Revisiting the link between trade openness and economic growth using panel methods

The link between trade openness and economic growth remains an open question due to the inconclusive findings provided by empirical studies. We argue that the primary reasons for such disagreement are the differences between previous studies regarding their methodologies, sample selection, measures of trade openness, and duration of analyses. Based on this argument, we reinvestigate the trade–growth nexus by applying newly developed methodologies on a robust sample of 82 countries for the period 1960–2019. Also, this study integrates five region-wise subsamples and use different measures of trade openness. Our econometric results corroborate the view that trade openness induces economic growth. These results emerge after employing a variety of panel methods to data, including common correlated effects mean group (CCEMG) estimator and a system generalized method of moments (system-GMM) estimator, which accounts for cross-sectional dependence, structural breaks and endogeneity between trade and growth. Based on our robust results, we can safely claim that trade openness may contribute to economic growth. The findings of the study question the validity of the empirical results of previous studies which argue that trade restrictions promote economic growth.


Background
The impact of trade openness on economic performance is an extensively discussed issue in the economics literature. 1 The literature on international trade explains three different perspectives regarding the effects of international trade on economic growth. First, the Ricardian-Heckscher-Ohlin model postulates that trade may cause a one-time increase in output due to comparative advantage of countries but will have no long-term impact on international trade. Second, neoclassical growth theory posits that technological change is exogenous and, therefore, the trade policy of a country does not affect its economic performance. Third, new growth theories suggest that continuously increasing trade openness leads to the adoption of new technologies, thus allowing an open economy to grow faster than a closed economy.
The empirical literature extensively analyzes the impact of international trade on economic growth (A detailed review of relevant literature is compiled in Table 1). The central question these analyses seek to answer is whether international trade affects longterm economic growth. The answers are inconclusive, as three separate theories have been put forth.
Some scholars propose that trade openness contributes to economic growth (see Table 1). Conversely, others believe that trade openness hurts economic growth. For example, Redding (1999) reports that trade openness can prevent economic growth and welfare if economic agents fail to internalize the potential for productivity growth. Redding (1999) also points out that if the private sector does not fully materialize the potential of productivity growth and the spillover effect, then trade openness may hinder economic growth and welfare. Some prominent studies, such as those conducted by Rodríguez and Rodrik (2000), Rodrik, Subramanian, and Trebbi (2004), and Wacziarg and Welch (2008), indicate that trade openness negatively impacts economic growth. Kim, Lin, and Suen (2011) and Ramzan et al. (2019) consider the relationship between economic growth and trade openness to be non-linear. Kim, Lin, and Suen (2011) report that trade openness positively affects economic activity in high-income countries but negatively affects economic activity in low-income countries. Ramzan et al. (2019) find that TFP influences the relationship between trade openness and economic growth such that it is negative when countries possess low level of TFP development while the relationship is positive for countries who have achieved a threshold level of TFP.
Finally, some researchers posit that the impact of trade openness on economic growth depends on the characteristics of the economy. For example, Cooray, Dutta, and Mallick (2017) document that trade openness is beneficial in a favorable economic and social environment. However, if the economic environment is not conducive to the benefits of trade openness, then it might harm the economy. Furthermore, Rodrik, Subramanian, and Trebbi (2004) find that the impact of trade openness on economic growth weakens when the regression is controlled via the measures of institutions and geography. Rigobon and Rodrik (2005) also report that the estimated coefficient of trade openness on economic growth appears to be negative when it is simultaneously introduced with institutions and geography as explanatory variables in regression.
The literature documents several reasons for these inconclusive findings. For example, Frankel and Romer (1999) point out that merely examining the correlation between trade and growth cannot identify and explain the direction of causation between the two variables. Specifically, they show that exogenous cross-country variations in trade openness are positively associated with economic growth across countries. They specify a growth equation by using 'distance' as an instrumental variable to identify causation. The findings of Frankel and Romer (1999) reinforce the notion that trade openness promotes economic growth. However, Rigobon and Rodrik (2005) examine that international trade harms economic growth when endogeneity is considered. Acemoglu, Simon, and Robinson (2001) and Rodrik, Subramanian, and Trebbi (2004) introduce the role of institutions in controlling the effect of endogeneity. These inconclusive findings reported despite controlling endogeneity invoke the need for further empirical investigations on the trade-growth nexus. Further, Rodríguez and Rodrik (2000), Levine and Renelt (1992), and Squalli and Wilson (2011) note that the findings of most of the studies are sensitive to measures of trade openness, the methodologies adopted, and the time-span considered.   Yanikkaya (2003) and Alcala and Ciccone (2004), among many others, use trade volume to proxy trade openness. Meanwhile, Wacziarg and Welch (2008), Kneller, Morgan, and Kanchanahatakij (2008), and Gries, Kraft, and Meierrieks (2009) examine trade policies.
