Do cultural dimensions affect the trade openness of the OECD countries? . 53

The first regression analysis aims to investigate the main hypothesis of this thesis:

“How do the cultural dimensions affect trade openness?”. The analysis is run on the sample of 36 OECD countries in the period 2010-2019. This analysis examines the general effect of six cultural dimensions on trade openness in the recent years. Table 6 represents the outcome of the first regression analysis with the use of the OLS estimator.

Since the estimates in terms of significance were comparable through the years, Table 6 presents just the outcome of analysis based on the average data from the period 2010-2019.

54

Table 6: The OLS regression analysis of the OECD countries between years 2010-2019

Variables Estimated

parameters Std. error P-value

Intercept -18.06972 9.64000 0.07309 .

log(pdi) 0.14271 0.18783 0.45477

log(idv) 0.21798 0.22723 0.34696

log(mas) 0.08008 0.09892 0.42613

log(uai) -0.21233 0.26980 0.43899

log(ltowvs) 0.57328 0.17498 0.00319 **

log(ivr) 0.24461 0.17131 0.16621

log(g.cap.form) 0.17183 0.39237 0.66536

log(work_pop) 5.42263 2.27767 0.02556 *

log(econ_size) -0.25293 0.04884 2.65e-05 ***

log(ex_rate) -0.03850 0.03311 0.25630

log(prox_HC) -0.54177 0.66973 0.42650

Multiple R-squared: 0.7226

Adjusted R-squared: 0.6116

Source: Own Elaboration With Use Of RStudio¹³ And Based On Respective Data.

The results summarized in Table 6 show that just one cultural dimension significantly affects trade openness. Specifically, the long-term vs short-term dimension is considered to be significant on a 1% level of confidence. Its positive effect on trade openness means, that the more long-term oriented is the country, the higher is its tendency to be open to international trade. The regression outcome indicates that a 1% increase in this dimension will cause a 0.57% rise in the country´s trade openness. Apart from the cultural determinants, there are two additional variables that affect trade openness significantly.

More specifically, the economic size of a country is significant on a 1% level of confidence and the labor size (proportion of the population in productive age) on a 5%

level of confidence. As expected, these two variables have an opposite effect on trade openness. Trade openness decreases with increasing economic size because the country´s dependence on international trade weakens. Specifically, the 1% increase in economic size will evoke a 0.25% decrease in the country´s trade openness. On the other hand, the 1% increase in labor size will provoke a 5.42% increase in trade openness. As already

13 The detailed description of the RStudio software is provided in the following webpage:

https://rstudio.com/ .

55

explained above, this is related to the fact that mainly the productive population demands the tradable goods, so in this way, they drive the trade openness as well.

Regarding the rest of the variables, the majority meets the above-mentioned hypothesis about their influence. Surprising is the opposite impact of power distance dimension and human capital level. In the case of the power distance dimension, it was expected that the higher corruption levels in more hierarchical countries will impact trade openness negatively. Nevertheless, the results suggest that the hierarchical countries are more open.

The possible explanation is that the strictly defined decision structure provides order and a clear strategy that in production firms increases efficiency and consequently their competitiveness in the international market. In the case of human capital, it was expected that the highly qualified workers will increase the competitiveness of national firms and in this way, the country´s participation in international trade. However, the regression analysis shows that increasing human capital leads to a decrease in trade openness. A possible explanation is offered by Giordani & Mariani (2020) who observed that the revival of protectionist tendencies in the last decade is related to the increase of the highly educated class. The authors suggest that the exponentially increasing globalization also enhanced the increase of the share of highly educated workers in the country that due to the two-stage political game is able to shape the trade openness preferences and the redistribution of benefits sourcing from free international trade. This biased redistribution causes that the lower educated workers are left out and do not enjoy the benefits of trade liberalization. Motivated by this injustice, they enhance the formation of the coalition (political representants) that supports protectionism (Giordani a Mariani 2020).¹⁴

Analysis yields another interesting outcome which is the value of adjusted R-squared. It indicates how much of the overall variance of the trade openness index is explained by the model. In this case, the used model specification explains 61.16 % of the overall variance in trade openness.

The last aspects to be checked, are the Gauss-Markov assumptions. The results of tests that were run have not found any violation of the assumptions. This holds for the

14 Moreover, the statement of these authors supports the USA political situation described in the theoretical part. Trump´s election program was built on the protection and preservation of manufacturing job positions.

56

endogeneity issue as well, since after running the regression analysis with the 2SLS estimator, the Hausman test indicated that the null hypothesis cannot be rejected. This means that the endogeneity was not proved and the OLS estimates are considered to be BLUE¹⁵.

