Verifying Tourism-Led Growth Hypothesis (TLGH) in Ethiopia: Time series analysis

Tourism-led growth hypothesis argues that international tourism is a source of economic advancement through generating foreign currency, creating employment opportunities, stimulating greater investments in infrastructure and inspiring the development of other economic sectors that ultimately improved the living standard of the citizens of a country. Knowing the importance of tourism sector, Ethiopia has taken this sector as a strategy to achieve its short and long run economic objectives. Therefore, the principal objective of this study is to validate the Tourism-Led growth hypothesis (TLGH) in Ethiopia over the annual period 1991-2018. To attain the envisaged objective, the standard Augmented Dickey–Fuller (ADF) and Phillips–Perron (PP) unit root tests were applied to check the stationarity of each series. To scrutinize the long run relationship between tourism and economic growth, ARDL bound test of cointegration was applied. In addition, pairwise Granger causality test was used to identify the direction of causality between tourism receipt and economic growth. The empirical findings confirm the existence of long run relationship between tourism receipt, economic growth and other control variables. The results of the pairwise Granger causality also predicted unidirectional causality running from tourism receipt to economic growth in long run. This validates the tourismled growth hypothesis for Ethiopia. Hence, policy makers should focus on promoting and development of the untapped tourism industry of the country to augment economic growth in Ethiopia.


Introduction
Tourism-led growth hypothesis (TLGH) assumes that the development of international tourism boosts economic growth. Similar to the export-led growth hypothesis (ELGH), it advocates that economic advancement can be promoted not only by increasing human resources and technology within the economy, but also by intensifying foreign exchange earnings (Ohlan, 2017). Consequently, different scholars have tried to analyze the effect of tourism on growth and development process of a given economy. Ekanayake and Long (2012) argued that tourism is vital for economic advancement, employment generation and thereby poverty reduction. In addition to providing direct employment creation in the hotel and restaurant industries, it also creates employment in other sectors through its externality on other sector such as agriculture, manufacturing, and food processing and also service sectors such as banking, foreign exchange transactions, transportation, communication (Kammas and Salehi-Esfahani, 1992). It can also improve the current account balance through generating foreign currency (Oh, 2005). International tourism directly inspires the development of local industries such as transport, hotels and restaurants, manufacturing, handicraft, agriculture and the trade (Lionetti and Gonzalez, 2012). Besides, Khalil, Kakar, and Malik (2007) argued that tourism industry can boost household's income and government revenues through multiplier effects.
However, many authors questioned the favorable effect of international tourism on economic growth as there might be leakages as well. According to Kammas and Salehi-Esfahani (1992), "imported goods that are demanded by the tourists, and the cost of goods and services that are needed for the infrastructural development of the industry" are the leakages. In addition, the expansion of the tourism infrastructure such as construction of hotels could cause the pollution of coastal waters, especially if sewerage capacity is inadequate (Sunlu, 2003). As a result, tourism expansion may adversely affect the economy. Therefore, it is important to conduct more vigorous research so as to evaluate the net effect of tourism industry on a given economy.
Even though the empirical findings have no clear-cut answer for the effect of tourism on economic growth, many countries give attention international tourism as a strategy to achieve their short and long run economic objectives. Ethiopia is one of the countries that used it as its development strategy. In Ethiopia, the perception about the importance of tourism industry is traced back to the early 1960s. Ethiopia has been working to revitalize the nation's tourism industry through restructuring the responsible organizations and introducing new systems (Ministry of Culture and Tourism, 2019). According to the United Nations World Tourism Organization report and Ministry of tourism and culture reports, the number of international tourists and tourism receipts in Ethiopia is gradually increasing from time to time. Therefore, the main objective of the study is to analyze the effect of tourism on economic growth in Ethiopia over the period 1991-2018. That means, this paper tests the validity of capital formation, and gross enrollment. The variables nominated in the study are based on the tourism-led growth hypothesis which posits that international tourism boosts economic growth through intensifying foreign exchange earnings, creating employment opportunity and positive externality (Ohlan,2017). The inclusion of other control variables (gross capital formation-proxy for physical capital formation and gross enrollment rate -proxy for human capital) in the analysis is to reduces the omitted variable bias. In addition, as the dependence of Ethiopian Economy on Agriculture is very high, drought dummy which captures the recurrent drought is included in the model. The dynamics of leadership change may have an economic implication on a given country . Therefor to capture the effect of leadership change on economic growth, leadership change dummy is added to in to the model. International tourism receipt is measured as a ratio of total export while government spending and remittance are measured as a percentage of GDP. Some of the variables included in the model were obtained from international official data sources (world development indicator and UN-stats) while the remaining variables were obtained from domestic sources (from ministry of culture and tourism and ministry of education).