In the present study, we attempt to investigate trade openness in three different dimensions. Specifically, we concentrate on trade volume, tariff rate, and globalization. Trade volume specifies a country's overall level of trade openness. Tariff rates indicate policymakers' stances toward reducing trade restrictions. Globalization is a synthetic measure intended to cover the overall openness environment.
Furthermore, this paper addresses several methodological issues. Two main econometric issues including cross-sectional dependence and endogeneity have been discussed in panel studies. We use a common correlated effects mean group estimator (CCEMG) to address cross-sectional dependence. Importantly, this estimator is robust even when faced with structural breaks. The issue of endogeneity is examined using a generalized method of moments (GMM) estimator. Recently, Mullings and Mahabir (2018) document ambiguous findings when examining endogeneity through GMM. Therefore, it would be interesting to address the issues of cross-sectional dependence and endogeneity jointly. For this purpose, we use CCEMG-GMM by using the lags of endogenous variables to account for cross-sectional dependence and endogeneity. Finally, we distribute the data into five subsamples to test the sensitivity of regional biases in testing the trade-growth nexus.
We find that trade openness promotes economic growth. Our findings are robust to various measures of trade openness, regional samples, and methodologies. We discuss the two most widely examined econometric issues (i.e. cross-sectional dependence and the endogeneity). We find similar findings in both cases. Furthermore, our findings are robust to the distribution of full sample into five region-wise subsamples. Therefore, we suggest that policymakers should promote trade openness to enhance the economic growth of their countries.
The rest of this paper is arranged in the following way. The next section will discuss the methodology and econometric issues of the study. The empirical findings of the paper and comparisons with the findings of previous studies are presented in Section 3. Section 4 concludes the article and provides some important policy implications. The details about construction of variables and data sources are presented in Appendix E.

Methodology and econometric issues
Following extant literature, we can specify a trade-growth nexus model using the following framework: where y is the natural log of per capita gross domestic product (GDP), open represents the various measures for trade openness including trade volume, trade policy and globalization, and Z represents a vector of control variables. A substantial body of literature documents that trade volume, which specify the overall level of a country's trade openness, positively affects economic growth. However, the impact of trade policies on economic growth is found to be controversial. On the one hand, lowering trade barriers can reduce the economic cost of production. On the other hand, there is a consensus among researchers that lowering trade barriers could negatively impact the performance of industries and, in turn, the economic growth of countries. This controversy also motivates us to reinvestigate the trade-growth nexus. Further, the overall level of openness and liberalization is associated with the rapidly evolving globalization process.
Several other factors have also been incorporated in this framework to broaden the scope of our study and to improve overall fit of regression. For example, financial sector development is a broadly discussed factor concerning the trade-growth nexus. Specifically, Misati and Nyamongo (2012) utilize the measure of financial sector development in their growth regression. Similarly, Lloyd and MacLaren (2000) and Jin (2006) have focused on foreign direct investment as crucial factor for economic growth whereas Arawatari, Hori, and Mino (2018) incorporate inflation in endogenous growth model with heterogenous R&D skills of agents. Further, Teixeira and Queirós (2016) introduce human capital in growth model to assess its direct and indirect impact on economic growth. In addition, following the comparative advantage theory, Buch and Toubal (2009) consider physical capital as a critical determinant of economic growth.
Following preceding discussion and taking insights from Yanikkaya (2003), we specify the following econometric equation to investigate the relationship between trade openness and economic growth: where y is the log of GDP per capita; open is a proxy for trade openness which is further divided into trade volumes (trade), tariff policy (tariff ), and a general index of globalization (Glob); k is the natural log of real per capita physical capital; fd denotes financial development; cpi is a measure of inflation; edu is used as a measure of human capital; and gov is governance. 2 There may be at least two types of econometric issues associated with an empirical model like ours, e.g. cross-sectional dependence and endogeneity bias.
