Research note: Tourism and economic growth in Latin American countries further empirical evidence

Tourism Economics, 2011, 17 (6), 1365 1373 doi: 10.5367/te.2011.0095 Research note: Tourism and economic growth in Latin American countries further empirical evidence BICHAKA FAYISSA Department of Economics and Finance, Middle Tennessee State University, Murfreesboro, TN 37132, USA. E-mail: bfayissa@mtsu.edu. (Corresponding author.) CHRISTIAN NSIAH Department of Accounting and Economics, Black Hills State University, Spearfish, SD 57799, USA. E-mail: ChristianNsiah@bhsu.edu. BEDASSA TADESSE Department of Economics, University of Minnesota-Duluth, 1318 Kirby Drive, Duluth, MN 55812, USA. E-mail: btadesse@d.umn.edu. Using panel data that span from 1990 to 2005, the authors investigate the impact of tourism on the economic growth of 18 heterogeneous Latin American countries within the framework of the conventional neoclassical growth model. Results from the empirical models show that revenues from the tourism industry contribute positively to both the current level and the growth rate of the per capita GDP of the countries in the region, as do investments in physical and human capital. The findings imply that Latin American economies may enhance their economic growth in the short run by strengthening their tourism industries strategically, while not neglecting the traditional sources of economic growth. Keywords: dynamic panel data; fixed effects; random effects; Arellano Bond models; quantile regression; Latin America JEL classification: C33, F14, L83, O40, O54 The contributions of tourism to the economic growth of developed countries have been widely documented in the empirical literature (see, for example, Dritsakis, 2004, for Greece; Balaguer and Cantavella-Jorda, 2002, for Spain; Oh, 2005, for Korea; Tosun, 1999, and Gunduz and Hatemi-J, 2005, for Turkey; Proenca and Soukiazis, 2008, for Portugal; and Fayissa et al, 2008, for Africa). Given the phenomenal rise in tourist arrivals and earnings from the

1366 TOURISM ECONOMICS sector, the potential contributions of tourism to the economic growth of countries in Latin America have also been noted in several studies. Using data from 1985 to 1998 for a cross section of Latin American countries, Eugenio- Martín et al (2004), for example, indicate a positive but marginal effect of tourism receipts on the economic growth of low- and medium-income countries in the region. Clancy (1999) and Brida et al (2008) for Mexico, Vanegas and Croes (2007) for Nicaragua, Divino and McAleer (2008) for Brazil and Divino and McAleer (2009) for Peru also indicate that tourism plays a significant role in the economic growth of the respective countries. However, given the heterogeneity of the economic structure of the countries in the region and the variations in the time periods covered in these studies, the results cannot be generalized to every country in the region. Thus, using a panel data of 18 Latin American countries over a relatively extended period of time (1990 2005) for which reliable data on variables of interest are available, we investigate the potential contribution of tourism to the economic growth of the countries in the region and compare the observed impact of tourism with that of the traditional sources of economic growth (such as investment in human capital and foreign direct investment). Our findings reveal that the tourism sector in Latin American countries contributes significantly to both the current levels of gross domestic product and economic growth rate of the countries, as do investments in physical and human capital, implying that these economies might augment their economic growth in the short-term by investing strategically in their tourism sector. The rest of the paper proceeds as follows. The next section presents the conventional neoclassical growth model, which incorporates tourism as one potential source of economic growth, and a brief discussion of the econometric framework that we use to drive the estimates. The subsequent section presents empirical results and the final section draws conclusions in light of those results. The economic growth model with tourism Following the existing studies, we describe the responsiveness of income levels to the revenues generated from tourism and other sources of economic growth as follows: lny it = β 0 + β 1 lntrp it + β 2 lngcf it + β 3 lnefi it + β 4 lnsch it + β 5 lntrd it + β 6 lnaid it + β 7 lnofi it + β 8 FDI it + β 9 EXR it + ξ it (1) where Y it is real GDP per capita; TRP it is tourist receipts per capita; GCF it is gross fixed capital formation as a percentage of real GDP (used as a proxy for investment in physical capital); EFI it is the economic freedom index, measuring the consistency of a nation s institutions and policies with economic freedom; SCH it is tertiary enrolment, capturing investment in human capital; and TRD it is the sum of imports and exports as a percentage of GDP for each country under consideration. AID it, OFI it, FDI it and EXR it, respectively, denote the sum of official development assistance, foreign portfolio investments and other foreign financial flows other than foreign direct investment, the stock of foreign

