The Political Economy of Tariff Revenue Dependence: Does Electoral District Magnitude Matter? Objective. This paper investigates whether electoral district magnitude in democratic developing countries influences government dependence on trade tax revenue through the aggregation of voters’ preferences over taxation. These preferences over the use of different tax instruments tend to vary with voters’ income levels. Method. The author uses fixed effects panel regression, with the dependent variable being tariff share of government revenue. Results and conclusion. This study finds that higher electoral district magnitudes are significantly associated with lower tariff revenue shares, controlling for confounders and alternative factors. It also finds that income inequality moderates the negative association between district magnitude and tariff revenue share. 1. Introduction Tariffs are an important source of government revenue. Developing countries, in particular, rely more on tariff revenues than their developed counterparts (Tanzi, 1987; Hitiris, 1990; Kubota, 2005). In fact, Tanzi (1987) found that in developing countries, import duties account for 25% of total tax revenue, making them ‘by far the single most important revenue source’. One proposed reason for this is that developing country bureaucracies are too ineffective to collect other forms of taxes (Aizenman, 1987; Gardner and Kimbrough, 1992; Khattry and Rao, 2002). On the other hand, political economy suggests that democratisation drives tariff revenue dependence up in developing countries by strengthening the influence of the poor median voter, who imports few high-quality goods and is less sensitive to hikes in tariffs than other taxes (Moutos, 2001; Adam, 2009). However, existing research has yet to consider whether democracies with different electoral systems will produce the same equilibrium level of tariff revenue dependence. Electoral systems with larger district magnitudes, which allow representation of more different voters than just the poor median voter, might be expected to produce lower tariff revenue shares in equilibrium. Thus, this paper investigates the effect of varying electoral district magnitude on tariff revenue dependence in developing countries. It finds significant evidence of a negative association between district magnitude and tariff revenue share in developing countries, particularly when the effect of income inequality and its interaction with district magnitude is accounted for. This study endeavours to provide a better understanding of the challenges faced by governments in developing countries when they undertake trade liberalisation policies and are constrained by fiscal needs.
25
Embed
The Political Economy of Tariff Revenue Dependence [Manuscript]-1
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
The Political Economy of Tariff Revenue Dependence:
Does Electoral District Magnitude Matter?
Objective. This paper investigates whether electoral district magnitude in democratic developing countries
influences government dependence on trade tax revenue through the aggregation of voters’ preferences over
taxation. These preferences over the use of different tax instruments tend to vary with voters’ income levels.
Method. The author uses fixed effects panel regression, with the dependent variable being tariff share of
government revenue. Results and conclusion. This study finds that higher electoral district magnitudes are
significantly associated with lower tariff revenue shares, controlling for confounders and alternative factors. It
also finds that income inequality moderates the negative association between district magnitude and tariff
revenue share.
1. Introduction
Tariffs are an important source of government revenue. Developing countries, in particular, rely more on tariff
revenues than their developed counterparts (Tanzi, 1987; Hitiris, 1990; Kubota, 2005). In fact, Tanzi (1987)
found that in developing countries, import duties account for 25% of total tax revenue, making them ‘by far the
single most important revenue source’. One proposed reason for this is that developing country bureaucracies
are too ineffective to collect other forms of taxes (Aizenman, 1987; Gardner and Kimbrough, 1992; Khattry and
Rao, 2002). On the other hand, political economy suggests that democratisation drives tariff revenue
dependence up in developing countries by strengthening the influence of the poor median voter, who imports
few high-quality goods and is less sensitive to hikes in tariffs than other taxes (Moutos, 2001; Adam, 2009).
However, existing research has yet to consider whether democracies with different electoral systems will
produce the same equilibrium level of tariff revenue dependence. Electoral systems with larger district
magnitudes, which allow representation of more different voters than just the poor median voter, might be
expected to produce lower tariff revenue shares in equilibrium. Thus, this paper investigates the effect of
varying electoral district magnitude on tariff revenue dependence in developing countries. It finds significant
evidence of a negative association between district magnitude and tariff revenue share in developing countries,
particularly when the effect of income inequality and its interaction with district magnitude is accounted for.