Interestingly, much of the openness-growth nexus ignores the issue of cross-sectional dependence which may cause biased results and misleading policy implications. With the high degree of financial integration and expanding waves of globalization, the economic interdependence among countries has substantially increased. Thus, estimating equation (2) without controlling for cross-sectional dependence may cause a downward bias in standard errors and affect the reliability of estimates. We test the level of cross-sectional dependence by using Pesaran's (2004) cross-sectional dependence test. Pesaran (2006) proposed CCEMG method to address cross-sectional dependence and heterogeneity in panels.
In addition, the possibility of the presence of omitted variables bias and simultaneity in similar empirical models as well as the evidence provided in existing literature about two-way causality between trade openness and economic growth may lead to potential endogeneity bias. We rely on system-GMM panel estimator as proposed by Arellano and Bover (1995) and Blundell and Bond (1998) to overcome this problem. System-GMM has gained popularity among researchers. For example, Nickell (1981) and Teixeira and Queirós (2016) document that it can provide solutions for several issues including heterogeneity among panels, endogeneity, lack of external instruments, short panel bias, dynamic panel bias and other types of measurement errors. Furthermore, GMM estimators can handle unbalanced panels and the bias induced by multiple endogenous variables.
It is important to ensure that our empirical results are not jointly influenced by crosssectional dependence and potential endogeneity bias. Therefore, these issues should also be addressed simultaneously. For this purpose, following Chudik and and Pesaran (2015) and Neal (2015), we employ CCEMG-GMM method to estimate equation (2) by using the lags of endogenous variables. The CCEMG method controls for the endogeneity bias that can arise due to common factors and reverse causation. However, to ensure that our results are reliable, we estimate a CCE equation for each country using GMM. The simple average of the slope parameters will provide CCEMG-GMM.

Empirical findings and discussion
The estimation process is conducted in several steps. Specifically, we estimate equation (2) using different panel estimators, accounting for cross-sectional dependence and endogeneity, for a panel dataset of 82 countries over the period 1960-2019. The natural start of the estimation process is to estimate equation (2) using ordinary least squares (OLS) method. We use country-fixed effects and time-fixed effects to control for heterogeneity among panels and contemporaneous correlation, respectively. The results of OLS method serve as baseline estimates. Then, we address the issues of cross-sectional dependence and endogeneity bias by estimating equation (2) using CCEMG and GMM methods, respectively. Finally, CCEMG-GMM method is employed to address crosssectional dependence and endogeneity jointly. In addition, several diagnostic tests are performed to test the robustness of our estimates. Further, we divide our dataset into two groups including full sample and five region-wise subsamples.
The findings for the full sample are presented in Table 2. Columns 1, 2, and 3 of Table 2 show the results of the pooled ordinary least square (POLS) estimator. The results are based on a priori expectations. However, we do not discuss POLS results due to the bias inherited from cross-sectional dependence and endogeneity problem. Statistic for Pesaran cross-sectional dependence (CD) test and corresponding p-values are reported in the lower panel of columns 4, 5, and 6 of Table 2. Based on the p-values, the null hypothesis of cross-sectional independence among the panels is rejected. It verifies our argument that high integration of financial development and expanding liberalization has increased economic interdependence among countries. Given the situation, POLS method provides biased estimates. Therefore, we use CCEMG as an alternative method to estimate equation (2). The upper panel of columns 4, 5, and 6 of Table 2 shows the CCMEG estimates.
We regress different measures of trade openness on log difference of per capita GDP. Three different measures of trade openness enter in equation (2) separately to avoid the problem of multicollinearity. The estimated coefficient of trade (the natural log of trade-to-GDP ratio) presented in column 4 of Table 2 is positive and significant, showing that trade volumes meaningfully enters in the growth regression. It reveals that an increase in trade volumes may boost economic growth of a country. This result is in line with the previous studies of Chang, Kaltani, and Loayza (2009), Rao, Arthur, and Chaitanya (2011), Cooray, Mallick, and Dutta (2014, Cooray, Dutta, and Mallick (2017), and Zahonogo (2017). However, our finding is not consistent with the predictions provided in O' Rourke (2000) and Gries, Kraft, and Meierrieks (2009). Also, it is evident that a country's trade dependency ratio and trade volume may not exhibit its political stance because several types of macroeconomic fluctuations emerging from exchange rate dynamics, technological progress and other factors may be reflected in trade volumes. Therefore, we look at other proxies for trade openness which highlight a country's position in the context of its trade policies. Furthermore, as mentioned earlier, trade policies and trade liberalization exhibit more controversial economic impacts than trade volume.
