Crecimiento económico y turismo en América Latina y el Caribe: un análisis comparativo mediante técnicas de simbolización y agrupamiento de datos

Resumen:El presente trabajo estudia la dinámica del crecimiento económico y el turismo, para 22 países de América Latina y el Caribe, entre 1995-2015. Los indicadores utilizados son la tasa de crecimiento del PIB per cápita, el gasto de los turistas por habitante y los arribos internacionales por habitante. Utilizando el concepto de régimen económico, se construyen series temporales simbólicas, y mediante un análisis de clúster se detectan grupos homogéneos de países. Se identifican dos grupos de países, uno de alto y otro de bajo turismo. La evolución temporal de estos grupos no muestra una trayectoria de convergencia.Abstract: The relationship between tourism and economic growth have been a common interest in the modern literature. The Tourism Led Growth Hypothesis was one of the most important contribution finding this relation where tourism development boost economic growth increasing investment, consumption, and employment levels. However, there are too many literatures that try to prove this relationship in different countries or regions; and they find different results. This paper studies the dynamics of economic growth and tourism evolution for 22 Latin America and the Caribbean countries during the period 1995-2015.The variables representing economic growth are growth rates of per capita GDP, and for tourism, international tourist arrivals per inhabitant and international tourist expenditure per inhabitant. With these variables obtained from different data sources, we carry out a comparative analysis for tourism performance considering the international tourist expenditure per inhabitant firstly, and international tourist arrivals per inhabitant, lately. Latin America and the Caribbean is composed for many countries which are not always similar in their performance. In particular, tourism sector is one of the region’s virtues, but it performance differs considering different countries and sub-regions. To this end, we propose an exploring study of the dynamic of a set of Latin America and the Caribbean countries to observe similarities and differences. We represent the dynamical economic performance of each country by the bi-dimensional time series of our variables: tourism and economic growth measured by its corresponding indicators. By the way that both variables are measured in different units, the classical ways of obtaining distances between series are not correct. To capture the dynamic of both variables, we introduce the notion of economic regime. In which each regime represents an economic performance situation, that differs from the others. In this case, we define four regimes combining different performance in tourism and economic growth. We determinate those regimes using the mean of each variable for all countries during the considered period. So, on one hand, we can obtain two regimes with higher or lower economic growth, with low levels of tourism. On the other hand, there are two additional regimes with high levels of tourism combined with higher or lower economic growth, as well. From the conception of economic regimes, we can describe the country’s dynamic as a sequence of economic regimes where each country computes a symbolic sequence which determinate de economic regimes where each one has belonged to. By using the concept of economic regime, this paper introduces a notion of distance between the different countries, measuring how close they are in their respective regime dynamics. To this end, we consider a binary distance which is higher when countries have belonged to different regimes during the period (so their behaviors have differed); and is lower when they are more similar. Then, a Minimal Spanning Tree and a Hierarchical Tree are constructed to detect groups of countries sharing similar performance. To build the Minimal Spanning Tree we use the Kruskal algorithm, and to build the Hierarchical Tree we use the nearest neighbor single-linkage cluster algorithm. In addition, considering sub-periods (overlapping windows) of ten years, we compute a measure of the Minimal Spanning Tree’s size to evaluate the dynamic of the distances between countries. An increase of this measure is linked with more heterogeneous countries; and in the other case, it represents countries with similar dynamic performance in terms of tourism and economic growth. For the 1995–2015 period, two clusters can be clearly detected. The two main clusters can be interpreted as two groups of countries with low and high performance in the tourism, not finding major differences according to the indicator used for tourism. From this nonparametric approach, two clear groups of countries with well differentiated behaviors are detected. The low tourism group consists mainly of countries of South and Central America, while the high tourism group is mainly made up of Caribbean islands. This result is independent of the indicator used to measure tourism, either the amount of arrivals per inhabitants or expenditure per inhabitants. Uruguay and Panamá (using international tourist expenditure per inhabitant) and Panamá (using international tourist arrivals per inhabitant) do not integrate into any group because their behavior is not as high as the group with best tourism performance and is neither as low as the other group. When considering tourism arrivals as the measure of tourism, Uruguay change its behavior into a well-performed country, jointed with the group with better levels of tourism. The fact that tourism indicators are measured at per inhabitant terms, implies that countries with few population register high levels of tourism per inhabitant (as the Caribbean islands and Uruguay cases), while other countries with high levels of tourism but too much population (such as Brazil) register low rates of tourism per capita. Characterizing both groups, we observe great differences in terms of tourism, on average. However, the performance observed in the higher tourism growth seems to be more unsettle than the other group. In terms of economic growth, the average dynamic is not clear in favor of one group, but observing the economic growth’s mean, the higher tourism group shows higher levels than the other one. For the analysis of the evolution of the heterogeneity in our set of countries, we define the evolution of the global distance inside the Minimal Spanning Tree as the sum their 21 links (corresponding to the 22 countries). The evolution of the global distance for a ten-year overlapping window, shows that the distance between the two groups increases in time, while the well-performed group’s distance seems to contract over time. Hence, in the dynamic, the groups detected do not show signs of convergence. The groups dynamic behavior shows some stylized facts. First, there are two groups of countries, characterized by their performance in tourism (high and low performance), where Panamá has an outlier behavior and Uruguay is the only country that moves through the groups. Additionally, the distance between the two groups increases over time, to a greater or lesser depending on the indicator used to measure tourism (greater increase in the case of arrivals than in expenditure). Countries with high tourism performance are countries with little population where international tourism may be the principal source of tourism flows. In addition, these countries have not too much industry sector development, so tourism may represent an important quote of its growth and development.