Table 7: The Hausman test (OECD, 2010-2019)

H0: There is no endogeneity issue H1: There is endogeneity issue

P-value: 0.2816564

Source: Own Elaboration With Use Of RStudio And Based On Respective Data.

7.2. Does the impact of cultural dimensions on trade openness differ in time?

The second regression analysis serves to evaluate whether the impact of culture on trade openness differs in time. The assumption is that the strength of cultural effect on trade openness will be distinct in the case of the 1980-1989 period and the 2010-2019 period. The main reason to define especially these years is related to the globalization evolution. Even though globalization started much earlier than in 1989, after this year it became a real global phenomenon. For that reason, I consider the years 1980-1989 to be a period before the globalization boom. On the other hand, the period 2010-2019 represents the time of the most intense globalization since, as chapter 2 showed, the level of globalization is continuously increasing. Nevertheless, the level of globalization is not the unique difference between these two periods. Another aspect to bear in mind is the distinct quantity of multilateral trade agreements. The quantity of trade agreements is in the period 2010-2019 much higher than it was in 1980-1989, (Mattoo et al. 2020).

Based on these differences, I assume that two possible opposing effects might be observed. The first of these possible effects is that the higher globalization levels in

2010-15 BLUE estimate is the best, linear, unbiased estimate obtained by the OLS estimation when the Gauss-Markov assumptions hold.

57

2019 will enhance the protectionist tendencies and strengthen the effect of cultural dimensions on trade openness compared to 1980-1989. The second possible effect is that the international trade agreements valid in the period 2010-2019 consider the importance of culture and prevent additional, unnecessary protectionism.¹⁶ Moreover, the trade agreements may represent the dominant view best practice that affects the trade strategies worldwide. Jointly it would mean that the trade agreements may lead to suppression of cultural effect in the period 2010-2019. To examine which of these effects is manifested empirically, I run a regression analysis on the reduced OECD sample containing 28 member countries¹⁷ in the period 1980-1989 and 2010-2019.

Table 8 contains the results from the OLS regression analysis aimed at the period 1980-1989. Even though the size of the coefficients was much more volatile through these years compared to previous regression analysis, in terms of the significance, the estimates were comparable. For that reason, Table 8 contains just the outcome from the analysis of the average values from the period 1980-1989.

Before I will progress in the evaluation of the results, I need to comment on one change in the regression function. After running the analysis with all the explanatory variables, the test for multicollinearity showed disturbing values. To prevent the multicollinearity bias I had to eliminate the human capital from the regressors.¹⁸ The outcome of the regression analysis after such a change is summarized in Table 8.

16 The example is the trade agreement signed between Colombia and Northern Triangle. The (217) Colombia-Northern Triangle (Annex III., Schedule) represents the sectoral carve-out oriented to regional television. More specifically, the trade agreement contains limitations on free trade in the television broadcasting industry, so the countries´ national culture will be protected (Mattoo et al. 2020).

17 The reduced OECD sample compared to the previous sample does not contain the former socialist states.

18 In the reduced OECD sample, appeared a strong correlation between the individualism vs collectivism dimension and the human capital measure. Since the cultural dimensions are the main explanatory variables of our interest, I decided to eliminate the human capital variable. Its exclusion resolved the multicollinearity issue. A possible explanation of this is that the individualistic tendency in a society enhances the individualistic approach in schools which leads students towards independence, good self-presentation, and outstanding individual achievements. This drive then motivates the students to achieve more prestigious jobs which are generally related to higher human capital.

58

Table 8: The OLS regression analysis of reduced OECD sample in period 1980-1989

Variables Estimated

parameters Std. Error P-value

Intercept 2.453434 7.982811 0.76232

log(pdi) 0.143075 0.230917 0.54374

log(idv) 0.125135 0.259901 0.63632

log(mas) 0.051030 0.107834 0.64207

log(uai) -0.295867 0.329991 0.38246

log(ltowvs) 0.527423 0.175088 0.00785 **

log(ivr) 0.054014 0.308031 0.86287

log(g.cap.form) -0.416503 0.501356 0.41763

log(work_pop) 1.019886 1.949221 0.60757

log(econ_size) -0.283683 0.052621 4.88e-05 ***

log(ex_rate) -0.008121 0.035073 0.81966

Multiple R-squared 0.758

Adjusted R-squared 0.6156

Source: Own Elaboration With Use Of RStudio And Based On Respective Data.

Similarly, as in the previous regression analysis, just one cultural dimension shows a significant effect on trade openness. The long-term vs short-term orientation seems to be the most important cultural indicator of trade openness, even in the period 1980-1989. Its significant positive effect is confirmed on a 1% level of confidence. The 1% increase in the long-term orientation dimension leads to a 0.53% rise in the country´s trade openness.