The Model
In order to verify the tourism-led growth hypothesis, the standard classical production function is used. Because it is the most widely used aggregate production function in macroeconomics analysis. For instances, Balaguer and Cantavella-Jordà (2002) have followed this functional form in order to prove the link between tourism and economic growth. In addition to the standard production inputs (human and physical capital) tourism was included in the model as non-standard type of export. Therefore, my research followed the following modified production function: Where: PCGDPgt = Per capita real GDP growth rate at time t.
TOUR t = Tourism receipt as a percentage of total export at time t RGFCgt = Real gross fixed capital growth rate at time t.

Cointegration test
In order to test the existence of a stationary equilibrium relationship among the variables of interest, the univariate cointegration approaches of Engle and Granger (1987) and Phillips and Hansen (1990), and the multivariate cointegration procedures of Johansen and Juselius (1990), have been used extensively in the literature. Recently, the cointegration approach by Pesaran et al. (2001), known as the ARDL bounds testing, has become popular among researchers. In this research, ARDL approach to cointegration is used to prove the existence of long run and short-run relationship among the variables of interest. This approach has several advantages over other cointegration techniques (Pesaran et al., 2001). First, it is applicable irrespective of whether the underlying regressors are purely I (0), purely I (1) or fractionally integrated. Second, the model uses a sufficient number of lags to capture the data-generating process in general to the specific modelling framework. Third, the error correction model is derivable from the ARDL through a simple linear combination, which integrates both shortrun adjustments with long-run information without losing the latter's information. Fourth, the small samples properties of the ARDL procedure are far superior to those of the multivariate cointegration techniques. Fifth, endogeneity and serial correlation problems are corrected through appropriate lag selection. (Umoh and Effiong, 2013). Therefore, this study applied Autoregressive Distributed Lag (ARDL) Model to analyzing the long run and short run effects of tourism receipt on economic growth in Ethiopia. To test whether there is a long run equilibrium relationship between the variables; bounds test for co-integration is carried out as proposed by Pesaran, Shin, and Smith (2001). Accordingly, the hypothesis of the relationship between tourism receipt and economic growth is tested based on the following ARDL bound test equation. ∆ = β +λ PCGDP +λ LnTOUR +λ RGCF + +λ ENROL Where: PCGDPgt = Per capita real GDP growth rate at time t.
TOUR t = Tourism receipt as a percentage of total export at time t RGFCgt = Real gross fixed capital growth rate at time t. ENROLg t = Growth rate of secondary school gross enrolment at time t. DROUT= Dummy variable for recurrent drought LEDCH= Dummy variable for leadership change et = Random variable at time t. According to Pesaran et al. (2001), if the computed Wald F-statistic is less than the lower bound critical value, there is no long run relationship between the variables included in the model (tourism receipt and economic growth and other control variables). That means, the null hypothesis (λ = λ = λ = λ = 0. + will not be rejected. On the other hand, if the calculated F-statistic is greater than the upper bound critical value, then there is a long run relationship between tourism receipt and economic growth accepted. That means, alternative hypothesis of long run relationship (λ . λ . λ . λ . 0. + will be accepted.
Further, pairwise Granger causality test is applied to determine whether there is uni-directional causality or bi-directional causality between tourism receipt and economic growth in Ethiopia.