This study endeavours to provide a better understanding of the challenges faced by governments in developing
countries when they undertake trade liberalisation policies and are constrained by fiscal needs.
2 THE POLITICAL ECONOMY OF TARIFF REVENUE DEPENDENCE
2. Existing Literature
2.1 Explaining tariff revenue dependence using bureaucratic capabilities
A longstanding explanation for developing countries’ significant tariff revenue shares involves weak
administrative and bureaucratic capabilities, which prevent governments in developing countries from
effectively collecting income tax and other domestic direct taxes. This explanation – termed the Administrative
Capabilities Hypothesis (ACH) in current research – posits that the costs of collecting income taxes are high in
developing countries, leading bureaucracies to collect more revenue from tariffs. Developing countries tend to
have archaic tax administrations which are unable to competently assess tax liabilities and monitor domestic
compliance to tax laws effectively (Khattry and Rao, 2002). Structural characteristics of taxpayers in developing
countries further weaken the bureaucracy’s ability to raise revenue through direct taxes. Firstly, given that
literacy is required for citizens to file income taxes (Riezman and Slemrod, 1987), low levels of education
amongst citizens of developing countries hamper the government’s efficacy in collecting income taxes. Kenny
and Winer (2006) showed that educational attainment and school enrolment were positively related to individual
income tax revenue share and negatively associated with trade tax rates. Secondly, income tax compliance
monitoring and collection is hindered when the rural taxable population is geographically dispersed. Meanwhile,
tariffs are less costly to collect because taxable trade activity tends to be centralised in a few ports (Aizenman,
1987; Gardner and Kimbrough, 1992).
Yet, the ACH fails to consider the costliness of tariff collection due to complex tariff codes (Adam, 2009).
Empirical evidence for it has also been mixed. In Mahdavi (2008)’s study of tax revenue mix in developing
countries, the association between urbanisation and trade tax revenue share of GDP was insignificant.
Additionally, while Kenny and Winer (2006) found a negative association between educational levels and trade
tax rates, the association of the former with trade tax revenue share was insignificant.
2.2 Explaining tariff revenue dependence using political economy
Alternative explanations for tariff revenue dependence built on political economy theories have also emerged.
Kenny and Winer (2006) and Mahdavi (2008) proposed that in countries with authoritarian institutions
(regardless of the country’s level of economic development), the regime lacks legitimacy in the eyes of the
citizenry. This makes it harder for the government to enforce direct taxation, pushing it to collect more revenue
from indirect taxes instead. However, this “democratic legitimacy” explanation is unable to account for the
insignificant relationship between democracy and trade tax share found in both studies, particularly when the
THE POLITICAL ECONOMY OF TARIFF REVENUE DEPENDENCE 3
sample is not restricted to developing countries as was the case for Kenny and Winer (2006). Adam (2009)
suggested an alternative political explanation by focusing on voters’ interests and using a theoretical model
developed by Moutos (2001). He showed that in an environment where a vertically-differentiated good is traded,
with developing countries exporting low-quality versions to the rest of the world and importing high-quality
ones (Schott, 2004), tariff revenues encompass greater proportions of developing countries’ government
revenues as they become more democratic. Democratisation expands the selectorate to shift decisive trade and
tax policy positions away from the rich elite to the poor median voter who consumes low-quality, locally-
produced goods (Fajgelbaum et al., 2011; Bils and Klenow, 2000). The prices of low-quality, domestically-
produced goods are not affected by tariffs, so the median voter consuming low-quality goods will be
‘impervious to the imposition of tariffs’ but not higher income taxes (Adam, 2009, p. 611). Adam thus predicted
that majority voting will result in higher tariff revenue shares. Using a sample of 64 countries labelled by the
World Bank as Low Income, Lower-Middle Income or Upper-Middle Income, which fitted his assumptions of
trade patterns, Adam found that democracy had the expected positive sign and remained significant even after
controlling for bureaucratic quality, while not all ACH variables were significant.
