Published online by Cambridge University Press: 13 June 2011
The likely endogeneity between democracy and trade is addressed with an instrumental variables strategy in this article about whether international trade fosters democracy. The authors use a measure of natural openness to obtain estimates of the causal impact of openness on democratization in three separate samples spanning the last 130 years. A positive impact of openness on democracy is apparent in the data over the long run. The post–World War II results suggest that with a rise in trade with other countries equal to a one standard deviation increase, countries such as Indonesia, Russia, and Venezuela could eventually become as democratic as the U.S., Great Britain, or France. There is some variation in the impact of openness by region that may be because trade seems to have a positive impact only when the capital-to-labor ratio is sufficiently high. This is consistent with the idea that openness promotes democracy when it strengthens the economic fortunes of the middle class.
2 See Rodrik, Dani, “Democracy and Economic Performance” (Manuscript, Kennedy School of Government, Harvard University, December 1997)Google Scholar; and idem “Institutions for High-Quality Growth: What They Are and How to Acquire Them,” Working Paper, no. 7540 (Cambridge, Mass.: National Bureau of Economic Research, February 2000)Google Scholar for an overview. On economic growth, see Helliwell, John F., “Empirical Linkages between Democracy and Economic Growth,” Working Paper, no. 4066 (Cambridge, Mass.: National Bureau of Economic Research, May 1992)Google Scholar; Barro, Robert J., “Democracy and Growth,” Journal of Economic Growth 1, no. 1 (1996)CrossRefGoogle Scholar; Rodrik, Dani, “Where Did All the Growth Go? External Shocks, Social Conflict, and Growth Collapses,” Journal of Economic Growth 4, no. 4, (1996)Google Scholar; and Lindert, Peter H., “Voice and Growth: Was Churchill Right?” Working Paper, no. 9749, (Cambridge, Mass.: National Bureau of Economic Research, June 2003)Google Scholar. On income and asset distribution, see Acemoglu, Daron and Robinson, James A., “Why Did the West Extend the Franchise? Growth, Inequality and Democracy in Historical Perspective,” Quarterly Journal of Economics 115 (2000)CrossRefGoogle Scholar; and again also in Lindert.
3 As a vivid illustration of such beliefs, former V.S. President William J. Clinton argued that “[j]ust as democracy helps make the world safe for commerce, commerce helps make the world safe for democracy. It's a two-way street.” www.issues2000.org/Celeb/BillClintonFreeTrade.htm (accessed November 27, 2004). His successor, George W. Bush, claimed that, “when we promote open trade, we are promoting political freedom. Societies that open to commerce across their borders will open to democracy within their borders, not always immediately, and not always smoothly, but in good time.” Office of the Press Secretary, The White House, Remarks, by the President to the Council of the Americas, May 7,2001; http://www.whitehouse.gov/news (accessed May 8, 2001).
4 See Gourevitch, Peter Alexis, “International Trade, Domestic Coalitions, and Liberty: Comparative Responses to the Crisis of 1 873–1896,” Journal of Interdisciplinary History 8, no. 2 (1977)Google Scholar. The impact of tariffs and reduced international trade on political outcomes was seen as ambiguous in this comparative study of France, Germany, Great Britain, and the United States between 1873 and 1896.
5 See Bussmann, Margit, “Examining Causality Among Conflict, Democracy, Openness, and Economic Growth” (Manuscript, Department of Political Science University of Alabama, 2001)Google Scholar.
12 Sachs, Jeffrey D., “Institutions Don't Rule: Direct Effects of Geography on Per Capita Income,” Working Paper, no. 9490 (Cambridge, Mass.: National Bureau of Economic Research, February 2003)Google Scholar.
16 Bhagwati, Jagdish, In Defense of Globalization (New York: Oxford University Press, 2004)Google Scholar.
17 Lin, Justin Yifu and Nugent, Jeffrey B., “Institutions and Economic Development,” in Behrman, Jerey and Srinivasan, T. N., eds., Handbook of Development Economics, (Amsterdam: Elsevier Science B. V., 1995), 2336Google Scholar.
18 Acemoglu and Robinson (fn. 2).
19 Acemoglu, Daron and Robinson, James A., The Economic Origins of Dictatorship and Democracy (New York: Cambridge University Press, 2006)Google Scholar.
20 For a recent review of the literature, see Goldberg, Pinelopi Koujianou and Pavcnik, Nina, “Trade, Inequality and Poverty: What Do We Know?” Brookings Trade Forum (Washington, D.C.: The Brookings Institution, 2004)Google Scholar.
21 Acemoglu and Robinson (fn. 19).
22 On Argentina, see Porto, Guido, “Using Survey Data to Assess the Distributional Effects of Trade Policy,” Journal of International Economics 70 (September 2006)CrossRefGoogle Scholar; and on Mexico see Ianchovi-China, Elena, Nicita, Alessandro, and Soloaga, Isidro, “Trade Reform and Household Welfare: The Case of Mexico, Policy Research,” Working Paper, no. 2667 (Washington, D.C.: World Bank, August 2001).Google Scholar
23 Wei, Shang Jin and Wu, Yi, “Globalization and Inequality without Differences in Data Definitions, Legal Systems, and Other Institutions,” (Manuscript, International Monetary Fund, October 2002)Google Scholar; and Borraz, Fernando and López-Córdova, J. Ernesto, “Has Globalization Deepened Income Inequality in Mexico?” Global Economy Journal 7, no. 1 (2007)CrossRefGoogle Scholar.