Tourism and economic growth in Latin America 1367 Table 1. Description of the variables in the empirical model and their a priori expected signs. Variable Description A priori Mean expected (std dev) signs PCI GDP per capita (constant 2000 US$) Dependent 3,044.247 variable (1,956.414) TRP International tourism receipts (current US$) per capita + 71.433 (79.841) GCF Gross fixed capital formation (% of GDP) + 19.064 (3.961) EFI Economic freedom index + 6.081 (1.020) SCH School enrolment, tertiary (% gross) + 23.445 (11.557) FDI Foreign direct investment, net inflows (% of GDP) +/ 2.921 (2.396) TRD International trade (% of GDP) + 99.950 (11.700) AID Official development assistance (% of GDP) +/ 0.026 (0.051) OFI Other foreign capital flows (% of GDP) +/ 0.045 (0.046) EXR Official exchange rate (foreign currency per US$) + 1,045.735 (3,818.604) Notes: Number of countries = 18. The values of the variables used are two-year averages from 1990 to 2005. All variables are log-transformed for the regression estimation. All data are from World Bank s World Development Indicators CD-ROM, 2007, except the economic freedom index (EFI), which is taken from the Freedom House Foundation database. direct investment and the official exchange rate of each of country i s currency against the US dollar. All financial values are in US$. Except for the economic freedom index, which is taken from Freedom House, and the other foreign financial flows, which are from the UNCTAD Statistical Handbook (UNCTAD, 2010), all data are from the World Bank Development Indicators (World Bank, 2007). All the variables are log transformed, as denoted by the prefix ln. Table 1 provides detailed definitions, a priori expectation of the signs and descriptive statistics of each of the variables. To estimate the parameters corresponding with the variables, we first employ the fixed and random effects panel data estimation methods. However, since some of the traditional growth factors included in the model are either predetermined or endogenous, or both, and the current period of growth of the GDP could depend on its values in the past, we specify a dynamic variant of our empirical model, as described in Equation (2) below, and evaluate the joint effects of tourism receipts and the traditional sources of economic growth by employing the Arellano Bond (1991) generalized method of moments (GMM) estimator:

1368 TOURISM ECONOMICS lny it = α lny it 1 + β lnx it 1 + γ lnz it + ν i + ε it (2) where Y it represents the first difference of the log of the per capita income in country i during time t; Y it 1 stands for its lagged difference; X it 1 is a vector containing the lagged and differenced values of the series of endogenous variables; and Z it is a vector of the variables that are exogenous. The scalars α, β and γ are parameters to be estimated, with ν i representing country-specific effects assumed to be independently and identically distributed over the countries and ε it denoting an independently distributed stochastic disturbance term. Although results from both the panel data estimation and the Arellano Bond (1991) estimation methods enable us to address our research question while accounting for the country-/time-specific variations and the dynamic nature of some of the variables, they do not control for the heterogeneity in the conditional distribution of the GDP of the respective countries. To account for such variation and to determine the extent to which Latin American countries at different levels of income could rely on receipts from tourism, we employ the quantile regression analysis following Koenker and Bassett (1978) and Buchinsky (1998), an empirical exposition of which could be described as follows: lny it = β θ lnx it + e θit with Q θ (lny it lnx it = X it β θ ) (3) where X it is the vector of exogenous variables discussed earlier and β θ is the vector of parameters to be estimated corresponding with the θth conditional quantile of the log of GDP per capita (lny it ) given lnx it (that is, Q θ lny it lnx it ). Empirical results and interpretations Table 2 presents both the fixed effects and random effects estimates of the variables in our empirical model. The results broadly reveal the expected relationship between the explanatory variables and the per capita income levels of the countries in our study. Accordingly, all the variables representing the traditional sources of growth have the expected signs. While these results, in which we account simultaneously for the heterogeneity of the countries and the fluctuations in their economic performance over time, are appealing, several of the traditional growth explanatory variables we include in the model are either predetermined (for example, schooling) or endogenous (for example, FDI and AID), or both, thus confounding the effects. Estimates of the effects of the variables included in the Arellano Bond dynamic panel GMM estimator (where one- and two-period lagged levels and first difference of the variables serve as instruments for the endogenous variables) are reported in Table 3. Column 2 reports results for one-year lags of the dependent variable series, column 3 reports results that incorporate two-year lags of the dependent variable (growth in GDP per capita) series. In both models, the Sargan test fails to reject the null hypothesis that the over-identifying restrictions are valid. While the Arellano Bond test rejects the null hypothesis of no first autocorrelation