2.3 What role can electoral institutions play?
However, Adam’s theory does not consider that democracies vary in form, particularly in terms of electoral
district magnitude. His theory presupposes a majoritarian system of democracy and is highly dependent on the
median voter to explain outcomes. Yet, according to the Database of Political Institutions (DPI) (Beck et al.,
2001), approximately 50% of countries classified as Low, Lower-Middle and Upper-Middle Income have some
form of proportional representation (PR) electoral system. When district magnitude exceeds 1, the median voter
may not be solely decisive, resulting in different conclusions regarding equilibrium tariff revenue share. No
study to date has investigated whether this is the case.
Much of existing research on the effect of electoral institutions on trade policy has uncovered a “protectionist
bias” in majoritarian systems: the majority party maximises the joint welfare of a majority of but not all districts,
and therefore adopts higher tariffs to redistribute income to industries linked to districts controlled by the
majority party (Grossman and Helpman, 2005). In contrast, in a PR system, the legislature is more insulated
from regional protectionist pressures (Rogowski, 1987), or more likely to represent the interests of the country
as a whole (Evans, 2009). Such theories focus on producers’ trade preferences, usually expressed through
narrow interest groups, instead of the preferences of ordinary voting consumers. An extension of Adam (2009)’s
4 THE POLITICAL ECONOMY OF TARIFF REVENUE DEPENDENCE
research might complement the majoritarian bias literature by providing an alternative, consumer-centric
perspective on how electoral institutions influence trade policy.
The broader literature on the representational consequences of electoral systems might similarly suggest that PR
systems exhibit lower tariff revenue shares. In developing countries, wealthy voters who demand high-quality
imports are the minority, and PR systems might lend these pro-trade minority interests greater influence over
policymaking. It has been argued that electoral systems with higher district magnitudes produce more
proportional electoral results (Taagepera and Shugart, 1989; Cox, 1997) that enhance political representation of
minority groups by lowering the vote share threshold required by a party to guarantee a legislative seat. Cross-
national and country-level studies of electoral district magnitudes in socially heterogeneous countries provide
evidence of a positive association between district magnitude and minority representation (Ordeshook and
Shvetsova, 1994; Clark, Gilligan and Golder, 2006). Since rich elites who import high-quality goods form the
minority of the electorate in developing countries, we might expect that higher district magnitudes enhance the
electoral representation of interests favouring imports and lower tariffs in developing countries, resulting in
lower tariff revenue shares.
2.4 Inequality, tariffs and voter preferences
Political economy explanations for tariff revenue dependence imply that tariffs serve a redistributive function
alongside other taxes: due to income-based differences in patterns of consumption and tendencies to import
high-quality goods, poor voters desire that the government raise more revenue via tariffs than income taxes so
that the tax burden is shifted from the poor to rich importing voters (Moutos, 2001; Adam, 2009). It has been
suggested that a rise in income inequality in a democracy – defined as an increase in mean income relative to
median income – increases voter demand for redistribution by taxation (Meltzer and Richard, 1981). Hence we
might expect inequality to influence how an electoral system aggregates voter preferences over tariff revenues in
a developing country.
To date, we know that inequality affects tariff revenue share directly, but the interaction between inequality and
electoral systems in determining tariff revenue share has yet to be investigated. Katsimi and Moutos (2010)
suggested that in a developing country engaging in vertically-differentiated trade, inequality is nonlinearly
related to tariff revenue share. In their model, higher inequality increases demand for redistribution through both
income taxes and tariffs. However, the optimal weight placed on each tax instrument depends on the existing
level of inequality. This is because the level of inequality determines the size of the tariff tax base and the extent
THE POLITICAL ECONOMY OF TARIFF REVENUE DEPENDENCE 5
to which it shrinks in response to higher tariffs (assuming changes in inequality are mean-preserving), which in
turn affects the revenue that can be collected from tariffs. Dutt and Mitra (2002) obtained similar findings in the
context of horizontally-differentiated trade: they found that the relationship between inequality and
protectionism depends on a country’s present capital endowment. These studies suggest that mean-preserving
increases in income inequality change the relative intensities of pro- and anti-tariff preferences among voters,
and may therefore influence the aggregation of policy preferences over tariffs.
