Introduction

Traditionally, the Gross Domestic Product per capita (GDP per capita) is considered the go-to variable to determine a population’s economic development and is restricted exclusively to an economic measure (Todaro and Smith, 2020). Recently, however, studies on development have begun incorporating other noneconomic factors, such as education and health. These factors, together with the economic criteria, provide a baseline for measuring a population’s development in broader terms (World Bank, 1991; Lee, 2017). In the search for economic activities that enable economic growth and improve the level of economic development, many countries have been especially interested in tourist activity since it is an economic activity that has a strong potential for job creation, the generation of foreign currency, and revenue increase. In short, it may be able to boost economic growth in host regions (Brida et al., 2020). In some cases, the development of tourism has been found to contribute to reducing inequality (Chi, 2020; Nguyen et al. (2021)) or reducing poverty (Garza-Rodriguez (2019); Folarin, Adeniyi (2019)).

In fact, what is actually important in economic policies is not only the promotion of a country’s economic growth but also, the channeling of this economic growth into improved economic development in the territory (Croes, 2012). This latter concept is much broader and it serves to satisfy the needs and demands of the resident population, improving its quality of life (Ranis et al., 2000).

In terms of the analysis of the relationship between tourism and economic growth, many studies have researched this connection. Most of them agree that a causal relationship exists between both variables, that tourism influences growth (Balaguer and Cantavella-Jordá, 2002; Brida et al., 2016), that the economic cycle influences the development of tourism (Antonakakis et al., (2015); Sokhanvar et al., 2018), and that there is a bidirectional relationship between tourism and economic growth (Bojanic and Lo, 2016; Hussain-Shahzad et al. (2017)).

Given that a relationship between tourism and economic growth has been proven in the economies of host countries and national governments, despite a lack of sufficient empirical evidence, various international organizations have been promoting tourism activity as a tool to facilitate the population’s development in those host regions that attract tourist flows to their territory (OECD, 2010; UNCTAD - United Nations Conference on Trade and Development (2011)). Such has been the case with the relationship between tourism and economic growth, with the suggestion that tourism is a tool for economic development (Cárdenas-García and Pulido-Fernández, 2019).

Many studies have already analyzed the relationship between tourism and GDP per capita, finding long-term equilibrium relationships between the expansion of tourism and economic growth, whereby a higher level of tourists received means higher levels of economic growth (Akadiri et al., 2017). As previously mentioned, the economic development of a population, in a broad sense, and in addition to the economic variables, has to be linked to additional variables with a multidimensional content (Wahyuningsih et al., 2020). In this scenario, although some studies have measured development in a broader sense (Andergassen and Candela, 2013; Banerjee et al. 2018; Bojanic and Lo, 2016; Li et al., 2018), there is a clear lack of analysis of the relationship between tourism and economic development as a multidimensional variable.

In this regard, human development, and its measurement through the Human Development Index (HDI), is a multidimensional variable related to the living conditions of the resident population (income, education, and health), which has been used on many occasions (more than level of poverty or income inequality) to measure a country’s level of development (Cárdenas-García et al., 2015; Chattopadhyay et al., 2021; Croes et al., 2021). The link between tourism and human development arises from the economic growth generated by the expansion of tourist activity. This economic growth is used to develop policies that will improve the education and health levels of the host population (Alcalá-Ordóñez and Segarra, 2023).

This article analyzes the relationship between tourism and economic growth, measuring the economic growth of the countries in the broadest possible sense, with a link to the concept of human development (Cárdenas-García et al., 2015). As a novelty, a wide set of countries is used for this analysis. This overcomes the limitations of prior works that analyzed the relationship between tourism and human development using small country samples (Chattopadhyay et al., 2021).

Although distinct works have already analyzed the relationship between tourism and economic development, they tended to focus on the application of econometric tests to determine the type of causal relationship existing between these variables (Alcalá-Ordóñez and Segarra, 2023). This work takes a distinct approach, analyzing the qualitative dynamic behavior arising between tourism and human development. Different country groups are identified that have similar behavior within the group and, simultaneously, with differences as compared to the other groups. Thus it is possible to verify the relationship existing between tourism and human development in each of these country groups, to determine if a higher level of tourism specialization is linked to a higher level of human development.

This approach does not attempt to determine if a causal relationship exists by which tourism precedes the level of development. Rather, this approach of grouping countries aims to determine if, at similar levels of development, the country groups with a higher level of tourism specialization display higher levels of human development. This would suggest that tourism activity is an economic activity that promotes human development to a greater extent than other economic activities.

In this context, this study analyzes the dynamic relationship between tourism and economic development, considering development as a multidimensional variable. It uses a data panel consisting of 123 countries for the period between 1995–2019 and considers the diversity of countries in terms of tourism development and their economic development dynamics. To perform this dynamic analysis, the concept of economic regime is introduced (Brida, 2008; Cristelli et al., 2015, Brida et al., 2020), and symbolic time series are used (Risso (2018)).

This article contributes to the empirical literature examining the relationship between tourism and economic development. It analyzes the qualitative dynamic behavior of the countries without considering any particular model. Therefore, this analysis enables the identification of groups of countries with similar dynamics, for which economic models of the same type can be identified. The results of this study indicate that there are different groups of countries displaying similar dynamic behavior in terms of both tourism and development. These groups are characterized by their level of tourism specialization and economic development. Therefore, it is interesting to note the heterogeneity existing in the relationship between tourism and development, as well as the consequences that this situation has for both the empirical analysis and the political implications.

The rest of the document is organized as follows: the following section reviews the literature on the subject under study, section “Data” presents the data used, section “Methodology” details the methodology applied, section “Results” presents the results obtained, section “Discussion” includes a discussion of the paper, and, finally, section “Conclusions and policy implications” outlines the final conclusions and policy implications of the work.