Overview of Tourism sector in Ethiopia
In Ethiopia, the perception about the importance of tourism industry is traced back to the early 1960s. Ethiopia has been working to revitalize the nation's tourism industry through restructuring the responsible organizations and introducing new systems (Ministry of Culture and Tourism, 2019). As clearly indicated in Figure-1, the number of international tourists has gradually increased over the past twenty-seven years. In 1991 the number of international tourists in Ethiopia was about 81581. This figure has continuously increased up to the year 1997 and reached 138856. In the next two consecutive it has recorded a poor performance. This could be due to the war between Ethiopia and Eritrea which took place from 1998 to 2000. After the year 2000, the total no of tourists arrived in Ethiopia has continued to grow until 2017. In comparison to the growth registered in previous years, the increment in tourism number in 2016 was insignificant. This slowdown in visitor arrivals in 2016 was due to the violent anti-government protests across the country started at the end of 2015 which led the Ethiopian government to pronounce a six-month state of emergency. Countries which are the major source of Ethiopian tourism market (United States, United Kingdom, Germany and others) cautioned their citizens not to travel to some regions of Ethiopia. This phenomenon adversely affected Ethiopia's holiday tourism sector as many international operators cancelled planned trips (Altes,2018). Then the figure of visitor arrivals has reimbursed to progress in 2017. However, the number of visitors again declined in 2018.
Similar to the trend of visitor arrival, the gross revenue generated by the tourism sector (tourism receipt) has recorded a continuous progress (except for the year 1998 and 1999) over the past twenty-seven years. However, the progress in tourism receipt showed a big jump since 2013. This structural break was happened due to the update of parameters used to calculate tourism revenue (Tourism Statistics bulletin, 2009-2015. This shows that tourism sector could be one potential sector of foreign currency generation (see Figure-2 for detail).

. Correlation matrix and Descriptive statistics
In order to have a general picture, the correlation matrix descriptive statistics about each variable over the period 1991-2018 is reported in Table 1. The mean of GDP per capita growth (PCGDPg), Gross Capital Formation growth (RGCFg), tourism receipt as a % of total export (TOUR) and secondary school enrollment growth (ENROLg) are about 4.14 %, 13.57%, 27.42 and 7.05% respectively. Over the period 1991-2018 real GDP per capita recorded a maximum growth of 10.41% and a minimum growth of -12.13%. On the other hand, tourism receipt as a% of total export reached a maximum value of 39.23% and a minimum value of 15.97%. The descriptive statistics also showed that only real per capita GDP growth is normally distributed with Jarque-Bera statistic of 7.66 and probability value of 0.023< 0.05. As displayed in the correlation matrix, the real GDP per capita, real gross capital Journal of Tourism, Hospitality and Sports www.iiste.org ISSN (Paper) 2312-5187 ISSN (Online) 2312-5179 An International Peer-reviewed Journal Vol.45, 2019 7 formation, tourism receipt as a% of total export and secondary school enrolment are positively correlated. But gross enrollment is negatively correlated with real gross capital formation.

Unit Root Test and Optimum lag selection
Before testing the long-run and short run association between the variables it is compulsory to examine the integrating properties of the variables. To do this, the standard Augmented Dickey-Fuller (ADF) and Phillips-Perron(PP) unit root tests were applied. Trend and intercepts are included in the tests. The results of the ADF test reported in Table-2 confirmed that PCGDPg and RGCFg are not subject to unit root problem at level with intercept at 1% level of significance. On the other hands TOUR and ENROLg are not stationary at level. This clearly shows that the integrating order of the variables is not unique rather a mixture of I (1) and I (0). The robustness of ADF unit root test is checked by applying PP unit root test. Accordingly, the results provided by ADF test are reliable and consistent. That means the PP unit root test validates that variables are a mixture of I (1) & I (0).
In order to determine the appropriate lag order, VAR lag order selection criteria was used. We have followed Schwarz information criterion (SIC) for choosing appropriate lag length due to its superior properties (Pesaran andSmith, 1998 andLiew, et.al. 2008). It has considerable high performance in selecting the true lag order, even if the sample size is small. The results reported in the Table-3 reflect that lag 1 is suitable for the sample size.