In light of the findings and shortcomings of existing literature, I develop a theory that explains the effect of
varying electoral district magnitude on the aggregation of tariff preferences and the eventual tariff revenue share
allocated by the government.
3. Theory and Hypotheses
3.1 Setup
I propose a theory which extends Adam’s (2009) framework, explaining how tariff preferences are aggregated
under electoral systems of varying district magnitude. To do so, I first make the same assumptions regarding
world trade and consumer preferences as Adam did.
I assume that world trade operates such that specialisation of production takes place within a product type and
across quality levels of a vertically-differentiated good. Schott (2004) found that factor endowments can explain
within-product specialisation but not across-product specialisation in trade between the United States and
various developing and developed country trading partners. Capital-abundant, developed countries use their
endowment advantage to produce and export high-quality good varieties, while developing countries conversely
export low-quality goods and import high-quality ones from the developed world.
I also use the model developed by Moutos (2001) to define individuals’ preferences regarding tariff revenue
dependence in a developing country based on income and consumption patterns. Consider a vertically-
differentiated good � with quality �. Each consumer purchases 1 unit of � or none of it, but can choose quality
� from a range of continuous values to maximise utility. There is a “dividing” level of quality ��, below which
the developing country produces � at lower marginal cost than developed countries. Since the consumer would
like to minimise the price he/she pays for the good and price varies positively with marginal cost, the consumer
purchases quality levels below �� domestically at the price of the locally-produced good ����� , and imports
quality above �� from developed countries at price ��� under free trade. The consumer thus faces the
following budget constraint:
6 THE POLITICAL ECONOMY OF TARIFF REVENUE DEPENDENCE
� = ���� + ������if� < ����� + ����if� ≥ ��� (1)
where � is pre-tax income, � is the number of units of a homogeneous good consumed and � is the quality of a
fixed unit of the vertically-differentiated good consumed. The consumer maximises utility �(�, �) subject to
this budget constraint. Low-income consumers are assumed to maximise utility by consuming � < �� , while
high-income consumers maximise utility by consuming � > ��1.
The government has 2 tax instruments at its disposal – tariffs and income taxes – and raises revenue either by
implementing only tariffs (tariff revenue share 100%) or only income taxes (tariff revenue share 0%). With
income tax of rate 0 ≤ � ≤ 1 the budget constraint becomes:
where �� ′ is the new dividing quality, and ��& > �� since high-quality imports have become more expensive.
Figure 1 illustrates these new budget constraints and the corresponding new in utility-maximising (�, �) bundles for low-income consumers in a developing country. It shows that low-income consumers have higher
utility when government revenue is collected from tariffs rather than income tax. Low-income consumers spend
a larger share of their incomes on low-quality, locally-produced goods instead of high-quality imports, and are
therefore less sensitive to tariff hikes. High-income consumers, however, are more likely so spend a larger
share of their incomes on high-quality imported goods where � > ��, and thus obtain higher utility when
government revenue is collected from income tax rather than tariffs, making them more sensitive to tariff hikes.
[Figure 1 here]
3.2 Tariff preference aggregation under different district magnitudes
Based on these assumptions and the context of a developing country, I first consider the outcomes of political
competition in terms of tariff revenue share when district magnitude equals 1. My reasoning is guided by
Hotelling (1929)’s framework of competition as applied to party competition by Osborne (1995). There are ' citizens, and each citizen # has a complete and transitive preference relation over tariff rates such that each
citizen has a unimodal, symmetric distribution of utility over tariff rates, and gains highest utility from a most
preferred tariff rate ()* . Assume that citizens vote sincerely for candidates contesting a district based on tariff rate
THE POLITICAL ECONOMY OF TARIFF REVENUE DEPENDENCE 7
preferences, which vary with income as described in Section 3.1. There is a one-dimensional policy space ,
identified by a real line, which contains a unimodal distribution of voters’ ideal tariff rates. By restricting our
analysis to the single dimension of tariffs, we assume that preferences over other taxes are implicitly determined
by preferences for a particular tariff rate; for instance, a voter that has a high ideal tariff rate will have a low
ideal income tax rate, since tariffs and income taxes are substitute sources of government revenue.
