1
Did the Enfranchisement of Workers Promote
Protectionism? An Empirical Investigation
during the Pre-World War II Era.
Master Thesis
Research Masters in Economics and Business
Rijksuniversiteit Groningen
Diogo Requena1
Supervisors: Tristan Kohl and Petros Milionis
Abstract
In this paper, I empirically investigate what role the expansion of franchise and electoral cycles had in two distinct eras, the globalization backlash in the late 19th century and the increased
economic isolationism during the interwar period. First, I analyze if the increasing share of voters explains the move towards protectionism. Second, I study if the electoral cycle influences the tariff choice made by policy makers. I capture this effect analyzing if being in a year prior to elections change how tariffs are set. The results indicate that, contrary to what the literature has found, increases in franchise actually moved tariffs away from the theoretical preference from the newly enfranchised workers. This preference is captured by how scarce or abundant labor is in a given country and year. Besides that, the electoral cycle was an important tariff determinant. Policy makers changed the tariff in the direction of the theoretical preference of the median voter in years preceding elections. I hypothesize that they do this to increase their reelection chances.
Keywords: Democracy, Trade Policy, Globalization, Electoral Cycle JEL codes: F13, D72, N70
2
1. Introduction
The late 19th century was an era characterized by increasing economic integration. During this period, the world underwent a similar process to the globalization from the late 20th century. Due to this reason, this era is also known as the first wave of globalization, in contrast to the similar globalization process that took place after the 1950s. Trade, capital, and labor flows increased drastically during the second half of the 19th century (Baldwin and Martin, 1999; O’Rourke and Williamson, 1999; O’Rourke and Williamson, 2002). Due to this similarity between these eras, a share of the literature has tried to understand what explained this economic globalization, especially the increased levels of trade (Jacks et al., 2008; Jacks et al., 2010). They found that one of the main reasons trade increased substantially was the decreasing trade barriers imposed by national states.
This initial increase in trade and factor flow later met resistance by political actors, leading to a political backlash to the globalization process (O’Rourke and Williamson, 1999; Williamson, 2005). This backlash meant increased trade barriers in most countries in Europe and in the New World. The economic isolationism was even more profound during the subsequent period, after World War I and before World War II. Countries kept on increasing their trade barriers and trade levels plunged. Therefore, the interwar era and the first wave of globalization experienced changing tariff levels.
3
more protectionist, like the United States, and some others liberalization their trade, like China. Therefore, what explains these different reactions?
Those two periods were also marked by the expansion of democratic institutions, especially by the enfranchisement of national citizens, i.e. expansion of voting rights. Countries where only the rich could vote ended up extending voting rights to include the poor and illiterate. As explained, this globalization process also produced losers, that is, a share of the people lost share of the national income during the globalization process. This process led to a decline in wages in many European countries (Williamson, 2002). The enfranchisement of workers unsatisfied with globalization could have been the reason for rising trade protection during these two eras. Thus, was the expansion of voting rights a crucial determinant in the later 19th century and interwar period trade backlash? It is still not clear if the inclusion of new voters explained the political backlash against globalization.
4
of these factors are not practical to examine due to the lack of data, such as campaign financing. However, others, such as electoral cycles and enfranchisement could explain partially how the globalization backlash took place and how tariffs were increasing in some countries before World War II.
In this paper, I study if this increase in the franchise over different nations explains different reactions to globalization and the general increase in protectionism. While similar studies focus on the effect of democracy expansion during these eras, I constraint my analysis on the expansion of franchise. To my knowledge, no papers used franchise expansion instead of democracy indicators. The reason for doing this is the arbitrariness of how democracy indicators are built, given different weights to different aspects of democracy, and the lack of a straightforward interpretation. Some democracy indicators such as the Polity IV, give maximum scores to countries where women could not vote. 2 Additionally, most papers argue that the expected effect is captured by the expansion of the voting pool, so using the voters’ franchise capture this effect directly. Finally, I also study if countries became more susceptible to the electoral cycle when franchise increased. That is if tariffs were more sensitive to the electoral cycle while franchise was expanding.
I perform an empirical study using data from 35 countries, both developed and developing, for two distinct eras. These two periods are the first wave of globalization, from 1865 to 1913, and the interwar period, from 1919 to 1938. This paper has two main contributions to the literature. First, franchise expansion was actually associated with moving away from the tariff level preference of the newly enfranchised class (workers). This result goes against the finding that democracy expansion moved tariffs in the
5
direction of workers’ preferences. I discuss alternatives theories that might explain this result, such as the expansion of alternatives means to influence policy by the elites and alternatives workers’ preferences such as the preference to protect the new industrial sector. Nevertheless, this result is surprising because it goes against what was found by the literature. Second, the electoral cycle was important for tariffs setting during these two eras. In other words, when franchise was low, policy makers shifted tariffs in the direction of land owners’ preferences in pre-electoral years. When franchise was high, policy makers shifted tariffs in the direction of workers’ preferences in pre-electoral years. This effect also increased its magnitude with the extension of voting rights and how protectionist (or pro-free trade) were the workers, captured by the land-labor ratio. I hypothesize that policy makers would do that to increase their re-election chances now that workers could also choose their representatives.
6
The paper is divided in the following way: Section 2 provides a review on the literature of tariff determinants and the political economy of trade policy; Section 3 provides the empirical strategy used in this paper; Section 4 describes the data and its sources used in this study; Section 5 reports the results and discuss them; Section 6 concludes this paper.
2. Literature Review
A large share of the literature has tried to understand what caused the changing tariffs levels before WWII and another share has tried to understand how politics and the political system affects trade barriers. However, few have tried to understand how politics explained the increased protectionism at the end of the 19th century and then the even greater increase in protectionism between the two world wars. Thus, if the political backlash to globalization was a product of countries’ internal politics, does the inclusion of the enfranchised workers explain it? During these two eras, tariffs became more restrictive while voting rights expanded. This evidence shows that some groups and individuals were against trade. Especially workers that could suddenly vote in nations were tariffs were increasing.
7
might be beneficial for the country as a whole, some groups within the country might oppose it.
The Stolper-Samuelson Theorem in a Hecksher-Ohlin framework helps to explain why some individuals/groups will oppose or support trade with other countries. This theorem states that a given country will export products that use intensively the factor which that country is abundant in. Thus, the price and, consequently, the factor return will increase. This will lead owners of the abundant factor to support trade since their factor return will increase. The converse will happen with owners of the scarce factor. The country will import products that use the scarce factor more intensively. In consequence, the price of these products and the return of the factor that is used more intensively in their production will decrease. Therefore, owners of the scarce factor will oppose free trade because this will reduce their factor return.
8
while in labor scarce countries, left-wing parties tend to be more protectionist. The opposite is true for right-wing parties. Thus, factor-endowment is an important economic determinant in explaining why some groups have anti-trade stances.
Other economic factors might play a role in determining a country’s tariffs such as its income inequality and economies of scale. If the government is concerned about the country’s income inequality level then tariffs generally exhibit an anti-trade bias (Limão and Panagariya, 2007). Capital abundant countries tend to protect trade more as inequality increases and labor abundant countries tend to protect trade less as inequality increases (Dutt and Mitra, 2006). That is, the government attend to workers’ interests more as inequality increases, since workers are usually the group with lower income (in relation to capital owners). Moreover, an increase in scale effects, i.e. plants and industries economies of scale, leads labor abundant countries to be more protectionist and capital abundant countries to be less protectionist (Djerdjian, 2009).
