Trade, Democracy, and the Gravity Equation 1Miaojie Yu China Center for Economic Research Peking University, China June 17, 2006 [Abstract] Democracy has scale and selection effects on trade. Importer’s democratization could change various trade barriers and hence affect trade flow (scale effect). Simultaneously, a democratic country would become a favorable exporter due to mutual ideological recognition and other concerns (selection effect). Thus, the net effect of democratization on trade remains an empirical issue. In this paper, we present a theoretical gravity model with democracy. Using a rich panel dataset while clearly controlling for the endogeneity of democracy, we find robust empirical evidence that democracy significantly fosters trade. Overall, democratization accounts for around 3% trade growth via scale effect and 8% via selection effect, ceteris paribus. Keywords: Trade, Democracy, Gravity Equation, Transportation Costs JEL Classification: F13, F14 1 China Center for Economic Research, Peking University, Beijing, 100871, China (e-mail: mjyu@). I thank Robert Feenstra, Joaquim Silvestre, Gordon Hansen and Justin Lin for their very helpful comments. This paper is also benefited from seminar participants at the University of Hong Kong and Peking University. However, all errors are mine.
Relatively little research has concentrated on effects of democratization on trade globalization —one of the most intriguing topics on international political economy. As Dani Rodrik (1995, ) noted, “Theoretical and empirical work relating institutional contexts to trade policy outcomes is in its infancy but should be a promising area of research”. Over the last four decades trade has grown dramatically. As shown in Figure 1, the average bilateral log export increased from in 1962 to in 1998, which increased around 16%. Simultaneously, global political liberalization emerged during this period. The average democracy index increased from in 1962 to in 1998 according to reports from Polity IV—a widely accepted database on democracy. Thus, such a phenomenon raises the question: does global democratization foster trade? More specifically, can an importer’s democratization promote trade? And can an exporter’s democratization encourage trade? The effect of an importer’s democratization on bilateral trade is theoretically ambiguous. An importer’s democracy could foster trade flow, which we call a positive “scale effect”, for at least two reasons. First, democracy guarantees the protection and enforcement of intellectual property rights in an importing country, and the corresponding reduction in uncertainty fosters international trade and economic growth (Rodrik, 2000). Second, an importer’s political liberalization restricts the ability of a government to use strategic trade policies (such as barriers) in labor-abundant countries in order to earn political support (Helen V. Milner and Keiko Kubota, 2005). However, democratization also has a negative scale effect on trade. An importer’s democracy could discourage trade for two reasons: First, democratic countries may engage in excessive redistribution programs, which could lower economic growth and 1
then harm trade (Robert J. Barro, 1996). Second, democratization implies transferring power from non-elected elites to the wider population group, most of whom will be workers. In line with the Heckscher-Ohlin-Stolper-Samuelson theorem, Kevin O’Rourke and Allen M. Taylor (2006) proposed that democratization would increase protectionism in capital-abundant countries in which workers may benefit from trade barriers. As a consequence, trade would shrink in those countries. In addition, democratization could also have a “selection effect” on trade. A country may prefer to import goods from countries with a high level of democracy for the following reasons: First, countries with a high democracy level may prefer to trade with other democratic regimes, ceteris paribus, due to mutual ideological recognition. One good example is that American people may prefer to consume products from the UK over those from China unless products made in China have striking price advantages. Second, countries with a low democracy level may prefer to import products from democratic countries. One possible reason is the attractive institutional stability such as the transparent transaction process in democratic countries. Least but not the last, it is also relatively easy for a democratic exporter to join a regional free trade agreement, ceteris paribus . For example, Thailand, a relatively highly democratic country, joined the Association of South Eastern Asian Nations (ASEAN) in 1992. In contrast, Vietnam, a relatively low democratic nation, joined the ASEAN in 1995, even though these two developing countries have very similar economic structures. To summarize, since the scale effect of an importer’s democratization on trade is ambiguous, the net effect of the democracy of trading countries on trade is uncertain and remains to be an empirical question. 2
Some researchers have made significant contributions on this issue. Barry Eichengreen and David Leblang (2006) provide an excellent survey for the related literature. As they pointed out, current literatures include, among others already mentioned, Bernard Grofman and Mark Gray (2000), Dennis Quinn(2000), Jan Fidrumc (2001), Francesco Giavazzi and Guido Tabellini (2005), and Miaojie Yu (2005). Particularly, Grofman and Gray (2000) suggest a negative effect of authoritarianism on trade by examining the impact on trade of the number of years a country was ruled by an authoritarian regime. Based on a larger country sample, Giavazzi and Tabellini (2005) obtained a similar result by using the widely accepted Polity IV dataset maintained by Monty G. Marshall and Keith Jaggers (2004). Fidrumc (2001) finds a strongly positive effect of democracy on economic growth in 25 transition countries. Quinn (2000) investigates the impact of democracy on international financial liberalization and concludes that democratization is more likely to remove various capital controls. All of these studies treat democracy as exogenously given; however, globalization could have an impact on democratization. Based on this concern, Yu (2005) finds a striking positive effect of trade on democracy, although the reverse effect is not always significant, when estimated by the simultaneous equation method. However, most of the previous studies are based on reduced-form estimations. The lack of a theoretical model could make estimation results volatile and biased. For this reason, in this paper we perform a structural form estimation based on an augmented gravity model, inspired by Paul R. Krugman (1980), Elhanan Helpman (1987), Scott L. Baier and Jeffery H. Bergstrand (2001), and Robert C. Feenstra (2002). The reason for 3