Literature review

Economic growth versus economic development

Traditionally, studies on development have focused on economic growth and have been based on the premise that the efficient allocation of resources maximizes growth and that the expansion of growth and consumption is a measure of population welfare (Easterly, 2002). However, the emergence of new studies at the end of the last century, beginning with the works by Sen (1990, 1999), resulted in a change of focus for studies on development. They moved from an exclusive view of development linked to economic growth to the inclusion of new factors that connect it to the population’s living conditions (Croes et al., 2018).

Economic growth and development are distinct concepts that do not need to be linked. In other words, increased economic growth does not necessarily imply improved economic development (Croes et al., 2021). However, it is also true that economic growth, and the revenue generated, can be used to improve a population’s living conditions through better health care, infrastructures, and education (Banerjee et al., 2018; Cárdenas-García and Pulido-Fernández, 2019).

In this regard, the first studies to analyze the relationship between tourist activity and the economies of host countries focused exclusively on the relationship between tourism and economic growth, using a traditional view of development that is linked to economic variables.

Tourism and economic growth

Numerous studies have analyzed the relationship between tourism and economic growth. Therefore, it is a highly relevant research area in the economic analysis of tourist activity, with three streams of perfectly defined results in which these works may be grouped (Alcalá-Ordóñez et al., 2023; Brida et al., 2016).

Firstly, different studies have determined that tourism development drives economic growth, identified under the tourism-led economic growth hypothesis. Both the first study to analyze this causal relationship (Balaguer and Cantavella-Jorda, 2002), as well as the later studies (Brida et al., 2016; Castro-Nuño et al., 2013; Lin et al., 2019, Pérez-Rodríguez et al., 2021; Ridderstaat et al., 2016), have confirmed the existence of this relationship.

Secondly, other studies determined that the evolution of the economic cycle has an influence on the development of tourism, identified under the economic-driven tourism growth. These studies indicate that those economies with a greater level of investment, stability in the price level, or lower level of unemployment determine the development of tourism (Antonakakis et al. (2015); Rivera, 2017; Sokhanvar et al., 2018; Tang, Tan (2018)).

Finally, a third wave of studies determined that the relationship between the development of tourism and economic growth has a bidirectional character. These studies note that the relationship between both variables is a causal bidirectional relationship (Antonakakis et al., 2019; Bojanic and Lo, 2016; Chingarande and Saayman, 2018; Hussain-Shahzad et al. (2017); Ridderstaat et al., 2013).

Human Development as a measure of development

Since the end of the last century, the scientific literature has shown that the concept of development cannot be linked exclusively to variables of economic content. Instead, development should be considered along with other non-economic factors that are related to the population’s living conditions. Therefore, it is a multidimensional concept (Alcalá-Ordóñez and Segarra, 2023).

When measuring development using a multidimensional perspective, this concept is often linked to human development (Cárdenas et al., 2015; Chattopadhyay et al., 2021). In this regard, the HDI is a multidimensional indicator that, in addition to considering variables of economic content, in this case per capita income, also incorporates other non-economic factors, specifically, life expectancy and educational level of the population (United Nations Development Program, 2022).

The HDI offers some major advantages as a measure of development over other indicators, providing a more complete vision of society’s progress and focusing not only on economic factors but also on factors related to the population’s living conditions. This makes it possible to identify inequalities that need to be addressed to promote more equitable and sustainable development (Sharma et al., 2020; Tan et al., 2019). Moreover, since it was created by the United Nations Development Program for a large group of countries, it permits homogenous comparison-making between a broad base of countries at a global level (Cárdenas-García and Pulido-Fernández, 2019).

Tourism and human development

The expansion of tourism activity can influence the level of human development (Croes et al., 2021). The common link between these two variables is the economic impact generated by the expansion of tourist activity since this is a linked process, whereby a higher level of tourists results in an increase in income generated and thus, a higher level of economic growth (Brida et al., 2016). Countries can take advantage of this higher level of economic growth to develop specific policies aimed at improving the living conditions of the host population, thereby improving human development (Eluwole et al., 2022).

This link between tourism and human development has also been highlighted by the United Nations Tourism in its Millennium Development Goals of 2000, which declared that factors such as health and education are very important in economic development. It was suggested that tourism may improve human development given that it has an influence on these non-economic factors (UN Tourism, 2006).

The triple component of the HDI, the most frequently used indicator to measure economic development, has been considered in most of the studies analyzing the relationship between tourism and economic development (Alcalá-Ordóñez and Segarra, 2023).

Distinct studies have attempted to determine whether tourism is a tool for economic growth in host countries, although most of the studies have exclusively used economic content to measure the concept of development (Wahyuningsih et al., 2020). Therefore, there is a major lack of empirical studies that consider whether tourism influences development and that do so while considering development to be a multidimensional variable encompassing other factors (beyond those associated with the economy).

Some of these studies have outlined that the expansion of tourism has led to an increase in the level of development for host countries. This suggests that tourism has a positive unidirectional relationship with the living conditions of the population (Meyer and Meyer, 2016). Fahimi et al. (2018), examining microstates, found evidence supporting the idea that the expansion of tourism leads to an improvement in human capital. Other studies have also noted that this causal relationship between tourism and development exists, but only in developed countries (Banerjee et al., 2018; Bojanic and Lo, 2016). Some studies have suggested that only the least developed countries have benefited from the tourism industry in terms of increased economic development ratios (Cárdenas-García et al., 2015).

However, although it has been indicated that tourism influences economic growth, some authors have noted that tourism does not have an influence on the development of host countries (Rivera, 2017), or simply, that the expansion of this activity does not have any effect on human development (Croes et al., 2021).

As an intermediate position between these two schools of thought, some works have suggested that tourism has a positive influence on the development of the resident population, but this causal relationship is only found when certain factors exist in the host countries, such as infrastructure, environment, technology, and human capital (Andergassen and Candela, 2013; Cárdenas-García and Pulido-Fernández, 2019; Li et al., 2018).

