Tourism growth and environmental sustainability: trade-off or convergence?

Abstract

Besides its socioeconomic benefits, tourism has been documented as one of the leading sectors with deleterious effects on the environment. This study investigates the relationship between tourism dynamics and environmental sustainability using biennial data for 148 countries over the period from 2006 to 2016. The first step develops a tourism growth index that encompasses various dimensions of tourism development and various panel cointegration techniques are then employed to characterize the dynamic association between environment sustainability and tourism growth. Empirical results reveal that tourism growth and environmental sustainability are indeed convergent not only for the full sample countries but also across geographical regions and socioeconomic clusters. In addition, a negative impact of tourism growth on the environmental welfare is evidenced in the long term; suggesting a trade-off between tourism activities and environment performance for the full sample over the past decade. At the regional level, similar finding is reported for Asia and Europe against a positive environmental impact for America and an inconclusive output for Africa. The observed difference might be attributed to the heterogeneity in the unsustainability level of regional tourism development with limited exposure for Africa and America. Interestingly, the convergence of tourism growth and environment well-being tends to exhibit varied speeds of adjustment across sample panels. The observed differences could be attributed to the country-level switching propensity from environment-harmful tourism practices as well as their socioeconomic characteristics. Consequently, policies geared towards minimizing the adverse environmental effects should be integrated with countries tourism management policies to enable the transition to sustainable tourism sector development. Thus, targeting nature tourism becomes a critical approach to tourism development rather than setting traditional goals such as number of visitors, income stream and employment.

Appendices

Appendix 1: Regional tourism growth

Appendix 2: Sustainability indices across regions

Appendix 3: Technical details on PCA

The PCA as introduced by Jolliffe (2002) consists of decomposing each observation from a sample into principal components.

Let X be a vector of p variables on a sample of n observations:

$$X = \left( {x_{1} ,x_{2} ,...,x_{p} } \right)$$

The first principal component of the sample is defined by the linear transformation:

$$z_{1} \equiv a_{1}^{T} X = \sum\limits_{i = 1}^{p} {a_{i1} x_{i} }$$

where the vector $a_{1} = \left( {a_{11} ,a_{21} ,...,a_{p1} } \right)$ is such that $Var\left( {z_{1} } \right)\,\,\,is\,\,\max imum$.

Similarly, the kth principal component of the sample is defined by the linear transformation:

$$z_{k} \equiv a_{k}^{T} X = \sum\limits_{i = 1}^{p} {a_{ik} x_{i} } \,\,\,\,\,\,k = 1,...,p$$

where the vector $a_{k} = \left( {a_{1k} ,a_{2k} ,...,a_{pk} } \right)$ is such that $Var\left( {z_{k} } \right)\,\,\,is\,\,\max imum$. subject to $\left\{ \begin{gathered} {\text{cov}} \left[ {z_{k} ,z_{l} } \right] = 0\,\,\,\,for\,\,\,k > l \ge 1 \hfill \\ and \hfill \\ a_{k}^{T} a_{k} = 1 \hfill \\ \end{gathered} \right.$

It is shown that ${\text{cov}} \left[ {z_{1} ,z_{2} } \right] = a_{1}^{T} Sa_{2} = \lambda_{1} a_{1}^{T} a_{2}$. where S is the covariance matrix and $\lambda_{1}$ the largest eigenvalue of S.

In general, the kth eigenvalue of S is the variance of the kth principal component; that is:$Var\left[ {z_{k} } \right] = a_{k}^{T} Sa_{k} = \lambda_{k}$.

Therefore, the kth principal component retains the kth greatest fraction of the variation in the sample.

For data compression, PCA reduces the dimensionality of the data from p to m by approximating $X \cong X^{m} = Z^{m} A^{mT}$.

Where.

• $Z^{m}$ is the $n \times m$ portion of $Z$ and $A^{m}$ is the $p \times m$ portion of $A$

• $Z = A^{T} X$ and A being an orthogonal $p \times p$ matrix.

• $Z = \left( {z_{1} ,z_{2} ,...,z_{p} } \right)$ and X as defined previously