Tourism and economic growth in Latin America 1369 Table 2. Fixed effects and random effects results. Variable Description Fixed effects Random effects coefficients coefficients TRP International tourism receipts (current US$) 0.0781 0.0894 per capita (0.0105) *** (0.0145) *** GCF Gross fixed capital formation (% of GDP) 0.0762 0.0634 (0.0227) *** (0.0321) ** EFI Economic freedom index 0.1814 0.1538 (0.0336) *** (0.0475) *** SCH School enrolment, tertiary (% gross) 0.0662 0.0920 (0.0168) *** (0.0232) *** FDI Foreign direct investment, net inflows 0.0113 0.0139 (% of GDP) (0.0048) ** (0.0066) ** TRD International trade as % of GDP 0.0104 0.04866 (0.0215) (0.0295) AID Official development assistance (% of GDP) 0.0231 0.0410 (0.0054) *** (0.0075) *** OFI Other foreign financial inflows (% of GDP) 0.0768 0.0825 (0.0084) *** (0.0119) *** EXR Official exchange rate (foreign currency 0.0233 0.0311 per US$) (0.0059) *** (0.0074) *** Constant Intercept 7.3458 7.4608 (0.2137) *** (0.2978) *** Observations 224 224 R-squared 0.75 0.73 Hausman specification test 124.8 *** Notes: Standard errors are in parentheses; *** p < 0.01; ** p < 0.05. All variables are log-transformed. in the differenced residuals AR(1), it fails to reject the null hypothesis of no second-order autocorrelation in the differenced residuals. Consequently, we can say that these coefficients reflect the true (efficient and unbiased) relationship between growth in the per capita GDP of the countries in Latin America, the receipts from tourism (our variable of interest) and the other growth determinants included in the model. Based on the results in Table 3, we observe that the lagged values of GDP per capita ( Y it 1 ) and changes in tourism receipts (TRP) have a significant and positive impact on the per capita income growth rate of Latin American countries. Accordingly, a 10% increase in tourism receipts would lead to a 0.4% growth in the GDP per capita of a typical country in Latin America. After accounting for the endogenous nature of the traditional growth-explaining factors, we also find that the lagged value of investment in human capital (SCH) and physical capital (GCF) are related positively and significantly with economic growth. The economic freedom index (EFI), which is used as a proxy for the institutions, openness to trade (TRD) and foreign direct investment (FDI), however, are not statistically significant. When comparing the relative influence of tourism receipts (TRP) on per capita income growth of the countries with that of the traditional sources of growth (that is, investment in GCF, SCH and FDI), our tests reject the null hypothesis of equality of the