Along these same lines, in a study using panel data from 133 countries, Chattopadhyay et al. (2021) determined that, although no global relationship exists between tourism and human development for all countries, the specific characteristics of each country (level of growth, degree of urbanization, or commercial openness) are determinants for tourism to improve human development levels.

Finally, other studies in the scientific literature have looked to determine whether the relationship between tourism and development is a bidirectional causal relationship, with papers affirming the existence of this relationship between tourism and development (Pulido-Fernández and Cárdenas-García, 2021).

Therefore, when examining the few studies that have analyzed the relationship between tourism and development, it may be concluded that contradictory and biased results exist. This may be due to the characteristics of the samples chosen, the variables used, and the methodology employed. Currently, there is no defined school of thought in the scientific literature with regard to the ability of tourism to improve living conditions for the resident population. This contrasts with the conclusions drawn regarding the relationship between tourism and economic growth.

This gap in the scientific literature provides an opportunity for new empirical studies that can analyze the relationship between tourism and development.

Data

In this study, data from different sources of information were used with the objective of analyzing the relationship between tourism and economic development, in accordance with the methodology proposed in the following section. The data used in the present study are available for a total of 123 countries, covering all geographical areas worldwide. The specific data for these countries are as follows, including a web link to the availability of the data to provide greater transparency:

  • Tourist activity. The number of international tourists received was used as a variable for measuring tourist activity. For those countries for which this data was unavailable, the number of international visitors received was used, based on annual information provided by the United Nations Tourism between 1995 and the present (UN Tourism, 2022).

    Data on international tourists received at a country level are available at https://www.unwto.org/tourism-data/global-and-regional-tourism-performance

  • Economic development. The HDI, developed by the United Nations Development Program and available annually from 1990 to the present day, was used as a variable for measuring economic development (United Nations Development Program, 2022).

    Data from the HDI for each country are available at https://hdr.undp.org/data-center/human-development-index#/indicies/HDI

  • Total population. The de facto population was used as a measurement variable and counts all residents regardless of their legal status or citizenship. This information was provided by the World Bank and is available from 1960 to the present day, on an annual basis (World Bank, 2022).

Data on the population of the distinct countries are available and accessible at https://data.worldbank.org/indicator/sp.pop.totl.

Based on the data indicated above, the initial variables are transformed, specifically, in the case of tourism, through the use of the relativized per capita variable. A descriptive summary of the variables used in the analysis is presented in Table 1. Finally, two variables have been used to analyze the relationship between tourism and economic development:

  • International tourists per inhabitant received in the country (number of international tourists / total population of the country), as a measure of tourism specialization. The unit of this variable is established at a relative value, by dividing the number of tourists by the population.

  • HDI of the country, as a measure of economic development. The unit of this variable is established at a relative value for each country, which, in all cases, is between 0 (lowest level of human development) and 1 (highest level of human development).

Table 1 Descriptive statistics of the set of 123 countries for the period 1995–2019.
Full size table

Regarding the tourist sector, the measurement of tourism is a subject that has generated great interest, and, on many occasions, the selection of different indicators leads to different results (Song and Wu, 2021). As a result, the results of the empirical analysis may be affected by the indicators used to represent the tourist demand (Fonseca and Sanchez-Rivero, 2020), with there being important differences between studies with respect to the tourism indicator. According to Rosselló-Nadal, He (2020), tourist arrivals or tourism expenditure are frequently used to measure tourist demand; however, when looking at the literature, differences in the results are found depending on the indicator considered. Indeed, in their study, which looked at 191 countries between 1998–2016, the authors found evidence that estimates may differ depending on the indicator used for the tourism demand of a destination (international tourist arrivals, or international tourist expenditure in this case). Other studies use indicators that do not measure the degree of tourist activity of a destination, as is the case for the number of tourist arrivals, the expenses, or the revenues. Instead, they consider an indicator that measures the degree of specialization that an economy has in tourism, for example, international tourist arrivals in per capita terms or expenditure or income as a percentage of GDP or exports. This work uses the number of international tourist arrivals, in relation to the population, and thus obtains the degree of tourism specialization of a destination (such as Dritsakis, 2012; Tang and Abosedra, 2016).

With regard to the measurement of economic development, the arrival of the HDI has resulted in a notable improvement in terms of GDP per capita, which is traditionally used to measure the progress of a country linked only to economic aspects (Lind, 2019). In fact, the HDI includes other noneconomic factors as it measures three key dimensions of development: a long and healthy life, being well-informed, and having a decent standard of living. This is why this index was created from the geometric mean of the normalized indices for each of the three dimensions indicated: (i) health: life expectancy at birth; (ii) education: years of schooling for adults and expected years of schooling for children; and (iii) standard of living: Gross National Income per capita (United Nations Development Program, 2022). Therefore, since the emergence of this index, there have been increasingly more studies that have incorporated HDI as a measurement of economic development. This variable has been shown to represent development better than other variables that are based exclusively on economic factors (Anand and Sen, 2000; Jalil and Kamaruddin, 2018; Ngoo and Tey, 2019; Ogwang and Abdou, 2003; Sajith and Malathi, 2020).

The time scale considered in this study covers the period between 1995–2019, in order to perform the broadest possible time analysis. On the one hand, there is an initial time restriction in terms of the data, given that the first data available on international tourist arrivals, provided by the United Nations Tourism, refer to the 1995 fiscal year. On the other hand, the data for the 2019 fiscal year are the latest in the time series analyzed. Therefore, the consequences of the COVID-19 crisis, which may have had a different impact at the country level, as well as the level of recovery in international tourist arrivals, do not affect the results of this work.

Methodology

In this work, an analysis is carried out involving the dynamics of two variables: tourism specialization and the HDI. Each of the countries considered in the analysis is represented by a two-dimensional time series of coordinates of these two variables.

