Rethinking coalition size and trade policies in authoritarian regimes: Are single-party dictatorships less protectionist?

Introduction

The role of the party in a dictatorship has been extensively investigated in the literature of comparative authoritarianism. One main conclusion of this research is that the dominant party contributes to authoritarian survival because it can co-opt its opponents via the sharing of power and resources (Greene, 2010; Reuter and Turovsky, 2014). Based on this insight, scholars find that single-party dictatorships are more likely to adopt liberal trade policies than other types of dictatorships to please more members of those coalitions (Hankla and Kuthy, 2013; Milner and Kubota, 2005). ¹ However, these studies on authoritarian trade politics assume that members of coalitions have a homogeneous preference for free trade. This assumption, as criticized by Gallagher and Hanson (2015: 377), makes the size of the coalition the only concern in analyzing politicians’ decision-making. In addition, politicians become less likely to maintain a homogeneous coalition as the required size of their coalition increases. More importantly, trade policies incur distributive effects that shape different trade policy preferences (Scheve and Slaughter, 2001). Thus, this assumption of homogeneous preference is problematic when we are examining how authoritarian parties make trade policies.

In this article, I argue that single-party dictatorships are no less protectionist than other types of dictatorships. As trade policies have distributive effects across different social groups, it is not always in the best interests of autocrats to exclude all protected groups when they have to build a large coalition. Instead, they make trade policies in a selective way. Authoritarian leaders subject to large coalitions, especially those under single-party dictatorships, can set ad hoc tariffs on different products in response to the heterogeneous demands of potential supporters (i.e. their selectorate). Thus, protectionism in single-party dictatorships is exercised via complex tariff schedules.

It is important to investigate tariff complexity under dictatorships for two reasons. First, a high level of tariff complexity creates economic inefficiency. It causes trade distortion because exporters and importers have incentives to get their products placed in categories subject to lower tariffs (Deardorff, 2006). Like high tariff rates, a complex tariff schedule also encourages corruption and impedes trade flows (Ehrlich, 2011; Gatti, 1999). Thus, tariff complexity is as important as tariff levels in trade policy analysis. Second, a complex tariff schedule indicates “greater efforts towards central planning of the economy’s production and consumption patterns” (Gwartney et al., 2011: 196). Although tariff complexity might result from the government’s intention to levy higher tariffs in sectors with low import elasticities of demand (Grossman and Helpman, 1994), as we will see below, it is more often an outcome of selective responses to demands for protectionism. Thus, investigating tariff complexity helps researchers understand how politicians respond to the particularistic trade policy preferences of their supporters and ensure their political survival.

Based on these discussions, I argue that the increased complexity of tariff schedules under single-party dictatorships results from the heterogeneous trade policy preferences among politicians’ potential supporters. Although politicians in single-party dictatorships need to maintain large coalitions to stay in power, they do not have to exclude all special interest groups from their coalitions. Instead, they need to decide which special interest groups should be included and which excluded. As some special interest groups demand protectionism and other free trades, authoritarian leaders subject to larger coalitions respond to these heterogeneous preferences in a more selective way when recruiting coalition members. Therefore, tariffs become lower but more complex under single-party dictatorships.

The case of Malaysia, which has been politically dominated by the National Front since 1974, provides a typical example of the coexistence of low and dispersed tariff rates. While Malaysia reduced its tariff rates from 17.5% in 1988 to 9.5% in 2002, its degree of tariff dispersion increased from 91% to 210% (Athukorala, 2005). According to Athukorala and Wai-Heng (2007), one main source of tariff dispersion in Malaysia is the protection of the automotive industry. The national automobile corporation Proton, in particular, has been granted a huge degree of protection since it was established by the government in 1983, including duty exemptions and discriminatory tariffs on parts for foreign cars (Jomo, 1994). During the financial crisis of 1997, the political connection between Proton and the Malaysian government prompted the government to stop buying imported cars for official use and to only purchase Protons (Pepinsky, 2009: 129–130). The overall trend toward reducing tariffs has not prevented the Malaysian government from protecting its automotive industry. Instead, trade liberalization exercised via tariff reduction has contributed to the complexity of tariffs on automotive imports in Malaysia.