24 Li and Reuveny (fn. 6).
26 Rigobon and Rodrik (fn. 7).
27 Rigobon, Roberto, “Identification Through Heteroskedasticity,” The Review of Economics and Statistics 85 (November 2003)Google Scholar.
31 Frankel and Romer (fn. 8).
32 See respectively Frankel, Jeffrey A. and Rose, Andrew K., “An Estimate of the Effect of Currency Unions on Trade and Growth,” Quarterly Journal of Economics 117 (May 2002)CrossRefGoogle Scholar; Wei, Shang-Jin, “Natural Openness and Good Government,” Working Paper, no. 7765 (Cambridge, Mass.: National Bureau of Economic Research, June 2000)Google Scholar; and Frankel, and Rose, , “Is Trade Good for the Environment? Sorting Out the Causality,” Review of Economics and Statistics 87, no. 1 (2005)CrossRefGoogle Scholar.
33 Frankel and Romer (fn. 8). An important point that the authors make is that controls for size are necessary. Smaller countries are naturally more proximate to other countries and also engage in more international and less intranational trade. If trade and exchange in general (i.e., the overall size of the economy), rather than international trade, affected the amount of democracy, our instrument could be negatively correlated with the error term if we omit such controls and our point estimate could possibly be biased downward. Therefore, key controls may be population and land area.
34 Anderson, James E., “A. Theoretical foundation for the Gravity Equation,” American Economic Review 69, no. 1 (1979)Google Scholar.
35 We assume a homoskedastic error process for equation 2. If we do not, then Ti. will be multiplied by a pair-specific residual, and this would affect our parameter of interest. Assuming homoskedasticity makes it so that a constant pre-multiplies all of the instruments and hence will not affect our instrumental variables regressions.
36 Our sample after World War II begins with 1960 because in unreported robustness checks we use lagged values of democracy in some regression specifications. To keep a consistent sample, it was necessary to begin with 1960 in the cross-sectional regressions.
37 Prior to 1913 we use data from López-Córdova and Meissner and López-Córdova, Barbieri. J. Ernesto and Meissner, Christopher M., “Exchange Rate Regimes and International Trade: Evidence from the Classical Gold Standard Era,” American Economic Review 93, no. 1 (2003)CrossRefGoogle Scholar; and Barbieri, Katherine, “Economic Interdependence and Militarized Interstate Conflict, 1870–1985” (Ph.D. diss., Binghampton University, 1996)Google Scholar. Data from Ritschl and Wolf and Barbieri are used for the interwar period. Ritschl, Albrecht and Wolf, Nikolaus, “Endogeneity of Currency Areas and Trade Blocs: Evidence from the Inter-war Period,” Working Paper, no. 4112 (London: CEPR, November 2003Google Scholar); and again also from Barbieri. For the post-1960 period, we use data from Rose and Glick. Rose, Andrew K. and Glick, Reuven, “Does a Currency Union affect Trade? The Time-Series Evidence,” European Economic Review 46 (June 2002)Google Scholar. The latter is available on line at http://faculty.haas.berkeley.edu/arose/RecRes.htm. For a full description of our sources, we refer the reader to the data appendix found in the working paper version of this article atwww.nber.org/papers/w11117.pdf. A copy is also available upon request from the authors.
38 Marshall, Monty G. and Jaggers, Keith, “Polity IV Project: Political Regime Characteristics and Transitions, 1800–1999” (Integrated Network for Societal Conflict Research Program, Center for International Development and Conflict Management, University of Maryland, College Park, December 2000)Google Scholar.
39 Alvarez, Mike, Cheibub, José Antonio, Limongi, Fernando, and Przeworski, Adam, Democracy and Development: Political Institutions and Material Well-Being in the World, 1950–1990 (New York: Cambridge University Press, 2000)Google Scholar.
40 See Tatu Vanhanen, The Polyarchy Dataset: Vanhanen's Index of Democracy, www.prio.no (accessed March 2003); and Freedom House Country Ratings, www.freedomhouse.com (accessed March 2003).
41 We are aware that this comparative static is only an approximation. One problem with this is that categorical coding of the dependent variable makes it technically incorrect to suppose that incremental changes are uniform across the interval. Ordered probit analysis we undertook, but do not report, helps us to be more precise about this.
42 One possibility is that the nature of the process of democratization has changed over time. This is plausible. Samuel P. Huntington's third wave did not really begin until the late 1970s. Huntington, Samuel P., The Third Wave: Democratization in the Late Twentieth Century (Norman: University of Oklahoma Press, 1991)Google Scholar.