In order to compare these dynamics and thereby find homogenous country groups sharing similar dynamics, it was first necessary to introduce a metric permitting this comparison. A fundamental issue in this analysis is that the units of measurement used for each variable are different and the relationship between them is unknown since tourism is measured in the number of tourists per inhabitant while the HDI is an index that varies between 0 and 1. Therefore, the frequently used Euclidean metrics are not valid for this analysis. For this reason, in this study, the problem was analyzed within the framework of complex systems by introducing the concept of “regimes”.

In economic literature, the term “regime” is used to characterize a type of behavior exhibited by one economy, which can be qualitatively distinguished from the “regime” that characterizes another economy. In this way, one regime is distinguished and differentiated from another, so that the economy as a whole may be considered a system of multiple regimes. Intuitively, an “economic regime” may be considered a set of rules governing the economy as a system and determining certain qualitative behaviors (Boehm and Punzo, 2001).

Regime changes, on the other hand, are associated with qualitative changes in the dynamics of an economy. Identifying and characterizing these regimes is a complex issue. For example, when working with mathematical models, a commonly used criterion is through Markov partitions (see Adler, 1998). Another widely used criterion when working with data is the division of the state space using various statistical indicators, such as the mean, median, etc. (see Brida and Punzo, 2003).

Firstly, a distance between countries was calculated to compare their trajectories; secondly, a symbolic time series analysis was used and the concept of “regime” was incorporated; as a result, the original two-dimensional series was transformed into a one-dimensional symbolic series. Then, a metric allowing for the comparison of the dynamic trajectories of the different countries was introduced; finally, a cluster analysis was performed to group the countries based on their dynamics.

The symbolic time series analysis methodology, still quite undeveloped in the field of economics, has been used in some previous works, such as that by Brida et al. (2020) that analyzes the relationship between tourism and economic growth. All analyses have been performed using RStudio software.

Time series symbolization

To identify the qualitatively relevant characteristics, the concepts of regime and regime dynamics were introduced (Brida, 2008; Brida et al., 2020). Each regime had its own economic performance model that made it qualitatively different from the rest. The partitioning of the space of tourism states and the development was established by means of annual averages of international arrivals per capita (x) and the HDI (y). The space was divided into four regions, which were determined by the annual averages of tourism and economic development, \({\bar{x}}_{t}\) and \({\bar{y}}_{t}\) respectively, with \(t=1,\ldots ,25\). Using this partitioning of the states space into regimes, two types of dynamics are distinguished: one within each of the regimes and one of change between regimes. While the dynamic observed in each regime determines a performance model that differs from the models that act in the others, the dynamics of change from one region to another indicate where an economy is at each temporal moment. This dynamic describes performance in terms of tourism specialization and economic development in a qualitative way.

A change of regime of course signals some qualitative transformation. To explore these qualitative changes for every country, let us substitute a bi-dimensional time series \(\left\{\left({x}_{1},{y}_{1}\right),\,\left({x}_{2},{y}_{2}\right),\,\ldots ,\,\left({x}_{{\rm{T}}},{y}_{{\rm{T}}}\right)\right\}\), by a sequence of symbols: \(s=\left\{{s}_{1},{s}_{2},\ldots ,{s}_{T}\right\}\), such that \({s}_{t}=j\) if and only if \(\left({x}_{t},{y}_{t}\right)\) belongs to a selected state space region,\(\,{R}_{j}\). It is defined four regions in the following way:

$${R}_{1}=\left\{(x,y):{x}_{t}\ge {\bar{x}}_{t},{y}_{t}\ge {\bar{y}}_{t})\right\}$$
(1)

Regime 1: countries with above-average HDI and tourism specialization. In this regime, the most developed economies specializing in tourism are expected to be found. The majority of European countries are expected to be found in this regime; countries in other regions with a high level of tourism specialization could also be included.

$${R}_{2}=\left\{\left(x,y\right):{x}_{t}\le {\bar{x}}_{t},{y}_{t}\ge {\bar{y}}_{t})\right\}$$
(2)

Regime 2: countries with high HDI and low tourism specialization. In this regime, the most developed economies, but in which tourism activity has a less important weight in their economic base, are expected to be found. Some large countries such as the US and Germany are expected to be found in this regime. Other countries may also be found here even if they do not present similar levels of development as European countries, for example, they have higher levels in relative terms (above the sample average).

$${R}_{3}=\left\{\left(x,y\right):{x}_{t}\le {\bar{x}}_{t},{y}_{t}\le {\bar{y}}_{t})\right\}$$
(3)

Regime 3: countries with low HDI and low tourism specialization. In this regime, economies with a lower level of development and where tourism activity is not relevant to their economic activity, are expected to be found. Countries such as China, other Asian countries, countries on the African continent, and countries in South America are expected to be included in this regime.

$${R}_{4}=\left\{\left(x,y\right):{x}_{t}\ge {\bar{x}}_{t},{y}_{t}\le {\bar{y}}_{t})\right\}$$
(4)

Regime 4: countries with low HDI and high tourism specialization. Countries with a lower level of development and a high level of tourism specialization, such as Caribbean countries and some island countries, are expected to be found in this regime.

Clustering

Once the one-dimensional symbolic series is obtained, a metric is introduced that allows comparing the dynamics of the countries, and which in turn allows for obtaining homogeneous groups. Given the symbolic sequences \({\left\{{s}_{{it}}\right\}}_{t=1}^{t=T}\) and \({\{{s}_{{jt}}\}}_{t=1}^{t=T}\) the distance between two countries, i and j is given by.

$$\begin{array}{l}d\left({s}_{i},{s}_{j}\right)=\mathop{\sum }\limits_{t=1}^{T}f\left({s}_{{it}}{,s}_{{jt}}\right),{where}\, f\left({s}_{i}{,s}_{j}\right)=\left\{\begin{array}{l}0\;{if}{s}_{{it}}={s}_{{jt}}\\ 1\,{if}{s}_{{it}}\ne {s}_{{jt}}\end{array}\right.\\\qquad\quad\;\;\;{with\; t}=1,\,\ldots\, ,T\end{array}$$
(5)

Intuitively, the distance between two countries measures the number of years of regime non-coincidence during the period. If the distance between two countries is zero, the countries have been in the same regime for the entire period. On the contrary, if the distance between two countries is T, the countries have not coincided for any time during the analyzed period. If the distance between two countries is α, it means that they have not coincided for α years during the period. In other words, they have coincided for T-α years.