In the studied period, the high long-term orientation in a developing country could lead to a faster opening towards international trade and related faster economic growth. An example is Hong Kong or Indonesia. Both states are long-term oriented (achieved values of 61 and 62) and, as was shown in chapter 1, during 1976-1985 belonged to highly open countries with high average GDP growth. Regarding the other cultural dimensions, the direction of their effect is consistent with the one from the previous regression analysis.

From the macroeconomic control variables, just the economic size has a significant negative impact on trade openness. The significance of economic size is confirmed on a 1% level of confidence, and its 1% increase weakens the trade openness by 0.28%. In the case of the remaining macroeconomic control variables, one surprising effect might be observed. The gross capital formation in the period 1980-1989 resulted to be a negative factor of trade openness. Since I have not found any specific reason for the gross capital

59

formation to have a negative impact on trade openness, I assume that its negative sign is related to the impact of the economic size. The statistical description of the dataset showed that the gross capital formation between the years 1980-1989 was considerably higher than the one between 2010-2019. This indicates that the investment in physical capital was at that time the main booster of economic growth. It means that by the construction of the trade openness index (trade volume/GDP) higher values of gross capital formation would be associated with greater GDP and consequently lower trade openness index.¹⁹ Regarding the adjusted R-squared, the model is able to explain 61.56%

of the overall variance in trade openness.

Regarding the Gauss-Markov assumptions, the results have not shown any violation of the assumptions. After running the 2SLS estimation, the Hausman test indicated again, that the null hypothesis cannot be rejected, so the endogeneity was not proved. It means that the OLS estimation is BLUE.

Table 9: The Hausman test (reduced OECD, 1980-1989)

H0: There is no endogeneity issue H1: There is endogeneity issue

P-value: 0.440003

Source: Own Elaboration With Use Of RStudio And Based On Respective Data.

Table 10 summarizes the results of the OLS regression analysis from the period 2010-2019. The presented estimates are based on the average data from this period since the significance of individual variables was constant through the observed years. Moreover, the regression function had to be modified equally as in the case of the 1980-1989 analysis. In order to deal with the multicollinearity issue, the human capital was eliminated from the regression function. The presented outcome already includes this modification.

19 However, with the increasing globalization and the international trade volume, technological progress and innovations are transmitted also through trade. This weakens the effect of gross capital formation on GDP growth. For that reason, I expect that in the next estimation oriented on the period 2010-2019 the gross capital formation will have the expected positive effect on trade openness.

60

Table 10: The OLS regression analysis of reduced OECD sample in period 2010-2019

Variables Estimated

parameters Std. error P-value

Intercept -8.76247 10.52232 0.416532

log(pdi) 0.43331 0.22731 0.073674 .

log(idv) -0.46519 0.28258 0.118078

log(mas) 0.11143 0.11717 0.354904

log(uai) -0.79539 0.33670 0.030347 *

log(ltowvs) 0.69278 0.17222 0.000883 ***

log(ivr) 0.26795 0.29024 0.368811

log(g.cap.form) 0.43714 0.40885 0.299929

log(work_pop) 3.68883 2.34845 0.134665

log(econ_size) -0.30202 0.06155 0.000133 ***

log(ex_rate) -0.14278 0.04603 0.006483 **

Multiple R-squared 0.7771

Adjusted R-squared 0.646

Source: Own Elaboration With Use Of RStudio And Based On Respective Data.

The estimation results show a much stronger effect of cultural dimensions than in the period 1980-1989. In this case, there are three cultural dimensions that are considered to have a significant influence on trade openness. First is, equally as before, the dimension of long-term vs short-term orientation. Its positive significant effect can be confirmed on the 1% level of confidence. Moreover, the 1% increase in the country´s long-term orientation enhances even a 0.69% increase in trade openness which is 0.16 percentage points higher effect than in the period 1980-1989. The second important cultural indicator of trade openness is uncertainty avoidance. Uncertainty avoidance is confirmed to have a significant negative impact on a 5% level of confidence. It means, that countries avoiding high uncertainty tend to prevent the vulnerability and insecurity that is brought up by opening. More specifically, if the uncertainty avoidance increases by 1% it provokes a 0.795% decrease in trade openness. Last, the power distance dimension can be proclaimed to be significant on a 10% level of confidence. The effect of power distance resulted again to be positive. By increasing the power distance by 1% the trade openness index rises by 0.43%. It means that in the period 2010-2019, the hierarchy and strict order are important to perceive a country´s trade openness. From the macroeconomic control variables, the economic size and exchange rate resulted to be significant for trade openness. Both explanatory variables have a negative impact and can be confirmed

61

significant on the 1% level of confidence. More specifically the 1% increase in economic size causes a 0.30% decrease in trade openness. In the case of the exchange rate, its 1%

rise enhances a 0.14% reduction in the country´s openness. The overall strength of the model specification is given by the adjusted R-squared. In this case, the model can explain 64.6% of the overall variance in the trade openness index.