9
agreements with third parties (Cheng and Joshi, 2010; Baldwin and Jaimovich, 2012; Baier et al., 2014). Therefore, the spread of trade agreements creates a domino effect.
10
Besides democratic institutions and the extension of who can vote, the electoral system also mattered in defining tariff levels. Electoral systems matter because they change the dynamics of policy formation. The electoral system encompasses how congress seats are divided, if the country has a separate election for the executive branch, how large are the electoral districts, the length of terms, among others characteristics. Countries that use proportional representation (PR) systems tend to have lower tariffs and non-tariff barriers than countries that use majoritarian systems (Evans, 2009; Hatfield and Hauk, 2014; Rickard, 2012). In PR systems, elected legislators try to maximize aggregate welfare because their election is not tied to specific geographic areas (Grossman and Helpman, 2005). However, what seems to be important in majoritarian systems is the factor endowments of swing districts (Muûls and Petropoulou, 2013; Wiberg, 2014). Industries concentrated in swing districts tend to have higher levels of protection. District size also matters for the electoral system, parties linked to larger electoral groups, measured by districts, tend to choose more pro-trade policies (Hankla, 2006). Presidential systems make parties to support more free-trade but it also increases partisan conflicts over trade policy (Milner and Judkins, 2004). Finally, other political institutions like the process that a policy/law is approved also matter. Countries with more institutional veto points and/or veto players change their trade policies less often (O’Reilly, 2005). The wide range of results found by the literature shows that political institutions are important factors in determining trade policy.
11
associated with votes in favor of free trade in the US (Baldwin and Magee, 2000). Besides that, re-electoral/electoral incentives faced by politicians also change the way tariffs are set. Being close to elections makes representatives less likely to support trade liberalization (Conconi et al., 2014). American presidents are also more likely to start trade disputes in years preceding elections, especially disputes involving industries that are important in swing states (Conconi et al., 2017). Results also indicate that politicians are more likely to reduce trade barriers right after elections (Frye and Mansfield, 2004). Thus, the incentive of getting elected seems to make politicians assume more protectionist stances.
However, is this also the case in a similar context of trade liberalization and a subsequent backlash? As explained, a similar globalization wave to the late 20th century occurred during the second half of the 19th century (Jacks et al., 2010; O’Rourke and Williamson, 1999; O’Rourke and Williamson, 2002; Williamson, 2005). An initial trade and factor mobility integration were later replaced by a political backlash in the European core and in the New World. How did this political backlash take place? Was it associated with the newly obtained voting rights by the working class in those countries or only due to external forces? Besides that, were the electoral cycle also important during that era?
12
of internal markets, the world tariff environment, revenues needs, geography, which served as a natural barrier, and, naturally, tariff policy autonomy were important in explaining why tariff levels were different during that era (Blattman et al., 2002; Clemens and Williamson, 2002; O’Rourke and Taylor, 2006).
This evidence indicates what determined each country’s preferred tariff. However, we are still missing the role that politics had during this era. If the backlash was indeed politically motived, what caused it? Thus, how important were the inclusion of newly enfranchised voters in this first political backlash during the late 19th century and during the isolationism of the interwar era? Additionally, how the electoral cycle, which is often ignored in these eras, explained the political reaction to globalization? To answer those gaps I study two different political aspects of these eras. First, I study the effect of the franchise indicator on the tariff level. This indicator is meant to capture how the inclusion of workers in the voters’ pool affected tariffs and if this political backlash was due to the inclusion of workers in this pool. Second, I study if the electoral cycle had an effect in tariff setting during these two eras.
3. Empirical Strategy
13 𝑡𝑎𝑟𝑖𝑓𝑓𝑖𝑡 = 𝛽0+ 𝛽1𝑓𝑟𝑎𝑛𝑐ℎ𝑖𝑠𝑒𝑖𝑡+ 𝛽2ln(𝑅𝑖𝑡⁄𝐿𝑖𝑡)+ 𝛽3𝑃𝑟𝑒𝐸𝑙𝑒𝑐𝑡𝑜𝑟𝑎𝑙𝑌𝑒𝑎𝑟𝑖𝑡 + 𝛽4[𝑓𝑟𝑎𝑛𝑐ℎ𝑖𝑠𝑒𝑖𝑡∗ ln(𝑅𝑖𝑡⁄𝐿𝑖𝑡)]+ 𝛽5[𝑃𝑟𝑒𝐸𝑙𝑒𝑐𝑡𝑜𝑟𝑎𝑙𝑌𝑒𝑎𝑟𝑖𝑡 ∗ ln(𝑅𝑖𝑡⁄𝐿𝑖𝑡)]+ 𝛽6[𝑃𝑟𝑒𝐸𝑙𝑒𝑐𝑡𝑜𝑟𝑎𝑙𝑌𝑒𝑎𝑟𝑖𝑡∗ 𝑓𝑟𝑎𝑛𝑐ℎ𝑖𝑠𝑒𝑖𝑡] + 𝛽7[𝑃𝑟𝑒𝐸𝑙𝑒𝑐𝑡𝑜𝑟𝑎𝑙𝑌𝑒𝑎𝑟𝑖𝑡∗ 𝑓𝑟𝑎𝑛𝑐ℎ𝑖𝑠𝑒𝑖𝑡∗ ln(𝑅𝑖𝑡⁄𝐿𝑖𝑡)] + 𝛽8ln(𝑔𝑑𝑝𝑝𝑐𝑖𝑡) + 𝛿𝑖 + 𝜀𝑖𝑡(1)
Where 𝑡𝑎𝑟𝑖𝑓𝑓𝑖𝑡 is the trade policy measurement for a country i in the year t and is measured by the ratio of duties to imports expressed in percentage. The 𝑓𝑟𝑎𝑛𝑐ℎ𝑖𝑠𝑒𝑖𝑡 variable is an index of voters’ enfranchisement. This variable ranges from 0 to 5 and it is added to capture how an increase on eligible voters affect the tariff level. This effect could be for two different reasons or a mixture of both. That is, it could be the case that the increase of voters leads to different politicians being elected that vote more in line with the median voter preference, which I name the direct effect of enfranchisement. Alternatively, it could be that the elected representatives shift their policies towards the median voter in order to get re-elected, which I name the indirect effect.
14
leading elected politicians to shift their policies’ preferences prior to elections. In other words, if tariff levels were susceptible to the electoral cycle.
The land-labor ratio ln(𝑅𝑖𝑡⁄𝐿𝑖𝑡) is measured by taking the log of the ratio of the agricultural land by the population. This variable is rescaled to have mean 0 and standard deviation (s.d.) equal to 1. As explained earlier, this variable captures the median voter policy preference. The larger (smaller) this variable and franchise are, the more (less) protectionist the median voter is. Alternatively, I also use capital-labor ratios dummies. However, I chose to use land-labor ratios in my main specification due to the higher importance of land than capital in many countries during the late 19th century and to the lack of data on capital for most countries. 3 While some countries such as the United Kingdom, Belgium and the United States industrialized early in the 19th century, some others such as Spain, Austria-Hungary and the Ottoman Empire industrialized late in the century (Berend et al., 1982; Henderson, 2013).