As for the election candidates, there are - of them, each with a different party affiliation for simplicity and a
preference relation over ,. When a candidate . takes the policy position (/, his/her vote share is 0/((/). Candidate .’s preferences over , are then structured such that he/she gains highest utility from a position that
yields him/her the highest vote share amongst all candidates (. wins by plurality), followed by a position that
yields him/her a vote share equal to another candidate with the next highest vote share (. ties), followed by a
position that does not give him plurality (. loses).
The game of district-level electoral competition is as follows. It is a simultaneous one-shot game in which there
are 2 players2, candidates . and 1, who choose tariff rate platforms (/ and (2 respectively from the distribution
of voters’ ideal tariff rates. Candidates earn payoffs in terms of vote shares 0/3(/4 and 02((2). It can be shown
that the outcome of this game is a unique Nash equilibrium where candidates . and 1 both adopt the median
voter ideal tariff rate 5 and tie in vote shares (see Online Appendix A). In a developing country, the proportion
of society that is poor outweighs the proportion of the rich. Hence, the median voter is poor, and as explained
earlier, has a greater propensity to consume low-quality goods. Low-quality goods are produced domestically in
a developing country and their prices are not affected by tariffs. Assuming citizens vote based on economic
preferences, the median voter is less sensitive to tariffs than his/her richer counterpart, and is more supportive of
the government raising tariffs than other taxes such as income tax. Therefore, when district magnitude is 1, the
elected legislator in each district represents the poor median voter’s tariff preferences, and will vote to increase
tariff revenue share in the legislature. Replicating this result across all single-member electoral districts in the
country, most if not all politicians with a seat in the legislature favour increased tariff collection, and the
legislature votes to collect a high proportion of government revenue from tariffs.
I now consider political competition over tariff policy positions when district magnitude exceeds 1. In this case,
several features of the aforementioned game are modified. Firstly, the number of electoral contenders in a
district is likely to increase beyond 2 candidates3. When there are more than 2 candidates, it can be shown that
there is no pure strategy Nash equilibrium where all candidates converge to the same position, at the median or
8 THE POLITICAL ECONOMY OF TARIFF REVENUE DEPENDENCE
elsewhere, even if each aimed to win a plurality of votes to guarantee winning a seat (see Online Appendix B).
Therefore, candidates may not pander to the poor median voter’s preferred tariff policy in their campaigns.
Some candidates may choose to represent the rich minority’s preference for lower tariffs, while other candidates
may continue to represent poor voters. This increases the likelihood of candidates contesting elections on
platforms advocating lower tariffs.
Secondly, on top of being more likely to contest elections, candidates representing the rich minority are also
more likely to guarantee a legislative seat in a multimember district under PR arrangements when they contest.
When more than 1 legislator can be elected in a district, the minimum vote share needed by any candidate to
guarantee a seat, or the exclusion threshold, decreases. To illustrate, consider a list PR system that allocates
seats in a multimember district using quotas. A quota 6 is the minimum number of votes that guarantees a
candidate a seat in a district ":
6 = 7�8� + 9 (4)
where 7� is the total number of valid votes cast in district ", 8� is the number of seats contested in district "
(i.e. district magnitude), and 9 is a modifier that varies in value with the type of quota used (Clark et al., 2009).
The exclusion threshold : is then given by the ratio of the quota to the total number of votes cast4:
: = 67� =
18� + 9 (5)
Equation (5) implies that : decreases as 8� increases. Hence, as district magnitude increases, the threshold for
guaranteed representation falls towards the vote share obtained by candidates representing wealthy minority
interests, and candidates advocating tariff rates lower than the median voter’s ideal rate are more likely to be
elected to the legislature. Overall, under large district magnitudes, wealthy pro-trade interests gain political
representation and moderate the poor majority’s preferences for high reliance on tariffs revenues. Hence, I
expect to see lower tariff revenue shares when district magnitudes are greater than 1, and I test the following:
Hypothesis 1: In comparing developing countries with democratic elections, higher electoral district
magnitude is associated with lower tariff revenue share of government revenue.