In this regard, the existing literature asserts that tariff rate is a direct indicator of trade openness and it serves as an accurate policy-related variable. However, different studies on the trade-growth nexus employ different measures as potential proxies of tariff rate. For example, Kanbur and Zhang (2005) use 'effective tariff rate' (i.e. the ratio of tariff revenue to total import revenue), while Dobson and Ramlogan (2009) incorporate 'average tariff rate'. Ma and Dei (2009) examine that different measures of tariff rate could have different impact on economic development. Therefore, it is critical to rely on multiple measures of tariff rate in order to expand our understanding about the relationship between trade liberalization and economic growth. We also incorporate multiple measures of tariff rate into our econometric specification and regress them on the dependent variable using CCEMG method. However, we do not observe any substantial differences or ambiguity within our estimated results originated from cross-country analysis. Thus, we present the estimated results of 'average tariff rates' only. It also helps us compare our results with those of notable previous studies.
The estimated coefficient of our second measure of trade openness, 'average tariff rate', provided in column 5 of Table 2 is significant and negative. This trend implies that government's policy of reducing trade restrictions has a positive impact on economic  Notes. The standard errors are presented in the parentheses. * * * , * * and * represent 1, 5 and 10 percent level of significance.
growth. Further, it reveals that the economic growth of developing countries could be enhanced by reducing the average tariff rate. Numerically, our results show that, economic growth could be increased by 0.801 percent due to a one percent decrease in tariff rate. Our results corroborate the findings of Wacziarg and Welch (2008), Kneller, Morgan, and Kanchanahatakij (2008), and Gries, Kraft, and Meierrieks (2009). Nonetheless, the use of tariff rate as a proxy of trade openness has been criticized on various grounds despite having an appropriate trade policy dimension. For example, some aggregation biases exist, especially in the case of average tariff rates. Furthermore, there are significant gaps between statutory tariff rates and the collected tariffs, especially in developing countries. In addition, Dollar and Kraay (2004) note that there is little correlation between the measures of trade volume and tariff rates. Therefore, trade volumes and trade policy measures might yield misleading results in terms of economic growth.
Keeping these problems in view, we incorporate a third proxy of trade openness by using the 'globalization index' as provided by Dreher, Gaston, and Martens (2008). This index gives proper weights to trade volumes and tariff rates, as well as to other flow and restriction measures of trade openness. This index is considered to be a more comprehensive measure of trade openness than trade volume and average tariff rate. The estimates provided in Column 6 of Table 2 show that globalization has a significant and positive impact on economic growth, confirming our view that trade openness contribute to economic growth.
As mentioned earlier, besides trade openness, there are several other variables which may affect economic growth and absence of these variables from empirical model may make the relationship spurious. In this regard, we select a few control variables including physical capital (k), financial development (fd), inflation (cpi), human capital (edu), and governance (gov). The construction of these variables is explained in Appendix E. Table 2 shows that in all three model specifications (e.g. Columns 4, 5 and 6 of Table 2) the impact of physical capital (k) on economic growth is significant and positive. This outcome is in line with a substantial body of previous literature which explains that investment activities and physical capital stocks have a positive impact on economic growth. Recently, financial sector development has received special attention in two dimensions. First, it has been discovered that a well-functioning financial system could enhance the economic growth of countries (Jalil, Feridun, and Ma 2010) and that trade-growth nexus could flourish via various channels of financial sector development (Baltagi, Demetriades, and Law 2009). Therefore, a measure of financial sector development must be incorporated into a growth regression. The positive and significant coefficient of the fd variable shows that financial sector development has a positive impact on economic growth. Moreover, the negative value of the estimated coefficients of inflation suggests that inflation hurts economic growth. However, the coefficients are statistically insignificant in all three model specifications. Further, the estimated coefficient of human capital variable is significant and positive. This finding supports the broadly accepted view about endogenous growth theory.
