Level of Economic Development and Political Democracy Revisited

Level of Economic Development and Political Democracy Revisited YONG U. GLASURE,* AIE-RIE LEE,** AND JAMES NORRIS** This article reassesses a democracy-economic development linkage for the period 1972 to 1990. The results show that economic development has a significant effect on democratic performance in the semiperiphery and periphery, yet in an opposite direction. For the core, there is no linkage from economic development to political performance. Moreover, the finding also suggests that democracy enhances economic development only in the core and semiperiphery. This finding thus contradicts Barro's [1996] contention that more democracy enhances economic growth at low levels of political freedom, but it depresses economic growth when a moderate level of freedom has already been attained. (JEL 010) Introduction As Burkhart and Lewis-Beck [1994, p. 903] note, there are few "iron taws" in comparative politics. Yet, one hypothesis that seems established beyond challenge is the causal link between economic development and democracy. "Perhaps the most common generalization...has been that democracy is related to the state of economic development. The more well-to-do a nation, the greater the chances that it will sustain democracy" [Lipset, 1963, p. 3t]. Lipset [1959] offers the first well-known empirical test of the economic development hypothesis. Other researchers (such as Cutright [1963], Olsen [1968], Cutright and Wiley [1969], and Helliwell [1994]), using multivariate correlations with larger numbers of countries, provide further support. Jackman [1973] also finds an association between economic development and democratic development, and he supports the threshold hypothesis that a certain level of economic development is necessary to support democratic development. However, Muller and Seligson [1994] discovered that economic development is primarily relevant through its effect of income equity. In his later work, Muller [1995] shows that there is a positive and significant relationship between economic development and democratization, with a much stronger negative relationship between income inequality and democratization. Thus, literature leaves a strong impression that economic development is at least a necessary, though perhaps insufficient, cause of democratization. However, the direction of causation has not been undisputed. Accordingly, over the last 40 years, researchers have tested this hypothesis with ever more sophisticated statistical techniques 03ollen [1979], Bollen and Jackman [1985], Arat [1988], Vanhanen [1990], and Hadenius [1992], to name a few). A useful technique involving the operational concept of causality was "Wayland Baptist University--U.S.A. and **Texas Tech University--U.S.A. 466

GLASURE ET AL.: ECONOMIC DEVELOPMENT AND DEMOCRACY 467 developed by Granger [1969]. Using this technique, Burkhart and Lewis-Beck [1994] attempt to identify the causal linkages between economic development and democratic performance by pooling 131 countries from 1972 to 1989 while taking into account the country's position in the world system, following Gonick and Rosh [1988].1 Burkhart and Lewis-Beck [1994] fmd that economic development significantly affects democratic performance for the countries in the core and for those countries in the semiperiphery and periphery. Unidirectional causality from economic development to democracy is also observed, thus supporting the modernization theory. Given Burkhart and Lewis-Beck's [1994] fmdings in support of the economic development thesis, along with the findings of others [Helliwell, 1994], it appears fruitless to further test this seemingly unchallengeable iron law. Nevertheless, Przeworski and Limongi [1997] contend that democracy is not a by-product of economic development. They found that the current wealth of a country is not a decisive factor in determining whether democracy can survive. Rather, a democracy is more likely to survive in a growing economy even in the poorest nations. This implies that a nation ought to have decades of sustained economic growth to enjoy stable and entrenched democracy that is effective in serving to create a basis of legitimacy [Lipset, 1963]. Moreover, in many nations (that is, South Korea, Taiwan, Singapore, and Argentina) during the 1970s and 1980s, rapid economic growth had been facilitated not by democracy but rather by authoritarianism, thus having an inverse relationship between economic growth and democracy [O'Donnell, 1973; Collier, 1979]. Going a step further, Kurth [1979] contends that democracy and economic development have an N-curve