Using the defined distance, the hierarchical tree was created using the nearest neighbor cluster analysis method (Mantegna, 1999; Mantegna and Stanley, 2000). Using the algorithm by Kruskal (1956), the minimum spanning tree (MST) was created. This tree was created progressively, joining all the countries from the sample using a minimum distance. According to this algorithm, in the first step, the two countries whose series had the shortest distances were connected. In the second step, the countries with the second shortest distance were connected. This pattern continued until all countries were connected in one tree.

Results

Symbolic time series analysis

Figure 1 shows the point cloud corresponding to 2019, with the respective averages of each variable. Each point represents a country in this year with its coordinates (Tourism, HDI). As is expected, the points are distributed in the four regions, showing that qualitatively the countries perform differently. A clustering in the second and third quadrants can be observed, indicating a clustering in the sections with a low level of tourism specialization, and, in turn, there are not many countries in the fourth quadrant. In other words, few countries have been considered to have a high level of tourism specialization but low levels of development, in the last year (Belize, Fiji, Jamaica, Saint Lucia, the Maldives, and Samoa).

Fig. 1
figure 1

Cloud of points of the 123 countries for the year 2019.

Full size image

Table 2 shows the percentage of time spent by each of the 123 countries analyzed in each of the previously defined regimes, showing that the large majority of the countries (80 countries) remained in the same regime for the entire period or, at least, for three-quarters of the period analyzed in the same regime (16 countries). In this regard, using the symbolization of the series, 4 clear groups were identified, made up of countries that remained in the same regime for the entire period:

  • Group 1: made up of countries that are in regime 1 for the entire period (high level of tourism specialization and high level of development): Austria, Bahamas, Barbados, Switzerland, Cyprus, Spain, France, Greece, Hong Kong, Ireland, Iceland, Italy, Luxembourg, Malta, Netherlands, Norway, Portugal, and Singapore.

  • Group 2: made up of countries that are in regime 2 for the entire period (low level of tourism specialization and high level of development): Germany, Argentina, Australia, Chile, South Korea, Costa Rica, Cuba, the United States, Russia, Iran, Japan, Kazakhstan, Kuwait, Mexico, Panama, United Kingdom, Romania, Trinidad and Tobago, and Ukraine.

  • Group 3: made up of countries that are in regime 3 for the entire period (low level of tourism specialization and low level of development): Azerbaijan, Benin, Bangladesh, Bolivia, Central African Republic, China, Congo, Algeria, Egypt, Gambia, Guatemala, Guyana, Honduras, Haiti, Indonesia, India, Cambodia, Laos, Lesotho, Morocco, Mali, Myanmar, Mongolia, Malawi, Namibia, Niger, Nicaragua, Nepal, Philippines, Papua New Guinea, Paraguay, Sudan, Sierra Leone, El Salvador, Togo, Tuvalu, Tanzania, Uganda, Vietnam, Zambia, and Zimbabwe.

  • Finally, Group 4, made up of Belize and the Maldives, which are in regime 4 for the entire period (high level of tourism specialization and low level of development):

Table 2 Percentage of time in each regime, of the set of 123 countries for the period 1995–2019.
Full size table

It is worth noting that according to the results obtained, regime changes can be difficult to observe. This could be a result of the fact that a regime change implies a structural change in the economy and in such a period as the one analyzed in this study (25 years), the observation of a structural change may be circumstantial in nature. In other words, the timing of structural changes seems to be slower than the tick of the chosen clock; in this case, an annual tick.

Within the group of countries that always remain in regime 1, two groups of countries can be identified. One of the groups is that in which tourism is an essential sector for the economy (like in the case of the Bahamas or Barbados, which have tourism contribution rates to GDP of above 25%), and in which tourism seems to have an influence in the high level of development. The other group is that in which, while tourism is not necessarily an essential sector for the economy, due to the existence of other economic activities, it is an important sector for development (such as Spain or Portugal, with tourism contribution rates to GDP of above 10%).

Within the group of countries that always remain in regime 2, there are fundamentally countries in which tourism has a marginal weight in relation to the level of population (like in the case of Germany, the US, and Japan), due to the lack of or little exploitation of the country’s tourism resources, which would result in development seeming to be related to other economic activities.

Within the group of countries that always remain in regime 3, there is a large group consisting of 41 countries (a third of the sample) that seem to be in a poverty trap, due to the low level of development and low level of tourism specialization. This is in such a way that the low level of development hinders the expansion of tourism activity, and, in turn, this lack of tourism development makes it difficult to increase the levels of development.

Finally, within the group of countries that always remain in regime 4, there are only two countries found, which are characterized by a high level of tourism specialization but have not transformed this into an improvement in development, possibly due to the existence of certain factors that hinder this relationship.

Therefore, the first issue to note is the little mobility that countries have in terms of their classification between the different regimes, given that 80 countries (two-thirds of the sample) remained in the same regime during the 25 years analyzed, which seems to show that the variables are somewhat stable, and thus justifies the fact that no major changes were observed during the period analyzed. This behavior reveals that the homogeneity in the tourism and development dynamic is the rule and not the exception.