adopting the gravity equation here is that it has already been shown to provide a powerful explanation for the world trade pattern. Krugman (1995) asks: why has trade grown so fast? Feenstra (1998) suggests three relatively obvious reasons: the growing GDP, trade liberalization, and falling transportation costs. Baier and Bergstrand (2001) find evidence confirming his idea using a gravity model. However, the institutional context could also have a significant effect on trade (Rodrik, 1995). The question is challenging partly because of the endogeneity of democracy. Previous studies usually avoid discussing this issue since it is difficult to find appropriate instruments. Because of this drawback, existing results might suffer from some estimation bias. In this paper we are able to clearly control for the endogeneity issue by adopting novel instruments for democracy. We estimate the net effect of democratization on trade using a structural specification derived from the gravity model with democracy. Our estimates are based on a rich panel data set for 134 IMF member countries over the last four decades. We find robust empirical evidence that democracy significantly fosters trade. Overall, democratization accounts for 3% of the growth of trade via the scale effect, and, for 8% via selection effect, ceteris paribus, controlling for the endogeneity of democracy. The rest of the paper is organized as follows. Section I presents a theoretical gravity equation. The key innovation here is to express it in a form that relates bilateral trade to country size, price level, various geographical factors, and democracy levels. Section II describes the procedure for estimating the theoretical gravity model. The main estimation 4
results and sensitivity analysis are discussed in Section III. Section IV closes with suggestions for future research. I. Theoretical Gravity Model Jan Tinbergen (1962) was the first to use a gravity equation to describe the trade pattern. In its simplest form, the gravity equation suggests that bilateral trade is directly proportional to the trading entities’ GDP. Based on this motivation, James Anderson (1979) provided a theoretical micro-foundation for the gravity equation based on a constant elasticity of substitution (CES) utility function, which has become a standard setup in subsequent work. An innovation of the present paper is our modification of the CES utility function by embedding a democracy index into the model. This modification is crucial in deriving a simple gravity equation which is operational for estimation. Suppose that each country produces unique product varieties, the export of good k from country ito countryjis identical to the consumption of good kin countryj. Democratization in countryj affects its imports and hence consumption via the change of tariffs and various non-tariff-barriers. For example, if countryjincreases tariff of good k due to pressure from labor unions, then the import of good k from countryi to countryjdecrease. Assume that country i=1,…,Iproduces Ncommodities, and consider the CES iutility function: σ−1NIiσ(1) U=[exp(z)⋅C], (σ>1) j∑∑iijki=1k=1 5
whereCdenotes the consumption in countryjof good k produced by country i. The ijkelasticity of substitutionσis assumed to be higher than one. The bilateral trade volume and hence consumption C, will be affected by tariffs and non-tariff barriers. ijkFurthermore, the selection effect implies that importerj’s aggregate welfare (utility)Ualso depends on the exporter’s democracy index z. This reflects the fact that jiimporting products from a high democratic exporter icould bring more satisfaction to the importerj. Thus, we model the aggregate utility function of country jas a strictly 2increasing function of the democracy indexzof exporter i(i=1,...I). We specify an i3exponential function here for ease of estimation. For brevity, and in agree with previous studies, we assume that, given iandj, i′p=p for all kand kin{1...N}, ., all the varieties imported by country jfrom ijk′ijkicountry have the same price p. Then consumption in countryjis also identical over ijithe entire line of products sold by countryi, .,C=C,∀k∈{1,...N}. Utility function ijkij(1) can then be expressed as: σ−1Iσ(2) U=N⋅[exp(z)⋅C]. j∑iiiji=1 2 Note ifi=j, Country jproduces but not import varieties from Countryi. 3 Since data on democracy index is scaled from -10 to 10, it is inappropriate to use a simple linear increasing function. 6
The representative consumer in the importing countryjmaximizes his/her utility (2) subject to his/her budget constraint: I(3) Y=NpC, j∑iijiji=1where Y is importerj’s GDP level. Observe that democracy is not included in the jbudget constraint (3) since democracy is not a commodity. Solving this maximization problem, we obtain the derived demand function for each productC: ij−σσ−1(4) C=(p/P)(Y/P)⋅(exp(z)), ijijjjjiwhere the aggregate price index Pis defined as: j1I1−σ1−σ(5) P=[N(p/exp(z))]. j∑iijii=1Finally, the total exportX from countryito countryj is: ijNi(6) X≡pC=NpC, ij∑ijkijkiijijk=1where the first equality follows the definition of export value, whereas the second is due to equal price assumption across varieties. Combining (4), (5) and (6), we obtain the export value from countryi to countryj: 1−σσ−1(7) X=NY(p/P)[exp(z)]. ijijijjiPaul Samuelson (1952) suggests that there exists an “iceberg” transport cost T across ijborders. In order to have one unit of the product reach the destination countryj, we need 7
T≥1units of the product shipped from the departure country. Hence, the price on a . ij(cost, insurance, freight) basepequals the product of the “iceberg” transport cost Tand ijijthe price on a . (free on board) basep. The iceberg transport cost is also a function iof an importer’s democracy index due to the scale effect we argued before, .,p=T(z)p. Thus, (7) can be written as: ijijji1−σ1−σσ−1(8) X=NY[T(z)/P]p[exp(z)]. ijijijjjiiClearly, in the gravity equation (8), the bilateral trade depends on the importing country’s GDP and the aggregate price index, the trading countries’ democracy levels, and the . price. However, bilateral trade is also affected by the number of varieties in the exporting countryN which is unfortunately unobservable. For estimation, we consider the imonopolistic competition model presented originally by Krugman (1979), which helps us eliminate the number of exporting varieties in our gravity equation (8). As in Krugman (1979), Baier-Bergstrand (2001), and Feenstra (2003), the representative firm in countryimaximizes profits. Specifically, the production of goods (y) incurs a fixed cost (κ) and constant marginal cost (φ) given that labor (l) is the ifirm’s unique input: (9) l=κ+φ⋅y. iiThe monopolistically competitive equilibrium implies two conditions for the representative firm. First, the firm’s maximization behavior requires that marginal revenue should be equal to marginal cost. Since the elasticity of demand equals the 8
elasticity of substitutionσwhen countryi’s number of varietiesNis large, we obtain the ifirst equilibrium condition: σ(10) p=()φ⋅w, iσ−1where wage is denoted asw. Second, the representative firm obtains zero profits due to free entry. Given that the firm’s profit function in countryi isπ=py−w(κ+φy), we obtain the equilibrium iiiiproduction levely for such a representative firm in countryi: i(11) y=(σ−1)κ/φ, iwhereyis a constant number given thatσ,κandφ are all constant parameters. It is also inoted that the GDP in countryiisY=Npy, and substituting this into (8), we have: iiiiYYT(z)ijijj1−σσ−1(12) X=⋅[]⋅[exp(z)]. ijiσ(p)yPiijTherefore, bilateral trade depends on the trading countries’ GDP, the iceberg cost, the trading countries’ democracy levels, the exporting firms’ fixed production, and various price indexes. For the readers’ convenience, we include the notation of the model in Table 1. II. Empirical Methodology To estimate the gravity equation (12), we specify the estimating equation by taking logs on both sides: (13) lnX=ln(YY)−σlnp+(1−σ)lnT+(σ−1)lnP+(σ−1)z−lny . ijijiijjii 9
