further extended to include US corporations and foreign banks that issued dollar-denominated assets (ibid).

More significantly, in March 2009, the Federal Reserve opened the Term Asset-Backed Securities Loan Facility (TALF), designed to provide dollar liquidity to eligible borrowers with highly rated asset-backed securities linked to small businesses and mortgages (Ashcraft 2012: 36). The Federal Reserve charged the securities at a higher rate than LIBOR and then enabled some of the eligible securities to be transacted in the financial market (Mehrling 2011: 134). In reality, the Federal Reserve’s acceptance of them as collateral at a high price merely affected dealers’ behavior, not banks. Therefore, the market price for the trading of securities was re-established. During the financial crisis, the Federal Reserve accumulated “more than $1 trillion of such securities” (ibid: 134). It moved “the wholesale money market onto its own balance sheet stepping in as a dealer of last resort” (ibid: 125).

To manage the external liquidity problem, the Federal Reserve extended dollar swaps with foreign central banks, including four emerging economies, Brazil, Mexico, South Korea, and Singapore. The selectivity of dollar swaps was meant to restore the pattern of financial globalization, which was deeply rooted in creating cross-border dollar debt. Sahasrabudhe noted that the dollar swap was “a systematic logic underlying the Fed’s choices in favor of economies that shared its policy preference for greater financial openness” (2019: 462). The four emerging economies shared a high level of financial openness as “an implicit condition to receive a Fed swap” (ibid: 467). But due to the high level of financial openness, the four emerging countries, in addition to European and Japanese countries, became more vulnerable to the dollar exposure of their banks operating internationally or foreign banks operating in those countries. While European and Japanese banks faced severe dollar funding shortages due to their active purchase of dollar-denominated assets such as mortgage-backed securities since 2000 (McGuire & von Peter 2009), emerging economies like Korea faced the instability of the domestic financial markets because foreign banks operating in Korea actively borrowed dollar-denominated assets abroad prior to the 2008 crisis. For instance, foreign banks in Korea made a large contribution to the cause of Korea’s overall short-term foreign dollar debt (Jeong 2009; Aizenman 2009). The Korean government restored the stability of the Korean won and the Korean financial market only after the announcement of dollar swaps with the United States (Park 2019), which was repeated in March 2020. Many Asian countries without access to Federal Reserve money depleted a considerable amount of foreign exchange reserves (Aizenman 2009).

Contemporary US monetary power is inherently linked to the monetary dynamics of financial globalization well beyond inter-state relations. Certainly, the United States has promoted financial liberalization to serve “US geopolitical interests” (Kirshner 2006: 16). But more importantly, the promotion of financial liberalization has contributed to the monetary capacity of the United States in a systemic way that produces dollar-denominated debts across borders. “Dollar-denominated assets of banks outside of the United States peaked at $10 trillion before the crisis, an amount equal to the total assets of the US commercial banking sector” (Shin 2012). When the US Federal Reserve operated domestic liquidity programs known as the Term Auction Facility (TAF) and the CPFF during the 2008–2009 crisis, foreign banks tapped more than half of these dollar sources through their branches in the United States. Broz (2015: 327) noted that “15 of the 30 largest borrowers at the discount window were branches or agencies of foreign banks.” Consequently, the United States is empowered in a systemic way that maintains US treasuries as “risk”-free assets. The financial crisis has led many countries to accumulate US dollar assets further to protect themselves against financial vulnerability (Aizenman et al. 2011).

CONCLUSION

This paper argues that the US dollar’s exorbitant privileges should be reconsidered with respect to the external role of the US dollar as a money of account. It shows that the Triffin Dilemma debate does not pay due attention to the US dollar’s increasing role as money of account in integrating global money and financial markets. The Triffin Dilemma is indeed grounded on the primacy of the US dollar as a safe international currency used in world trade. Therefore, Robert Triffin was misled in his understanding of US deficits as dangerous to the US dollar and the US economy. Others hold different views on US deficits but did not constructively engage in the US dollar’s external role, which produced external dollar claims seen as US deficits. The US dollar’s privilege is not adequately understood in the economic literature primarily due to a lack of conceptual understanding of money as money of account.

This paper developed two subsets of argument to characterize the US dollar’s global role beyond its reserve status. First, the contemporary pattern of financial globalization has institutionalized the infrastructural role of the US dollar as money of account in the offshore market. Foreign states and private actors have actively used the dollar as money of account to issue various forms of debt offshore. These dynamic processes of creating dollar debts make foreign actors dependent on US dollars as means of payment. The second argument is that the inherent monetary process of financial globalization has necessarily extended the US Federal Reserve’s role as a world monetary authority. The US monetary authority has played a major role in the development and growth of global financial markets, underpinned by dollar-denominated debts. Therefore, contemporary US monetary power can be better understood well beyond inter-state relations.