In fact, only 27 countries, out of the 123 countries analyzed, are in a different regime for at least a quarter of the period: Albania, Armenia, Bulgaria, Brazil, Botswana, Canada, Colombia, Slovakia, Eswatini, Finland, Fiji, Hungary, Jamaica, Jordan, Lithuania, Latvia, Moldova, Malaysia, New Zealand, Peru, Saint Lucia, Sweden, Thailand, Tonga, Tunisia, Turkey, and Samoa.

In this regard, Fig. 2 shows the time evolution of the symbolic series for some selected countries. As can be noted, there are some countries, like Brazil, that always have a low level of tourism specialization and alternate between periods of high and low economic development, with it seeming as though there is consolidation as being a low HDI country in recent years (until 2002, Brazil had an above average level of development but, after it was hit by a crisis, the country moved to the low development regime. Then, in 2013, it managed to return to the high HDI regime, albeit temporarily as in 2016, in the midst of a political and economic crisis, it returned to the low development regime, where it currently remains). This is similar to what happened in Fiji, insofar as it was almost always specialized in tourism and alternated HDI, consolidating itself in Regime 4 of the low HDI. As such, it seems as though certain countries define their behavior according to the degree of tourism specialization; in this case, not particularly specialized countries.

Fig. 2: Example of the trajectory of some of the countries that change regimes in the 1995–2019 period.
figure 2

Top panel: Brazil (left) and Fiji (right). Bottom panel: Latvia (left) and Eswatini (right).

Full size image

However, the behavior of Latvia or Eswatini seems to be determined by HDI and not by tourism specialization. As to be expected, Latvia remained always in regimes 1 and 2 with a high HDI while Eswatini remained in regimes 3 and 4 with a low HDI. In both cases, they alternated periods of high and low specialization in tourism.

Grouping homogeneous countries

In the case analyzed, there are many countries with zero distance. These are the countries that have the same symbolic representation, that is, the regimes dynamics are coincidental given that these countries always remain in the same regime. Therefore, there are three groups that start to form with countries that have zero distance (countries that are always placed in regimes 1, 2, and 3), and a small group, formed by Belize and the Maldives, which are the only countries that remained in regime 4 for the entire period analyzed. According to this algorithm, 6 groups were obtained, while some countries were not included in any of the groups as they were considered to be “outliers”.

Specifically, there was a graph with 123 nodes corresponding to each country and 122 links; however, given that there were several countries with the same dynamic (the distance between these countries is zero), each of these groups is represented in a single node; that is, the countries that always remained in regime 1 were considered together as one single node, with the same happening for the remaining three groups of countries with identical dynamics (groups 2, 3, and 4). Therefore, in this case, there is a node representing 18 countries from group A and another node (both pink) that represents multiple countries; the Czech Republic, Estonia, Croatia, Mauritius, and Slovenia, which all share the same dynamic (they always remain in regime 1, except in 1995). There is a node representing 19 countries from group B (light blue), another node representing 41 countries from group C (green), and a final node representing Belize and Maldives in group D. In this way, 80 countries are represented in four nodes. To complete the tree, 38 other nodes, each corresponding to a country, were established. Using Kruskal’s algorithm (1956), the MST is built, in which all nodes are connected in a single tree from the minimum distances. In this way, a tree is created having links that connect the nodes to represent the minimum distances between them (a longer arrow indicates a longer distance).

Figure 3 shows the MST. It is worth noting the central position that these multiple nodes have within the groups, that is, nodes that represent a group of countries with the same dynamics. The structure of the MST seems to be almost linear; moreover, while group C (green) is the most numerous, it is also the most compact of the large groups.

Fig. 3: Minimal spanning tree of the set of 123 countries for the period 1995–2019.
figure 3

(Nodes: Pink group A/Light blue group B/Green group C/Yellow group D/Orange group E/Blue group F/Red Outliers. Distances according to arrow color: black 1/red 2/light blue 3/green 4/blue 5/orange 6/pink 7/gray 8/violet 9).

Full size image

Figure 4 shows the geographic distribution of the different groups. There are 6 groups (3 large and 3 small), while some countries are not included in any of these groups, as they are considered to be “outliers”:

  • Group A: Albania, Austria, Belgium, Bulgaria, Bahamas, Barbados, Switzerland, Cyprus, Czech Republic, Denmark, Spain, Estonia, Finland, France, Greece, Hong Kong, Croatia, Hungary, Iceland, Ireland, Italy, Lithuania, Luxembourg, Latvia, Malta, Mauritius, Malaysia, Netherlands, Norway, New Zealand, Portugal, Qatar, Singapore, Slovakia, Slovenia, and Uruguay. This group is made up of countries that predominantly remained in regime 1, that is, in general, these are countries with a high tourism specialization and high economic development.

  • Group B: Argentina, Australia, Brazil, Chile, Colombia, Costa Rica, Cuba, Germany, Ecuador, United Kingdom, Iran, Israel, Jordan, Japan, Kazakhstan, South Korea, Kuwait, Sri Lanka, Mexico, North Macedonia, Panama, Peru, Poland, Romania, Russia, Tonga, Trinidad and Tobago, Ukraine, United States. This group is made up of countries that predominantly remained in regime 2, that is, in general, these are countries with a low tourism specialization and high economic development.

  • Group C: Azerbaijan, Benin, Bangladesh, Bolivia, Central African Republic, China, Congo, Dominican Republic, Algeria, Egypt, Gambia, Guatemala, Guyana, Honduras, Haiti, Indonesia, India, Cambodia, Laos, Lesotho, Mali, Morocco, Myanmar, Mongolia, Malawi, Namibia, Niger, Nicaragua, Nepal, Philippines, Papua New Guinea, Paraguay, Sudan, Sierra Leone, El Salvador, Togo, Tuvalu, Tanzania, Uganda, Vietnam, South Africa, Zambia, and Zimbabwe. This group is made up of countries that remained the majority of the time in regime 3, that is, in general, these are countries with a low tourism specialization and low economic development. With the exception of the Dominican Republic and South Africa (96% and 92%, respectively), all countries remained in regime 3 for the entire period.