The main interactions of interest are 𝑓𝑟𝑎𝑛𝑐ℎ𝑖𝑠𝑒𝑖𝑡∗ ln(𝑅𝑖𝑡⁄𝐿𝑖𝑡) and 𝑃𝑟𝑒𝐸𝑙𝑒𝑐𝑡𝑜𝑟𝑎𝑙𝑌𝑒𝑎𝑟𝑖𝑡∗ 𝑓𝑟𝑎𝑛𝑐ℎ𝑖𝑠𝑒𝑖𝑡∗ ln(𝑅𝑖𝑡⁄𝐿𝑖𝑡). The former captures how voting right expansions affect tariffs in a different fashion depending on the median voter preference. The latter captures if politicians change tariffs before elections in the direction of the median voter preference. This interaction measures if the indirect effect of enfranchisement is taking place, i.e. if politicians are changing their preferred policy in order to please voters and increase their chances of getting re-elected.
15
The variable 𝑔𝑑𝑝𝑝𝑐𝑖𝑡measures the GDP per capita. The term 𝛿𝑖 is a country level fixed effect and 𝜀𝑖𝑡 is the error term clustered at the country level to account for serial correlation between the error terms.
Additionally, I also consider using dummies for franchise, and high capital-labor and high land-labor ratio instead. For franchise, I consider two different cut-offs, the first when the franchise variable is equal or higher than 3 and the second when it is equal or higher than 4. The reason for this is that an increase of a magnitude of 1 in the franchise indicator does not always mean an increase of the same magnitude in the number of voters. For example, when franchise increases from 1 to 2, the increase in the number of people that can vote is expected to be lower than when franchise increase from 2 to 3. For the capital-labor and land-labor ratios, the use of dummies serves as a robustness check to the unreliability and lack of data. The labor dummy is equal to 1 when the land-labor ratio is higher than the mean and 0 otherwise. For the capital-land-labor dummy, the intuition is the same plus I consider countries with missing capital-labor ratios to be scarce in capital. I assume this because data is mostly available for countries from the European core, which are usually abundant in capital (Rogowski, 1989). Thus, countries with missing data for capital have the dummy equal to 0.
16
I also consider year fixed effects and mean Rest of the World (ROW) tariffs as robustness checks.
4. Data
The first wave of globalization was a period marked by an initial decrease in protectionism from 1840 until 1870 (Tena-Junguito, Lampe and Tâmega, 2012) and a subsequent globalization backslash after 1870, especially by the European center, and an even greater backlash during the interwar period (O’Rourke and Williamson, 1999; Williamson, 2005). The data on tariffs used in this paper was collected by Clemens and Williamson. The dataset is an unbalanced panel that covers the years from 1865 to 1938. It has the ratio of duties to imports expressed in percent for a sample of 35 countries, both developed and developing. The countries present in this panel are: Argentina, Australia, Austria-Hungary, Brazil, Burma, Canada, Ceylon, Chile, China, Colombia, Cuba, Denmark, Egypt, France, Germany, Greece, India, Indonesia, Italy, Japan, Mexico, New Zealand, Norway, Peru, Philippines, Portugal, Russia, Serbia, Spain, Sweden, Thailand, Turkey, the United Kingdom, the United States, and Uruguay. I exclude the World War I years from the sample.
17
property possession (or a similar alternative such as income or profession) or property AND literacy; 2 indicates when voting was based on literacy; 3 indicates when any “economically independent” individual could vote (i.e. anyone but servants or debtors); 4 indicates when suffrage was universal (for men); and 5 indicates when suffrage was universal and also included women (both on narrower bases than men and on the same bases). Information on franchise is present for most countries, except Australia, China, Egypt, India and Indonesia. Thus, these 5 countries are only presented in the regressions where the democracy score is used instead of the franchise variable.
The information on which countries had tariff autonomy was taken from Clemens and Williamson (2002). This variable is a dummy variable which is equal to 1 if the country has autonomy to set its own tariff and 0 otherwise. The data on the democracy indicator was taken from the Polity IV dataset (Marshall et al., 2016). I rescale the democracy score in the same fashion as O’Rourke and Taylor (2006), where democracy is equal to 0 if the country does not have autonomy in setting its own tariff and the score is scaled from 0 to 1. Thus, when the tariff autonomy variable is equal to 0, the democracy score is also equal to 0.
18
(Pahre, 2005). Data on gold standard membership came from Kramer & Milionis (2018). Finally, the data on capital stocks were gathered from Bergeaud et al. (2016) estimations.
Figure 1 presents the average tariff level in the World by year. The figure confirms that protectionism was on the rise after 1870 and between WWI and WWII. The data show that the increase before WWI was slower and less constant, while the increase after WWI was steeper. However, the first abrupt jump, around 1870, might be explained by the rise in the sample from 27 to 35 countries. Due to this reason, I also include a dashed line that only shows the tariff for the initial 27 countries of the sample. The inclusion of this dashed line indicates that this first abrupt jump was indeed due to the inclusion of more countries in the sample. However, the dashed line still shows that tariffs were increasing in both periods, especially in the interwar era.
20
The descriptive statistics are displayed in Table 1. The mean franchise throughout the sample was 2.97 with a standard deviation of 1.5. Figure 3 shows two histograms for the franchise variable, one for the pre-1914 period and another for the interwar period. Before World War I, most countries had voting laws based on property (franchise = 1) or universal voting restricted to men. During the Interwar period, most countries had universal voting restricted to men or universal voting. Thus, the distribution of the franchise variable is more scattered before World War I than after. In the sample, the highest average tariff observed is 58.2 percent and the lowest is 1.8 percent with a mean value of 15.2 percent and a standard deviation of 10.1 percent.
Table 1 - Summary Statistics
Variable Obs Mean Std.Dev. Min Max
21
The dummies variable can only assume values equals to 0 or 1, so their mean corresponds to the percentage of observations that have the dummy equal to 1. Therefore, 13.8 percent of the country-year observations are countries in a pre-electoral year, 83.2 percent of the observations had more than one party/candidate running for the legislative/executive during last elections, 15.1 percent of the observations are countries taking part in a war in a given year and 59.2 percent of country-year observations were countries adhering to the gold standard.
22
this issue, I also ran regression using random effects instead. I do not present them in the tables because the results were similar.
5. Results
5.1 Franchise and Re-Election Incentives
The results for the main specifications are presented in Table 2. In Column 1, I perform a simple regression including only the pre-electoral year dummy, the franchise variable and GDP per capita. At first glance, both franchise and re-electoral incentives, captured by the pre-electoral year dummy, do not seem to affect tariffs. Only the effect of GDP per capita is statistically significant. A 100 percent increase in GDP per capita, reduces the tariff by 8.9 percentage points (p.p.). In the next regression, I include the interaction between the land-labor ratio R/L and the franchise variable to capture how an increase in voting rights will have different effects on tariffs depending on the median voter preference. I also include the land-labor ratio so the interaction does not capture any linear effect of R/L on the tariff. In other words, a direct effect of the land-labor endowments on the tariff level. Such an effect could be the different competition in the international markets in labor-intensive goods and land-intensive goods. The former are typically less homogenous than the latter. For example, agricultural products are typically more homogenous. Thus, the necessity to protect in these markets could be higher.