Note that Hypothesis 1 reflects a “majoritarian bias” similar to that found by Grossman and Helpman (2005).
Therefore, my hypothesis may provide an alternative explanation for the “majoritarian bias”, one that considers
the role of consumer preferences in influencing protectionist policies.
3.3 The effect of inequality on tariff preference aggregation under different district magnitudes
THE POLITICAL ECONOMY OF TARIFF REVENUE DEPENDENCE 9
A second hypothesis concerns the effect of income inequality on the aforementioned institutional explanation
for tariff revenue dependence. Extending Katsimi and Moutos’ (2010) reasoning, I propose that a mean-
preserving increase in inequality causes the number of voters who can afford high-quality imports to fall relative
to voters who cannot afford high-quality imports. As a result, the proportion of voters with high incomes that
would vote for lower tariffs and tariff revenue share falls. This reduces the probability that a candidate
representing wealthy pro-trade minority interests wins a seat. Using threshold : defined in Section 3.2, suppose
that the probability that candidate . wins a seat in an 8-seat district " with - contesting candidates is �;<=>: �;<=> = � 1if0/ ≥ :
ℎ30/ , -,8�4if0/ < :� (6)
where 0/ is .’s vote share, and ℎ&(0) > 0, ℎ&(8�) > 0 (i.e. �;<=> increases with own vote share and district
magnitude respectively). Section 3.2 showed that increasing 8� to 8�′ reduces the exclusion threshold to, say,
:’ < :. The expression for �;<=> becomes:
�;<=> = � 1if0/ ≥ :′ℎ30/ , -,8�′4if0/ < :′� (7)
where ℎ30/ , -,8�4 < ℎ30/ , -,8�′4; that is, holding all else constant, the probability that . wins a seat increases
as district magnitude increases, as shown in Section 3.2.
Now consider the effect of a mean-preserving increase in income inequality on �;<=> . If . were a candidate
representing wealthy pro-trade minority interests, increased inequality lowers his vote share to 0/′, and �;<=> falls since ℎ&(0) > 0. Figure 2 illustrates how �;<=> falls with increased inequality. If inequality is severe, �;<=> could even fall below the probability of winning under a lower district magnitude. Consequently, high inequality
reduces the efficacy of large district magnitudes in increasing the political influence of minority interests
through the electoral mechanism. The extent to which large district magnitudes lower the equilibrium tariff
revenue share is thus diminished with high inequality.
[Figure 2 here]
This leads me to test:
Hypothesis 2: In comparing developing countries with democratic elections, an increase in district
magnitude should lower tariff revenue share of government revenue by less when inequality is high.
4. Methodology and Data
4.1 Model
10 THE POLITICAL ECONOMY OF TARIFF REVENUE DEPENDENCE
I empirically test the aforementioned hypotheses using the following fixed effects regression model:
A-_:CD'EE*> = FG8H8�*> + FIJ'-'*> + FKJ'-'I*> + FL8H8�*> ∗ J'-'*> + NOPQ--QRR*> + ST*> +UV*> + WX*> + YZ*> + [* + \*> where i indexes the country and t the year. LN_TARIFF measures natural log of tariff revenue share, MDMH
measures district magnitude, the main treatment for Hypothesis 1, and MDMH*GINI measures the interaction
between the Gini inequality index and district magnitude to test Hypothesis 2. Data used to construct these
variables are described in Section 4.2.1. OPENNESS measures trade share of GDP (see Section 4.2.3). C is a
vector of controls for MDMH. A is a vector of variables for ACH explanations. E is a vector of variables for
elite dominance explanations. N is a vector of variables for nontax revenue sources as alternative explanations.
[* denotes country fixed effects. Standard errors are clustered by country.
