*3.1. Data*

This research was based on the concept of competitiveness of Porter [10], the World Economic Forum [12], and Baumann, Cherry, and Chu [28], previously commented, where the focus is on the influence of the state and companies to generate conditions for competitiveness, well-being, and human development. Thus, to perform the comparative analysis of the competitiveness of Central American countries, the Global Competitiveness Index (GCI) published by the organization mentioned above for 2019 was used [12]. This index is designed to measure micro- and macroeconomic variables associated with the competitiveness of each country, constituting one of the most widely used references for measuring competitiveness [30].

The Global Competitiveness Index published in 2018 [12] incorporates a new methodology by including concepts associated with the Fourth Industrial Revolution (4IR) and evaluates a set of variables that collectively determines the level of a country's productivity, which nowadays influences the long-term improvements in living standards of millions of people around the world [12].

This version is a compilation of 98 indicators or observed variables that capture concepts that matter for productivity and long-term prosperity. These indicators are collected and grouped into 12 pillars or latent variables (Institutions, Infrastructure, ICT adoption, Macroeconomic stability, Health, Skills, Product market, Labor market, Financial system, Market size, Business dynamism, and Innovation capability). These 12 pillars are grouped into four categories or components (Enabling environment, Human capital,

Markets, and Innovation ecosystem) [12]. This structure of indicators, pillars, and categories or components is shown in Figure 2.

**Figure 2.** Composite indicators comprise the Global Competitiveness Index. Source: Framework adopted from WEF, modified by authors.

The first category or component, Enabling Environment, ensures an environment conducive to economic activity in each country, reducing transaction costs and accelerating the exchange of information, thereby increasing business confidence and productivity. The second component, Human Capital, measures how the physical, mental, and productive capabilities of individuals, as well as interpersonal skills and the ability to think critically and creatively, influence the competitiveness of countries. The third component, Market, measures the characteristics that enable the arrival of new products to a market while attracting, incentivizing, and retaining talent while providing an efficient paymen<sup>t</sup> system. Finally, the Innovation Ecosystem component is responsible for creating innovative products and services, fostering collaboration, creativity, diversity, confrontation; and the ability to turn ideas into new goods and services [12].

In addition, the calculation of the Global Competitiveness Index is based on successive aggregations of scores, from the indicator level (the most disaggregated level) to the overall GCI score (the highest level), thus in each aggregation level, there are different variables, being the score of each aggregation the arithmetic average of the variables that compose it and the overall GCI score will be the average of its 12 pillars [12]. In addition, each variable individually, before aggregation, presents values ranging from 0 to 100, with 100 being the highest value, as well as the ideal state of each variable. This score coincides with the highest level (GCI).

In turn, the Global Competitiveness Index comes from different sources. Of the 103 indicators composing the GCI, part of these is based on statistics provided by reliable external sources suppliers that adequately capture the identified concepts, is derived from external statistics from reputable organizations that collect high-quality data that will be regularly updated in the future, and have wide geographical coverage and are available for at least 75% of the economies covered by the GCI. Another part of the indicators is sourced from the Executive Opinion Survey (EOS), which for almost 40 years has been fundamental in providing critical aspects of the indicator for variables that are impossible or extremely difficult to measure statistically. The goal of the survey is to capture reality as best as possible, and business leaders are arguably the best at assessing these aspects [12,13].

For the 2019 publication, the opinion of 16,936 business executives in 41 different languages was taken between January and April 2019. In turn, the EOS comprises 78 questions divided into 10 sections, where most of the questions are answered on a scale of 1 to 7, with

7 being the highest and therefore considered the best in the world on specific aspects of the business environment of the country where the respondent operates [15].