Finally, the quality of institutions has recently gained enormous attention as an important determinant of economic growth. Acemoglu, Simon, and Robinson (2001), among others, provide evidence that the quality of institutions explains the differences among countries in economic growth. However, different studies specify different methods to measure the quality of institutions. It can be measured through incorporating different factors including law and order, the quality of formal institutions, the corruption and accountability of public officials and others. Therefore, we develop an index 'GOV' by using a principal component analysis. The method we adopted to construct the variable is presented in Appendix E. The estimated coefficient of 'GOV' is positive and significant, finding that the quality of institutions is a significant determinant of economic growth. This result is in line with that of Acemoglu, Simon, and Robinson (2001).
Besides cross-sectional dependence, we also control for the presence of endogeneity bias into our empirical model and estimate equation (2) using system-GMM method. The upper panel of columns 7, 8, and 9 of Table 2 shows the GMM estimates. The lower panel of these columns provide specification and validity tests statistic for overidentifying restrictions for instrumental variables. The test values prove that the hypothesis for exogeneity of instruments is not rejected (see corresponding Sargan and Hansen J values provided in Table 2). From Table 2, it can be verified that GMM estimates are broadly consistent with that of previously discussed CCEMG estimates. For example, the coefficient estimates of the Trade and Glob variables are significant and positive while that of the Tariff variable is significant and negative which is in line with our CCEMG estimates. It reveals that our estimates are robust to changes in estimation methods.
So far, we have addressed two methodological issues individually cross-sectional dependence and endogeneity bias by employing a CCMEG estimator and a GMM estimator. Witnessing the estimated results, we are convinced that the positive impact of trade openness on GDP growth is not derived from cross-sectional dependence or endogeneity bias. However, we argue that jointly addressing these issues could be more enlightening. For this purpose, following Chudik and Pesaran (2015) and Neal (2015), we use CCEMG-GMM method by using the lags of endogenous variables. The estimation results are shown in columns 10, 11, and 12 of Table 2. The results show that the signs and significance levels of estimated coefficients has not changed considerably. It implies that our main results are not individually or jointly affected by endogeneity or cross-sectional dependence. Drawing on these results, we argue that previous empirical studies which have suggested that trade openness has either a negative impact or no impact on economic growth should be reconsidered.

Sensitivity analysis
A number of studies test the convergence hypothesis by using the growth regression. The convergence hypothesis means that an economy's initial conditions have no significant implications on its long-run per capita income. The poor countries will tend to grow faster than the rich and eventually catch up with them. Thus, in the long-run, countries will converge to a common level of income per capita. The present study doesn't focus on the convergence discussion. However, keeping the sensitivity analysis in view, we put the initial value of the per-capita income on the independent side to test the convergence. The magnitude of the initial value is negative and significant (see Appendix B). However, the lower magnitudes show the slow speed of convergence. Notably, the inclusion of the initial value does not alter the signs of prime variables.
For robustness and to check the sensitivity of our estimates, we distribute our full sample into region-wise subsamples (for distribution of countries in subsamples see Appendix A) and re-estimate equation (2) for each of five subsamples using CCEMG, GMM, and CCEMG-GMM 3 panel estimators. First, we estimate equation (2) using CCEMG and GMM methods individually and then employing CCEMG-GMM method jointly. Like the results mentioned in Table 2, these estimates also passed through various Table 3. Impact of trade openness on economic growth: CCEMG-GMM estimates (region-wise sample).   Table 3.
The sign and significance of the estimated coefficients are broadly consistent with the full sample estimates provided in Table 2. For example, the coefficient estimates of the Trade and Glob variables are significant and positive while that of the Tariff variable is significant and negative. It shows that expanding trade volume, lowering trade restrictions, and opening to the rest of the world are critical factors that promote economic growth. The following results can be discerned easily from the estimates provided in Table 3. First, as a whole, trade openness accelerates economic growth both in developing and developed countries. Second, since the magnitude of the estimated coefficient of the 'trade' variable is greater for Sub-Saharan Africa than other regions, committing to trade openness through expanding trade volume is relatively more significant policy for these countries. Third, the estimated results of the 'tariff '' variable show that trade restrictions dampen economic growth and this effect is relatively stronger for emerging countries than other developing or developed countries. Thus, eliminating trade barriers is an important policy for emerging countries. Fourth, the estimated coefficient of the 'Glob' variable is greater for Sub-Saharan Africa which reveals that these countries can accelerate economic growth through opening-up and expanding exposure to rest of the world.