1370 TOURISM ECONOMICS Table 3. Arellano Bond dynamic panel data estimation results. Variables One-step and one-year lag (1) One-step and two-year lag (2) PCI (LD) 0.5214 0.5881 (0.0896) *** (0.0981) *** PCI (L2D) 0.0796 (0.0553) TRP (D(1)) 0.0413 0.0401 (0.0155) *** (0.0150) *** GCF (D(1)) 0.0808 0.0708 (0.0256) *** (0.0273) *** EFI (D(1)) 0.0237 0.0279 (0.0152) (0.0128) * SCH (D(1)) 0.0030 0.0052 (0.0183) (0.0189) SCH (LD) 0.0328 0.0319 (0.0199) *** (0.0119) *** FDI (D(1)) 0.0032 0.0042 (0.0003) (0.0035) FDI (LD) 0.0003 0.0028 (0.0026) (0.0030) TRD (D(1)) 0.0014 0.0042 (0.0188) (0.0176) AID (D(1)) 0.0077 0.0077 (0.0055) (0.0056) AID (LD) 0.0014 0.0015 (0.0042) (0.0039) OFI (D(1)) 0.0409 0.0407 (0.0149) *** (0.0149) *** EXR (D(1)) 0.0073 0.0081 (0.0069) (0.0090) Constant 3.5853 3.7387 (0.8002) *** (0.7581) *** Number of observations 176 164 Number of countries 17 17 Wald Chi-square 2,970.50 *** 5,945.51 *** Arellano Bond test of the null of no AR(1) residual errors 2.75 *** 2.52 *** Arellano Bond test of the null of no AR(2) residual errors 1.15 1.14 Sargan test of the validity of the null of over-identifying restrictions 52.04 44.87 Notes: Standards are in parentheses. *** and * indicate significance at the p < 0.01 and p < 0.1 levels, respectively. The suffix D(1) after each variable denotes the number of times the specific variable was differenced. LD denotes the lagged difference. The variable SCH is treated as predetermined, while FDI and AID are treated as endogenous variables. relative effect of TRP and FDI at a p-value of 0.0001. However, they fail to reject the equivalence of the effect of tourism receipts and increased human capital formation and investment in physical capital on the per capita income

Tourism and economic growth in Latin America 1371 0.14 0.12 0.1 Coefficient 0.08 0.06 0.04 0.02 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Quantile Figure 1. Patterns of the effects of tourism receipts on the economic growth of Latin American countries at different income levels. Note: The solid grey horizontal line denotes the OLS estimate from the fixed effects model. The dotted grey horizontal lines denote the upper and lower error bands (95% confidence interval) for the fixed effects estimate. The solid black curve and black dotted lines denote the bootstrap confidence interval for the quantile regression estimate. growth. The implication is that while tourism receipts in Latin American economies have the potential to generate a significantly larger per capita income effect than FDI, similar outcomes in terms of enhanced income growth (at least in the short run) could be obtained by exerting as much effort on increased educational opportunities and through further build-up of physical capital. Finally, Figure 1 presents our results from the quantile regression analysis in which we examine the relative effects of tourism receipts on economic growth while accounting for the differences in the conditional distributions of the per capita GDP of the countries in the region. With the black solid and dotted lines, respectively, denoting the quantile regression estimates and their corresponding bootstrapped confidence bands, the grey solid and dotted horizontal lines, respectively, denoting the OLS estimate from the fixed effects model and the upper and lower error bands (95% confidence interval) for the fixed effects estimate, the results clearly indicate that tourism receipts have a positive impact at all quantiles of income distribution, but they have a larger positive impact at the lower quantiles of income distribution than at the higher quantiles of income distribution. Given that the estimated coefficients for the quantile regression fall outside of the confidence interval area for the corresponding panel data estimates denoted by the dotted lines, we can also say that the positive impact of receipts from tourism on the economic growth of the countries in the region that are at a relatively lower GDP level is significantly greater than at the higher GDP