relationship with an upper threshold of economic development. Here, an increasing economic development does not affect a higher level of democracy because democracy is stabilized beyond that threshold. Lipset et al. [1993] also assert that for the short run, there may be a negative association between democracy and economic development despite increasing levels of economic development. However, there is a clear long-run trend in the relationship between democracy and economic development. More recently, Barro [1996] shows from a 1960-90 panel of about 100 countries that the overall effect of democracy on economic growth is weakly negative, but improvements in the standard of living, measured by gross domestic product, substantially foster the chance that political freedom will grow. This paper attempts to reassess the linkage between democracy and economic development from the methodological side by using a larger, more recent, pooled sample and alternative specifications with the hope of providing a more current resolution. Unlike the foregoing research, this paper tests not only a direction of the causal linkage, but also the existence of both short-run dynamic and long-run relationships. If democracy and economic development are cointegrated, that is, the two series drift together, then a vector error correction model (VECM) is used because not incorporating this cointegrating property into the model is a misspecification. However, if the two series drift apart, then democracy and economic development are not cointegrated. Thus, a vector autoregressive model (VAR) in first differences is used. For determining the

468 IAER: NOVEMBER 1999, VOL. 5, NO. 4 optimal lag length for the model, a multivariate generalization of the Akaike information criterion and the Schwartz-Bayesian criterion are used here. This paper is organized as follows. The second section presents data sources and variables used in the models. The third section discusses the cointegration and error correction modeling and how it differs from the standard Granger technique used in Burkhart and Lewis-Beck [1994]. The fourth section reports the empirical results, and the fifth section presents the conclusions. Data and Variables Energy consumption per capita comes from World Bank's 1995 data diskettes. This series for energy consumption per capita gives much more consistent figures than from the United Nations' publications. Also, the current measure of energy consumption per capita is in oil equivalent rather than coal equivalent as in the United Nations' publications. Political data came from Freedom House [various] for 1972 to 1992, excluding 1973, 1975, 1977, and 1979. States are categorized as not free (authoritarian), partly free (semidemocratic), and free (democratic) on the basis of political rights and civil liberties, that is, whether citizens can turn out incumbent governments through free elections and whether they can organize political parties and express critical views without government interference. Freedom House ranks each nation on separate 7-point scales for political rights and civil liberties) Following Burkhart and Lewis-Beck [1994], each nation in this study acquires an overall democracy index, ordered from a low of 2 to a high of 14 (for details, see Burkhart and Lewis-Beck [1994, p. 908]). The sample period is from 1972 to 1992, excluding 1973, 1975, 1977, and 1979. For the variables in the models, ec is energy consumption per capita, pc is the democracy index, dec is the first difference of ec, dpc is the first difference of pc, s or (s * Aln(ect) ) is the semiperiphery status multiplied by the first difference of ec (semiperiphery equals 1; otherwise, 0), and p or (p A ln(ect) ) is the periphery status multiplied by the first difference of ec (periphery equals 1; otherwise, 0). The worldsystem-position dummy variables are constructed following Burkhart and Lewis-Beck [1994]. The Method When dealing with time series data, researchers must evaluate each series to determine whether they are stationary or nonstationary because the statistical properties of regression analysis using nonstationary time series are dubious. Also, the regression equation may look quite interesting even though it represents a totally spurious relationship. More technically, if the series is stationary, then the effects of shocks to the series are temporary. Thus, the series reverts to its long-run mean level. However, if the series is nonstationary, then the effects of shocks to the series are permanent because the mean or variance of the series are time-dependent. Hence, a series must be successfully differenced until stationarity is achieved.