  • Group D: Belize and the Maldives. This group is made up of the two countries that always remained in regime 4, that is, in general, these are countries with a high tourism specialization and low economic development.

  • Group E: Armenia, Moldova, Thailand, and Turkey. This group has the particular characteristic of having low tourism specialization throughout the period but alternating between a high level of development (regime 2) and a low level of development (regime 3).

  • Group F: Botswana, Jamaica, and Tunisia. This group is made up of countries that fundamentally remained in regime 4, that is, these are countries with a high tourism specialization and low economic development, however, unlike group D, they moved during the period analyzed through other regimes.

  • Outliers: Canada, Fiji, Saint Lucia, Sweden, Eswatini, and Samoa. These countries presented different dynamics and were not integrated into any of the previously-defined groups.

Fig. 4: Geographical distribution of the different groups.
figure 4

(Note: Pink: group A/Light blue: group B/Green: group C/Yellow: group D/Orange: group E/Blue: group F/Red: Outliers).

Full size image

As can be seen, group A, which consists of countries with a high tourism specialization and high economic development, is basically made up of European countries, some Asian countries, and Uruguay (the only country in the Americas to be part of this group).

The countries in group B, that is, those countries with a good level of economic development, but a low specialization in the sector, are more geographically dispersed. This group consists of some European countries (in particular, Eastern European countries), a large part of Latin America and the Caribbean, as well as the US, Australia, and some Asian countries.

Group C, that is, those countries with a low tourism specialization and low economic development, consists of the vast majority of African countries, as well as a significant number of Asian countries, in addition to Bolivia and Paraguay in Latin America, as well as some countries in Central America.

The countries in Group D, that is, those countries that had a high tourism specialization but a low level of economic development throughout the period analyzed, as well as those in Group F, which were also in this regime for most of the period, do not have a uniform geographic pattern, since they are located on different continents.

Finally, the countries in Group E, that is, those countries with a low tourism specialization and alternating levels of economic development, are also geographically dispersed between Europe and Asia.

As can be seen in Table 3 both Group A and Group B are made up of countries with a high level of development; however, the countries in Group A, which also have a high level of tourism specialization, on average, have a significantly higher level of development than the countries in Group B, where the level of tourism specialization is low. These results appear to show that in terms of those countries specialized in tourism (Group A), the link with development is higher than for those countries that have achieved high levels of development due to the development of other economic activities.

Table 3 Descriptive data of the six groups obtained for the 1995–2019 period.
Full size table

Similar results can be found when comparing the data from Group C (countries with a low level of tourism specialization and low level of economic development) with the data from Groups D and F (countries with a high level of tourism specialization and low level of economic development). This is because, despite the level of development being low in all the countries, in the Group D and F countries, the level of development is significantly higher than in Group C countries. This appears to show that for those countries specialized in tourism (Groups D and F), the link with development is greater than for those countries that rely on other sectors as the basis of their economy.

Discussion

Tourism’s relevance lies not only in its contribution to economic growth but also in the fact that the improved economic growth generated by the expansion of tourism activity may translate into improved living conditions for the host population. Due to this chained process, many countries have opted for this economic activity with the aim of improving income, education, and health. In short, they hope to increase their levels of human development.

Although distinct works have analyzed the relationship between tourism and human development by applying causality tests to determine the type of relationship between these variables, this study adopts a different approach. It analyzes the qualitative dynamic behavior between tourism and human development, to identify clusters of countries that display similar behavior with regard to this relationship.

Firstly, it is necessary to note the little movement there is of the countries between the different regimes, which indicates great stability, given that 80 countries (two-thirds of the sample) remained in the same regime throughout the entire period analyzed (1995–2019). These results regarding the stability of the countries in the different regimes differ significantly from the results obtained in other studies that have used the same technique for the analysis of the dynamic relationship between variables (Brida et al., 2020). This is because even when there is a movement of the countries between regimes, this happens, at most, between two or three regimes (Jordan and Samoa are the only exceptions, passing through all four regimes).

Furthermore, the results appear to show that groups of countries with a higher level of tourism specialization have higher levels of human development. Therefore, tourism is configured as an effective tool to improve development levels, as previously stated in works such as that of Cárdenas-García et al. (2015) conducting a joint analysis with data from 144 countries or Bojanic and Lo (2016), whose global analysis referred to a sample of 187 countries.

Specifically, these results are found both in the group of countries with the highest level of development, (countries of Group A versus the countries of Group B), as previously revealed in works such as that of Meyer and Meyer (2016) analyzing South Africa and that of Tan et al. (2019) analyzing Malaysia. These results were also found in the case of countries with a lower level of development (countries of Group D and F as compared to the countries of Group C), as previously suggested by works, such as that of Sharma et al. (2020) examining India or Croes (2012) analyzing Nicaragua.

However, despite these majority results, countries have been identified that, despite having an important tourism specialization (Belize, Botswana, Jamaica, Maldives, and Tunisia), had a low level of human development. This has not allowed for the high level of tourism specialization to become a tool to improve the living conditions of the population in these countries.

This exception may be due to the link between tourism and human development, which, in addition to being affected by the level of tourism specialization, also depends on the destination’s characteristics. These characteristics include the provision of infrastructure, the level of education, and the existing investment climate in the receiving countries, as previously suggested by Cárdenas-García and Pulido-Fernández (2019), or by the level of economic growth, the development of the urbanization process, or the degree of commercial openness of the receiving countries, as identified by Chattopadhyay et al. (2021).

Conclusions and policy implications

Distinct international organizations have shown that what is really important is not the contribution of tourism to economic growth, but rather, that this economic growth generated by the expansion of tourism activity permits the improvement of living conditions of the host population (EC, 2018; IADB - Inter-American Development Bank (2020); UNCTAD - United Nations Conference on Trade and Development (2020)).