Egypt makes another case for tariff complexity under single-party dictatorships. Under the authoritarian rule of the National Democratic Party, Egypt’s simple average tariff decreased from 31% to 21% during the 1990s. Yet, its tariff dispersion increased in the same period because many sectors, such as alcoholic beverages and automobiles, were still overprotected (World Trade Organization (WTO), 1999). According to Acemoglu and Robinson (2012: 395–396), these sectors received protection because big “businessmen close to the regime” had “persuaded the government to restrict entry through state regulation.” Thus, tariffs become more complex as coalitions become larger because politicians’ protection of coalition members is more selective.

Figure 1 is a box plot that illustrates the variation in tariff complexity across authoritarian regimes. I plot tariff complexity, measured by the index of tariff dispersion (Gwartney et al., 2011), for each authoritarian country as of 2009 over the typology of dictatorships proposed by Geddes et al. (2012). Tariff complexity, as shown in Figure 1, is greater in single-party dictatorships than in other types of dictatorships. More importantly, Egypt, Malaysia, and Zimbabwe, which are coded as single-party authoritarian regimes, also have the highest index of tariff dispersion (10).

OPEN IN VIEWER

Figure 1.

Tariff dispersion under different types of dictatorships in 2009. The boxplot displays the median (horizontal line within the boxes), the 25th and 75th percentiles (the lower and upper limits of the boxes), and the lower and upper adjacent values (as the whiskers) of the tariff dispersion in each type of the authoritarian regimes. The adjacent values are the most extreme values that are still within 1.5 times the range between the 25th and 75th percentiles.

This article makes several contributions to the emerging literature on authoritarian trade politics. It demonstrates that tariffs become more complex as the preferences of rulers’ coalitions become more heterogeneous. Therefore, this article offers a caveat to the previous finding that single-party dictatorships are less protectionist than other dictatorships. It is similar to Kono’s (2006) finding regarding the difference in the pattern of protectionism between democracies and dictatorships: Although democracies have lower tariffs, they have higher nontariff barriers (NTBs) than dictatorships. The NTBs enable democracies to achieve “optimal obfuscation that allows politicians to protect their markets while maintaining a veneer of liberalization” (Kono, 2006: 369). This article demonstrates how dictators adopt a similar economic statecraft via tailoring their tariff schedules. In other words, the dispersion of tariff schedules is indicative of authoritarian leaders’ selective responses to the demands of members of their coalitions.

This article also contributes to the study of authoritarian consolidation. Since modernization theory predicts that economic growth facilitates democratization (Lipset, 1959), it is puzzling that many dictatorships are still able to resist democratic transition despite economic prosperity. What is more puzzling is that some autocracies intentionally establish institutions that generate economic growth (Wright, 2008). One prevalent approach to the analysis of both puzzles focuses on institutionalization. For one thing, dictators share power and resources with opposition forces within political institutions in exchange for their support (Gandhi, 2008; Svolik, 2009). For another, political institutions generate higher economic growth that in turn encourages more compliance with the status quo among citizens (Gallagher and Hanson, 2009). Since single-party dictatorships are generally regarded as more institutionalized than other types of authoritarian regimes (Hankla and Kuthy, 2013; Reuter and Turovsky, 2014), this article advances this research agenda by showing how adeptly tailored tariff schedules can be a feasible policy tool for dictators in party dictatorships.

This article proceeds as follows. In the next section, I discuss how the sizes of constituency and coalition affect trade policies. Then, I elaborate on why large coalitions under single-party dictatorships lead to tariff dispersion. In the empirical section, I discuss the operationalization of variables and conduct a series of statistical analyses to test my hypothesis; the final section concludes.

Constituency size and trade liberalization

Previous studies on trade politics focus disproportionally on democracies (Kono, 2015). One important theoretical perspective on this issue is how the size of constituency, defined by formal political institutions, affects politicians’ choices of trade policies. According to this view, politicians subject to a larger constituency are more likely to adopt free trade policies that can benefit more citizens. For example, constituency sizes under a system of proportional presentation (PR) are larger than single-member districts (SMDs), encouraging countries that use PR to adopt more open trade policies than those that use SMD (Nielson, 2003; Rogowski, 1987). Meanwhile, Lohmann and O’Halloran (1994: 599) demonstrate that in the United States, the President tends to favor free trade more than the Congress because the President “has a national constituency and cares about the losses incurred” by the protectionism demanded by congressional districts. Nielson (2003) reports similar findings in developing countries with presidential systems. Hankla (2006: 1134) also finds that strong parties tend to choose free trade policies because they have “stable linkages to large groups of the electorate.” Thus, a larger constituency induces more trade liberalization in democracies.