GLASURE ET AL.: ECONOMIC DEVELOPMENT AND DEMOCRACY 469 The model here uses the augmented Dickey-Fuller (ADF) and Phillip-Perron tests to detect whether series have unit roots. As shown in Table 1, the estimated statistics of the log levels of ec and pc are not significant at the.05 level in both tests. Thus, nonstationarity cannot be rejected. When the variables are in first differences, both tests reject nonstationarity at the.01 level. Hence, the appropriate transformation of the variables is first differences. Burk_hart and Lewis-Beck [1994] use fractional unit roots (p --.8) in their variable transformation (albeit, not spelled out in their paper) to correct the problem of autocorrelation. Although stationarity of ec andpc is obtained by applying first differences, information on the long-run properties between ec and pc is lost. The desire to evaluate a model that combines both short-run and long-run properties while maintaining stationarity in ec and pc prompts a reconsideration of the problem of regression using ec and pc measured in their levels. This leads to the concept of cointegration. TABLE 1 Unit Root Tests ADF Phillip-Perron Variable Level Difference Level Difference ec -0.10-27.75 * -0.20-44.55 * pc -2.79-33.31" -2.49-47.50* Notes: * denotes significance at the.01 level. One tag is used. As Granger [1981] and Engle and Yoo [1987] point out, variables that become stationary only after differencing may have a linear combination of the variables (measured in their levels) that are stationary. Put differently, if two variables are cointegrated, they are stationary of the same order and there is a long-run relationship between the variables. Moreover, the residuals produced by a linear combination of the two variables are also stationary. In that situation, the two variables are said to be cointegrated. In this paper, if there is a long-run relationship between democracy and economic development, which are nonstationary in their levels, then deviations from this long-run trend are stationary if ec and pc are cointegrated. Incorporating information from the cointegrated properties of time series, Engle and Granger [1987] introduce a two-step procedure known as a cointegration and error correction model. One of the advantages that a cointegration and error correction model has over the conventional technique used in Burkhart and Lewis-Beck [1994] is that the procedure specifically allows for a causal linkage between ec and pc. Stemming from a

470 IAER: NOVEMBER 1999, VOL. 5, NO. 4 long-run relationship, this characterizes the long-run alignment that persists beyond the short-run adjustment. The Burkhart and Lewis-Beck [1994] work, however, ignores the cointegrated properties of ec andpc and only uses near first differencing to deal with the nonstationary time series variables. Unfortunately, although this technique may remove the nonstationarity problem of ec and pc, data regarding the long-run levels of ec and pc are lost in the process (see Durlauf and Phillips [1988], Nelson and Plosser [1982], Maddala [1992, pp. 584-90], and Perman [1991, pp. 11-2]). As stated above, even though ec and pc are nonstationary in their levels, the residuals from a linear combination of ec and pc can be stationary. Therefore, tests for cointegration are performed and the results are reported in Table 2. As shown in Table 2, the null hypothesis of no cointegration for the pc-on-ec system can be rejected at the.01 level but cannot be rejected for the ec-on-pc system. Further, comparing with the MacKinnon [1991] critical value for cointegration tests, which is -3.92 at the.05 level without a trend, nonstationarity of residuals is also rejected for the pc-on-ec system. Consequently, the pc-on-ec system is estimated using the VECM. The ec-on.-pc model is estimated using the VAR in first differences. The optimal lag length for each system is determined by the multivariate generalization of the Akaike information criterion and Schwartz-Bayesian criterion. TABLE 2 Cointegration Tests Variables ADF pc on ec -5.85' ec on pc -0.66 Notes: * denotes significance at the.01 level. Three lags are used. For the pc-on-ec system, a cointegration and error correction model is constructed to examine the short-run and long-run relationships from economic development to democracy: Ain(PCt) = ~x + ~_~=, ~,Ain(PCt_ ) + ~_,~=~,f, Ain(ec,_,) + 6t(s* A ln(ect) ) +et(p* Ain(ect))+ ~'zt-1 +vt ' (1) where Ain(PCt) is the democracy index, Aln(Pct_i) is lag i of the democracy index, A In (ect_ i) is lag i of energy consumption per capita, (s A In (ect)) is the semiperiphery status multiplied by energy consumption per capita, (p A In(ect) ) is the periphery status