23
are abundant in land and, consequently, export products that are intensive in land are more protectionist than countries that are abundant in labor.
There are three possible explanations for the land-labor ratio coefficient. First, it could be the case that workers pressure the policy makers through different means other than voting. Given the setting of the period, this seems unlikely. Second, as explained before, it could be that the competition in international markets for land intensive good was higher than the competition for labor intensive goods. Since countries with land (labor) abundance are typically exporters of land (labor)-intensive goods, a high land-labor ratio means that the country would protect more because its exporting goods are more homogenous. As a third alternative, this variable could also be capturing the necessity to protect infant industries. If countries abundant in land were about to industrialize, this could indicate a demand for protectionism so that their infant industries could develop. This desire to industrialize could be explained by the necessity of a cheap factor for manufacturing since in these countries labor was expensive. Then, industrializing and (partially) substituting labor by capital could be an alternative for manufacturing. Then, this would explain why countries scarce in labor would aim to protect their infant industries to industrialize more than countries abundant in labor.
24
would find different means to influence the trade policy as their voting power decreased. Their votes lost weight in determining who would govern as lower classes entered the voting pool. Thus, the negative effect of franchise * R/L could be explained by an increase in bribes or campaign financing by the elites that just had their weight in electing representatives undermined. Unfortunately, it is not possible to check this theory with the data I have.
Another aspect that I am not considering here is the sectoral composition. If these newly enfranchised workers were employed by the growing industrial sector in opposition to the agricultural sector, it makes sense that they would demand more protection even when labor was abundant. This last scenario seems plausible because of the increasing urbanization and industrialization that took place during these eras. That is, if protectionism expanded employment and returns in the industrial sector because capital was scarce, then it could make sense that workers seek the capital-owners interest instead of their policy preference predicted by the theory. However, I do not have data on were the workers where employed at the time that franchise was expanding.
25
statistically significant. Policy makers tend to reduce the tariff by 3.1 p.p. during years preceding the elections. It is rough to give a proper interpretation of this negative coefficient of the pre-electoral year dummy. However, we are still missing one additional interaction between our three main variables, land-labor ratio, franchise and the pre-electoral dummy. Therefore, these missing interactions might be biasing the coefficient of the pre-electoral year dummy.
Table 2 - Main Regressions
(1) (2) (3) (4) (5) Baseline Franchise * R/L Pre Electoral Year Interactions Main Specification Main Specification with FE Dependent Variable: Average Tariff
Pre Electoral Year -0.023 -0.005 -0.031* -0.039*** -0.006 (0.01) (0.01) (0.02) (0.01) (0.01) Franchise 0.004 0.000 -0.001 -0.001 -0.004 (0.01) (0.01) (0.01) (0.01) (0.00) GDP per capita -0.089*** -0.075*** -0.075*** -0.075*** -0.014 (0.02) (0.01) (0.01) (0.01) (0.03) R/L 0.083*** 0.082*** 0.084*** 0.068** (0.02) (0.02) (0.02) (0.03) Franchise * R/L -0.010** -0.010** -0.010** -0.012*** (0.00) (0.00) (0.00) (0.00) Pre Electoral Year * R/L -0.001 -0.022 -0.016*** (0.01) (0.02) (0.01) Pre Electoral Year * Franchise 0.008 0.010** 0.001 (0.00) (0.00) (0.00) Pre Electoral Year * Franchise * R/L 0.006 0.005** (0.00) (0.00) Country Fixed Effects No No No No Yes No. Observations 1216 1216 1216 1216 1216 R-Square 0.239 0.505 0.507 0.508 0.077
26
In the following regression, I add this remaining interaction. Policy makers will only care about changing tariffs to increase their reelection odds if people can vote. This last condition is captured by the franchise variable. Additionally, this effect will also depend on the median voter policy preference, captured by the land-labor ratio. Column 4 displays the results for this regression. All the coefficient have statistical significance at conventional levels, except for the franchise variable, the interactions between the pre-electoral year dummy and the land labor ratio and the triple interaction. The interaction between franchise and the land-labor ratio still has an unexpected negative effect. However, in this regression, the pre-electoral year dummy and its triple interaction with franchise and the land-labor ratio has the expected positive sign. To put into perspective, in a country where voting is universal (franchise = 5) and land is one s.d. more abundant than the mean (R/L = 1), policy makers increase the tariff by 1.1 p.p. in years preceding elections.4 In this case, the median voter has preferences closer to the worker’s preference because voting is universal. It is also interesting to consider the case where the median voter has preferences closer to the land-owners’ preferences, i.e. when the vote is more restricted. Thus, in a country where the vote is based on property (franchise = 2) and land is one s.d. more abundant than the mean (R/L = 1), policy makers decrease the tariff by 1.9 p.p. in years preceding elections. These results indicate that policy makers shifted tariffs in the direction of the median voter’s preference, to increase their odds to get reelected.
One could argue that our specifications are still missing important unobserved tariff determinants that vary by countries. To fix this issue, I include country fixed effects
27
in the specification. Column 5 reports the results including the fixed effects. The results are robust, most coefficients still have statistical significance. The main differences in statistical significance are the coefficient of the pre-electoral year dummy, GDP per capita, and the interaction between the pre-electoral year dummy and franchise, which all lose their significance. The results indicate that policy makers changed tariffs in years prior to elections to increase their chances of getting re-elected. With the inclusion of fixed effects, in a country where the vote is universal (franchise = 5) and land is one s.d. more abundant than the mean (R/L = 1), policy makers increase the tariff by 0.9 p.p. in years preceding elections. In a country where the vote is based on property (franchise = 2) and land is one s.d. more abundant than the mean (R/L = 1), policy makers decrease the tariff by 0.6 p.p. in years preceding elections.
The interaction between franchise and the land-labor ratio still has the opposite sign as the theory would predict. Why this is the case in most regressions is (fairly) intriguing. It could be the case that, as voting rights expanded, policy makers only attended to workers’ interests before elections. In different years, policy makers would still attend to the elite interests and would only pay attention to workers’ interests right before elections. It is also possible that the land-owning elites could use different means. Means such as bribes and coercion enabled by the higher financial power of the elites. However, it is hard to confirm such hypotheses with the data I have.
28
5.2 Non-Pooled Regressions and Alternatives Measures for Franchise and R/L
So far the results presented come from pooled regressions, that is, regressions using data from both periods. This might be an issue since both periods were substantially different from each other. While the first was a time of increasing integration, economic growth and imperialism, the second was a period of globalization backlash, economic crisis and decrease in colonial relations.5 Due to this reason, I perform the regressions presented in Column 5 of Table 2 using two different samples, one for each period. Naturally, the sample sizes reduce substantially, from 1216 in the pooled regression to 797 in the pre-WWI era and 419 in the interwar era. Hence, the coefficients have less statistical significance than in the pooled regression.