The constituency-size argument on trade politics is further extended by Bueno de Mesquita et al (2003) in their selectorate theory. Unlike previous studies that mainly focus on constituency size, selectorate theory posits that political leaders’ incentives to invest in public goods are determined by two groups of citizens: the selectorate (S) and the winning coalition (W). ² The selectorate is the group of people within a country who have the power to select the leaders. Thus, the concept of selectorate is not only inclusive of constituency in democracies but also applicable to autocracies where political leaders are not chosen via free and fair elections. Meanwhile, the winning coalition is the set of individuals within selectorate whose support is essential for a leader to stay in power. The core idea of selectorate theory is that the size of W relative to S determines policy outcomes, especially the provision of public goods. According to Bueno de Mesquita et al. (2003: 196–197), selectorate theory can also explain trade policies because trade liberalization, functioning as a public good, increases social welfare by reducing domestic prices and stimulating economic growth for the overall society. Although politicians still have an incentive to protect some social groups facing foreign competition, the support of protected groups may be insufficient for politicians who need to build a large winning coalition. Accordingly, politicians subject to large coalitions need to cater to those who can benefit from trade liberalization. In short, a large winning coalition promotes free trade.

Milner and Kubota (2005: 115) offer a more sophisticated application of selectorate theory. Instead of assuming that free trade is a public good, Milner and Kubota (2005) emphasize its impact on the income of laborers in developing countries. According to the Heckscher–Ohlin model and Stolper–Samuelson theorem, the owners of abundant factors benefit from free trade due to increases in their wages. Thus, owners of abundant factors support free trade while those with scarce factors oppose it. Because democratization enlarges the size of selectorate, the identity of the median voter also shifts from capital owner to laborer in most developing countries. Consequently, politicians in developing countries subject to larger selectorates have a stronger incentive to embrace free trade to benefit more owners of abundant factors, such as laborers.

Milner and Kubota (2005) also apply selectorate theory to the analysis of variations in tariff levels across authoritarian regimes. Following Geddes (1999), Milner and Kubota (2005: 120) argue that “autocracies vary according to the size of their selectorate and winning coalition.” Specifically, the selectorate is largest in single-party dictatorships and smallest in personalist ones. Meanwhile, they claim that a larger selectorate implies “an increase in the size of the winning coalition.” Accordingly, single-party dictatorships have the highest level of trade openness and lowest tariff barriers when compared with other types of dictatorships with smaller winning coalitions.

Hankla and Kuthy (2013) further strengthen the findings of Milner and Kubota (2005) with deeper investigations into the relationship between selectorate theory and trade policies across authoritarian regimes. Assuming that a more institutionalized autocracy has a larger selectorate, Hankla and Kuthy (2013) contend that multi- and single-party dictatorships have larger selectorates than other types of dictatorships. Because institutionalized dictatorships “incorporate more social actors into policymaking, expanding the incentive for dictators to provide public goods such as free trade” (Hankla and Kuthy, 2013: 494), party dictatorships are more open to trade.

Nevertheless, current studies fail to explain why single-party dictatorships have more dispersed tariffs, another form of protectionism. In the next section, I elaborate on a theory of heterogeneous coalition to disentangle this puzzle.

Heterogeneous coalition, authoritarian regime types, and tariff complexity

One key assumption of previous studies on constituency/coalition size and trade policy is that members of the selectorate and the coalition share the same preference for the political leaders’ policies. Thus, their memberships are interchangeable (Bueno de Mesquita et al., 2003). ³ This assumption of homogeneous preference, according to Gallagher and Hanson (2015: 378), implies that “the composition of the membership of the winning coalition thus has no bearing on the ruler’s policy outputs.” The only thing that affects the politicians’ decision on the bundle of public and private goods is the size of the coalition. However, this assumption is disputable because members of the selectorate may have different policy preferences. It also ignores the possibility that policy changes can occur without a change in the size of W or S (Gallagher and Hanson, 2015). Politicians can reshuffle the members of their coalitions without changing the size of their selectorates or winning coalitions, and they can make new policies in response to the preferences of new members of their coalitions. Therefore, the composition of preferences is as important as the size of the winning coalition.

