A qualitative dynamic analysis of the relationship between tourism and human development

The paper investigates whether greater tourism specialization is associated with higher human development, moving beyond traditional growth metrics to a multidimensional development perspective. Motivated by extensive evidence on tourism–growth linkages and the policy relevance of channeling growth into improved living conditions, the study frames development via the Human Development Index (HDI) and explores the qualitative dynamics between tourism specialization (international tourist arrivals per capita) and human development across 123 countries (1995–2019). The research asks if, at similar development levels, countries with higher tourism specialization exhibit higher HDI than those relying on other activities. Using a non-parametric, model-free dynamic grouping approach, the study seeks to uncover heterogeneous country clusters, informing both empirical modeling and policy design.

The review distinguishes economic growth from economic development, emphasizing development’s multidimensional nature (health, education, income) and the limits of GDP per capita as a welfare proxy. It synthesizes three well-documented strands on tourism–growth: tourism-led growth, economy-driven tourism growth, and bidirectional causality. The HDI is presented as a widely adopted, comparable, multidimensional indicator capturing health, education, and income. On tourism–human development, the literature is sparse and yields mixed findings: some studies suggest tourism boosts development/human capital (often context-dependent or in developed countries), others find no effect, and some support bidirectionality. Heterogeneous country characteristics (infrastructure, education, technology, environment, urbanization, openness) likely condition tourism’s impact on human development, underscoring the need for broader, comparative analyses like the present study.

- Data: Panel of 123 countries, 1995–2019. Tourism measured by international tourist arrivals (or international visitors when necessary) per capita; development measured by HDI (0–1). Population from World Bank used to compute per capita. Indicators sourced from UN Tourism (arrivals), UNDP (HDI), and World Bank (population). - Symbolic time series and regimes: For each year t, compute sample annual averages of tourism per capita (x̄_t) and HDI (ȳ_t). Partition the state space into four regimes using these annual sample means as thresholds: (R1) high tourism, high HDI; (R2) low tourism, high HDI; (R3) low tourism, low HDI; (R4) high tourism, low HDI. Each country’s 2D trajectory {(x_t, y_t)} is mapped to a 1D symbolic sequence s_t ∈ {R1,R2,R3,R4}. - Distance metric: For countries i and j, define d(s_i, s_j) = Σ_{t=1}^T 1[s_it ≠ s_jt], the count of years their regimes differ (a Hamming-type distance on symbols). Intuitively, it measures regime non-coincidence over time. - Clustering via MST: Construct a hierarchical tree with nearest-neighbor clustering using the defined distance. Apply Kruskal’s algorithm to build a Minimum Spanning Tree (MST) linking nodes (countries or merged nodes for identical symbolic sequences) by minimal distances. Clusters/groups are identified from the MST structure. Analyses conducted in RStudio.

- Regime occupancy and stability: A large majority of countries show high regime stability. 80 countries remained in the same regime for the entire period; 16 countries stayed at least three-quarters of the time in one regime. Only 27 countries switched regimes for at least a quarter of the period. Structural regime changes appear infrequent on an annual timescale. - Regime composition (illustrative): In 2019, few countries occupied high-tourism/low-HDI space (e.g., Belize, Fiji, Jamaica, Saint Lucia, Maldives, Samoa), with most clustering in low-tourism quadrants. - MST-based clusters (Figure 3/4): Six groups plus outliers identified: • Group A (n=36): Predominantly Regime 1 (high tourism specialization, high HDI); mainly European plus some Asian and Uruguay. Average HDI 0.8438; tourists per capita 1.3609. • Group B (n=29): Predominantly Regime 2 (low tourism, high HDI); geographically dispersed including parts of Europe, the Americas, and Asia. Average HDI 0.7830; tourists per capita 0.2211. • Group C (n=43): Predominantly Regime 3 (low tourism, low HDI); many African and Asian countries, plus some from Latin America. Average HDI 0.5473; tourists per capita 0.0947. • Group D (n=2: Belize, Maldives): Regime 4 throughout (high tourism, low HDI). Average HDI 0.6741; tourists per capita 1.4550. • Group E (n=4: Armenia, Moldova, Thailand, Turkey): Low tourism but alternating between high and low development (Regimes 2 and 3). Average HDI 0.7146; tourists per capita 0.2084. • Group F (n=3: Botswana, Jamaica, Tunisia): Mostly Regime 4 (high tourism, low HDI) but with movements among regimes. Average HDI 0.6751; tourists per capita 0.6466. • Outliers (n=6): Canada, Fiji, Saint Lucia, Sweden, Eswatini, Samoa. - Core relationship: At comparable development tiers, country groups with higher tourism specialization exhibit higher HDI. This holds at the top (Group A vs B) and bottom ends (Groups D/F vs C). Conversely, a large cluster (Group C) shows low tourism specialization and low development, indicative of a poverty trap. - Country trajectory examples: Some countries’ dynamics appear driven by HDI (e.g., Latvia, Eswatini), others by tourism specialization (e.g., Brazil, Fiji).

The findings address the research question by showing that, within comparable development levels, greater tourism specialization is associated with higher human development. This supports tourism as a potentially effective avenue for improving living conditions beyond its role in economic growth. The strong regime stability suggests persistent structural configurations in tourism–development dynamics over 1995–2019. The MST clusters reveal heterogeneity: countries specialized in tourism tend to display higher HDI than similarly developed peers with low tourism specialization, while many low-specialization economies are stuck in low-HDI regimes. Exceptions (e.g., Belize, Botswana, Jamaica, Maldives, Tunisia) underscore that tourism alone is insufficient; enabling conditions such as infrastructure, education, investment climate, urbanization, and openness likely mediate the tourism–development linkage. These insights align with strands of prior literature while clarifying context-dependent effects and offering a dynamic, non-parametric perspective.

The study introduces a non-parametric, symbolic time series and regime-based approach to compare tourism–development dynamics across 123 countries. It identifies stable, heterogeneous clusters and shows that, at similar development levels, higher tourism specialization tends to coincide with higher HDI, implying tourism can promote human development more than other activities under conducive conditions. Policy implications include investing in tourism infrastructure, transport, accommodations, and destination attractiveness to foster tourism specialization, alongside strengthening enabling factors (infrastructure, education, regulatory and investment climate, urbanization, openness) to translate tourism gains into human development. International organizations should help finance tourism-related development projects, especially for countries in a low-tourism/low-HDI poverty trap. Future research should: (i) replicate analyses at regional/local scales; (ii) consider the type and characteristics of tourism (motivation, accommodation, expenditure) across clusters; (iii) incorporate additional determinants; and (iv) test alternative symbolization thresholds (e.g., median, truncated means).

- Measurement choices: Tourism measured by arrivals per capita; results may differ with alternative indicators (e.g., expenditures, revenues, shares of GDP/exports). Development proxied by HDI; other dimensions (poverty, inequality, quality of life) were not incorporated. - Period coverage: 1995–2019 constrained by data availability; excludes COVID-19 and may limit observable structural changes. - Methodological sensitivity: Symbolization thresholds based on annual means may be affected by outliers and high variability in tourism; alternative thresholds (median, trimmed mean) could yield different partitions. - Omitted conditioning factors: The analysis does not explicitly model destination characteristics (infrastructure, education, investment climate, urbanization, openness) that may condition the tourism–development relationship. - Dynamic resolution: Annual data may be too coarse to capture timing of structural regime changes.