The post-Bretton Woods monetary system is not Bretton Woods II, which primarily focuses on the narrow relationship between surplus East Asian countries and US current account deficits in the sphere of international trade. This relationship is an updated version of the Triffin Dilemma during the Bretton Woods period. The US dollar is narrowly defined as a safe medium of exchange in world trade. As discussed in section 2, it is somewhat problematic to assume that an expanding world trade determines the demand for dollar accumulation. The US dollar reserve role explains the supply and demand of monetary relations without revealing much about the US dollar’s external role as money of account in global money and financial markets. According to the Bretton Woods II thesis, the sustainability of the post-Bretton Woods system depends on countries’ willingness to purchase US official debts. Rather, the dynamic process of dollar debt creation offshore increases the world demand for US Treasury Securities as safe assets in financial globalization. Specifically, the process of dollar debt creation in global finance enforces surplus East Asian countries to accumulate dollar assets to manage exchange rates and domestic financial markets. Thus, Bretton Woods II ignores the significance of the US dollar as money of account in global financial markets.

Contrary to the conventional IR/IPE monetary power, which was as a comparable property within inter-state relations, Susan Strange noted that hegemonic power was better qualified as the financial capacity to provide international liquidity and act as an international lender of last resort than dominance in world trade (1987: 563). Financial power rather than trade dominance is a superior characteristic of a hegemonic state. Fields and Vernengo argued that the US as a monetary hegemon provides “a default risk-free asset to facilitate global accumulation” (2013: 747). To complement Strange’s ideas on US monetary power, the infrastructural power of the US dollar as money of account, shared between US actors and foreign actors, not only enables the United States to greatly influence the valuation of dollar debts in global financial markets but also enforces other states to accumulate dollar assets such as US treasuries. In this way, the United States is a “structurally advantaged hegemon” (Stoke 2018: 141). Therefore, unlike other states and private banks, the United States does not face a dollar liquidity problem and, therefore, no default risk on sovereign debt.

What does US monetary power mean to countries like China holding a large number of US treasuries? China’s holding of US treasuries is often seen as a direct threat to the US dollar’s reserve role if China were to dump them on global markets (Arrighi 2005; Cohen 2008: 462). They underestimate the capacity of the US Federal Reserve to influence the valuation of US dollars and dollar assets/debts. The US monetary authority can directly purchase as many US treasuries as dumped by the Chinese state, which is likely to risk Chinese financial markets. In other words, China’s extensive holdings of US treasuries can be seen as a “dollar trap” (Prasad 2014).

Furthermore, as the case of Korea in 2008 and 2020 demonstrates, the considerable accumulation of dollar reserves does not guarantee the stability of the Korean won and its financial markets. Unlike the sphere of world trade, China’s integration into global money and financial markets does not guarantee the secured process of RMB internationalization. In global money and financial markets, underpinned by the US dollar’s dominance, there is “no level playing field” for banks and financial institutions from the developing world (Park 2009).

REFERENCES

Adrian, T et al. (2010) The Commercial Paper Funding Facility. Federal Reserve Bank of New York Staff Reports 423, Revised June

Aizenman, J (2009) Hoarding International Reserves Versus a Pigovian Tax-Cum-Subsidy Scheme. NBER Working Paper 15484. Cambridge.

Aizenman, J et al. (2011) International Reserves and Swap Lines: Substitutes or Complements? International Review of Economics and Finance 20(1): 5–18.

Altman, O. L (1961) Professor Triffin on International Liquidity and the Role of the Fund. IMF Staff Papers 8(2): 151–191.

Altman, O. L (1963) Recent Developments in Foreign Markets for Dollars and Other Currencies. IMF Staff Papers 10 (1): 48–96.

Amato, M & Fantacci, L (2012) The End of Finance. Cambridge: Polity Press.

Amato, M & Fantacci, L (2014) Back to which Bretton Woods? Liquidity and Clearing as Alternative Principles for Reforming International Money. Cambridge Journal of Economics 38(6): 1431–1452.

Andrew, D (1994) Capital Mobility and State Autonomy: Toward a Structural Theory of International Monetary Relations. International Studies Quarterly 38(2): 193–218.

Andrew, D (2006) Monetary Power and Monetary Statecraft. In D Andrew (ed.) International Monetary Power. Ithaca: Cornell University Press.