The assumption of homogeneous preference among coalition members is particularly problematic for trade policy analysis. Trade liberalization creates winners and losers, depending on economic actors’ endowments, skill levels, and mobility (Hiscox, 2002; Rogowski, 1989). Yet, studies on authoritarian trade policies tend to downplay the distributive effect of trade. Bueno de Mesquita et al. (2003: 196) argue that trade is a public good because it “ensures consumers—and everyone is a consumer—that they can have access to quality goods at competitive prices.” Similarly, Hankla and Kuthy (2013: 494) regard free trade as a public good because it “provides benefits to society as a whole.” Although Milner and Kubota (2005) acknowledge that there are heterogeneous preferences regarding trade among citizens due to differences in the abundance of endowments, they seem to assume that all members of a large winning coalition are owners of abundant factors (e.g. labor in developing countries). These studies assume that all members of the coalition prefer free trade.

Yet, not everyone prefers free trade (Scheve and Slaughter, 2001). Heterogeneous trade policy preferences result in the formation of special interest groups, which create protectionist bias via lobbying (Ehrlich, 2011; Grossman and Helpman, 1994). Unfortunately, studies applying selectorate theory or other constituency-size arguments to explain trade politics do not consider the role of special interest groups. Instead, they assume a homogeneous preference for free trade and expect that the entrenched special interests will disintegrate when politicians need to build large coalitions. However, politicians are less able to build homogeneous coalitions as the required size of their coalitions increases. In other words, it is not always the best strategy for politicians to exclude all special interest groups when they have to build larger coalitions. According to selectorate theory, politicians must balance the marginal costs of public goods and private privileges, and members of the winning coalition enjoy more benefits than nonmembers. This difference in benefits creates incentives for special interest groups to seek policy favors. For instance, import-competing industries prefer high tariffs on final goods, while producers of intermediate products prefer lower tariffs on raw materials. Both groups may ask politicians to satisfy their demands when they are, or when they would like to be, included in the winning coalition. When politicians need to reduce tariffs to increase the aggregate welfare for members of their large coalitions, it may not always be in their best interests to exclude all interest groups. They can be selective in response to requests for policy favors from protectionists or supporters of free trade. Some of them may still get protection due to their political or economic importance to politicians’ winning coalitions.

Based on these discussions, I argue that heterogeneous preferences regarding trade liberalization in the selectorate result in more tariff complexity for dictatorships with larger coalitions. When dictators need to build a large coalition, they just need to exclude special interest groups whose contribution to their incumbency is relatively minor. Accordingly, the requirement to maintain a large coalition under a single-party dictatorship makes authoritarian leaders less able to build a coalition whose members hold homogeneous preferences regarding trade policies. Instead, the size of the coalition increases tariff complexity due to politicians’ selective responses to special interest groups among their selectorates. Thus, the rest of this article tests the following hypothesis: Tariff schedules are more dispersed and complex under single-party dictatorships than under personalist, military, or monarchical dictatorships.

It has to be noted that my argument concerning heterogeneous coalitions might also be applied to democracies because they usually have large constituencies. However, I focus on dictatorships alone in this article for two reasons. First, various levels of institutionalization result in different levels of tariff complexity in democracies. Ehrlich (2011) argues that democracies with more “access points,” operationalized by the establishment of formal political institutions, have more complicated tariffs. Betz (2017) demonstrates that “narrow interest institutions,” operationalized by the use of plurality rule, lead to tariff dispersion when countries sign preferential trade agreements. Although democracies are more institutionalized than dictatorships, some institutions increase constituency size while others dilute the influence of heterogeneous policy preferences. Second, tariffs may not be an ideal policy tool for democratic leaders to protect their constituents. A high level of tariff dispersion is still a sign that tariffs are being manipulated. As Kono (2006) forcefully argues, electoral competition increases the transparency of trade policies and discourages politicians from using tariffs rather than nontariff trade barriers in democracies. Accordingly, the difference in tariff complexity between democracies and dictatorships may not be attributed to differences in constituency size but rather to other institutional factors.

