International tourism and short-run growth: Reviewing the trade-off in the age of COVID-19
Since its onset, the coronavirus pandemic has exerted huge negative impacts on economies throughout the world (see the coverage on Vox here). Many countries have witnessed unprecedented drops in their GDP as well as skyrocketing levels of unemployment, not to mention widespread illness and many hundreds of thousands of deaths. With the COVID-19 pandemic still raging at the beginning of the summer, governments of countries that depend heavily on international tourism were confronted with the dilemma of whether or not to let travel restart. Tourism involves the mixing of different populations and travel through congested airports (as well as other transportation hubs). This could have the potentially deleterious effects of seeding further infections in spite of protective measures that are by now better understood. In light of this, we explore data on international tourism and short-run economic growth in order to quantify the impact of restarting tourism.
Over the last two decades, tourism exports have become increasingly important throughout the world. Their rapid growth has been fueled by lower transportation costs and higher incomes.
From 1995 to 2017, the world average share of tourism exports in GDP increased by 31%, from 5.42% to 7.11% over the period (see Figure 1a). Over the same period, the world average share of manufacturing exports in GDP increased by 27%, from 11.91% to 15.17%. Compared with manufacturing exports, tourism exports have grown more leading up to (and then stagnated less following) the 2008 global crisis. As shown in Figure 1b, manufacturing exports shares in terms of total GDP are higher than the world average for OECD countries. However, since 2010, tourism exports have been growing fast: their share in GDP has increased by 32% from 2010 to 2017.
Figure 1 Average share of tourism and manufacturing exports in GDP
There have not been any recent studies of the impact of tourism exports on growth. Sequeira et al. (2008) use panel data for 1980-2002 and find that a 1% increase in the tourism exports shares in total GDP accounts for a 0.03-0.11% increase in GDP per capita in the short run. Arezki et al. (2009) estimate the average growth rate of GDP per capita over 1980-2002, using cross-sectional data along with the instrumental variable method, and find that a one percentage point increase in the tourism exports as a share of total exports is associated with 0.012-0.017 percentage point increase in the growth of GDP per capita. Recently, Faber and Gaubert (2016, 2019) use rich data for Mexico with a quantitative spatial equilibrium model and find that tourism causes large and significant local economic gains that are driven in part by significant positive spillovers on manufacturing. However, these local spillovers are offset by reduced ‘agglomeration’ economies among less touristic regions.
We seek to quantify the short-run growth impact of tourism by means of a novel econometric approach with up-to-date data on bilateral tourism, manufacturing flows, and importers’ GDP per capita, which allow us to obtain reliable causal estimates with international panel data.
Data and methodology
Our data come from several sources, in particular the World Bank’s ‘World Development Indicators’ (WDI), the UN Comtrade database, and the Yearbook of Tourism Statistics of the World Tourism Organization. The period of study covers 1995 to 2017.
For the purpose of econometric analysis, we consider the determination of tourism exports from the perspective of demand and assume that supply is not limited by capacity. We define a country’s tourism specialisation as tourism exports as the share of the sum of ‘tourism exports’ plus ‘manufacturing exports’. For a particular exporting country (denoted in our model by ‘i’), this magnitude may be recognised as a weighted average of the preference over tourism and manufacturing goods of each importer country (denoted ‘j’) other than the chosen exporter. The chosen importer country’s income is weighted as relative to all other countries importing tourism from the chosen importer country. Supposing that the preference contains an endogenous component, our identification can rest on the importer countries’ lagged preferences and the current year’s weight. Bilateral manufacturing flows are obtained from UN Comtrade, but bilateral tourism flows from UN WTO are given in terms of numbers of arrivals and need to be adjusted.
We estimate the growth in GDP per capita as a function of tourism specialisation, while controlling for fertility, general government final consumption expenditure, openness to trade, and population growth. We allow for fixed country-specific characteristics and year dummies. Estimations using ordinary least squares yield that, for the sample of all countries, tourism specialisation does not (on average) have a statistically significant effect on the growth rate of GDP per capita. However, there is a positive and significant effect for the OECD countries. A 1% increase in tourism exports as the share of tourism and manufacturing exports is associated with 0.01 percentage point increase in the growth rate of GDP per capita for the OECD group (all else being equal). With the mean of GDP per capita growth rate being approximately two percentage points, the elasticity is around 0.5.