1372 TOURISM ECONOMICS levels. The implication here is that countries in the region that are at relatively lower income levels could benefit more in the short run by investing strategically in their tourism sector. Conclusion This study examines the effect of international tourism on the economic growth of a heterogeneous set of economies in Latin America, compares the potential contribution of receipts from tourism with the traditional sources of growth and identifies whether there exists a difference in the extent to which countries that are at different levels of per capita GDP could rely on incomes generated from their tourism sector. Our results show that receipts from tourism impact positively on the economic growth of Latin American countries (with a 10% increase in the spending of international tourists leading to an average of a 0.78% increase in the real GDP per capita income of a typical country in the region). A salient conclusion that can be drawn from this study is that tourist spending is important for the economic growth of all Latin American economies, especially in those countries which are at the low end of the income distribution spectrum. Our results suggest that Latin American countries offer a considerable potential for seaside tourism, environmental and ecotourism, cultural tourism, sports tourism and discovery tourism, the potential of which has yet to be exploited fully. The policy implication is that Latin American countries may improve their economic growth performance not only by investing in the traditional sources of growth such as investment in physical and human capital and trade, but also by directing resources (security, infrastructure and marketing skills) strategically toward the development of the tourism industry and improving their governance. References Arellano, M., and Bond, S. (1991), Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations, Review of Economic Studies, Vol 58, pp 277 297. Balaguer, J., and Cantavella-Jorda, M. (2002), Tourism as a long-run growth factor: the Spanish case, Applied Economics, Vol 34, No 7, pp 877 884. Brida, J.G., Carrera, E.S., and Risso, W.A. (2008), Tourism s impact on long-run Mexican economic growth, Economic Bulletin, Vol 3, No 21, pp 1 8. Buchinsky, M. (1998), Recent advances in quantile regression models: a practical guide for empirical research, Journal of Human Resources, Vol 33, No 1, pp 88 126. Clancy, M.J. (1999), Tourism and development: evidence from Mexico, Annals of Tourism Research, Vol 26, No 1, pp 1 20. Divino, J.A., and McAleer, M.J. (2008), Modeling sustainable international tourism demand to the Brazilian Amazon, Econometric Institute Report No EI 2008-22, Erasmus University, Rotterdam (http://econpapers.repec.org/paper/dgreureir/1765013773.htm, accessed 26 October 2011). Divino, J.A., and McAleer, M.J. (2009), Modeling and forecasting daily international mass tourism to Peru, CIRJE F-Series CIRJE-F-651, CIRJE, Faculty of Economics, University of Tokyo. Dritsakis, N. (2004), Tourism as a long-run economic growth factor: an empirical investigation for Greece, Tourism Economics, Vol 10, No 3, pp 305 316. Eugenio-Martin, J.L., Martin-Morales, N., and Scarpa, R. (2004), Tourism and economic growth in Latin American countries: a panel data approach, FEEM Working Paper No 26.2004. Fayissa, B., Nsiah, C., and Tadesse, B. (2008), Impact of tourism on economic growth and development in Africa, Tourism Economics, Vol 14, No 4, pp 791 806.

Tourism and economic growth in Latin America 1373 Gunduz, L., and Hatemi-J, A. (2005), Is the tourism-led growth hypothesis valid for Turkey?, Applied Economics, Vol 12, pp 499 504. Koenker, R., and Bassett, G. (1978), Regression quantiles, Econometrics, Vol 46, pp 33 50. Oh, C. (2005), The contribution of tourism development to economic growth in the Korean economy, Tourism Management, Vol 26, No 1, pp 39 44. Proenca, S., and Soukiazis, E. (2008), Tourism as an alternative source of regional growth in Portugal: a panel data analysis at NUTS II and III levels, Portuguese Economic Journal, Vol 7, No 1, pp 43 61. Tosun, C. (1999), An analysis of contributions of international inbound tourism to the Turkish economy, Tourism Economics, Vol 5, pp 217 250. UNCTAD (2010), UNCTAD Handbook of Statistics 2010 (http://unctad.org/templates/ Page.asp?intItemID=2364&lang=1, accessed 26 October 2011). Vanegas, M., and Croes, R. (2007), Tourism, economic expansion and poverty in Nicaragua: investigating cointegration and causal relations, Staff Working Paper No 7306, Department of Applied Economics, University of Minnesota. World Bank (2007), World Development Indicators 2007, World Bank, Washington, DC.