Following James Anderson and Eric van Wincoop (2003), the bilateral iceberg cost Tcan be broken into the bilateral distance costgand other border factors. Such factors ijijcould include: (a) the exporter’s democracy level z; (b) the indicator of a common land jborder: whether the trading countries share a common land border is important in reducing transportation cost; (c) the number of countries landlockedL; and (d) the ijnumber of island countriesI. Intuitively, countries far from sea trade less; countries ijwhich are farther apart trade less, while countries with many islands benefit from the convenient transportation and thus trade more. Hence, we have: (14) lnT=α+ρz+ρlng+ρB+ρL+ρI+µ, ijij0j1ij2ij3ij4ijijwhereB is a dummy variable which is unity if country iand countryjshare a common ijborder and zero otherwise. The constant termαcaptures any other border effects which ijare not specified in (14). Note that tariffs are not included here since global tariffs data are still currently unavailable. Thus, the effect of tariffs on transportation cost is partially absorbed by the importer’s democracy indexz, as we discussed above. jNow we can obtain the estimating equation for each periodt, moving the logarithm products of the GDP to the left side and substituting (14) into (13): (15) ln(X/YY)=(σ−1)z+(1−σ)ρz−σlnp+(1−σ)ρlng ijtitjtit0jtit1ij +(1−σ)ρB+(1−σ)ρL+(1−σ)ρI 2ij3ij4ij +[(1−σ)α−lny+(σ−1)lnP+(1−σ)µ]. ijtitjtijt 10
In this specification, bilateral trade openness—the logarithm of export value relative to the trading countries’ GDP—mainly depends on the exporter’s democracy level (z), ithe importer’s democracy level (z), the bilateral log distance (lng), the exporter’s . price index (lnp), and the importer’s log aggregate price index (lnP). In addition, ijbilateral openness is also affected by various borders’ effects (α,B,LandI) and ijijijijexporter’s firm productiony. i However, the importer’s aggregate price indexP in Specification (15) is junobservable since it depends on the unobservable exporter’s varieties numberNaccording to (5). To address this empirical challenge, Anderson and van iWincoop (2003) present an implicit price index based on the market equilibrium condition. Feenstra (2002) on the other hand recommends the fixed effects method to take account of this unobservable price index since it is relatively simple. The idea is that the aggregate price index term, the exporter’s fixed production level, and the various unspecified border effects can be absorbed into the error terme, .: ijte=(1−σ)α−lny+(σ−1)lnP+(1−σ)µ. ijtijtitjtijtThis error term can be decomposed into a country-pair random variableϕ , a year ij2specific effectωand an idiosyncratic effectε with normal distribution:ε~N(0,σ). tijtijtijAccordingly, we first perform fixed-effect estimations for the following specification: Xijt(16) ln()=β+βz+βz+βlnp+βlng 01it2jt3it4ijYYitjt 11
+βB+βL+βI+ϕ+ω+ε, 5ij6ij7ijijtijtwhereβ=σ−1,β=(1−σ)ρ,β=−σ,β=(1−σ)ρ,β=(1−σ)ρ,β=(1−σ)ρ1203415263andβ=(1−σ)ρcompared to Specification (15). Our main interests are the signs of the 74coefficientsβandβ. Of course, the time-invariant geographical variables will be 12dropped out automatically in the estimations. However, we cannot rely much on the fixed-effect estimates given here that the democracy indexes in most of the developed countries do not change over time. For example, within our sample, the democracy index in the USA remains at the same level over the years. As a consequence, such observations will be dropped out automatically as 4this might cause an estimation bias. Hence, the alternative random-effect estimation makes much more sense here assuming that the country-pair random variable ϕ in ij5Specification (16) is uncorrelated with other regressors. III. Data, Econometrics and Results In this section we first describe datasets used in the paper, followed by a discussion of econometric methods. We then address the possible endogeneity problem. Finally, we close the section with various robustness checks. 4 This is true regardless of the estimation methods used here. The within fixed-effect and the between fixed-effect are the same when the years covered in the data are more than two (Wooldridge, 2002). 5 One can perform the Hausman test (1978) here to compare the random-effect and fixed-effect estimations. To save space, we do not report the results here. 12
A. Data The regressand of (16) is the logarithm of bilateral aggregate openness which is the bilateral exports from country ito countryjrelative to the product of both countries’ GDP. Note that our definition of trade openness is different from the traditional one, which is the sum of exports and imports relative to a country’s GDP. The nominal export data comes from the NBER-UN Trade data maintained by Feenstra et. al.(2005). Information related to the log product of real GDP data (in constant US dollars) between trading countries are directly from Andrew K. Rose (2004). Since his dataset ends in the year 1998, we obtain 185,125 observations for 134 countries during the years 1962-1998. Table A in the Appendix lists all countries used in the estimations. We use the exporter’s consumer price index (CPI) to measure the exporteri’s price levelp. Such data can be accessed from the World Development Indicator (WDI, 2002) iby the World Bank, which specifies the base year of the CPI as 1995. The trading countries’ democracy levels—the key variables in Specification (16)—are taken from the Polity IV dataset by Marshall and Jaggers (2004), which is a widely accepted dataset to measure world democratization. Many previous studies (Quinn, 2000, Milner and Kubota, 2005, Yu, 2005, and Eichengreen and Leblang, 2006) use this dataset to construct the democracy index. Precisely, Polity IV includes annual composite indicators measuring both “institutionalized autocracy” and “institutionalized democracy” for just about every independent entity with a population over 500,000. The political liberalization index is defined as the difference between the democracy indicator and the authoritarian indicator. Since each indicator is an additive eleven-point scale (0- 13