Columns 1 and 2 of Table 3 present the results for the pre-WWI era and the interwar era, respectively. For the pre-WWI era, only the coefficients for the Pre-electoral year dummy, the franchise variable and the interaction between these two terms were statically significant at conventional levels. The first two coefficients have a negative effect on average tariff while the third had a positive. For this era, in a country where voting was universal (franchise = 5), policy makers would increase the tariff by 0.8 p.p. in years preceding elections. In a country where voting was restricted to property (franchise = 1), policy makers would reduce the tariff by 0.8 p.p. in years preceding elections. These results indicate that on the one hand enfranchisement reduced tariffs, but on the other it made policy makers increase the tariffs right before elections.
In the regression for the interwar period, only the coefficient for the land-labor ratio is significant. If a country becomes 1 s.d. deviation more land abundant, its tariffs decrease by 40 p.p. This coefficient seems unrealistic and its large value could be due to
29
the small variance of R/L within countries over time. Since I am using country fixed effects, the effect of this small variance is inflated. Even though the sample reduction makes these two regressions less reliable, they indicate that the political-economic factors played a more important role during the first wave of globalization (i.e. pre-1914 era) than during the interwar period. That is, franchise and the electoral cycle had a greater impact during the second wave of globalization. One possible reason that explains this result is that tariffs were a product of internal politics during that first era, while it was a product of external factors during the interwar era. In other words, between the World Wars, changes in tariffs were a result of partners changing their own tariffs while during the first wave of globalization, it was a consequence of the changing political landscape within the country. 6 Nevertheless, most of the effect of the land-labor ratio comes from the interwar period. This might be an indication that the infant industry protectionist was more present between the two world wars than during the first wave of globalization. The infant industry argument also explains why countries’ tariffs during the interwar era were less sensitive to internal politics and more to changes in other countries’ tariffs.
As explained earlier in this paper, it is not a straightforward task to translate how much a change in franchise changes the number of voters. A change from property based voting laws to literacy based voting laws will increase the electorate size by a different magnitude than a change from male voting only to universal vote. The effect also depends on the country and year, since literacy, property possession, and other economic definitions vary substantially throughout place and time. This is a clear limitation of the franchise variable. Due to that reason, its effects should be thought in qualitative rather than in quantitative terms. Thus, I consider three different measures of voting rights. The first and the second are two dummies with different voting right cut-offs, to check if a
30
specific increase in franchise, which included a high amount of new workers in the voting pool, had an effect. The third is the polity democracy indicator to check if the results are similar.
The results for these 3 different franchise measures are reported in Columns 3, 4 and 5 of Table 3. The first cut-off considered is voting laws that allow only the “economically independent” to vote. This broad definition refers to countries where people had to have a job, could not have any debts, any public assistance, had to be on time with their taxes, could not be under legal bankruptcy, and could not be slaves or personal servants. The definition varies between countries, but the idea is that this threshold excludes a share of the population based on economic determinants. Thus, in this first specification, this dummy is equal to 1 when only the “economically independent” could vote (or less restrictive laws) and 0 otherwise.
31
In column 4, I consider an alternative cut-off, the universal voting for men. Universal voting, as in modern times, refers to laws that enable everyone to vote but the people convicted of a crime, the people that are legally incompetent or do not have the required nationally/ has been living for a short time in the country. Thus, this dummy is
Table 3 - Non-Pooled Regressions and Alternative Franchise/Democracy Measures
(1) (2) (3) (4) (5)
Pre-1914 Sample
Post-1914 Sample
1st Cutoff 2nd Cutoff Democracy Indicator Dependent Variable: Average Tariff
Pre Electoral Year -0.012** 0.000 -0.005 -0.003 -0.001 (0.01) (0.03) (0.01) (0.00) (0.01) Franchise -0.008* -0.025 -0.021 -0.014 (0.00) (0.02) (0.01) (0.01) GDP per capita -0.023 0.069 -0.005 -0.012 -0.002 (0.03) (0.05) (0.03) (0.02) (0.03) R/L -0.066 -0.406*** 0.064** 0.066** 0.076*** (0.05) (0.13) (0.03) (0.03) (0.03) Franchise * R/L -0.006 -0.011 -0.038*** -0.046*** (0.00) (0.01) (0.01) (0.01)
Pre Electoral Year * R/L
-0.005 -0.005 -0.010* -0.007 0.012 (0.01) (0.02) (0.00) (0.00) (0.01)
Pre Electoral Year * Franchise
0.004** 0.000 0.006 0.004 (0.00) (0.01) (0.01) (0.01)
Pre Electoral Year * Franchise * R/L 0.001 0.000 0.018** 0.014*** (0.00) (0.00) (0.01) (0.00) Democracy -0.014 (0.02) Democracy * R/L -0.006 (0.02)
Pre Electoral Year * Democracy
0.004 (0.01)
Pre Electoral Year * Democracy * R/L
-0.010 (0.02) Country Fixed Effects Yes Yes Yes Yes Yes No. Observations 797 419 1216 1216 1244
R-Square 0.045 0.206 0.076 0.091 0.072
32
equal to one if voting is universal (both for men only and also including women) and zero otherwise. Column 4 displays the results for the regression including this dummy. The results are similar to the previous regression’s results with the exception that the coefficient of Pre Electoral Year * R/L loses its statistical significance. For a country where voting is universal (at least, for men) and that is one s.d. more land abundant than the mean (R/L = 1), policy makers increase tariffs by 1.4 p.p. in years before elections. For countries without universal voting, the re-election incentives do not play a role. Policy makers do not increase or decrease tariffs due to elections being scheduled in the following year.
Finally, I consider using a democracy indicator instead of the franchise variable. Column 5 reports the results for this last specification. All the coefficients but the coefficient of the land-labor ratio are statistically non-significant. This coefficient means that if a country is one s.d. more land abundant than the mean, its tariff would be 7.6 p.p. higher than if the country had a mean land-labor ratio. I do not believe this is an issue because the democracy indicator includes a wide variety of institutions that measure how democratic is a country. The democracy indicator is built by measuring different dimensions of democracy: openness and competitiveness of executive recruitment, executive constraints, regulation and competitiveness of participation. Therefore, it captures a wide range of factors that I am not focusing on this study. Moreover, there are also some issues with the democracy indicator. The correlation between the democracy indicator and the franchise variable is less than 0.5 (around 0.476) and there are 63 observations (country-year) where the democracy score is the highest but vote is not universal for women.
33
I am using agricultural land from 1961 as a proxy. Due to this reason, even though land is practically time-invariant, the land-labor ratio might be wrongly specified. Hence, the next step I take is to use a land-labor ratio dummy instead of the measure I have and a capital-labor ratio dummy as an alternative. This dummy is time002Dinvariant for each country and it is built in two steps. First, I create a variable to indicate if the country is above or below each yearly mean land-labor ratio. Second, if the country has more years where it is land abundant than land scarce I consider the dummy to be equal to 1, otherwise the dummy is equal to zero. The capital-labor ratio dummy is built in the same way as the land-labor ratio but I consider the countries which I do not have data to be scarce in capital.7
Table 4 reports regressions without and with country fixed effects for each of the factor of production dummies. Column 1 reports the specification using the land-labor ratio dummy without including fixed effects. Only the coefficients of the pre-electoral year dummy, GDP per capita, the land-labor ratio dummy and the interaction between the pre-electoral year dummy and franchise are statistically significant. The results indicate that land abundant countries have tariffs 7.5 p.p. higher than land scarce countries. As discussed in the last sub-section, the less differentiability in land-intensive good markets or the willingness to protect infant industries might explain this result. Besides that, the interaction between franchise and the pre-electoral year dummy is positive. This means that the higher the franchise is, the more policy makers increase (or less they decrease) the tariff in years before the elections.