GLASURE ET AL.: ECONOMIC DEVELOPMENT AND DEMOCRACY 471 multiplied by energy consumption per capita, k is the number of lags, and v t is an error term. The log transformation of the variables ec and pc is used to solve the problem of heteroskedasticity that may arise from cross-sectional data with time series data [Hanke and Reitsch, 1989]. Lastly, nt_ t is the lagged value of an error term from the cointegration equation: ln(pc)t = q)o + tpl ln(ect) + nt (2) For the ec-on-pc system, the following VAR in first differences is used to examine the short-run dynamic relationship from democracy to economic development: Aln(ect) = ~ + ~=1 ~ialn(ect_i) + ~=1 yialn(pct_,) +6t(s*Aln(ect))+et(p*Aln(ect))+v t (3) Note that the inclusion of nt_ 1 in (1) (the error-correction model) gives an extra avenue through which the effects of causality can occur. According to Engle and Granger [1987], economic growth (ec) causes democracy (pc) if either the coefficients (Yi) of the lags of ec as a group or the coefficient (~) of nt_ 1 is statistically significant. In the standard Granger causality test, however, ec causes pc if the coefficients (yi) of the lags of ec as a group are statistically significant, excluding 7zt_ 1 from (1). Thus, the error correction model provides two avenues through which causal effects can emerge. Moreover, the error correction model can be viewed as a constrained VAR system where the information contained in the undifferenced data is-reintroduced to the first differenced VAR system through the error correction term. The Findings Constructing the error correction term (nt_l) from (2) and including it in a regression model containing only differenced variables provides direct tests of both the short-run and long-run relationships from economic development to democracy. Table 3 shows that for the years 1972 through 1992, the estimated coefficient of the error correction term for the full sample is negative and significant at the.01 level, and the F-statistic for dec is also significant at the.05 level. This indicates that there are short-run and long-run causal relationships from economic development to democracy when s and p are controlled for. Also note that both s and p are significant at the. 10 and.05 levels, thus implying that semiperiphery and periphery status in the world have negative effects on democracy. Because of the negative signs of the lags of dec, the VECM is reestimated based on the world status. Their respective results are reported in Table 4. Regarding the individual coefficient estimates of the lags of dec, Table 4 shows that the signs of the lags of dec for the full sample are negative and dec_l and dec_2 are significant at the. 10 and.05 levels.

472 IAER: NOVEMBER 1999, VOL. 5, NO. 4 TABLE 3 VECM: The @c-on-dec System Number of Countries Variables F-statistic t-statistic Full Sample 138 Core 17 Semiperiphery 29 Periphery 92 dpc 0.22 dec 3.67** S P Xt- 1 dpc 1.46 dec 0.18 7~t_ 1 dpc 0.71 dec 1.88*** 7~ t-1 dpc 0.51 dec 2.85** t-1-1.73*** -2.32** -6.23* -3.90* -2.54*** -5.15" Notes: *, **, and *** denote levels of significance are.01,.05, and. 10, respectively. Three lags of dec and dpc are used. As shown in Table 4, the signs of the coefficients of dec are negative across the core countries but, more importantly, they are statistically insignificant, indicating that economic development has no significant effect on democratic performance in the short run. The core in Table 3 shows that the lags of dec have no effects on dpc, confirmed by the insignificance of the F-statistic. However, the error