10), the index is scaled between -10 and 10 consequently. The higher the number, the higher is the level of political liberalization of a country. The traditional gravity equation suggests that bilateral trade is also affected by various geographic factors. In particular, we include the great-circle bilateral distance, the number of islands that the trading countries have, the dummy of the common land border, and the number of land-locked countries. Again, this data can be accessed directly from Rose (2004). To control for endogeneity, we also include data on death penalty and on judicial independence for each country over the last four decades. The death penalty data are from Amnesty International (2005). For precision, we consider four death-penalty regimes, according to whether the death penalty is: (a) absolutely outlawed; (b) allowed in extreme cases; (c) de facto banned (., death penalty is sanctioned by law but has not been practiced for ten or more years.); or (d) permitted. Thus, the death penalty variable is zero if the death penalty is allowed at a particular year in a country; and one otherwise. Table B in the Appendix lists the countries in which the death penalty is prohibited. Data on judicial independence for trading countries can be obtained from the “Economic Freedom of the World” (James Gwartney and Robert A. Lawson, 2005). This 6is also an additive scale between 0 and 10 and spans every five years from 1970 to 2000. Panel A of Table 2 presents descriptive statistics for each variable while Panel B describes their partial correlations. As we can observe, the exporter’s democracy has almost no correlation with the importer’s democracy. Furthermore, the trading countries’ 6 Note that we substitute the data of year 2000 to that of year 1998 since the latter is unavailable. 14
7democracy variables are not highly correlated with all other gravity variables. These imply that the multicollinearity is not a problem for the coefficient of interest. B. Estimates We first perform a pooled ordinary least square (OLS) estimation to obtain benchmark results, and then report in Table 3 the estimates for fixed-effect and random-effect. Since the random-effect estimate is the most appropriate one given our data structure, we rely it for our analysis. Column (3) in Table 3 evidences the good properties of the random-effects estimates. 2Overall, the factors explain 20% of the growth of trade openness (R=). All geographic factors are economically and statistically significant. Countries with long ˆdistance trade less (β=−). Countries with common land borders trade more 3ˆˆ(β=), and countries with island also trade more (β=). In contrast, countries 46ˆwhich are land-locked trade less (β=−). All of these results are consistent with 5many previous related studies like Rose (2004) and Arvind Subramanian and Shang-Jin Wei (2003). The coefficient of the exporter’s democracy—one of the key variables of our ˆinterest—is β= with a high t-value. This means that a one scale increase in the 1exporter’s democracy leads to around a two percentage point increase of bilateral ˆopenness. The estimated constant elasticity of substitution isσ=, sinceβ=σ−1, 1 7 We did not report these to save space. Readers who are interested in such correlations can contact us directly. 15
which is also consistent with our theoretical assumptionσ>1. Analogously, the ˆimporter’s democracy, another key variable, has an estimate ofβ=. This suggests 2that the semi-elasticity of the importer’s democracy on trade is about a unit. ˆLet us turn to the price index. The random estimate turns out to beβ= 1ˆˆˆandβ=, which are inconsistent with the theoretical requirementβ≠−(β+1). 331ˆHowever, the coefficient of β can not be taken too seriously since it is statistically 3insignificant. This economically and statistically insignificant estimate comes from the measurement error of the exporter’s price index (lnp), which is a common shortcoming iof using a published price index data. As pointed out by Feenstra (2003), most of the published price index data are measured relative to an arbitrary base period (we choose 1995 in our estimation), which usually undermines the accuracy of the estimates. We then identify the coefficients in Specification (14) based on our estimation results. For example, in the random-effect estimate, the coefficient of distance ˆˆˆisρ=β/(1−σ)=− We now obtain coefficients for other 14geographic variables using the same method, which are presented in Table 4. All numbers have the anticipated economic meanings. The rise of an importer’s democracy reduces the iceberg cost due to the removal of trade barriers. The coefficient is a half for the random-effect estimation. Geographical distance significantly increases the iceberg transportation cost. Moreover, countries which are farther from sea suffer from the high iceberg cost as well. More interestingly, countries with common land borders have iceberg cost times lower than those without common land borders. This strikingly 16
large number could be biased due to the endogeneity problem. After controlling for that, we find that the effect of common land border becomes reduced to times only. Our final step is to offer a more economic explanation for these two key variables. Comparing 1962 with 1998, the world average exporter’s democracy index increased by points, while the average log bilateral openness increased by points. This means that the exporter’s democracy explains around % of the growth of bilateral openness since we * Similarly, given that the average importer’s democracy increases by about points, it accounts for about % of the growth in bilateral trade, ceteris paribus, * C. Endogeneity Issues We now explore the issue of endogeneity issues for the democracy indexesz iandz. Such democracy indexes are correlated with the error term (cov(z,ε)≠0, jitijtcov(z,ε)≠0 ) for two reasons. First, a source of endogeneity is the problem of jtijtomitted variables (Jeffery M. Wooldridge, 2002). Note that the error termε includes the ijtimporterj’s aggregate price indexP, which, in turn, includes the unobservable number jof varietiesNaccording to (5). In addition to this, the unobservable exporter’s fixed iproductionyis also absorbed into the error termε as well. These two omitted variables iijtthus lead to the endogeneity of the democracy index. Second, the democracy indexes zand zof the trading countries are also obviously correlated with the error term since ijboth variables are included into the importerj’s aggregate price indexP. j 17
The two-stage least squares (2SLS) estimation is a standard econometric method to address the endogeneity problem (Wooldridge, 2002). However, to our limited knowledge, few previous works perform such estimation since researchers immediately face the challenge of choosing good instruments for democracy—it is very difficult to find variables that affect only democracy but not trade. We introduce in the present paper two novel instruments for the 2SLS estimations: the dummies of death penalty abolition and the judicial independence. Clearly, these two variables are highly correlated with a country’s democracy level. Yet the intriguing matter is that they are not necessarily related to a country’s trade activities. A country with a high level of trade openness may well maintain the death penalty and have low judicial independence. Moreover, these two instruments are excluded from the democracy index of the Polity IV dataset. As we mentioned before, the political liberalization index is defined as the difference between institutionalized democracy indicator and institutionalized autocracy indictor. Both of them are derived from the coding of the openness and competitiveness of executive recruitment, competitiveness of political participation, and constraints on the chief executive using some specific weights. Table C in the Appendix offers the 8formation of the polity index in the Polity IV dataset. As one can observe, the two instruments we use here are not included into such components. 8 Readers who are interested in the components of political liberalization can refer to the dataset user’s manual of Polity IV project maintained by Marshall and Jaggers (2003). 18