I would also like to note that the assumption of heterogeneous preferences within party dictatorships has been applied in other studies of authoritarian political economy, especially in studies of exchange rate policies that have the same significant distributive effects as trade policies do. Steinberg and Malhotra (2014) contend that civilian dictatorships, which include single-party and limited multi-party regimes, are more likely to adopt flexible exchange rate regimes because they have larger selectorates than military and monarchical dictatorships. Importantly, the larger selectorates under civilian dictatorships lead to “a larger number of interest groups” empowered by legalized elections, opposition parties, and legislatures (Steinberg and Malhotra, 2014: 499). Thus, special interest groups in civilian dictatorships with a variety of institutional channels are more likely to demand that their political leaders adopt flexible exchange rate regimes than their counterparts in other authoritarian regimes.

Empirical analysis

Empirical results

Table 1 presents the estimation results. Model 1 estimates a baseline model using OLS with robust standard errors clustered at the country level. The coefficient Single Party is positive and statistically significant at p < 0.1 level, suggesting that tariffs are more dispersed under single-party dictatorships. Model 1 also shows that mean tariffs are positively associated with tariff dispersion, so protectionism may be implemented via tariff level and tariff dispersion at the same time. Thus, I estimate a seemingly unrelated regression model to consider the simultaneous effects of Single Party on tariff dispersion and mean tariffs. The results in model 2 indicate that single-party dictatorships have lower but more dispersed tariffs.

OPEN IN VIEWER

Table 1.

Single-party dictatorships and tariff dispersion.

	Model 1	Model 2		Model 3	Model 4
	(OLS)	(XTSUR)		(PCSE)
DV	Dispersion	Mean	Dispersion	Dispersion
Single party	0.975* (0.535)	−0.377*** (0.068)	0.266** (0.121)	0.348* (0.186)	0.965** (0.458)
Lagged DV		0.316*** (0.021)	0.809*** (0.026)	0.871*** (0.075)
Mean tariff	0.902*** (0.186)			0.063 (0.100)	0.376** (0.149)
Logged import	−0.306 (0.478)	0.147*** (0.013)	0.016 (0.065)	−0.058 (0.116)	−0.012 (0.232)
Logged GDP per capita	0.091 (0.362)	−0.190*** (0.033)	0.026 (0.026)	0.067 (0.081)	−0.262 (0.214)
Military size	0.754** (0.342)	−0.173*** (0.034)	−0.044 (0.068)	0.084 (0.087)	0.492** (0.199)
Institutionalized autocracy	−0.087 (0.115)	0.063*** (0.014)	0.084*** (0.027)	0.042** (0.020)	0.031 (0.071)
Durability	0.128 (0.090)	0.012* (0.007)	−0.017 (0.015)	0.006 (0.012)	0.109*** (0.041)
Durability 2	−0.002 (0.002)	−0.000*** (0.000)	0.000 (0.000)	−0.000 (0.000)	−0.002*** (0.001)
GATT/WTO	0.459 (0.670)	−0.350*** (0.096)	−0.209 (0.199)	−0.114 (0.174)	0.131 (0.444)
Constant	5.204 (9.969)			0.988 (2.449)	2.290 (4.820)
Ar1	No	No		No	Yes
N	248	240		240	248
Number of countries	42	41		41	42
R 2	0.439	NA		0.861	0.326

Note: PCSE: panel-corrected standard errors; OLS: ordinary least squares; NA: not available; DV: dependent variable; XTSUR: seemingly-unrelated regression model on panel data. The dependent variable is the index of tariff dispersion except model 2. The dependent variables in the first and second columns of model 2 are the mean and dispersion of tariff rates, respectively. Standard errors are included in parentheses. Model 1 reports OLS estimates with robust standard errors clustered at the country level. Models 3 and 4 use PCSEs. All independent variables are lagged for 1 year.

*p < 0.1; ** p < 0.05; *** p < 0.01.

Neither model 1 nor model 2 considers the identification issues arising in panel data, including serial correlation and contemporaneous correlation. Therefore, I lag the dependent variable for one period to address serial correlation. I also use PCSEs to correct for heteroskedasticity and contemporaneous correlation among panels (Beck and Katz, 1995). In model 3, the coefficient of Single Party is 0.348 and it is statistically significant at p < 0.1 level. Since the result of the test for serial correlation indicates that there is first-order autocorrelation (p < 0.001), I follow the advice of Keele and Kelly (2006: 203) and correct for AR1 in model 4. The coefficient of Single Party becomes larger and its statistical significance increases to p < 0.05 level. These results support the hypothesis of this article.