Although this estimation does account for country-specific time-invariant unobservable factors, in principle it still suffers from endogeneity bias. This is because there may exist time-varying unobserved factors that could cause reverse causality. For example, a country’s income per capita growth rate could affect government policies that influence resource allocation between the tourism and manufacturing sectors. Similarly, income growth and tourism exports might be jointly affected by other unobserved shocks, including changes in political risk and exchange rates, occurrence of conflicts and disasters, and numerous other factors (Eilat and Einav 2004). These issues interfere with causal inference.
As a result, we employ two-stage least square estimation methods, as detailed in Chen and Ioannides (2020) (our main estimation results are given in the appendix). As with ordinary least squares (OLS) estimation, tourism specialisation does not have a significant impact on GDP per capita growth rate for all countries (on average) but does have a positive effect for the OECD countries. The point estimate is the same as that in the OLS estimation (but with a slightly larger standard error). We examine the robustness of our findings by means of alternative measures of the growth rate of GDP per capita and obtain similar results. Our findings are in line with the existing research on this topic. However, we use panel data which are more recent and cover a time period that is characterised by considerable variations in both tourism and manufacturing exports.
We explore how tourism specialisation can affect a country’s short-run economic growth. A positive effect is found for the OECD countries, where a 1% increase in tourism exports as the share of total tourism and manufacturing exports is (on average) associated with 0.01 percentage point (or 0.5% in terms of elasticity) increase in the growth rate of GDP per capita.
Our findings of positive effects shed light on the potential policies governments may employ to restart their economies after the coronavirus pandemic has been successfully controlled. With resources being limited, encouraging the opening up of tourism can be an effective policy, as tourism specialisation does have short-run positive effects, especially in the context of excess capacity. However, these effects are small.
We have restricted our attention to quantifying the effects of tourism specialisation on short-run economic growth for OECD countries versus all countries and does not investigate the underlying mechanisms. We do not address the potential effects on the importing economies of diversion of travel and tourism from domestic to international destinations. Furthermore, tourism involves mixing populations and restarting it could have potentially deleterious effects of seeding further infections. The need to revive economies is, of course, pressing. We hope that our estimations will provide some guidance and provoke further research beyond this admittedly very simple framework. We do not address in our paper whether such benefits are worth the risks. Small (but still beneficial) short-run effects must be contrasted with potentially adverse long-run ones. For example, we underscore the subtlety of findings by Faber and Gaubert (2016, 2019) in their studies of Mexico as well as those by Chen (2020), who uses data for 64 developing countries and shows that the expansion of tourism exports has negative effects on individuals’ long-term educational attainment when they are considered during schooling ages.
Authors’ note: Tufts University funds only have supported this research. Thanks go to the World Tourism Organization for letting us access their detailed data and to Marcelo Bianconi and Chih Ming Tan for helpful comments, but we are solely responsible for the content.
Arezki, R, R Cherif and J Piotrowski (2009), “Tourism specialization and economic development: Evidence from the UNESCO world heritage list”, IMF Working Papers: 1-24.
Boustan, L, F V Ferreira, H Winkler and E M Zolt (2013), “The effect of rising income inequality on taxation and public expenditures: Evidence from US municipalities and school districts, 1970-2000”, Review of Economics and Statistics 95(4): 1291-1302.
Chen, Y (2020), “Does Tourism Expansion Discourage Skill Acquisition in the Developing Countries?”, MS in Economics Thesis. Tufts University.
Chen, Y and Y M Ioannides (2020), “International Tourism and Short-Run Growth”, working paper, Tufts University.
Eilat, Y and L Einav (2004), “Determinants of international tourism: a three-dimensional panel data analysis”, Applied Economics 36(12): 1315-1327.
1 For a similar approach see Boustan et al. (2013), who construct an instrument for the Gini index by fixing initial income distribution and predicting Gini index with national patterns of income growth.