We then report the partial correlations for each variable, shown in Panel B of Table 2. The good news is that the exporter’s dummy of death penalty abolition is highly correlated with its democracy level (corr=), but weakly correlated to its trade openness (corr=). Similarly, the exporter’s judicial independence variable is strongly correlated with its democracy level (corr=), but relatively weakly correlated to its trade openness (corr=). Similar observations apply to the importers. The last two columns in Table 3 present various 2SLS estimates using both dummies of death penalty abolition and judicial independence as instruments. Our sample is reduced to 21,257, and only 123 countries are included due to the availability of such instruments data. Column (4) shows the fixed-effect estimates while Column (5) shows the random-effect estimates. In comparing Column (5) to Column (3), one can observe that the coefficient of the exporter’s democracy in the 2SLS estimate () is higher than that obtained by OLS (). In contrast, the coefficient of an importer’s democracy in column (5) is identical to that in column (3), which is . Such coefficients are all significant at less than 1% in the conventional statistical sense. We now turn to the economic meaning of our measurements. As we mentioned, the world average exporter’s democracy index increases by points, while the average log bilateral openness increases by points. Thus, the exporter’s democracy explains around % of the growth of bilateral openness, * Similarly, the importer’s democratization accounts for around % of the growth in bilateral trade, * 19
In a nutshell, all our results are robust using various econometric methods. Since the impacts of democracy on trade are economically and statistically significant, we can safely conclude that global democratization fosters world trade. D. Further Robustness Checks Given that a country’s income varies, is the effect of democratization on trade sensitive to income? Previous studies suggest an affirmative answer. Milner and Kubota (2005) find evidence that democratization in developing countries leads to higher trade flow since politicians in democratic countries cannot use strategic trade policies to win political support from special interest groups like labor unions. In contrast, O’Rourke and Taylor (2006) suggest that democratization could discourage trade in capital-abundant countries because workers, who get more political power from the democratization, may prefer protectionism along the lines of the Heckscher-Ohlin-Stolper-Samuelson theorem. We thus perform our estimations by dividing the sample by country groups according to their income per capita. We do not restrict ourselves to two groups (developing and developed countries); instead, we split all countries into five categories according to the 2004 Gross National Income (GNI) per capita level reported by the World Bank: (a) low income countries ($825 or less); (b) lower middle income countries ($826 - $3,255); (c) upper middle income countries($3,256 - $10,065); (d) high income non-OECD countries($10,066 or more); and (e) high-income OECD countries ($10,066 or more). The list of countries included in the estimation is reported in Table D in the Appendix. Note that here we do not use the dummy variables to capture the income difference. This is mainly because we estimate structural forms which are based on a theoretical model, and hence we cannot add extra variables arbitrarily. 20
Table 5 presents the results, separating exporters and importers into each income group and using random-effect OLS and 2SLS estimations. Only the exporter’s and importer’s democracy variables are reported here to save space. Briefly speaking, we find that the exporter’s democratization fosters trade for most of the income groups. The coefficients of democracy vary between and for both OLS and 2SLS estimates, confirming our previous results at the global level: a high democratic country will export more due to the mutual ideological recognition and the intrinsic institutional stability. Put it in another way, a highly democratic entity will become an exporting source in global trade, regardless of its income level. Two exceptions remain here: (a) the random-effect OLS coefficient of low income exporters is . However, it swings to be significantly positive after controlling for the endogeneity; (b) the estimated 2SLS coefficient for the high income non-OECD countries is negative. However, we do not need to worry about that much since it is insignificant at the conventional statistical level. Now let us turn to the importers’ side. We find two different scenarios from the random-effect OLS estimates: (a) rich countries’ own democracy levels discourage trade, which is exactly consistent with O’Rourke and Taylor (2006); (b) poor countries’ own democracy levels foster trade, which is in line with the findings of Milner and Kubota (2005). The two scenarios remain broadly robust after controlling for the endogeneity issue. Yet one may observe two exceptions for the 2SLS estimates: low income importers () and high income importers (). Their coefficients from the 2SLS estimates are different from those from the random-effect OLS estimates. However, this should not be a problem since both of them are statistically insignificant. 21
Furthermore, we run the random-effect OLS and 2SLS estimations disaggregating by geographical regions. Our sample is separated into seven groups: (a) East Asian countries; (b) South Asian countries; (c) Mid-east and North African countries; (d) Sub-Saharan countries; (e) European countries; (f) North American countries; and (g) Latin American and Caribbean countries. As shown in Table 6, Most of the impacts of the exporter’s democratization on trade are significantly positive in the OLS estimates. One exception remains for North American countries (). However, this is probably due to the severity of the endogeneity problem in such regions. We then use 2SLS to mitigate this problem and find that the effect shown in Table 6 is positive again though still insiginficant. 9We also consider the effects of democratization for 20 transition countries. Broadly speaking, the transitional exporter’s democracy promotes trade as other countries. In contrast, the transitional importer’s democracy decreases trade, though such an effect is insignificant after controlling for endogeneity. One might also worry about the interaction effect of trading partners’ democracy levels. The correlation reported in Table 2B suggests that they might be mutually exclusive. This is confirmed again by the exercise of including the interaction term of 9 Such transition countries include: Albania, Armenia, Azerbaijan, Belarus, Bulgaria, Croatia, Czech Republic, Estonia, Georgia, Hungary, Kazakhstan, Kyrgyz Republic, Latvia, Lithuania, Poland, Russia, Slovenia, Tajikistan, Turkmenistan, and Uzbekistan. 22