correction term (Xt_x) is statistically significant at the.01 level, thus concluding that there is a long-run relationship from economic development to democracy in the core. The semiperiphery in Table 3 shows only lag 1 of dec is significant and has a negative sign. The F-statistic shows the lags of dec are marginally significant at the. 10 level, and the error correction term also is marginally significant at the.10 level. Hence, there are short-run and longrun relationships from economic development to democracy. The findings of the periphery are similar to those of the semiperiphery. In Table 3, the F-statistic shows that

GLASURE ET AL.: ECONOMIC DEVELOPMENT AND DEMOCRACY 473 the lags of dec together are significant at the.05 level, and the error correction term is significant at the.01 level. Also, in the periphery, short-run and long-run relationships from economic development to democracy exist, but in an opposite direction as shown by lag 2 of dec in Table 4. TABLE 4 Coefficient Estimates of the dpc-on-dec System Using a VECM Coefficients Full Sample: Core: Semiperiphery: Periphery: Variables Lags 138 Countries 17 Countries 29 Countries 92 Countries Constant 0.0380 0.0003 0.0188"* 0.0162* (3.5200) (0.3400) (1.9700) (2.6800) dpc 1 0.0062-0.1240"** -0.0293 0.0142 (0.2600) (- 1.9900) (-0.5700) (0.4800) dpc 2-0.0151-0.0082 0.0637-0.0294 (-0.6200) (-0.1300) (t.2600) (-0.9800) dpe 3-0.0112-0.0002 0.0214-0.0167 (-0.4500) (-0.0100) (0.3900) (-0.5500) dec 1-0.0602*** -0.0013-0.1527"** -0.0556 (-1.7000) (-0.0600) (-1.7500) (-1.2900) dec 2-0.0894** -0.0152 0.0692-0.1068" (-2.5700) (-0.6600) (0.7400) (-2.5600) dec 3-0.0306-0.0032-0.1055-0.0247 (-0.9000) (-0.1500) (-1.1500) (-0.6100) s -0.0033*** (-1.7300) p -0.0049** (-2.3200) n -0.0542* -0.1429" -0.0396*** -0.0596* t-1 (-6.2300) (-3.9000) (-2.5400) (-5.1500) Notes: *, **, and *** denote levels of significance are.01,.05, and. 10, respectively. Table 5 reports the results of the VARs in first differences, which measure a reverse causation from democracy to economic development. As shown, all the variables in the

474 IAER: NOVEMBER 1999, VOL. 5, NO. 4 full sample are statistically significant at the.01 level exceptp. In their respective world status, democratic performance has effects on economic development, given their significance at the.01 level. Moreover, note that the effect is much stronger for the semiperiphery than for the core. This is not surprising. Naturally, since there is very little variance in dpc for the core countries over the period 1972 through 1992, a strong relationship is not expected. TABLE 5 VARs in First Difference: The dec-on-dpc System Number of Countries Variables F-statistic t-statistic Full Sample 138 Core 17 Semiperiphery 29 Periphery 92 dpc 5.62* dec 18.71" S P dpc 2.77** dec 4.22* clpc 6.53* dec 19.27" dpc 1.46 dec 13.27' 2.56*** 0.94 Notes: *, **, and *** denote levels of significance are.01,.05, and. 10, respectively. Three lags of dec and dpc are used. Conclusions The primary goal of this paper was to revisit the linkage between democracy and economic development by estimating a bidirectional causal linkage between economic development and democratic performance for the years 1972 through 1992 across 138 countries and evaluating whether short-run and long-run relationships exist. It was observed that for developing countries (periphery) and newly industrializing countries (semiperiphery), the level of economic development contains useful information about the direction of democracy. The results clearly show that economic development has a significant effect on democratic performance for the countries in the semiperiphery and

