**4. Results**

The hierarchical cluster analysis performed on the 139 countries in the sample, considering the values of the 12 composite indicators of the 2019 GCI, formed five groups or clusters (Table 2).

**Table 2.** Distribution of countries by the group.


Cluster 1 is mainly made up of countries from the African region. Precisely 23 of the 26 countries belong to the region mentioned above, while only 2 Asian countries (Yemen and Pakistan) and one from the Central America and Caribbean region (Haiti) are included in the group. It is also important to note that the performance of the countries belonging to this cluster is lower than the others concerning the GCI. The cluster shows an average score of 42.48 points in the GCI, a value below the world average. Therefore, this cluster is made up of the countries with the lowest competitive levels. Slightly improving cluster 1, in terms of competitive levels, is cluster 2, comprising 16 Asian countries, 13 African countries, 10 from the rest of Europe, eight from Central America and the Caribbean, six from South America, and two belonging to the European Union. This country has an average GCI score of 56.03, slightly below the world average.

Cluster 3 shows a better picture than the two previous clusters, with an average score of 66.63 in the GCI. This cluster is made up of 13 countries from the European Union, 11 Asian countries, three from South America, and one from Africa, North America, Central America and the Caribbean, and the rest of Europe.

On the other hand, cluster 4 has the highest average score in the study, with 79.49 ICG points. This cluster is mainly made up of countries belonging to the European Union (12 countries), followed by 6 Asian countries, three from the rest of Europe, and two from North America and Oceania.

Finally, cluster 5 is made up of only two countries, South Africa and India, with an average score of 61.39 in the GCI. This cluster presents a particular behavior. It can be inferred that these countries are characterized by trade, having on average a good debt managemen<sup>t</sup> (79.08) and a relatively low inflation rate (4.92).

With the above, it is corroborated that the behavior of the countries in terms of competitiveness presents different nuances, visualized in various groupings, which show better standards than others in terms of competitiveness. Thus, according to this analysis, the Central American region presents differences in the behavior of its countries, since they were classified into two groups or clusters. Panama belonged to one group (cluster 3) along with 30 other countries, which presented similar characteristics in terms of competitiveness. On the other hand, Costa Rica, El Salvador, Guatemala, Honduras, and Nicaragua presented similar characteristics, belonging to the same cluster (cluster 2) that they shared with 50 other countries.

Table 3 provides the average values of the composite indicators of the global competitiveness of the five clusters.


**Table 3.** Average values of the composite indicators of 2019 GCI in each cluster.

Source: Authors' calculation.

The performance of cluster 4 stands out in 11 out of 12 pillars, concerning the other clusters, thus being the cluster made up of the most competitive countries analyzed. Cluster 2 (Costa Rica, El Salvador, Guatemala, Honduras, and Nicaragua) shows a medium level in the variables that make up the GCI; however, its competitive performance is surpassed by cluster 3 (Panama) in all the latent variables or pillars of the GCI.

Figure 4 shows, graphically, the comparison of the values of the composite indicators of the five groups, where the differences mentioned above can be observed.

**Figure 4.** Graphical profiles of average 2019 GCI of the groups. Source: Authors' elaboration.

In turn, the Bonferroni and Games–Howell tests [39] show the significant differences of each composite indicator of the 2019 GCI and determine the pillars that present the highest degree of heterogeneity among the defined groups. The values obtained are presented in Table 4.


**Table 4.** Bonferroni and Games–Howell test the composite indicators for the groups or clusters.

Note: Ci: Significant differences at 5% according to Bonferroni and Games–Howell tests with clusters 1 to 5. Source: Authors' calculation.

Thus, differences are observed between cluster 1 and cluster 2 in 11 of 12 pillars, being Market size the only one that does not show significant differences. In comparison, cluster 1 and cluster 3 have different behavior in the 12 pillars, as well as with cluster 4. Comparing cluster 1 with cluster 5, significant differences are observed in the macroeconomic stability and infrastructure pillars.

When doing the same analysis with cluster 2, significant differences are found in the 12 pillars, both with clusters 3 and 4, and significant differences are observed in the Macroeconomic Stability pillar with cluster 5.

While cluster 3 and cluster 4 present significant differences in the pillars of Macroeconomic Stability and Market Size. In addition, cluster 4 compared to cluster 5 presented significant differences in 4 pillars, Infrastructure, Skills, Labor market, and Business dynamism.

This result confirmed the division in competitive terms among the Central American countries. In addition, the same is visually ratified in Figure 5, where a comparison between clusters 2 and 3 is presented.

**Figure 5.** Comparative performance between clusters 2 and 3. Source: Authors' elaboration.

The descriptive statistical values of the Global Competitiveness Index 2019 for the two groups, including the seven Central American countries, have the following descriptive statistical values:


To carry out a comparative analysis exclusively among the Central American countries, Table 5 provides the average values of the composite indicators of the global competitiveness of the Central American countries.

The purpose of this is to find the variables with the greatest potential for differentiation in the region, without considering the other countries in each respective cluster, and to better approximate the study problem.

Table 6 presents the geometric means and the Mann–Whitney–Wilcoxon test for each set of countries' pillars or latent variables, where the largest statistically significant differences are found. This is to compare the cluster analysis results and evaluate the importance of the variables that most influence the grouping of countries in a given way since this test allows the binary comparison of two independent samples to determine the existence of differences between the populations.


**Table 5.** Average values of the composite indicators of 2018 and 2019 GCI in each Central American country.

Source: Authors' calculation.

**Table 6.** Prioritization of the 2019 GCI composite indicators for the two groups containing Central American countries.


Note: \* Variables prioritized according to ANOVA, Mann–Whitney–Wilcoxon, and Kruskal–Wallis tests in the complete groups. \*\* Panama point values for the 12 pillars of the 2019 GCI. \*\*\* Test conducted with all countries belonging to clusters 2 and 3. Source: Authors' calculation.

From the above, it is observed that there are statistically significant differences between the means of the two groups of Central American countries in the 12 composite indicators of the 2019 GCI, with higher mean values for all indicators associated with Cluster 2, and therefore with Panama, compared to the mean values for Cluster 3, to which Costa Rica, El Salvador, Guatemala, Honduras, and Nicaragua are associated.

In addition, it is possible to prioritize certain pillars of the 2019 GCI according to the Mann–Whitney–Wilcoxon test and the difference between the geometric means of the Central American countries, whereby the pillars with the largest significant differences are Macroeconomic stability (16.585), Infrastructure (9.752), Health (9.382), IT Adoption (9.182), and Financial system (9.081). Therefore, these can be considered key factors determining Panama's best performance in terms of competitiveness.