Given the importance of economic development for the host countries, empirical studies that analyze the relationship between tourism and economic development have begun to emerge. These works mainly link the multidimensional concept of development with human development, measured by the HDI. Here, the link between tourism and human development is produced through the economic growth generated by the expansion of tourism activity. This economic growth is used to develop policies to improve the host population’s education and health levels.

However, few such studies exist, and the scientific literature does not reveal a defined trend with regard to this relationship. Furthermore, most of these existing works rely on causality analyses to determine whether there is a relationship between tourism and human development. They do not analyze whether having a higher degree of tourism specialization, for groups of countries with similar levels of development, implies a higher level of human development, which would suggest that tourism promotes development to a greater extent than other economic activities.

Due to the methodology used, this empirical work cannot determine the type of relationship existing between tourism and development, that is, whether there is a unidirectional or bidirectional relationship between both variables. However, it does allow us to determine if countries with a higher level of tourism specialization have a higher level of development than those specializing in other productive activities.

This study aimed to contribute to the empirical discussion about the relationship between tourism and development through the use of a non-parametric and non-linear approach; specifically, the qualitative dynamic behavior of these two variables was compared using the definition of economic regime and clustering tools based on the concept of hierarchical and MST (Mantegna, 1999; Kruskal, 1956).

The results seem to indicate that tourism is an economic activity that can promote human development more than other economic activities. Indeed, at similar levels of human development, both in the case of countries with a high level of development (countries in Group A versus countries in Group B) and in the case of countries with a low level of development (countries in Groups D-F versus countries in Group C), the country groups with a higher level of tourism specialization have higher human development values than those countries specialized in other productive activities.

Therefore, public administrations should develop specific actions to increase the level of tourism specialization since tourism is a strategic tool that improves human development levels, as compared to other economic activities. It is necessary to invest in the improvement and expansion of tourism infrastructure, including the improvement of transportation systems in host destinations, increasing and improving the supply of accommodations and basic tourism-related services. Moreover, an attractive offer should be provided, both in terms of resources and attraction factors. This includes complementary services to attract a greater number of tourist flows, while developing destination promotion campaigns and, therefore, ensuring greater tourism specialization.

It should also be noted that, of the identified country groups, the most numerous one is that which includes countries from Group C, which is made up of 43 countries (approximately a third of the sample). This cluster is characterized by low tourism specialization and a low level of economic development, which seems to translate into a poverty trap, given that the low level of development prevents the expansion of the tourism activity, and, in turn, this lack of tourism development makes it difficult to increase the levels of development.

Policies should be developed that consider the lack of financial resources of these countries to carry out investment projects. International organizations and institutions linked to development, such as the United Nations Development Program, Inter-American Development Bank, or World Bank, should finance specific projects so that these countries may receive investments related to the improvement and expansion of tourism infrastructure, so as to improve human development through this activity. Suitable regulatory frameworks should be established in these countries, to encourage public-private collaboration for the development of tourism projects. In this way, private investments could make up for the lack of public financing in these destinations.

The analysis performed in this work has also identified groups of countries that, despite their high degree of tourism specialization, do not have high levels of human development (Belize, Botswana, Jamaica, Maldives, and Tunisia). This highlights the importance of identifying factors or characteristics that provide the destination with ideal initial conditions to permit the economic impacts generated by the expansion of tourism to be channeled into an improvement in human development. In addition to being conditioned by the host country’s level of tourism specialization, the link between tourism and human development also depends on infrastructure provision, education level, investment climate, urbanization level, and the degree of commercial openness. Although this current of scientific literature has not been widely studied, it has been addressed by some works analyzing the relationship between tourism and human development (Cárdenas-García and Pulido-Fernández, 2019; Chattopadhyay et al., 2021).

Policies established by public administrations should consider a dual objective: on the one hand, investing in the improvement and expansion of the tourism infrastructure and, on the other hand, increasing and improving the factors found to be determinant in configuring tourism as a tool for human development. Given that there are entities investing in projects linked to tourism aimed at improving the living conditions of the resident population, the failure to act on the determinant factors of this relationship could result in inefficient policies in terms of the allocation of resources linked to improved development.

Finally, this study has certain limitations, including the variables used to measure tourism specialization and economic development. With regard to tourism, it has been shown that changing the indicator used leads to differences in the results obtained. In terms of economic development, while other factors such as poverty level, quality of life, or income inequality are related to development, human development, and its measurement through HDI, is the most frequently used indicator to measure it. Moreover, the short period analyzed (1995–2019) is another limitation. There is a restriction in the initial period used since it is the first year in which data were available on development and this may determine the small variability between countries among the different regimes. Another limitation lies in the fact that it does not analyze the characteristics of the destination as a determinant in the relationship between tourism and human development, in accordance with the new current of the scientific literature. In terms of methodology, the choice of the measure used for the symbolization of the series can affect the results. For example, the mean may be influenced by outliers in the data, and this can be relevant for certain variables, such as tourism, which displays a high degree of variation. It would be interesting to perform the same exercise using other measures for the symbolization of the series, such as the truncated mean, the median, or some type of threshold.

Future lines of research may highlight the fact that this study consists of an analysis at the country level, although it is clear that the impacts of tourism are produced in the territory at the regional and local levels. As a result, it may be interesting to replicate this work at the regional level using different countries as an analysis, depending on the availability of such data.

Moreover, as a continuation of this study, in addition to the degree of tourism specialization, it may be interesting to analyze the type of tourism received by each of the groups of countries that have been identified. In other words, to examine whether the characteristics of the type of tourism received (accommodations, motivations, or level of expenditure) in each cluster also determine the relationship between tourism and human development. Furthermore, it may be interesting to introduce the influence of other factors on the relationship between tourism and development into the analysis of this relationship, as discussed previously in the limitations.