These results might be biased due to unobserved country-level variables. In column 2, I fix this issue by including country fixed effects. With the inclusion of fixed
34
effects, the dummy for land-labor ratio disappears since it is time-invariant. In this regression, only GDP per capita and the interaction between franchise and the land-labor
Table 4 - Alternative R/L and K/L measures
(1) (2) (3) (4) R/L Dummy R/L Dummy with FE K/L Dummy K/L Dummy with FE Dependent Variable: Average Tariff
Pre Electoral Year -0.046** -0.004 -0.073* -0.003 (0.02) (0.01) (0.04) (0.01) Franchise -0.001 -0.006 -0.004 0.005 (0.01) (0.01) (0.01) (0.02) GDP per capita -0.075*** -0.041* -0.093*** -0.023 (0.02) (0.02) (0.02) (0.02) R/L 0.075** (0.03) K/L -0.006 (0.07) Franchise * R/L 0.003 -0.011** (0.00) (0.01) Franchise * K/L 0.006 -0.009 (0.02) (0.02) Pre Electoral Year * R/L -0.007 -0.002
(0.03) (0.02)
Pre Electoral Year * K/L -0.017 -0.008 (0.06) (0.02) Pre Electoral Year *
Franchise
0.012** 0.001 0.014 -0.002 (0.00) (0.00) (0.01) (0.01) Pre Electoral Year *
Franchise * R/L
0.000 0.001 (0.00) (0.00) Pre Electoral Year *
Franchise * K/L
0.007 0.006 (0.02) (0.01) Country Fixed Effects No Yes No Yes No. Observations 1216 1216 1216 1216 R-Square 0.440 0.051 0.255 0.022
35
dummy have coefficients with statistical significance. This result indicates that increasing franchise in land abundant countries actually decreases the average tariff, moving the policy away from workers’ preference. For land-scarce countries, franchise does not have an effect.
In the regressions in column 3 and 4, I use the capital-labor ratio dummy instead of the land-labor ratio dummy. In the first regression, only the coefficients of the pre-electoral year dummy and GDP per capita have statistical significance. This result indicates that in years preceding elections, policy makers decrease the tariffs by 7.3 p.p. In the specification of column 4, I add country fixed effect to check if the results are robust. With the inclusion of fixed effects, all the coefficients lose their significance.
5.3 Robustness Checks
36
policy makers do not increase the tariff (or reduce in land scarce countries) to please voters, they just do not change it.
As discussed at the beginning of this section, we were ignoring if there was political competition or not in the country. To account for this problem, I built the pre-electoral year dummy differently. The new pre-pre-electoral year dummy is equal to 1 only when it was a year preceding elections and when the opposition dummy was equal to 1. The opposition dummy was equal to 1 when voters had more than one option to choose from during last elections. Thus, this new dummy is supposed to capture the effect of policy makers reacting to re-election incentives when they faced opposition during the last elections. Column 2 displays the results of this specification. The coefficients’ sign and significance remain similar, except for pre-electoral year dummy. The results show that franchise still lead policy makers to move tariffs on the opposite direction of workers’ preferences but during years preceding elections, they changed tariffs in the direction of workers’ interests when franchise was high. As an example, if the country had universal voting (franchise = 5) and was 1 s.d. more land abundant than the mean, policy makers would increase tariffs by 1.4 p.p. in years before elections.
37
Table 5 - Robustness Checks
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Time FE Opposition Dummy
Countries with Tariff Autonomy
Lagged Franchise Variable
War Variables TAs Gold
Standard
1-Year Lag mean ROW tariff
5-Years Lag mean ROW tariff Dependent Variable: Average Tariff
Pre Electoral Year -0.012* -0.003 -0.006 -0.005 -0.007 -0.011* -0.012** -0.012* -0.015**
(0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) Franchise -0.003 -0.003 -0.003 -0.003 -0.004 -0.009** -0.008** -0.004 -0.005 (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.01) GDP per capita -0.027 -0.020 -0.016 -0.018 -0.017 -0.028 -0.025 -0.029 -0.029 (0.03) (0.03) (0.03) (0.02) (0.03) (0.03) (0.03) (0.02) (0.02) R/L 0.101** 0.065** 0.070** 0.066** 0.066** -0.064 -0.062 0.092*** 0.067** (0.04) (0.03) (0.03) (0.03) (0.03) (0.05) (0.04) (0.03) (0.03) Franchise * R/L -0.010** -0.013*** -0.013** -0.013*** -0.012*** -0.005 -0.006 -0.012*** -0.016*** (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Pre Electoral Year * R/L -0.023*** -0.016** -0.016*** -0.016*** -0.016*** -0.006 -0.005 -0.022*** -0.022***
(0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.00) (0.00)
Pre Electoral Year * Franchise
0.003 -0.000 0.001 0.001 0.002 0.003* 0.004** 0.003 0.003
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Pre Electoral Year * Franchise * R/L
0.007*** 0.006** 0.005** 0.006** 0.005*** 0.001 0.001 0.007*** 0.007***
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Number of Wars involved 0.010
(0.03)
Involved in Any War -0.027
(0.04)
Number of TAs 0.002
(0.00)
Use Gold Standard 0.006
(0.02)
Mean ROW tariff 0.695*** 0.383***
(0.15) (0.12)
Year Fixed Effects Yes No No No No No No No No
Country Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes
No. Observations 1216 1210 1158 1176 1216 797 797 1181 1044
R-Square 0.279 0.081 0.080 0.080 0.080 0.054 0.047 0.198 0.150
38
Franchise might take a while to affect the policy decision, either by asymmetric information or due to the slow pace of policy setting. To fix this issue, I use lagged franchise values instead of the contemporaneous ones. Column 4 reports the results for this specification. The results are robust to this check. There are also some additional control variables that might be relevant for these two eras which I have not included so far. Therefore, I include the number of wars, a dummy that indicates if the country was involved in any war, the number of trade agreements that the country had and a dummy to indicate gold standard membership. Column 5 reports the regression including the war variables, column 6 includes the number of trade agreements (TAs) and column 7 includes gold standard membership. The results are robust to the inclusion of the war dummy and the number of wars variable.
For TAs and gold standard I only have data for the era before 1914, so the sample decreases substantially and most coefficients lose their significance. Column 6 reports the regression including the number of TAs. Only the coefficients of the pre-electoral year dummy, franchise and their interaction are statistically significant. While franchise reduces the tariffs, its effect in pre-electoral years is positive. However, since the number of TAs coefficient is non-significant, the change in significance should be due to the smaller sample. Column 7 presents the regression including the gold standard membership variable. The conclusion is similar to the regressions including the number of TAs, most coefficients lose their significance due to the smaller sample size.