GLASURE ET AL.: ECONOMIC DEVELOPMENT AND DEMOCRACY 475 periphery, yet in an opposite direction. This is puzzling because what seems in theory to be a clear-cut relationship is not obtained upon opinion. Economic development actually leads to lower levels of democracy for countries in the semiperiphery and periphery. It is expected that any effect of economic conditions on democratic performance would be positive. The sign reversal may stern from the possibility that as nations strive for economic development, the nations tend to trade off democracy for economic development. This finding is similar to what Huntington [1968] assessed, overturning one of the basic tenets of the modernists, that economic (and social) development gives rise to a political culture more favorable to social peace. According to Huntington, the opposite occurs in reality due to the conflicts in society and politics that do not diminish through modernization but are usually sorely aggravated. This phenomenon is more noticeable when the traditional, ingrained conditions change suddenly. Change often leads to an exacerbation of the diverse rifts in the population. It is in this phase, according to Huntington, that ethnic and other divisions are strengthened. In traditional societies, ethnic and other divisions seldom cause major problems. Here, the fmdings for the periphery and semiperiphery support the assessment by Lipset et al. [1993], without empirical results, of a likely reversal of the relationship in the form of authoritarian regimes, in spite of increasing levels of economic development in the short-run (see the F-statistics in Tables 3 and 4). Furthermore, the fmdings also support their contention that there is a clear long-term trend in the relationship between economic development and political democracy (see the significance of the error correction terms in Table 3). What conclusion can be drawn for the core? In the core, there is no short-run causal relationship from economic development to democracy, but there is a long-run relationship. This implies that when the levels of economic growth decrease in the core societies, the effects of these on the relationship from economic development to democracy are temporary and it eventually reverts to its long-run relationship. In other words, democracy is not threatened in the long run due to economic downturns in the core. An additional hypothesis was further explored-democratic performance affects economic development across countries. Support for this was found for the core and semiperiphery, but not for the periphery. The evidence clearly suggests that democracy enhances economic development across core and semiperiphery countries. Thus, it may well be that democratizing countries have a greater capacity to adopt new technologies and to reallocate resources in response to changing economic conditions, thereby increasing the rate of economic growth. It also seems likely that democratic countries are more creditworthy compared to other countries, such as the periphery, and they are safer places to invest. Footnotes. Building on the work of Snyder and Kick [1979] by classifying countries as core, semiperiphery, and periphery, Gonick and Rosh [ 1988] demonstrate that there are distinct paths to democracy. According to Gonick and Rosh [1988, p. 195], "The trajectory of core capitalist

476 IAER: NOVEMBER 1999, VOL. 5, NO. 4. states proceeds from a higher value on the democracy index than does that of semi-peripheral states, which in turn proceed from a higher level than peripheral states." The scale for political rights assesses the right to vote, election meaningfulness, multiple political parties, opposition power, and government independence from foreign or military control, whereas the scale for civil liberties covers the freedoms of speech, assembly, and religion and freedom from terrorism or blatant inequality. References Arat, Zebra F. "Democracy and Economic Development: Modernization Theory Revisited," Comparative Politics, 21, 1, 1988, pp. 21-36. Barro, Robert J. "Democracy and Growth," Journal of Economic Growth, 1, March 1996, pp. 1-27. Bollen, Kenneth A. "Political Democracy and the Timing of Development,"American Sociological Review, 44, 4, 1979, pp. 572-8. Bollen, Kenneth A.; Jackman, Robert W. "Political Democracy and the Size of Distribution of Income," American Sociological Review, 48, 1985, pp. 468-79. Burkhart, Ross E.; Lewis-Beck, Michael S. "Comparative Democracy: The Economic Development Thesis," American Political Science Review, 88, 4, 1994, pp. 903-10. Collier, David, ed. The New Authoritarianism