both countries’ democracy levels into the estimations. We find that the coefficients of 10such an interaction term are highly close to a nil for all estimations. The last concern is about the missing data problem in our estimates. Usually, small countries have trade data, but not CPI data. As a consequence, most of the missing data comes from the regressors’ side and hence should not be a severe problem for our estimation (Rose, 2004). IV. Concluding Remarks Democratization could affect trade in various matters. In this paper we present a theoretical gravity model to consider democratization’s “scale effect” and “selection effect” on trade. Democracy could promote trade via various channels such as the removal of various trade barriers and the reduction of trade uncertainty. Furthermore, a country with high democracy level will be a favorable exporter in international trade. This could be because of ideological recognition among democratic countries, or because of the possible intrinsic and attractive institutional stability from democratic countries. In contrast, democratization could also have a negative effect on trade due to excessive 10 We did not report the coefficients for such a term in the text. Interested readers can refer to us directly. 23
redistribution programs and other concerns. Thus, the net effect of democracy on trade remains an empirical question. Since the “gravity” model has been very successful in explaining trade patterns, we estimate a structural gravity equation with democracy based on a gravity theoretical model. We find robust evidence that democratization significantly fosters trade. Overall, the importer’s democratization accounts for % of the growth of trade via the scale effect; and the exporter’s democratization explains around % via the selection effect. After controlling for the endogeneity of democracy, importer’s democratization still explains % while exporter’s democratization accounts for %. These striking findings are in line with many studies on international political economy such as Milner and Kubota (2005) and Eichengreen and Leblang (2006). More importantly, they are consistent with the trade literature that analyzes growing trade flows. For example, by examining the bilateral trade flows among 16 OECD countries, Baier and Bergstrand (2001, ) point out: “We found that approximately 67-69% of this (trade) growth could be explained by real GDP growth, 23-26% by tariff-rate reductions and preferential trade agreements, 8-9% by transport-cost declines, and virtually none by real GDP converge.” Here we go further to identify that trading countries’ democratization could account for a total 6%-11% of trade flows within the 23%-26% of trade liberalization and preferential trade agreements. The present paper offers two main contributions: First, it is the first one, to our limited knowledge, to include a democracy variable into a theoretical gravity model. Because of this, we are able to perform a structural estimation to investigate the impact of democratization on trade. We believe that this exercise helps reduce the potential 24
estimation biases that may be present in previous analysis due to their adoption of the reduced-form method. Second, we present two novel instruments to mitigate the endogeneity of democracy, a problem not addressed by previous researchers because of the lack of good instruments. Fortunately, we find that the dummies of death penalty abolition and judicial independence are good instruments, since they are highly related to democracy but weakly correlated with trade volume. A fruitful direction for future research might be to consider the reverse impact of growing trade on democratization, following Margit Bussmann (2001), Quan Li and Rafael Reuveny (2003), Ernesto Lopez-Cordova and Christopher Meissner (2005), Giavazzi and Tabellini (2005), Nita Rudra (2005) and Yu (2005). One should aim at a complete empirical investigation grounded on a well-developed micro-foundation model. 25
REFERENCES Anderson, James. “A Theoretical Foundation for the Gravity Equation.” American Economic Review, 1979, 75 (1), pp. 178-190. Anderson, James and van Wincoop, Eric. “Gravity with Gravitas: A Solution to the Border Puzzle,” American Economic Review, 2003, 93(1), pp. 170-192. Baier, Scott L. and Bergstrand, Jeffery H. “The Growth of World Trade: Tariffs, Transport Costs, and Income Similarity.” Journal of International Economics, 53(2001), pp. 1-27. Barro, Robert J. “Democracy and Growth.” Journal of Economic Growth, 1996, 1(1), pp. 1-27. Bussmann, Margit. “Examining Causality among Conflict, Democracy, Openness, and Economic Growth.” Unpublished Paper, University of Alabama, 2001. Eichengreen, Barry and Leblang, David. “Democracy and Globalization.” Unpublished Paper, University of California, Berkeley, 2006. Feenstra, Robert C. “Integration and Disintegration in the Global Economy.” Journal of Economic Perspectives, 1998, 12, -50. Feenstra, Robert C. “Border Effects and the Gravity Equation: Consistent Methods for Estimation.” Scottish Journal of Political Economics, 2002, 49, pp. 491-506. Feenstra, Robert C. Advanced International Trade: Theory and Evidence. Princeton University Press, 2003. 26
Feenstra, Robert C., Robert E. Lipsey, Haiyan Deng, Alyson Ma and Hengyong Mo. “World Trade Flow: 1962-2000.” National Bureau of Economic Research, Inc., NBER Working Papers, , 2005. Fidrumc, Jan. “Economic Reform, Democracy and Growth during Post-Communist Transition.” Unpublished Paper, Center for European Integration Studies, University of Bonn, 2001. Giavazzi, Francesco and Tabellini, Guido. “Economic and Political Liberalizations.” Journal of Monetary Economics, 2005, 52, pp. 1297-1340. Grofman, Bernard and Gray, Mark. “Geopolitical Influences on Trade Openness in Thirty-One Long-Term Democracies, 1960-1995.” Unpublished Paper, University of California, Irvine, 2000. Gwartney, James and Lawson, Robert A. “Economic Freedom of the world: 2004 Annual Report.” Cato Institute, Washington ., 2005. Hausman, Jerry A. “Specification Tests in Econometrics.” Econometrica,1978, 46, -1271. Helpman, Elhanan. “Imperfect Competition and International Trade: Evidence from Fourteen Industrial Countries.” Journal of the Japanese and International Economics, 1987, 1, -81. Li, Quan and Reuveny, Rafael. “Economic Globalization and Democracy: An Empirical Analysis.” Unpublished Paper, Pen State and Indiana University, 2003. 27