The GMM method in the dynamic panel mainly solves the endogeneity caused by the lag term of the dependent variable as the explanatory variable. To solve the endogeneity bias from other sources, such as the two-way causality of economic growth and openness, there is a need to introduce additional instrumental variables of openness. Therefore, we follow Frankel and Romer (1999) in using geographic variables as the instruments. These variables are population, land area and distance from the equator. It is argued that the countries' geographic characteristics with impacts on trade are not affected by income, government policies, and other factors that influence growth. Several other prominent studies like Irwin and Tervio (2002), Dollar and Kraay (2003), Alcala and Ciccone (2004), Borrmann, Busse, and Neuhaus (2006) and Kim, Lin, and Suen (2011) also follow the Frankel and Romer (1999) method of choosing external instruments. The findings are reported in Appendix C. 4 The major outcomes of the study don't alter both in sign and significance. Furthermore, as our data period is long, we also replicate our results in Appendix C using 5-year averages of our full sample and report them in Appendix D. There are some minor differences, but they are not strong enough to alter our overall conclusion.

Conclusion
The impact of trade openness on economic growth is an extensively discussed issue in economic literature. However, previous studies exhibit inconclusive results about the impact of trade openness on economic growth. Several studies suggest trade openness to be a significant determinant of economic growth, while many others predict that trade openness dampen growth and decrease welfare of a society. Also, a few studies find non-linear relationship between trade and growth whereby the degree of non-linearity depends upon income-level, level of TFP and other factors. The reasons behind previous studies' inconclusive findings are their sensitivity to measures of trade openness, the methodologies adopted, and the timespan considered.
This article revisits the trade-growth nexus using heterogeneous panel methods that consider cross-sectional dependence and endogeneity in empirical models. The empirical analysis presented in this paper is based on a panel dataset of 82 countries for the period 1960-2019. We use three different dimensions to measure trade openness to test the robustness of the empirical results. Specifically, we concentrate on trade volume, tariff rate, and a synthetic measure of globalization. We address the issues of crosssectional dependence by employing the CCEMG method and we control for endogeneity bias using the system-GMM method. Further, the CCEMG-GMM method is employed to address cross-sectional dependence and endogeneity jointly. Also, we control for regional heterogeneity among panels by distributing our full sample into five regionwise subsamples and re-estimating it using previously discussed panel methods. Several diagnostic tests are performed to test the robustness of our estimates.
The findings of this article corroborate the thesis of trade-led growth. Specifically, we suggest that expanding trade volume, eliminating trade restrictions and opening up to the rest of the world can contribute to economic growth. Our findings are robust to several econometric issues including cross-sectional dependence and endogeneity that can arise in a longer panel. Also, these results are robust to alternative estimation methods. Therefore, we can safely conclude that different countries' levels of trade openness explain differences in their economic growth. Furthermore, we contradict the idea that restrictions on trade promote economic growth. Some authors have made such an argument to motivate the protectionist policy measures being undertaken in some emerging economies. Our estimates suggest that trade restrictions are likely to hamper the economic growth of countries that have already begun to boost their economic growth. Therefore, policymakers are encouraged to implement trade policies based on lower tariffs to attract trade and investment opportunities. Studies which postulate that trade restrictions favor economic activity should be revised considering these econometric issues, as their findings could be affected by cross-sectional dependence and endogeneity bias.
Our findings provide a clear message to policymakers that more trade openness can bolster the economic performance of countries. Furthermore, reductions of trade restrictions through free trade agreements are expected to help countries to improve their economies. Other factors, such as physical capital, human capital, financial sector development, and the quality of institutions, also play essential roles in economic growth.
Though the findings of this study are straightforward and useful for policymakers, our panel covers several decades, and many events have occurred during this period. Therefore, researchers could investigate whether the impact of trade has varied between smaller periods within the period of 1960-2019. An analysis of this nature could provide meaningful insights. Furthermore, there is a serious need for an empirical study to shed light on the robust findings of the trade-growth nexus and the channels through which trade impacts countries' economic performance.
The index of globalization is prepared by Dreher, Gaston, and Martens (2009) which is a sum of three different indices, that is, social globalization, economic globalization and political globalization. The authors allocate the different weights to a number of openness and liberalization indicators.