in Latin America, Princeton, N J: Princeton University Press, 1979. Cutright, Phillips. "National Political Development: Measurement and Analysis," American Sociological Review, 28, 2, 1963, pp. 253-64. Cutright, Phillips; Wiley, James A. "Modernization and Political Representation: 1927-1966, Studies in Comparative International Development, 5, 2, 1969, pp. 23-41. Durlauf, Steven N.; Phillips, Peter C. B. "Trends Versus Random Walks in Time Series Analysis," Econometrica, 1988, pp. 1333-54. Engle, Robert F.; Granger, Clive W. J. "Cointegration and Error-Correction: Representation, Estimation and Testing," Econometrica, 1987, pp. 251-76. Engle, Robert F.; Yoo, B. S. "Forecasting and Testing in Cointegrated System," Journal of Econometrics, 35, 1987, pp. 143-5. Freedom House. Freedom in the World: Annual Survey of Political Rights and Ovil Liberties, various issues. Gonick, Lev S.; Rosh, Robert M. "The Structural Constraints of the World Economy on National Political Development," Comparative Political Studies, 21, 2, 1988, pp. 171-99. Granger, C. W. J. "Investigating Causal Relations by Econometric Models and Cross-Spectral Methods," Econometrica, 1969, pp. 424-38. Granger, C. W. J. "Some Properties of Time Series Data and Their Use in Econometric Model Specification," Journal of Econometrics, 16, 1981, pp. 121-30. Hadenius, Axel. Democracy and Development, Cambridge, MA: Cambridge University Press, 1992. Hanke, John E.; Reitsch, Arthur G. Business Forecasting, Boston, MA: Allyn and Bacon, 1989. Helliwell, John F. "Empirical Linkages Between Democracy and Economic Growth, "British Journal of Political Science, 25, 1994, pp. 225-48. Huntington, Samuel P. Political Order in Changing Societies, New Haven, CT: Yale University Press, 1968. Jackman, Robert W. "On the Relation of Economic Development to Democratic Performance," American Journal of Political Science, 17, 3, 1973, pp. 611-21.

GLASURE ET AL.: ECONOMIC DEVELOPMENT AND DEMOCRACY 477 Kurth. James R. "Industrial Change and Political Change: A European Perspective," in David Collier, ed., The Authoritarianism in Latin America, Princeton, NJ.: Princeton University Press, 1979. Lipset, Seymour Martin. "Some Social Requisites of Democracy," American Political Science Review, 53, 1959, pp. 69-105. m. Political Man, Garden City, NY: Anchor, 1963. Lipset, Seymour Martin; Seong, Kyoung-Ryung; Torres, John Charles. "A Comparative Analysis of the Social Requisites of Democracy," International Social Science Journal, 45, 2, 1993, pp. 155-75. MacKinnon, James G. "Critical Values for Cointegration Tests," in R. F. Engle; C. W. J. Granger, eds., Long-Run Economic Relationships, Oxford, United Kingdom: Oxford University Press, 1991, pp. 267-76. Maddala, G. S. Introduction to Econometrics, New York, NY: Macmillan, 1992. Muller, Edward N. "Economic Determinants of Democracy," American Sociological Review, 60, 6, 1995, pp. 966-82. Muller, Edward N.; Seligson, Mitchell A. "Civic Culture and Democracy: The Question of Causal Relationships," American Political Science Review, 88, 3, 1994, pp. 635-52. Nelson, C. R.; Plosser, C. L. "Trends and Random Walks in Macroeconomic Time Series: Some Evidence and Implications," Journal of Monetary Economics, 1982, pp. 570-97. Olsen, M. E. "Multivariate Analysis of National Political Development," American Sociological Review, 33, 1968, pp. 699-712. O'Donnell, Guillermo. "Modernization and Bureaucratic Authoritarian: Studies in South American Politics," Politics of Modernization Series No. 9, Institute of International Studies, Berkeley, CA: University of California Press, 1973. Perman, Roger. "Cointegration: An Introduction to the Literature," Journal of Economics Studies, 18, 3, 1991, pp. 3-30. Przeworski, Adam; Limongi, Fernando. "Modernization: Theories and Facts," World Politics, 49, 2, 1997, pp. 155-83. Snyder, David; Kick, Edward L. "Structural Position in the World System and Economic Growth, 1955-1970: A Multiple-Network Analysis of Transnational Interactions," American Journal of Sociology, 84, 5, 1979, pp. 1096-126. Vanhanen, Tatu. The Process of Democratization, New York, NY: Crane Russak, 1990.