Lopez-Cordova, Ernesto and Meissner, Christopher. “The Globalization of Trade and Democracy, 1870-2000.” National Bureau of Economic Research, Inc., NBER Working Papers, No. 11117, 2005. Krugman, Paul R. “Increasing Returns, Monopolistic Competition, and International Trade.” Journal of International Economics, 1979, 9, pp. 469-479. Krugman, Paul R. “Growing World Trade: Causes and Consequence.” Brooking Papers on Economic Activity, 1, pp. 327-377. Mansfield, Edward D., Helen V. Milner and B. Peter Rosendorff “Free to Trade: Democracies, Autocracies and International Trade.” American Political Science Review, 2000, 94 (1), pp. 305-322. Marshall, Monty G. and Jaggers, Keith “Polity IV Project.” dataset available at , 2002. Milner, Helen and Kubota, Keiko. “Why the Move to Free Trade? Democracy and Trade Policy in the Developing Countries.” International Organization, 2005, 59 (1), -93. O’ Rourke, Kevin and Taylor, Alan M. “Democracy and Protectionism in the Nineteen Century.” National Bureau of Economic Research, Inc., NBER Working Papers, , 2006. Quinn, Dennis, “Democracy and International Financial Liberalization,” Unpublished Paper, Georgetown University, 2001. 28
Rodrik, Dani. “Institutions for High-Quality Growth: What They Are and How to Acquire Them.” Center for Economic Policy Research Inc., CEPR Discussion Paper, No. 2370, London, 2000. Rodrik, Dani. “The Political Economy of Trade Policy.” In Handbook of International Economics, ed. Gene M. Grossman and Kenneth Rogoff. Amsterdam: Elsevier Science BV. Vol. III, 1995, pp. 61-88. Rose, Andrew K. “Do We Really Know That the WTO Increases Trade?” American Economic Review, March 2004, 94(1), pp. 98-114. Rudra, Nita. “Globalization and the Strengthening of Democracy in the Developing World.” American Journal of Political Science, 2005, 49, pp. 704-730. Samuelson, Paul. “The Transfer Problem and Transport Costs: The Terms of Trade When Impediments are Absent.” Economic Journals, 1952, 62, pp. 278-304. Subramanian Arvind and Wei, Shang-Jin. “The WTO Promotes Trade, Strongly but Unevenly.” National Bureau of Economic Research, Inc., NBER Working Papers, No. 10024, 2003. Tinbergen, Jan. “Shaping the World Economy.” New York: Twentieth Century Fund, 1962. Wooldridge, Jeffery M. “Econometric Analysis of Cross Section and Panel Data.” MIT Press, 2002. Yu, Miaojie. “Trade Globalization and Political Liberalization: A Gravity Approach,” available at SSRN: 2005. 29
Table 1: Main Notation for Models Symbol Definition Panel A: Theoretical Model Amount of good k produced in countryiand consumed in countryj C ijkNumber of varieties produced in countryi N iExporteri’s democracy level z iImporterj’s democracy level z jσ Elasticity of substitution, σ>1 Countryiand country j’s GDP level , respectively Y,Y ijPrice on a . (cost, insurance, freight) base p ijPrice on a . (free on board) base p iBilateral iceberg transportation cost T ijValue of exports from country ito countryj X ijAggregate price index in importing countryj P jw Wages Labor input for the representative firm in country i l iy Output ofcountryi’s representative firm, which is fixed in equilibrium: y=y iiiκ Fixed cost for the representative firm in country i Constant marginal cost for the representative firm in country i φ Panel B: Empirical Model Unspecified bilateral border effect α ijBilateral distance cost g ijDummy variable which is unity if countryiand countryjshare common border B ijNumber of countries landlocked L ijNumber of island countries I ijError term in transport cost specification (14) µ ijtCountry-pair random variable ϕ ijYear-specific random variable ω tIdiosyncratic random variable ε ijt 30
Table 2: Descriptive Statistics of Variables Panel A: Basic Statistics Variable List Mean Std. Dev. Min Max Log Exports Log Product GDP Log Openness Democracy Index of Exporter -10 10 Log Consumer Price Index (1995:100) Log Distance Land Border 0 1 Number Landlocked 0 2 Number Islands 0 2 Exporter’s Death Penalty Abolition 0 1 Exporter’s Judicial Independence Importer’s Death Penalty Abolition 0 1 Importer’s Judicial Independence Panel B: Key Correlations (1) (2) (3) (4) (5) (6) (7) (8) (1) Log Openness (2) Exporter’s Democracy (3)Importer’s Democracy (4) Log CPI (5) Exporter’s Death Penalty Abolition (6) Exporter’s Judicial Independence (7) Importer’s Death Penalty Abolition (8) Importer’s Judicial Independence Note: we obtain 185,125 observations for 134 IMF member countries from the years 1962 to 1998. We chose 1995 as the base year to calculate CPI following WDI (2002). For data on the dummies of death penalty abolition and judicial independence, we have 21,257 observations, spanning every five years between 1970 and 2000. Sources: Amnesty International (2005), Feenstra et al. (2005), Rose (2004) and WDI (2002). 31
Table 3: Effects of Democracy on Trade Log Openness Pooled Fixed Random 2SLS Estimate OLS Effects Effects Fixed Effects Random Effects (1) (2) (3) (4) (5) Exporter’s Democracy ** ** ** ** ** () () () () () Importer’s Democracy ** ** ** ** ** () () () ()()Log CPI ** ** * () () () () () Log Distance ** ** ** () () () Land Border ** ** * () () () Number Landlocked ** ** () () () Number Islands ** ** ** () () () # of Observations 185,125 185,125 185,125 21,257 21,257 # of Groups 6,080 6,080 3,494 3,494 Prob.>F or Prob.>χ Notes: Regressand is log openness, which is defined as the difference between log bilateral trade and log product of GDP in both trading countries. Double-star and single- star mean significance at 1% and 5% level for a two-tailed test, respectively. The z-statistics are in parenthesis and clustered by trading countries’ pair-id for random effects and fixed effects. Year effects are not reported here to save space. 32
Table 4: Calculated Coefficients for Iceberg Cost Specification Iceberg Transport Cost Pooled Fixed Random 2SLS Estimate OLS Effects Effects Fixed Random (1) (2) (3) Effects Effects (4) (5) Importer’s Democracy (ρ) Log Distance (ρ) Land Border (ρ) Number Landlocked (ρ) Number Islands (ρ) Notes: Numbers reported in this table are calculated from Table 3 using method discussed in the text. 33
Table 5: Estimates Varied by Income Groups Log Openness Random Effects 2SLS Random Effects Exporter’s Importer’s Exporter’s Importer’s Democracy Democracy Democracy Democracy Categories of Exporters Low Income Countries ** ** () () () () Lower Middle Income Countries * ** ** () () () () Upper Middle Income Countries ** ** ** ** () () () () High Income Non-OECD ** ** Countries () () () () High Income OECD Countries ** ** ** () () () () Categories of Importers Low Income Countries ** ** () () () () Lower Middle Income Countries ** ** ** () () () () Upper Middle Income Countries ** ** ** * () () () () High Income Non-OECD ** ** ** ** Countries () () () () High Income OECD Countries ** ** () () () () Notes: Regressand is log openness, which is defined as the difference between log bilateral trade and log product of GDP in both trading countries. Double stars and single star mean significance at 1% and 5% level for a two-tailed test, respectively. The z-statistics are in parenthesis and clustered by the trading countries’ pair-id for random effects. Year effects are not reported here to save space. 34