Regarding the direction of the effects of variables, the results show that the gross capital formation again affects the trade openness positively. This supports the assumption made above, that the negative effect in regression analysis for 1980-1989 was related to its strong positive impact on GDP growth. In the case of the remaining variables, there is just one surprising change compared to previous results. With the change of the sample, individualism vs collectivism resulted to have a negative impact on trade openness. To properly analyze this change, it is important to look closer to countries that form the studied sample. It is formed by economically highly significant developed countries that in their majority were members of GATT far before 1980. It means that there is a long trade history between them. Based on the study of Hofstede et al. (2008), the collectivistic societies, after building mutual relations, tend to have significant trust in their trade partners. This enhances mutual collaboration and trade. On the other hand, based on the study of Hajikhameneh & Kimbrough (2019) the individualistic countries look all the time for new potentially more lucrative trade partners. Nevertheless, this weakens the mutual trust between them and their recent partners. In this light, it would be understandable that the highly individualistic and economically strong countries can be considered untrustful partners, that may any time change their position in order to achieve a more beneficial deal. For that reason, mainly in trade organizations, strong individualism can have a negative effect on trade openness.

Nevertheless, to be sure that the distinct sign is not a consequence of biased estimation the Gauss-Markov assumptions must be tested. The results have not shown any violation of the assumptions. The Hausman test, after the 2SLS estimation, once again indicated that the null hypothesis cannot be rejected, so the estimation is unbiased and efficient.

62

Table 11: The Hausman test (reduced OECD, 2010-2019)

H0: There is no endogeneity issue H1: There is endogeneity issue

P-value: 0.3400789

Source: Own Elaboration With Use Of RStudio And Based On Respective Data.

In addition to regression analysis, I made a graphical representation of the evolution of cultural effects during the studied periods since the values of its coefficients change over time despite the stable significance. Figure 11 allows a deeper comparison of strength and stability of effects of cultural dimension between 1980-1989 and 2010-2019.

Source: Own Elaboration With Use Of RStudio And Based On Respective Data.

The first difference between the studied periods is in the stability of the cultural effect. In the period 1980-1989, the strength of the cultural dimension´s impact is quite volatile compared to the period 2010-2019. The indulgence vs restraint dimension had the most variable influence on trade openness in 1980-1989. In 1984 it achieved even negative values. On the other hand, the masculinity vs femininity dimension and the long-term vs short-term orientation have the most stable impact. The long-term vs short-term orientation has in addition the strongest effect in between the cultural dimensions. This is

-1

Figure 11: Evolution of cultural influence - comparison between periods 1980-1989 and 2010-2019

63

visible also from the results of the estimation because the long-term vs short-term dimension is the unique cultural indicator that resulted significant for all the subsamples.

Interesting is also the evolution of uncertainty avoidance and the individualism vs collectivism dimension. Their effect was quite stable until the year 1985 when individualism vs collectivism started to impact the trade openness negatively and the negative influence of uncertainty avoidance was even stronger. The year 1985 is also the year when the USA pursued its first free trade agreement signed with Israel. It was also the start of the USA´s own trade negotiations, (Office of the United States Trade Representative 2021). This fact supports the assumption that individualistic societies tend to look for more lucrative trade partners and that can cause a loss of trust and consequent reduction of international trade.

The negative impact of individualism vs collectivism continues and is even stronger in the 2010-2019 period. The more intensive cultural influence is actually observable for all the dimensions. In addition, their effect is more consistent through all the years from the period. These observations indicate that the first expected effect seems to be true.

At the beginning of this chapter, I specified two possible effects that can result from the estimation. The first assumed a possible intensification of the cultural effect on trade openness through the increasing level of globalization. The second, on the other hand, assumed that the quantity of trade agreements that regulate the state´s decision-making reduces the influence of culture in general. The fact that more cultural dimensions have a

At the beginning of this chapter, I specified two possible effects that can result from the estimation. The first assumed a possible intensification of the cultural effect on trade openness through the increasing level of globalization. The second, on the other hand, assumed that the quantity of trade agreements that regulate the state´s decision-making reduces the influence of culture in general. The fact that more cultural dimensions have a

In document Hlavní práce78395_kert00.pdf, 1.4 MB Stáhnout (Stránka 61-71)