43 One might wonder why the coefficient on distance increases over time when conventional wisdom would have it that transportation costs fall over time. If short-distance transportation costs and other trade costs have fallen more quickly over time, as Hummels argues, then trade may have become more 'regionalized' over time. This would yield a rise on the distance coefficient in a series of gravity regressions. See David Hummels, “Have International Transport Costs Declined?” (Manuscript, Graduate School of Business, University of Chicago, 1999).
44 In the working paper version, we also explain how we tested the exclusion restriction that the geographic information incorporated in the gravity equation was uncorrelated with the error term with Hansen's J-statistic (a test for overidentifying restrictions that is robust to heteroskedasticity). The findings were overall encouraging. We found that when we include population, land area, number of borders, and the landlocked indicator directly, and use the average distance to countries and the island indicator as excluded instruments, we can never reject the null hypothesis of exogeneity.
45 We do not present year-by-year ordered probit results although the results are available in the longer version of this paper. We are comfortable using continuous dependent variable techniques be-cause as the number of categories becomes large, ordered probit and linear models produce very similar results and because the predictions we get from our model are very nearly within the bounds of the actual outcomes. In any case the qualitative results are identical to the results of the linear model.
46 It would appear that the general positive correlation is not too sensitive to the way in which we code the democracy variable. As previously mentioned, we used the binary data from Alvarez et al. in an instrumental variables probit model including openness and size. See Alvarez et al. (fn. 39). Their data is available from 1960 to 1990. The results show a highly significant and positive relationship between democracy and openness.
47 Our regressions use generated instruments as instruments rather than the geographic information itself. Nevertheless, we simulate the change in the standard error of the openness coefficient for small changes we impose in the underlying geographic data. This allows us to gauge how much sampling error could be affecting the standard errors. To save on computation time, we chose to carry out such a simulation for 1960 and 1995. Doing so barely altered the standard errors on the coefficient on openness. For example, our simulations yielded an increase of 0.001 in the standard error of the openness coefficient in both years.
48 Our data for the interwar period on trade were kindly made available by Albrecht Ritschl. We selected the years that comprise the interwar period based on the years for which data were available in this dataset.
49 The countries in our 1910 sample are Argentina, Austria, Australia, Belgium, Brazil, Canada, Chile, Denmark, France, Germany, Great Britain, Greece, Italy, Japan, Mexico, New Zealand, Norway, Portugal, Russia, Spain, Sweden, Switzerland, and the U.S.
50 We tried other specifications of these regressions. Unreported ordered probit results, including controls for population and size, find slightly more years with positive and statistically significant coefficients on openness. Nevertheless, in many years these are only statistically significant at about the 85 or 90 percent confidence level. One might attribute these imprecise results to the fact that the number of degrees of freedom is so small or to outliers.
51 The countries in the Eastern European region for which we have data prior to 1990 are Hungary, Poland, Romania, and Turkey.
52 Data underlying Clemens, Michael and Williamson, Jeffrey G., “Wealth Bias in the First Global Capital Market Boom, 1870–1913,” The Economic Journal 495 (2004)Google Scholar. The data were kindly made available to us by Michael Clemens and Jeffrey Williamson.
53 Acemoglu and Robinson (fn. 19).
54 Measures of the capital stock and the labor force come from Loayza, Fajnzylber, and Calderon. Loayza, Norman, Fajnzylber, Pablo, and Calderon, Cesar, Economic growth in Latin America and the Caribbean: Stylized Facts, Explanations, and Forecasts (Washington, D.C.: World Bank, 2005)CrossRefGoogle Scholar, These are not available for all countries in Table 5. In addition, because data is only available until 1999, we substitute this year for 2000 in the sample. Reliable data on the capital stock ire not available for earlier periods. Our results in Table 5 do not change when we restrict attention to this smaller sample. Also information on endowments of skilled versus unskilled workers is too spotty to be of any practical use.
53 This evidence also suggests that high capital/labor ratios alone only promote democracy when a country” is sufficiently open to international trade. This may or may not be consistent with the argument presented in Acemoglu and Robinson that capital abundance deters coups and makes repressions costly. Acemoglu and Robinson (fn. 19).
56 Arellano, Manuel and Bover, Olympia, “Another Look at the Instrumental Variable Estimation of Error Components Models,” Journal of Econometrics 68 (1995)CrossRefGoogle Scholar; and Blundell, Richard and Bond, Stephen, “Initial Conditions and Moment Restrictions in Dynamic Panel Data Models,” Journal of Econometrics 87 (1998)CrossRefGoogle Scholar.
58 Land area must be omitted since it is not time varying.
59 Eichengreen and Leblang find evidence of a virtuous circle between openness and democracy that might imply a larger positive impact. Eichengreen and Leblang (fn. 25).
60 Acemoglu, Daron, Johnson, Simon, Robinson, James A., and Yared, Pierre, “Reevaluating the Modernization Hypothesis,” Working Paper, no. 13334 (Cambridge, Mass.: National Bureau of Economic Research, August 2007)Google Scholar.