The control variables are worth discussing. The results suggest that politics has an impact on tariff dispersion. Although the statistical significance of Military Size, Institutionalized Autocracy, Durability, and Durability 2 does not drop below the conventional level in some models, the results suggest that a dictatorship with a higher capacity to tax may have a more dispersed tariff schedule. For instance, the coefficient of Military Size is 0.492 and statistically significant at p < 0.05 level in model 4, implying that a dictatorship with more coercive capacity to levy taxes has a more dispersed tariff schedule. Similarly, the coefficients of Durability and Durability 2 in models 2 and 4 suggest an inverted U-shaped relationship between authoritarian duration and tariff dispersion.

Robustness checks

Varieties of dictatorships

One may argue that it is problematic to lump monarchical, military, and personalist dictatorships together as the baseline category. Therefore, I include three dummy variables for single-party, monarchical, and military dictatorships in model 5 of Table 2. I use personalist regime as the baseline category because scholars argue that this kind of regime has the smallest winning coalition (Chang and Golden, 2010). All coefficients of regime dummies in model 5 are positive, suggesting that personalist dictatorships have the lowest level of tariff dispersion. This finding corroborates that a larger coalition contains more heterogeneous preferences and leads to tariff complexity. ¹⁰

OPEN IN VIEWER

Table 2.

Robustness checks.

	Model 5	Model 6	Model 7	Model 8	Model 9
	(PCSE)				(OLS)
DV	Dispersion				Particularism
Single party	1.320*** (0.372)		1.155*** (0.353)	1.130** (0.480)	0.934* (0.521)
No party		−4.060*** (1.116)
Multi-party		−2.654*** (0.793)
Monarchy	1.049*** (0.328)
Military	0.576 (0.720)
Mean tariff	0.392*** (0.147)	0.363** (0.148)	0.362** (0.153)	0.398*** (0.146)	0.145*** (0.043)
Logged import	0.057 (0.229)	−0.071 (0.245)	0.195 (0.195)	0.062 (0.255)	0.170 (0.377)
Logged GDP per capita	−0.387* (0.225)	0.155 (0.262)	−0.354* (0.187)	−0.400* (0.220)	−0.190 (0.379)
Military size	0.410* (0.230)	0.232 (0.226)	−0.018 (0.203)	0.328 (0.211)	0.473 (0.464)
Institutionalized autocracy	0.007 (0.068)	0.030 (0.078)	−0.001 (0.053)	0.089 (0.082)	−0.047 (0.124)
Durability	0.113*** (0.039)	0.073* (0.038)	0.071** (0.036)	0.126*** (0.044)	0.040 (0.052)
Durability 2	−0.002*** (0.001)	−0.002** (0.001)	−0.001** (0.001)	−0.002*** (0.001)	−0.002** (0.001)
GATT/WTO		0.357 (0.454)	−0.435 (0.317)	−0.158 (0.467)	−1.436 (0.854)
Number of PTAs				0.144* (0.076)
Constant	1.423 (4.723)	5.243 (4.947)	−0.309 (4.628)	0.892 (5.452)	−0.539 (6.529)
AR1	Yes	Yes	Yes	Yes	No
N	248	272	166	211	73
Number of countries	42	47	30	41	36
R 2	0.339	0.273	0.443	0.351	0.640

Note: PCSE: panel-corrected standard error; OLS: ordinary least squares; WTO: World Trade Organization. The dependent variable is the index of tariff dispersion in models 5–8. The dependent variable of model 9 is the index of tariff particularism constructed by Kono (2007). Standard errors are included in parentheses. Models 5–8 report PCSEs. Model 8 excludes Middle Eastern countries. Model 9 uses robust standard errors clustered at the country level. All independent variables are lagged for 1 year except model 9.

* p < 0.1; ** p < 0.05; *** p < 0.01.

In addition to the data set of Geddes et al. (2012), I use another data set of dictatorships established by Cheibub et al. (2010). In their data sets, Cheibub et al. construct a variable to measure how many parties exist in a dictatorship. There are three categories of this variable: No Party, One Party or multiple parties that belong to the regime front, and Multiple Parties. I create three dummy variables for each category and use One Party as a baseline category in my analysis.