4.2 Data
The overall dataset describes 139 developing countries between the years 1990 and 2013. The dataset is
restricted to 139 countries as I only use countries classified by the World Bank as Low Income, Lower-Middle
Income or Upper-Middle Income to satisfy the assumption that developing countries export low-quality
products. Only the period between 1990 and 2013 is covered by this study due to data availability.
4.2.1 Treatment and outcome variables
The outcome variable is LN_TARIFF, which describes the log of customs and import duties as a share of tax
revenue from the World Development Indicators (WDI). The treatment testing Hypothesis 1 is district
magnitude, captured by the variable MDMH. MDMH describes mean district magnitude in the lower house from
the DPI. Only lower house data is used because not all countries in the sample have multi-chamber legislatures.
Country-years where district magnitude is coded as Not Applicable or Indirect are excluded5. The treatment
testing Hypothesis 2 is income inequality, as captured by the variables GINI, its square GINI2 and the interaction
MDMH*GINI. These variables are constructed based on Gini indices from the WDI. While Gini indices are
reported only for 740 out of 7367 country-year observations in the dataset, its strength lies in its common usage
by many researchers, which facilitates comparison across studies.
4.2.2 Control variables (C)
As controls, I include variables that potentially cause the non-random assignment of district magnitudes to
national economies and confound the relationship between political institutions and trade and tax policy
outcomes. The first set of controls comprises dummy variables that indicate whether a country has based its
THE POLITICAL ECONOMY OF TARIFF REVENUE DEPENDENCE 11
legal system on English, French, German, Socialist or Scandinavian law: LEGOR_UK, LEGOR_FR,
LEGOR_GE, LEGOR_SO and LEGOR_SC respectively (La Porta et al., 1999). The second set of controls
comprises dummies for structural adjustment agreements that potentially affect variations in both political
institutions and trade and tax policies: IMF_SBA and IMF_EFF indicate whether a country signed an IMF
Stand-By Agreement or Extended Fund Facility agreement respectively in a year (Dreher, 2006). Finally, the
level of economic development as measured by real per capita income from the WDI, and captured by the
variable INCOME, is controlled for.
4.2.3 Alternative explanations: Tariff tax base
A country may be highly dependent on tariff revenues simply because its taxable tariff base is large. I use the
variable OPENNESS, which measures total trade (exports plus imports) share of GDP from the WDI, to proxy
the size of its tariff base in a given year.
4.2.4 Alternative explanations: Nontax revenue sources (N)
A government can rely less on tax revenues if it obtains funds from other sources. To account for these
alternative revenue sources, I firstly include variables OIL and MINERAL, which measure oil and mineral rents
respectively as a percentage of GDP. I also include the variable ODA, which measures the amount of net official
development assistance received as a share of GDP, as well as IBRD, which measures loans from the
International Bank for Reconstruction and Development and credits from the International Development
Assistance as a share of GDP. Data was obtained from the WDI.
4.2.5 Alternative explanations: Administrative Capabilities Hypothesis (A)
As suggested by the ACH literature in Section 2.1, numerous factors affect the government’s administrative
capabilities of collecting directing taxes and the extent it might have to rely on indirect taxes like tariffs as a
substitute. I therefore include BUREAUQUAL as a measure of bureaucratic quality according to the
International Country Risk Guide (ICRG), measures of population size POP and population density POPDEN,
rural population share as a measure of urbanisation and secondary school enrolment rates SECONDARY to
measure the population’s education level. I also include the ICRG corruption index CORRUPT since corruption
also negatively affects the government’s ability to collect various taxes (Bai and Wei, 2001; Gatti, 1999).
4.2.6 Alternative explanations: Elite dominance (E)
My proposed theory requires that legislatures are ‘mature and autonomous’, capable of ‘injecting the interests
and concerns of their constituencies into the policy process’ (Diamond, 1997). Yet, numerous developing
12 THE POLITICAL ECONOMY OF TARIFF REVENUE DEPENDENCE
country democracies suffer from weak legislatures that face obstacles in representing voters’ interests. This