Arrighi, G (2005) Hegemony Unravelling. New Left Review 32: 23–80.

Aschcraft, A et al. (2012) The Federal Reserve’s Term Asset-Backed Securities Loan Facility. Economic Policy Review: 29–66. Federal Reserve Bank of New York November.

Avdjiev, S et al. (2015) Breaking Free of the Triple Coincidence in International Finance. BIS Working Papers 524: 1–39.

Bergsten, F (2009) The Dollar and the Deficits. Foreign Affairs 88(6).

Bernanke, B. S (2005) The Global Saving Glut and the U.S. Current Account Deficit. Speech 77, Board of Governors of the Federal Reserve System.

Bernanke, B. S (2013) The Crisis as a Classic Financial Panic: Remarks. Fourteenth Jacques Polak Annual Research Conference. Washington: IMF.

BIS (1959) Annual Report

BIS (1964) Annual Report

BIS (1966) Annual Report

BIS (1973) Annual Report

Black, S & Munro, A (2010) Why Issue Bonds Offshore? BIS Working Papers 334. Switzerland: BIS.

Bordo, M & McCauley, R (2017) Triffin: Dilemma or Myth. BIS Working Papers 684: 1–41.

Borio, C & Disyatat, P (2011) Global Imbalances and the Financial Crisis: Link or No Link? BIS Working Papers 346. Switzerland: BIS.

Bowman, A et al. (2013) Central Bank-Led Capitalism? Seattle University Law Review 36(2): 455–487.

Braun, B et al. (2020) Financial Globalization as Positive Integration: Monetary Technocrats and the Eurodollar Market in the 1970s. Review of International Political Economy, https://doi.org/10.1080/09692290.2020.1740291

Broz, L. J (2015) The Politics of Rescuing the World’s Financial System: The Federal Reserve as a Global Lender of Last Resort. Korean Journal of International Studies 13(2): 323–351.

Burn, G (1999) The State, the City and the Euromarkets. Review of International Political Economy 6(2): 225–261.

Burn, G (2006) The Re-Emergence of Global Finance. London: Palgrave.

Chey, H-K. (2012) Why did the US Federal Reserve Unprecedentedly Offer Swap Lines to Emerging Market Economies During the Global Financial Crisis? Can We Expect Them Again in The Future? GRIPS Discussion Paper: 1–15.

Chinn, M & Frankel, J (2005) Will the Euro Eventually Surpass the Dollar as Leading International Reserve Currency? NBER Working Paper Series 11510. National Bureau of Economic Research.

Cohen, B. J (1971) The Seigniorage Gain of an International Currency: An Empirical Test. The Quarterly Journal of Economics 85(3): 494–507.

Cohen, B. J (1998) The Geography of Money. Ithaca: Cornell University Press.

Cohen, B. J (2006) The Macrofoundations of Monetary Power in International Monetary Power. In D Andrews (ed.) International Monetary Power. Ithaca, NY: Cornell University Press.

Cohen. B. J (2008) The International Monetary System: Diffusion and Ambiguity. International Affairs 84(3): 455–470.

Cohen, B. J (2011) The Benefits and Costs of an International Currency: Getting the Calculus Right. Open Economic Review 23: 13–31.

Cohen. B. J (2016) The IPE of Money Revisited. Review of International Political Economy 24(4): 657–680.

De Cecco M (1987) Financial Innovation and Monetary Theory. In M De Cecco (ed.) Changing Money: Financial Innovation in Developed Countries. Oxford: Basil Blackwell.

Ding, D et al. (2019) China’s Offshore Corporate Dollar Bonds. In A Schipke et al. (eds) The Future of China’s Bond Market. Washington: IMF.

Dooley, M. P et al. (2004) The Revised Bretton Woods System. International Journal of Finance and Economics 9(4): 307–313.

Dooley, M. P et al. (2009) Bretton Woods II Still Defines the International Monetary System, Pacific Economic Review 14(3): 297–311.

Duncan, R (2005) The Dollar Crisis: Causes, Consequences, Cures. Singapore: John Wiley & Sons.

Eichengren, B (1996) The European Payments Union: An Efficient Mechanism for Rebuilding Europe’s Trade. In B Eichengreen (ed.) Europe’s Post-War Recovery. Cambridge: Cambridge University Press.

Eichengreen, B et al. (2003) Currency Mismatches, Debt Intolerance and Original Sin: Why They Are Not the Same and Why It Matters. NBER Working Paper 10036.

Eichengreen, B (2011) Exorbitant Privilege. Oxford: Oxford University Press.