Thus, and as previously commented, this research uses the Global Competitiveness Index data published in 2019 by the World Economic Forum [15]. Specifically, 139 records were selected and utilized corresponding to countries that did not present missing data in any pillar, including most Latin American countries and countries in the Central American region. Although the analysis focuses on the Central American countries (Costa Rica, El Salvador, Guatemala, Honduras, Nicaragua, and Panama), it was necessary to use the complete information of the countries mentioned to have a sufficient sample size to apply the analysis and classification methods to analyze the competitive differences and similarities between countries.

Table 1 presents the values of the descriptive statistics for the 12 composite indicators or pillars that make up the 2018 and 2019 GCI for the countries in the sample.


**Table 1.** Descriptive statistics of the 12 composite indicators of the 2018 and 2019 GCI.

Note: \* Non-normal distributions according to the Kolmogorov–Smirnov test. Source: Authors' calculation.

The descriptive statistical values of the Global Competitiveness Index 2019 of the 139 countries have the following descriptive statistical values: Average = 60.639, Median = 60.929, Standard Deviation = 12.427, Skewness coefficient = 0.058, Kurtosis coefficient = −0.741. The average 2019 GCI indicates a medium level of competitiveness.

From this brief characterization, it became evident that the latent variable with the highest average value corresponds to the Macroeconomic stability pillar (80.083), followed by Health (75.406), Infrastructure (65.397), and Financial system (62.441) pillars, with performance levels at the international level above the average value of the 2019 GCI (60.639).

While the pillars ICT adoption (55.330), Product market (55.262), Institutions (55.074), Market size (54.504), and Innovation capability (42.981) present not only values below the average value of the 2019 GCI but are the pillars with low performance in terms of competitiveness for the set of countries analyzed.

#### *3.2. Multivariate Analysis*

A cluster analysis [31] was performed to group countries with similar values and behaviors in terms of competitiveness and in order to differentiate from other countries with different values and behaviors [32]. A clustering proposal was made, and as previously mentioned, using the 12 pillars or latent variables of the 2018 GCI as evaluation variables, applying the inter-group linkage method through the squared Euclidean distance, different analyses were obtained (grouping in 4, 5, 6, and 7 clusters, respectively). The analysis with grouping in 5 clusters was selected because it shows better values of intragroup linkage and intergroup differences, at the same time that the natural association observed with the dendrogram shows that using 5 clusters will allow a positive interpretation of the results [33]. The dendrogram, or tree diagram, is a graphical representation of the clustering procedure, where the nodes represent the clusters. The stem lengths represent the distances at which the clusters are joined [34]. In addition, this number of clusters allows associating the countries in such a way that in each group, some observations statistically can be worked with other methodologies used later.

The ANOVA statistical technique and the Bonferroni and Games–Howell tests were applied to evaluate the existence of statistically significant differences between the five groups of countries. The ANOVA technique is a method for analyzing the equality of means of variables with normal distributions between different populations through the analysis of sample variances [35], determining the influence of some variables on others and their significance [36]. The Kruskal–Wallis non-parametric test [37] was used to analyze differences in the means of variables with non-normal distributions. Complementarily, the Bonferroni and Games–Howell tests were applied to determine, more concretely, the differences between the four groups [38].

Finally, the non-parametric Mann–Whitney–Wilcoxon test was applied to the subsample of data from the Central American countries to compare the cluster analysis results and determine whether differences exist [39]. The influence of the variables on the grouping of the countries was also evaluated, based on the geometric means of each variable, which is the average of the rate of change of a variable [40], averaging percentages, indices, or relative figures. From the results of the Mann–Whitney–Wilcoxon test, the variables with the greatest differences were identified.

The statistical analysis described was conducted using IBM SPSS Statistics (v. 26) and Microsoft Excel (v. 2019) software. Figure 3 summarizes the statistical analysis performed throughout the paper.

Thanks to the identification of significant differences in the competitiveness of Central American countries, groups with similar characteristics to the countries of the region are identified and the variables with the most significant potential for differentiation among the countries are identified, which in turn can have an impact on support when making decisions on public policy issues.

**Figure 3.** Summary statistical analysis.
