Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Edwin Modelling International Tourism Demand For Zimbabwe 20081101


Published on

Published in: Travel, Business
  • Be the first to comment

Edwin Modelling International Tourism Demand For Zimbabwe 20081101

  1. 1. Modeling international tourism demand for Zimbabwe <ul><li>Edwin Muchapondwa and Obert Pimhidzai </li></ul><ul><li>School of Economics, University of Cape Town, South Africa </li></ul><ul><li>4 November 2008 </li></ul>
  2. 2. Presentation Plan <ul><li>introduction </li></ul><ul><li>background to international tourism in Zimbabwe </li></ul><ul><li>theoretical foundations for international tourism demand studies </li></ul><ul><li>the methodology used in this investigation </li></ul><ul><li>the empirical results </li></ul><ul><li>conclusion and policy implications </li></ul>
  3. 3. Introduction <ul><li>in Africa, tourism is (potentially) a major source of foreign currency; significant contributor to employment; key sector for achieving shared economic growth and poverty alleviation (Mitchell & Ashley, 2006; World Bank, 2006) </li></ul><ul><li>authorities need to know the key drivers of tourism demand especially the determinants of international tourism demand as it tends to dominate tourism in Africa </li></ul><ul><li>sadly, there has been little research on international tourism demand in Africa (Xiao & Smith, 2006; Rogerson, 2007) </li></ul><ul><li>this paper investigates the determinants of international tourism demand for Zimbabwe </li></ul>
  4. 4. Significance of Zimbabwe tourism <ul><li>Zimbabwe hosts scenery, biodiversity, warm climate, Victoria Falls, Zimbabwe ruins and man-made Lake Kariba </li></ul><ul><li>tourism has been important with international tourism receipts accounting for 7.3% of total exports in 1997 </li></ul><ul><li>since then international tourism receipts have been declining reaching as low as 3.3% of total exports in 2003 </li></ul><ul><li>agricultural and mining exports have also been declining leaving tourism as the only major source of foreign currency </li></ul><ul><li>agriculture and mining can only improve from the medium term due to the structural changes needed to revive them </li></ul><ul><li>rejuvenation of tourism could revive the economy sooner </li></ul>
  5. 5. Trends in international arrivals (1) <ul><li>Source: UNWTO (2006) </li></ul>
  6. 6. Trends in international arrivals (2) <ul><li>during 1994-2005, international tourism arrivals to the African continent were rising </li></ul><ul><li>for Zimbabwe, international tourist arrivals rose rapidly since economic liberalisation in 1990 </li></ul><ul><li>the highest international tourist arrivals growth rate was in 1995, when Zimbabwe hosted the All-Africa Games </li></ul><ul><li>in fact, international arrivals to Zimbabwe were growing at a faster rate than the continent until 1997 </li></ul><ul><li>average international tourist arrivals growth for 1990-2000 (12%) was above that of the continent (6.4%) (UNWTO, 2006) </li></ul>
  7. 7. Events affecting tourism (1) <ul><li>Zimbabwe’s tourism fortunes started changing in 2000 due to three factors: </li></ul><ul><ul><li>referendum on the controversial draft new constitution which failed </li></ul></ul><ul><ul><li>the “fast-track” land reform implemented immediately after defeat </li></ul></ul><ul><ul><li>parliamentary election in which the opposition won a number of seats </li></ul></ul><ul><li>tourist arrivals have generally continued to fall save in 2001/2 </li></ul><ul><li>in June 2001 there was a total solar eclipse (also) visible from Zimbabwe attracting a 12.7% rise in number of visitors </li></ul><ul><li>2002 was another odd year due to a presidential election: </li></ul><ul><ul><li>the opposition alleged the election was stolen by the ruling party </li></ul></ul><ul><ul><li>the West imposed sanctions on key figures in the ruling party </li></ul></ul>