39
country will also increase theirs as a retaliation. However, it does not measure the extent of the effect since it might also include other time-varying omitted variables. Therefore, to account for these external effects and to get a proper measure of them, I run two additional robustness checks that include the lagged mean tariff of the Rest of the World (ROW). 8 I use the lagged value because the reaction is expected to take some time and not to be contemporaneous. I consider two different lags. In column 8, I use one-year lags, and in column 9, I use 5-years lags.9
Column 8 shows that the results are robust to the inclusion of 1-year lagged mean ROW tariff. Additionally, the coefficient of the mean ROW tariff is positive and statistically significant. A 10 p.p. increase in the mean ROW tariff increases the country’s tariff by 7.0 p.p. in the following year. Column 9 displays the regression using 5-years lag mean ROW tariff instead of 1-year lag. The results are still robust and the coefficient of the mean ROW tariff is also positive and statistically significant. The main difference is that the magnitude of this coefficient reduces, showing that the effect of an increase of the ROW tariff decays with time. This result means that if the other countries increase their tariffs by 10 p.p., on average, the country will have tariffs 3.8 p.p. higher 5 years later.
In the appendix, I present a table similar to Table 5 for each period separately. The results of Table 5.1 are similar to the results of column 1 of Table 3 and the results of Table 5.2 are similar to the ones of column 2 of Table 3. There are two major differences between these two tables that were not perceivable in Table 3. First, being involved in a war increased the average tariff during the pre-WWI era while it decreased the tariff in
8 Rest of the World refers to all countries for which I have tariff data in a given year with the exception of the country which tariff is in the left-hand side of the equation.
40
the interwar period. Second, during the pre-WWI era, the tariff of other countries only affected the tariff of a country with a substantial delay. A country did not respond to changes in tariff by other countries in the preceding year, this only happened if the change was 5 years before. In contrast, during the interwar period, this effect was faster. Countries responded to changes in tariffs by other countries in a year time, but changes from 5 years ago did not matter. As discussed earlier in this paper, this difference might indicate the greater importance of protecting infant industries during the interwar era. The quicker response to other countries’ tariffs could indicate a higher preoccupation with international competition in manufacturing sectors. Additionally, as mentioned in columns 1 and 2 of Table 3, internal political factors played a more determinant role during the pre WWI era while the external factors, such as the ROW tariff, were more important during the interwar era.
6. Conclusion
In this paper, I analyzed two main questions regarding how political institutions impacted trade policy during two eras before the current wave of globalization. First, I studied to what extent the franchise expansion contributed to the move towards protectionism during that era and if it explains why some countries did not become more protectionist. Second, considering the increasing franchise in most countries, how did the electoral cycle affected tariffs. Did the electoral cycle amplify or attenuate the effect of voter enfranchisement?
41
theorem, workers in labor scarce countries have the preference for protectionism, while workers in labor abundant countries prefer free-trade. To capture this preference I used land-labor or capital-labor ratios.
My results indicate that this relation was the inverse for these two eras. Increases in franchise were associated with shifting the trade policy in the direction of the land and/or capital owners’ interests. I theorized that this might be due to the use of different means by the elite as their voting weight deteriorates. The larger financial power owned by elites could let them influence policy through bribes or campaign financing, for example. This result goes against the one found in part of the literature that democracy pushed trade policy in the direction of workers’ interests. However, the literature usually uses democracy indicators instead of a direct franchise measure. While franchise is one important aspect of democracy, it is not the only one. It could be that the increase in franchise were constricted by retractions in other democratic institutions, such as political competition or executive constraint. Nevertheless, this does not seem to be the case because throughout the periods both franchise and democracy indicators were increasing.
Another possibility for finding the inverse relation between franchise and tariff is that the Stolper-Samuelson theorem does not fully capture the workers’ true preference. For example, workers could demand protectionism even if labor was abundant. If countries were urbanizing and workers were moving from the agricultural to the manufacturing sector and capital was scarce, workers could demand protectionism to increase employment and returns in the industrial sectors. Unfortunately, data on the sectoral composition for these eras is not available to further test these theories.
42
makers pushed trade policy in the direction of land or capital owners’ interests. In countries where franchise was already high, policy makers pushed tariffs in the direction of workers’ interests. Therefore, the expected effect of franchise was present in years close to elections. While the actual increase in franchise seemed detrimental to workers, this effect was attenuated when policy makers needed to be re-elected. This result is consistent with a view that voters are myopic, that is, they only look at policy made close to the elections when deciding their vote.
Thus, did politics, that is, political institutions and incentives, played a role in the changing trade liberalization? The answer seems to be yes. On one hand, the expansion of voting rights made policy makers set more (less) protectionist policies in labor abundant (scarce) countries, going against the interests of the newly enfranchised workers. On the other hand, politicians reacted to the voters’ interests next to elections as the newly enfranchised workers were included in the voting pool. Despite the results being robust to a wide variety of checks, when non-pooled regressions were performed, the results were different for each era. However, it indicated that internal political factors were indeed relevant during the pre-WWI era, while external factors, such as partners’ tariffs were more relevant during the interwar era.
43
remediate this issue by using country fixed effects and later time fixed effects. Finally, one additional limitation of the data is that there is no information on what month the election took place. If elections were held in December, it is expected that the year before was still too far from actual elections. Thus, changing tariffs in the preceding year would not help politicians to get re-elected.
44
7. References
Baier, S. L., Bergstrand, J. H., & Mariutto, R. (2014). Economic determinants of free trade agreements revisited: Distinguishing sources of interdependence. Review of International
Economics, 22(1), 31-58.
Baldwin, R., & Jaimovich, D. (2012). Are free trade agreements contagious?. Journal of
international Economics, 88(1), 1-16.
Baldwin, R. E., & Magee, C. S. (2000). Is trade policy for sale? Congressional voting on recent trade bills. Public Choice, 105(1-2), 79-101.
Baldwin, R. E., & Martin, P. (1999). Two waves of globalisation: superficial similarities,
fundamental differences (No. w6904). National Bureau of Economic Research.
Berend, T. I., Ránki, G., Pálmai, É., & Ránki, G. (1982). The European periphery and industrialization 1780-1914 (pp. 24-24). Cambridge: Cambridge University Press.
Bergeaud, A., Cette, G., & Lecat, R. (2016). Productivity trends in advanced countries between 1890 and 2012. Review of Income and Wealth, 62(3), 420-444.
Bolt, J., Inklaar, R., de Jong, H., & van Zanden, J. L. (2018). Rebasing ’Maddison’: new income comparisons and the shape of long-run economic development. Tech. rep. Maddison Project Working Paper 10.
Blattman, C., Clemens, M. A., & Williamson, J. G. (2002). Who Protected and Why? Tariffs the World Around 1870-1938. Harvard Institute of Economic Research Discussion Paper
No. 2010
Chen, M. X., & Joshi, S. (2010). Third-country effects on the formation of free trade agreements. Journal of International Economics, 82(2), 238-248.
Clemens, M. A., & Williamson, J. G. (2002). Closed jaguar, open dragon: comparing
tariffs in Latin America and Asia before World War II (No. w9401). National Bureau of Economic
Research.
Clemens, M. A., & Williamson, J. G. (2004). Why did the tariff–growth correlation change after 1950?. Journal of Economic Growth, 9(1), 5-46.