Model 6 presents the new results. Single-party dictatorships have more dispersed tariffs than dictatorships without any parties or those with multiple parties. This finding is consistent with my theory regarding the size of winning coalition and heterogeneous preferences. Dictatorships allow the formation of other parties either to generate legitimacy or to split opposition forces (Gandhi, 2008). If opposition parties cannot form a united front to overthrow the regime, dictators are able to maintain smaller winning coalitions to secure their incumbency. The existence of multiple parties outside the authoritarian government splits the elites and decreases the size of winning coalitions for dictators, who in those circumstances do not have to respond to many requests from special interest groups. As a result, tariff rates are more dispersed under single-party dictatorships than under multi-party ones. ¹¹

My sample includes several Middle Eastern monarchical dictatorships that rely on oil exports as their main source of revenues. The trade policies and the logic of political survival of these countries may differ from those of other dictatorships. For this reason, I exclude Middle Eastern countries from my sample in model 7, in which Single Party remains statistically significant at p < 0.01 level.

To investigate how preferential trade agreements (PTAs) affect tariff schedules, I control for the cumulative number of PTAs a country has signed in model 8. ¹² While the positive coefficient of Number of PTAs suggests that having more PTAs results in greater tariff dispersion, Single Party still has a statistically significant association with the dependent variable.

Conclusion

This article joins the emerging literature on authoritarian trade politics. Previous studies conclude that single-party dictatorships have lower tariff rates than other types of dictatorships, but they fail to explain why tariffs are more dispersed and complex under single-party dictatorships. I answer this question by addressing the heterogeneity of preferences concerning trade policies under single-party dictatorships. When politicians need to maintain large coalitions, they do not need to exclude all special interest groups from their coalitions as they can respond selectively to demands for protectionism. In short, the heterogeneous preferences within politicians’ coalitions lead to tariff complexity, an alternative form of protectionism. This argument is supported by the data on 42 authoritarian countries during the period 2000–2009.

This article contributes to the literature of international political economy in two ways. First, it advances the understanding of authoritarian trade politics. Current studies find that the larger coalitions of single-party dictatorships result in greater trade openness and lower tariff rates. However, this finding does not imply that politicians are more willing to wholeheartedly renounce protectionism. Instead, authoritarian leaders become more adept at implementing more complex trade policies that better balance the forces of protectionism and free trade within their coalitions.

Second, this article clarifies the varieties of protectionism that exist under dictatorships. The early literature attributes protectionism to the influence of special interests (Grossman and Helpman, 1994), but recent studies find that some forms of protectionism depend on other political variables. For example, Kono (2006) demonstrates that democracies tend to use NTBs to protect industries, because voters are less able to detect qualitative NTBs. Similarly, this article investigates how special interest politics affects the degree of tariff dispersion when mean tariffs are reduced under single-party dictatorships. Authoritarian leaders may rely on sophisticated forms of protectionism just as their counterparts in democracies do.

Taking a broader view, this article contributes to the literature on how political institutionalization affects authoritarian survival (Gandhi, 2008; Wright, 2008). This research line suggests that dictators establish binding political institutions, such as parties and legislatures, to consolidate their incumbency. This article finds an unintended consequence of political institutionalization, a more dispersed tariff schedule.

Before concluding, I would like to note that the argument of this article parallels what Ehrlich (2011) has investigated regarding access points theory and tariff complexity in democracies. An access point, according to Ehrlich (2011: 25), “is any policymaker who provides meaningful access to interest groups trying to influence policy.” Using this conceptualization, Ehrlich (2011) argues that political institutions in democracies can be regarded as “access points” through which special interest groups can distort trade policies in their favor. Accordingly, tariff rates are more complex when there are more access points. Nevertheless, Ehrlich (2011: 184–185) is cautious about applying access point theory to authoritarian regimes for two reasons. First, political systems under dictatorships are either closed or less institutionalized, so they might not have clear access points. Second, not all political institutions under dictatorships can hold authoritarian leaders accountable to their people, so special interest groups may not affect policies via institutionalized access points. Therefore, access point theory cannot be directly applied to analysis of authoritarian trade politics unless we can successfully identify what the meaningful and functional access points under dictatorships are. It might be the case that a larger coalition implies more access points under a dictatorship, because coalition size is positively related to political institutionalization. Future studies may investigate this topic with detailed analysis.

Supplemental material