  8. 8. Events affecting tourism (2) <ul><li>still in 2002, Zimbabwe hosted the semi-finals and finals for Miss Malaika and another (partial) solar eclipse in December </li></ul><ul><li>international arrivals to Zimbabwe declined at an average annual rate of 4.5% during 2000-2005 </li></ul><ul><li>in contrast, international tourist arrivals to the continent as a whole grew by 5.7% between 2000 and 2005 </li></ul><ul><li>of course, general slowdown for Zimbabwe and the continent could also be attributed to the September 11 terror attacks in the United States </li></ul><ul><li>Market share trends and disaggregated data corroborate the poor performance of international tourism in Zimbabwe </li></ul>
  9. 9. Relative market share trends <ul><li>Source: UNWTO (2006) </li></ul>
  10. 10. Trends in monthly holiday visits <ul><li>Source: Zimbabwe Central Statistical Office (various years) </li></ul>
  11. 11. Events to capture in modeling <ul><li>generally, there are seasonality peaks in international holiday visits to Zimbabwe, with massive inflows every December </li></ul><ul><li>the sharp peak observed in June 2001 relates to the influx of visitors to witness the total solar eclipse from Zimbabwe </li></ul><ul><li>a trough is observed for the September 11 2001 terror attacks </li></ul><ul><li>another peak is observed in September 2002, the month in which the country hosted the semi-finals for the Miss Malaika </li></ul><ul><li>the final was held in December, the same month as another solar eclipse that could be partially viewed from Zimbabwe </li></ul><ul><li>a trough is observed for the March 2002 presidential election </li></ul><ul><li>modelling should also pay attention to these specific events </li></ul>
  12. 12. Determinants of foreign tourism <ul><li>several studies review determinants of international tourism demand (Crouch, 1994a, 1994b; Lim, 1997; Li et al., 2005) </li></ul><ul><li>in fact, the standard theory of demand forms the theoretical foundations for modelling international tourism demand </li></ul><ul><li>analysis is carried out using single-equation models in which explanatory variables are based on standard demand theory </li></ul><ul><li>this has been criticised for ignoring the interdependence of budget allocations to different destinations and the ad hoc selection of variables (Li et al., 2005) </li></ul><ul><li>an alternative approach would be a systems one e.g. the Almost Ideal Demand Systems (AIDS) </li></ul>
  13. 13. Determinants of foreign tourism <ul><li>nonetheless, the single estimation framework remains the more dominant approach hence its use in this study </li></ul><ul><li>in this framework, the key determinants of international tourism demand are price, income and taste formation </li></ul><ul><li>the magnitude of their influence varies widely and meta-analysis shows that the influence depends on the destination hence the need for destination specific studies (Crouch, 1995) </li></ul><ul><li>the price of tourism includes the cost of commodities and access to tourism facilities in the destination country </li></ul><ul><li>exchange-rate-adjusted relative consumer prices of the destination and origin are normally used </li></ul>
  14. 14. Determinants of foreign tourism <ul><li>some studies model the exchange rate separately a rguing that tourists have up-to-date information on foreign exchange rates than commodity prices hence the exchange rate is likely to have a separate impact from that of commodity prices </li></ul><ul><li>“ wanderlust” destinations are price inelastic while “sunlust” destinations are price elastic (Crouch, 1994a) </li></ul><ul><li>transport cost is also an element of price but empirical models typically include it separately </li></ul><ul><li>a general finding among a few studies that use a proxy for transport cost is that it is insignificant (Martin & Witt, 1987; Witt & Witt, 1992; Kim & Song, 1998; Li et al., 2005) </li></ul>
  15. 15. Determinants of foreign tourism <ul><li>income is one of the most important determinants of international tourism demand </li></ul><ul><li>the general finding is that international tourism is a luxury good (Crouch, 1994b) </li></ul><ul><li>Gauci et al. (2004) attribute the slowdown in global tourism, prior to 2001, to a slowdown in global economic activity </li></ul><ul><li>estimated income elasticities range from 1.0 to 2.0 vindicating that international tourism is a luxury good </li></ul><ul><li>this implies that international tourism is pro-cyclical hence countries in which tourism makes a significant contribution to the economy are the worst affected by recessions </li></ul>