Conconi, P., DeRemer, D. R., Kirchsteiger, G., Trimarchi, L., & Zanardi, M. (2017). Suspiciously timed trade disputes. Journal of International Economics, 105, 57-76.
Conconi, P., Facchini, G., & Zanardi, M. (2014). Policymakers' horizon and trade reforms: The protectionist effect of elections. Journal of International Economics, 94(1), 102-118.
Decker, J. H., & Lim, J. J. (2009). Democracy and trade: an empirical study. Economics
of governance, 10(2), 165-186.
Djerdjian, D. O. (2009). Economies of Scale, Ideology and Trade Policy. International
Economic Journal, 23(1), 1-21.
Dreher, A. (2006). Does globalization affect growth? Evidence from a new index of globalization. Applied economics, 38(10), 1091-1110.
Dutt, P., & Mitra, D. (2005). Political ideology and endogenous trade policy: an empirical investigation. Review of Economics and Statistics, 87(1), 59-72.
45
Eichengreen, B., & Leblang, D. (2008). Democracy and globalization. Economics &
Politics, 20(3), 289-334.
Evans, C. L. (2009). A protectionist bias in majoritarian politics: An empirical investigation. Economics & Politics, 21(2), 278-307.
Feyrer, J. (2009). Trade and Income--Exploiting Time Series in Geography (No. w14910). National Bureau of Economic Research.
Frankel, J. A., & Romer, D. H. (1999). Does trade cause growth?. American economic
review, 89(3), 379-399.
Frye, T., & Mansfield, E. D. (2004). Timing is everything: Elections and trade liberalization in the postcommunist world. Comparative Political Studies, 37(4), 371-398.
Goldberg, P. K., & Maggi, G. (1999). Protection for sale: An empirical investigation. American Economic Review, 89(5), 1135-1155.
Grossman, G. M., & Helpman, E. (1992). Protection for sale(No. w4149). National Bureau of Economic Research.
Grossman, G. M., & Helpman, E. (2005). A protectionist bias in majoritarian politics. The
Quarterly Journal of Economics, 120(4), 1239-1282.
Hankla, C. R. (2006). Party strength and international trade: A cross-national analysis. Comparative Political Studies, 39(9), 1133-1156.
Hatfield, J. W., & Hauk Jr, W. R. (2014). Electoral regime and trade policy. Journal of
Comparative Economics, 42(3), 518-534.
Henderson, W. O. (2013). Industrial Revolution on the Continent: Germany, France, Russia 1800-1914. Routledge.
Irwin, D. A., & Terviö, M. (2002). Does trade raise income?: Evidence from the twentieth century. Journal of International Economics, 58(1), 1-18.
Jacks, D. S., Meissner, C. M., & Novy, D. (2008). Trade Costs, 1870-2000. American
Economic Review, 98(2), 529-34.
Jacks, D. S., Meissner, C. M., & Novy, D. (2010). Trade costs in the first wave of globalization. Explorations in Economic History, 47(2), 127-141.
Jäkel, I. C., & Smolka, M. (2017). Trade policy preferences and factor abundance. Journal
of International Economics, 106, 1-19.
Kramer, B. S., & Milionis, P. (2018). Democratic Constraints and Adherence to the Classical Gold Standard. (GGDC Research Memoranda; Vol. GD-175). University of Groningen. Limão, N., & Panagariya, A. (2007). Inequality and endogenous trade policy outcomes. Journal of International Economics, 72(2), 292-309.
Mansfield, E. D., Milner, H. V., & Rosendorff, B. P. (2000). Free to trade: Democracies, autocracies, and international trade. American Political Science Review, 94(2), 305-321.
Marshall, M., Gurr, T. & Jaggers, K. (2016). POLITY™ IV PROJECT Political Regime Characteristics and Transitions, 1800-2015 Dataset Users’ Manual. 1st ed. Centre for Systemic Peace.
Mayda, A. M., & Rodrik, D. (2005). Why are some people (and countries) more protectionist than others?. European Economic Review, 49(6), 1393-1430.
46
Milner, H. V., & Judkins, B. (2004). Partisanship, trade policy, and globalization: Is there a left–right divide on trade policy?. International Studies Quarterly, 48(1), 95-119.
Milner, H. V., & Kubota, K. (2005). Why the move to free trade? Democracy and trade policy in the developing countries. International organization, 59(1), 107-143.
Milner, H. V., & Mukherjee, B. (2009). Democratization and economic globalization. Annual Review of Political Science, 12, 163-181.
Muûls, M., & Petropoulou, D. (2013). A swing state theory of trade protection in the Electoral College. Canadian Journal of Economics/Revue canadienne d'économique, 46(2), 705-724.
O’Reilly, R. F. (2005). Veto points, veto players, and international trade policy. Comparative Political Studies, 38(6), 652-675.
O'Rourke, K. H., & Taylor, A. M. (2006). Democracy and protectionism (No. w12250). National Bureau of Economic Research.
O'Rourke, K. H., Sinnott, R., Richardson, J. D., & Rodrik, D. (2001). The determinants of individual trade policy preferences: International survey evidence [with comments and discussion]. In Brookings trade forum (pp. 157-206). Brookings Institution Press.
O'Rourke, K. H., & Williamson, J. G. (1999). Globalization and history: the evolution of a
nineteenth-century Atlantic economy. MIT press.
O’Rourke, K. H., & Williamson, J. G. (2002). When did globalisation begin?. European
Review of Economic History, 6(1), 23-50.
Pahre, R. (2005), Trade Agreements Database (TAD). https://netfiles.uiuc.edu/pahre/www/tad.html
Przeworski, A., Newman, S., Park, S. K., Queralt, D., Rivero, G., & Shin, K. J. (2013). Political Institutions and Political Events (PIPE) Data Set. Department of Politics, New York
University.
Rickard, S. J. (2012). A non-tariff protectionist bias in majoritarian politics: Government subsidies and electoral institutions. International Studies Quarterly, 56(4), 777-785.
Rogowski, R. (1989). Commerce and coalitions: How trade affects domestic political alignments. Princeton University Press.
Scheve, K. F., & Slaughter, M. J. (2001). What determines individual trade-policy preferences?. Journal of International Economics, 54(2), 267-292.
Stinnett, D. M., Tir, J., Diehl, P. F., Schafer, P., & Gochman, C. (2002). The correlates of war (cow) project direct contiguity data, version 3.0. Conflict Management and Peace
Science, 19(2), 59-67.
Swinnen, J. F. (2009). The growth of agricultural protection in Europe in the 19th and 20th centuries. World Economy, 32(11), 1499-1537.
Tena-Junguito, A., Lampe, M., & Tâmega, F. (2012). How much trade liberalization was there in the world before and after Cobden-Chevalier?. The Journal of Economic History, 72(3), 708-740.
Wiberg, M. (2014). Comparative trade policy. Review of International Economics, 22(2), 410-421.
Williamson, J. G. (2003). Was It Stolper-Samuelson, Infant Industry or Something Else?
47
Williamson, J. G. (2005). Winners and losers over two centuries of globalization. In Wider
perspectives on global development (pp. 136-174). Palgrave Macmillan, London.