Table 6: Estimates Varied by Regions and Transition Countries Log Openness Random Effects 2SLS Random Effects Exporter’s Importer’s Exporter’s Importer’s Democracy Democracy Democracy Democracy Categories of Exporters East Asian Countries ** ** ** () () () () South Asian Countries ** ** () () () () Mideast & North African ** ** ** * Countries () () () () Sub-Saharan Countries ** ** * () () () () European Countries ** ** ** ** () () () () North American Countries ** ** () () () () Latin American & Caribbean ** ** ** ** Countries () () () () Transition Countries ** ** () () () () Categories of Importers East Asian Countries ** ** ** () () () () South Asian Countries ** () () () () Mideast & North African ** ** ** * Countries () () () () Sub-Saharan Countries ** ** * () () () () European Countries ** ** ** ** () () () () North American Countries ** ** () () () () Latin American & Caribbean ** ** ** Countries () () () () Transition Countries ** ** () () () () Notes: Regressand is log openness, which is defined as the difference between log bilateral trade and log product of GDP in both trading countries. Double stars and single star mean significance at 1% and 5% level for a two-tailed test, respectively. The z-statistics are in parenthesis and clustered by trading countries’ pair-id for random effects. 35
Figure 1: World Trade Flow and Democracy Level World Average Trade and Democracy Log ExportAverage Democracy Level Sources: Export data from Feenstra, et. al. (2005); Democracy index from Polity IV by Marshall-Jaggers (2004). 36Log ExportAverage Democracy Index (from -10 to 10)
Appendix Table A: List of Countries for estimations East Asia & Pacific Europe Latin America Sub-Saharan Australia Albania Argentina Angola Cambodia Armenia Bolivia Benin China Austria Brazil Burkina Faso Fiji Azerbaijan Chile Burundi Indonesia Belarus Colombia Cameroon Japan Belgium Costa Rica Chad Korea Bulgaria Dominican Republic Equatorial Guinea Malaysia Croatia Ecuador Ethiopia Micronesia, Fed. Sts. Cyprus El Salvador Gambia, The New Zealand Czech Republic Guatemala Ghana Papua New Guinea Denmark Guyana Guinea Philippines Estonia Haiti Kenya Singapore Finland Honduras Liberia Thailand France Jamaica Malawi Georgia Mexico Mali South Asia Germany Nicaragua Mauritania Bangladesh Greece Panama Mauritius India Hungary Paraguay Mozambique Nepal Iceland Peru Niger Pakistan Ireland Trinidad and Tobago Nigeria Sri Lanka Italy Uruguay Rwanda Kazakhstan Venezuela, RB Senegal Mid-East Latvia Sierra Leone Algeria Lithuania North America South Africa Bahrain Macedonia, FYR Canada Sudan Djibouti Netherlands United States Tanzania Egypt, Arab Rep. Norway Togo Iran, Islamic Rep. Poland Uganda Iraq Portugal Zambia Israel Romania Zimbabwe Jordan Slovak Republic Kuwait Slovenia Lebanon Spain Libya Sweden Malta Switzerland Morocco Tajikistan Oman Turkey Qatar Turkmenistan Saudi Arabia Ukraine Syrian Arab Republic United Kingdom Tunisia Uzbekistan United Arab Emirates Yugoslavia, Fed. Rep. Source: World Development Indicator CD-Rom (2002), the World Bank. 37
Appendix Table B: List of Countries without Death Penalty Outlawed (year) Permitted in Extreme Cases (year) Andorra (1990) Malta (1971) Albania (2000) Angola (1992) Marshall Islands (1986) Argentina (1984) Armenia (2003) Mauritius (1995) Bolivia (1997) Australia (1984) Mexico (2005) Brazil (1979) Austria (1950) Micronesia (1986) Chile (2001) Azerbaijan (1998) Moldova (1995)CookIslands(.)Belgium (1996) Monaco (1962) El Salvador (1983) Bermuda (1999) Mozambique (1990) Fiji (1979) Bhutan (2004) Namibia (1990) Israel (1954) Bosnia-Herzegovina (1997) Nepal (1990) Latvia (1999) Bulgaria (1998) Netherlands (1870) Peru (1979) Cambodia (1989) New Zealand (1961) Canada (1976) Nicaragua (1979) De Facto Ban on Death Penalty (year) Cape Verde (1981) Niue (.) Algeria (1993) Colombia (1910) Norway (1905) Benin (1987) Costa Rica (1877) Palau (.) Brunei Darussalam (1957) Côte d'Ivoire (2000) Panama (1903) Burkina Faso (1988) Croatia (1990) Paraguay (1992) Central African Republic (1981) Cyprus (1983) Poland (1997) Congo (Republic) (1982) Czech Republic (1990) Portugal (1867) Gambia (1981) Denmark (1933) Romania (1989) Grenada (1978) Djibouti (1995) Samoa (2004) Kenya (.) Dominican Republic (1966) San Marino (1848) Madagascar (1958) East Timor (1999) São Tomé and Príncipe (1990) Maldives (1952) Ecuador (1906) Senegal (2004) Mali (1980) Estonia (1998) Serbia and Montenegro (2002) Mauritania (1987) Finland (1949) Seychelles (1993) Morocco (1993) France (1981) Slovak Republic (1990) Myanmar (1993) Georgia (1997) Slovenia (1989) Nauru (1968) Germany (1987) Solomon Islands (1966) Niger (1976) Greece (1993) South Africa (1995) Papua New Guinea (1950) Guinea-Bissau (1993) Spain (1978) Russia (1999) Haiti (1987) Sweden (1921) Sri Lanka (1976) Honduras (1956) Switzerland (1942) Suriname (1982) Hungary (1990) Turkey (2002) Togo (.) Iceland (1928) Turkmenistan (1999) Tonga (1982) Ireland (1990) Tuvalu (1978) Tunisia (1990) Italy (1947) Ukraine (1999) Kiribati (1979) United Kingdom (1973) Liechtenstein (1987) Uruguay (1907) Lithuania (1998) Vanuatu (1980) Luxembourg (1979) Vatican City (1969) Macedonia (1991) Venezuela (1863) Data Source: Amnesty International (2005). 38
Appendix Table C: Formation of Polity Index in the Polity IV Project Categories Scale Weight Democracy Coding Authority Coding Competitiveness of Executive Recruitment (1) Selection +2 (2) Transitional +1 (3) Election +2 Openness of Executive Recruitment (1) Closed +1 (2) Dual/Designation +1 (3) Dual/Election +1 (4) Election +1 Constraint on Chief Executive (1) Unlimited authority +3 (2) Intermediate category +2 (3) Slight to moderate limitations +1 (4) Intermediate category +1 (5) Substantial limitations +2 (6) Intermediate category +3 (7) Executive parity or subordination +4 Competitiveness of Political Participation (1) Repressed +2 (2) Suppressed +1 (3) Fractional +1 (4) Transitional +2 (5) Competitive +3 Note: According to the Polity IV project, the polity indicator is defined as difference between the institutionalized democracy and the institutionalized autocracy. Both of them are derived from coding of competitiveness of political participation, the regulation of participation, the openness and competitiveness of executive recruitment, and constraints on the chief executive using the weights shown above. Readers can refer to the Polity IV Project Dataset Users’ Manual by Marshall-Jaggers (2004) for details. 39
Appendix Table D: List of Countries Varied by Income High Income OECD Upper Middle Income Lower Middle Income Low Income Australia Argentina Algeria Bangladesh Austria Bahrain Bolivia Benin Belgium Brazil Bulgaria Burundi Canada Chile China Cameroon Denmark Costa Rica Colombia Chad Finland Croatia Dominican Republic Ghana France Czech Republic Ecuador Guinea-Bissau Germany Gabon Egypt, Arab Rep. Haiti Greece Hungary El Salvador India Iceland Malaysia Equatorial Guinea Indonesia Ireland Mauritius Fiji Kenya Italy Mexico Guatemala Madagascar Japan Oman Guyana Malawi Netherlands Panama Honduras Mali New Zealand Poland Iran, Islamic Rep. Nepal Norway South Africa Jamaica Nicaragua Portugal Trinidad and Tobago Jordan Niger Spain Turkey Latvia Nigeria Sweden Uruguay Lithuania Pakistan Switzerland Venezuela, RB Morocco Rwanda United Kingdom Namibia Sierra Leone United States Papua New Guinea Tanzania Paraguay Togo High Income Non-OECD Peru Uganda Cyprus UkraineIsrael Romania Zambia Korea Sri Lanka Zimbabwe Kuwait Syrian Arab Republic Malta Thailand Singapore Tunisia Slovenia United Arab Emirates Source: World Development Indicator CD-Rom (2002), the World Bank. 40