Cointegration results
After checking the unit root properties and identifying the appropriate lag length of the variables, the bounds test of cointegration is conducted in order to analyze the long-run relationship between the variables. Table 4 presents the calculated F-statistics and critical values for bound test. As can be seen from the table the calculated F-statistic = 5.895 and is greater than the upper bound critical value of 6.36 provided by Pesaran et al. (2001) at the 2.5% level of significance. Accordingly, the null hypothesis of no long-run relationship is rejected. Therefore, the conclusion is that there is cointegration or a long-run relationship between RGDPg, TOUR, and the other control variable included in the model. After confirming the existence of a long-run relationship between the four variables included in the growth model, the long-run equation is estimated. Table 5 presents the results of the estimated long-run coefficients of the estimated growth model. The coefficients of all of the variables included in the model (except ENROg) are statistically significant at 1%. Tourism receipt to export ratio (TOUR) and growth in gross capital formation (RGCFg) positively and significantly affects economic growth at 1% and 5% respectively. As expected, the recurrent drought dummy variable (DROU) and leadership change dummy (LEDCH) negatively and significantly affects real per capita GDP growth at 1% level of significance. The coefficient of tourism receipt to export ratio (TOUR) is about 0.964, suggesting that as the ratio of tourism receipt to export increases by 1percent, economic growth will increase by about 0.964 percent per year. This finding is consistent with the argument of tourism-led growth hypothesis (TLGH) which argued that international tourism is a source of economic growth through its contribution to foreign exchange revenues, stimulating greater investments in infrastructure, and creating employment opportunities that ultimately improve the living standard of the citizens of a country (Balaguer & Cantavella-J, 2002 ;UNWTo, 2017 andRisso &Bonapace, 2009). Since tourism is part of the current account balance, it can improve the persistent current deficit in Ethiopia and promote economic growth, if the current marketing and promotion strategy is further improved.

Short run dynamics and ECM
The error-correction model presented in Table 6 shows that the one period lagged error-correction term is within acceptable levels. As expected, its magnitude is less than one and has a negative sign. Specifically, the coefficient of ECT which measures the speed of adjustment is about -0.895 and is statistically significant at 1%. This implies that approximately 89.5% of shocks from the previous period are corrected within the current year. This result implies that economic growth converges to its long-run equilibrium by 89.5% in one year with the speed of adjustment via the channel of tourism receipt and other control variables. As clearly indicated in Table 6, the short run effect of the explanatory variables is similar to the long run effect. The results show that tourism has a statistically significant positive impact on economic growth. This signifies that tourism acts as an engine of economic growth in the short-short-run as well.

Diagnostic test
The soundness of the results is dependent on the fit and stability of the model. Therefore table 7 summarizes the results of the various diagnostic and stability tests of the TLGH model. Based on the Breusch-Godfrey serial correlation LM Test, there is no serial correlation in the model. AS proved by the Jarque-Bera normality test, the residuals are normally distributed in the model. The model also appears not be heteroscedastic as it passes the Breusch-Pagan-Godfrey heteroskedasticity tests. The Ramsey RESET test results also confirm that the model is correctly specified and stable. Further, the level of multicollinearity is also tested in terms of Variance Inflation Factor (VIF). Accordingly, the VIF (centered) values reported in Table 8 clearly shows that there is no multicollineariy problem.

Pairwise Granger causality results
In order to identify the direction of causality between international tourism receipt and economic growth in Ethiopia, pair wise Granger-causality test was conducted. The test result indicated that causality runs from tourism receipt to economic growth. At a lag length of one and two there is a uni-directional Granger causal relationship from tourism receipt to economic growth. This implies that the tourism-Led economic growth hypothesis is valid in Ethiopia.

Conclusion and Recommendation
The main objective of this study was to validate the Tourism-Led growth hypothesis (TLGH) in Ethiopia over the period 1991-2018. The cointegration test result indicates that there is a long run relationship between tourism receipt, economic growth and other control variables. The results of the pairwise Granger causality test also predicted unidirectional causality running from tourism receipt to economic growth. This validates that the tourism-led growth hypothesis works for Ethiopia. This result is consistent with the finding of Cantavella-Jorda (2002) for Spain, Grillon (2013) for Dominican, Ay, Kartal, & Arslan (2015) for Turkey, Ohlan, (2017) for India, Shih and Do (2016) for Vietnam. Hence, appropriate resources and efforts should be directed towards promoting and developing the untapped tourism industry of the country to augment economic growth. Besides, any intervention related to the sector has to be integrated into the broad national development plan and other universal value chains. In addition, a comprehensive primary survey on the bottlenecks of the sector should be undertaken so as to solve the real challenges of the sector at the grass root level. This enables to further maximize the effect of tourism on the socio-economic development of the country.