The model specification

To specify the model necessary for empirical analysis, I combined the six-dimensional Hofstede´s cultural model (2010) and the econometric model for trade openness used by Tahir et al. (2018).

Hofstede´s model is used to capture the characteristics of the national cultures of studied countries. As was described in chapter 3.2. the model uses six dimensions to describe the national culture. These dimensions represent in the model six exogenous variables that allow separate analysis of the influence of specific cultural aspects on trade openness.

8 The lists of all the samples used in the analytical part are included in the Appendix.

41

Empirically, they are expressed by a value from a scale between 0-100 that indicates the relative position of a country´s culture inside of the dimensions. More specifically, the dimensions are bounded by two opposite extremes (for example hierarchical country vs egalitarian country) and the country´s received numerical value indicates to which of these extremes the country´s culture approximates. Nevertheless, not all the data fitted the 0-100 scale at first. Slovakia, Greece, and Portugal achieved in the originally collected data higher value in some dimensions. To prevent the inconsistency with the theoretical framework that requires the 0-100 scale, I used the adjusted dataset from 2015 available on Hofstede´s webpage (Hofstede 2021). The specific adjustments were:

• Slovakia: For Slovakia, the necessary changes occurred in two dimensions. The adjustment from 104 to 100 was done in the case of power distance and the change from 110 to 100 was made in the masculinity vs. femininity dimension.

• Greece: In the case of Greece, the values were exceeding the scale just for the uncertainty avoidance dimension. The value was changed from 112 to 100.

• Portugal: Portugal similarly to Greece reported a higher value in the case of uncertainty avoidance dimension. The original value of 104 was adjusted to 99.

(Hofstede 2021)

Regarding the character of the cultural data, they are secondary since they were originally collected by Hofstede through the survey studies. In addition, the data have a cross-sectional character. It means that the differentiation of cultural characteristics between countries is possible but its evolution throughout time is not. The cross-sectional character of the data is related to the cultural definition. The culture was assumed to be a set of stable national characteristics, whose negligible changes occur in decades and not in years (Hofstede 2011).⁹

The second model used for proper specification of macroeconomic determinants of trade openness is from the study of Tahir et al. (2018). I chose this study since its purpose was to investigate the macroeconomic determinants of trade openness. It is true that their analysis was oriented on the member countries of the South Asian Association for

9 However, as I discussed in the theoretical part, the question remains whether this assumption still holds despite the intensive cultural integration provided by globalization. In my point of view, the remeasurement of Hofstede´s cultural dimensions could bring interesting results for potential further studies.

42

Regional Cooperation, however, their aim was to investigate which of the generally recognized indicators of trade openness matters the most for their sample countries.

Inspired by Tahir et al. (2018) econometric model, I defined the endogenous variable as the logarithm of the trade openness index measured from the commodity market perspective. It means that the trade openness index is calculated as the proportion between the trade volume (sum of exports and imports) and the GDP. This form of expressing trade openness belongs to the most used ones and since it represents the importance of international transactions relative to domestic transactions, it is a suitable indicator of a country´s integration level into the world economy. The index is measured in percentage with values ranging between the years 1980-1989 and 2010-2019. The source of the trade openness data is The World Bank (a) (2021). Apart from the cultural explanatory variables, the model contains a set of macroeconomic exogenous variables. For each of these variables, the dataset runs from 1980 to 1989 and from 2010 to 2019. The first is a logarithm of gross capital formation as a percentage of GDP. This variable approximates the gross domestic investment and consists of outlays on an increment of fixed assets and the net changes of inventory levels. The data were also retrieved from The World Bank (b) (2021). The following determinant is the logarithm of the labor force that is expressed as a proportion of the population between 15 to 64 years to the total population. The source of this data is The World Bank (c) (2021). Next is the indicator of economic size, which in previous researches resulted to be a significant determinant of trade openness.

In the model is this parameter expressed through the logarithm of real GDP at constant national prices of 2017 measured in millions of $. The data were retrieved from Penn World Table, version 10.0 (Feenstra et al. 2015). The penultimate determinant that will be considered is the logarithm of the exchange rate measured in terms of national currency per US dollar that was obtained from the database of OECD (2021). The last macroeconomic indicator that I decided to use is the logarithm of human capital that is approximated by the mean years of schooling across all the educational levels of the population aged 25 or more. The majority of data was taken from Roser & Ortiz-Ospina (2016). Nevertheless, the last date available in this dataset was the year 2016. For that reason, the 2017-2019 data were retrieved from The World Bank (d) (2021). The inclusion of all these variables should prevent the omitted variable bias that would otherwise occur because significant determinants of trade openness would be missing.

43

The logarithmic form of all macroeconomic indicators was suggested by Tahir et al.

(2018) article and based on the linearity diagnostics I decided to use the logarithmic form in the case of cultural variables as well.

All in all, the econometric model that is used to examine the research questions of this thesis is as follows:

log(open) = β0 + β1log(pdi) + β2log(idv) + β3log(mas) + β4log(uai) + β5log(ltowvs) + β6log(ivr) + β7log(g.cap.form) + β8log(work_pop) + β9log(econ_size) + β10log(ex_rate)

+ β11log(prox_HC) + u

Table 2: Description of variables from an econometric model

The variable The description of the variable Endogenous

variable

log(open) Logarithmic form of trade openness index Exogenous cultural

variables

log(pdi) Logarithm of power distance dimension

log(idv) Logarithm of individualism vs collectivism dimension log(mas) Logarithm of masculinity vs femininity dimension log(uai) Logarithm of uncertainty avoidance dimension

log(ltowvs) Logarithm of long-term vs short-term orientation dimension log(ivr) Logarithm of indulgence vs restraint dimension

Exogenous macroeconomic variables

log(g.cap.form) Logarithm of gross capital formation (% of GDP)

log(work_pop) Logarithm of the labor force (% of 15 to 64 years old population) log(econ_size) Logarithm of economic size (GDP in constant prices 2017) log(ex_rate) Logarithm of the exchange rate (national currency per US dollar) log(prox_HC) Logarithm of human capital (average years of schooling) Source: Own Elaboration.

(1)

44