  16. 16. Determinants of foreign tourism <ul><li>international tourism is subject to changes in tastes over time and this has been generally captured by using a time trend </li></ul><ul><li>the time trend is criticised because it assumes a constant rate of change in tastes and it also captures a host of other factors clouding its precise interpretation </li></ul><ul><li>an alternative measure of taste formation, used especially in dynamic models, is the lagged value of the international tourism demand variable </li></ul><ul><li>this is based on the notion that once tourists like a place, they will return to it and will surely recommend it to family, friends and colleagues; may be more effective than advertising </li></ul>
  17. 17. Determinants of foreign tourism <ul><li>improving the quality of facilities and standards of hospitality to tourists can help attract more visitors </li></ul><ul><li>price of complements or substitutes affects international tourism demand </li></ul><ul><li>just like the demand for any commodity, international tourism is subject to both positive and negative shocks hence specific events and political risk are also important determinants of international tourism demand </li></ul><ul><li>high political risk discourages visits to that destination, except by the adventurous </li></ul>
  18. 18. Determinants of Zim tourism <ul><li>this paper focuses on international holiday visits to Zimbabwe instead of total international visits </li></ul><ul><li>this distinction is made because the explanatory variables differ by purpose of visit and influence may also vary by purpose of visit </li></ul><ul><ul><li>a destination’s economic performance should certainly affect business visitors while its effect on holiday visitors may not be so obvious </li></ul></ul><ul><ul><li>price variability is likely to have a greater impact on business visitors than holiday visitors in a “wanderlust” destination </li></ul></ul><ul><li>this paper uses a single-equation model </li></ul><ul><ul><li>appropriate for study of demand from all countries rather than from particular countries of origin </li></ul></ul><ul><ul><li>we have no access to monthly data disaggregated by both purpose of visit and country of origin for tourists to Zimbabwe </li></ul></ul>
  19. 19. Determinants of Zim tourism <ul><li>the study uses monthly time series data from 1998 to 2005 mainly from IMF (2008), Zimbabwe Central Statistical Office (various years) </li></ul><ul><li>the dependent variable is international holiday visits to Zimbabwe </li></ul><ul><li>the explanatory variable are taste formation, domestic prices, foreign prices, income and transport costs </li></ul><ul><ul><li>proxies are used (the lagged dependent variable; Zimbabwe’s CPI (2000=100) divided by the parallel market exchange rate; South African CPI (2000=100) divided by the parallel market exchange rate; monthly United States unemployment rate and international oil prices) </li></ul></ul><ul><ul><li>dummies are included to capture the solar eclipse, Miss Malaika, September 11 terror attacks, the festive season and post-2000 political instability </li></ul></ul>
  20. 20. Estimation technique <ul><li>the autoregressive distributed lag (ARDL) approach to cointegration is used to estimate the international tourism demand function for Zimbabwe </li></ul><ul><ul><li>this approach allows the estimation of both short-run and long run elasticities of international tourism demand </li></ul></ul><ul><ul><li>it offers explicit tests for identifying a unique cointegration vector rather than assuming it </li></ul></ul><ul><ul><li>the ARDL is applicable even when some explanatory variables are endogenous </li></ul></ul><ul><ul><li>the existence of a long run relationship is independent of the variables’ order of integration </li></ul></ul><ul><li>the paper opts for a log-linear model as in (Crouch, 1992, 1994b; Vanegas & Croes 2000; Tan et al., 2002; Song et al., 2003) </li></ul>
  21. 21. The ARDL estimation technique <ul><li>the ARDL(p,q,r,s,m) is given by equation (1) where the disturbances are assumed to be serially uncorrelated </li></ul><ul><li>can also be presented as in equation (2) with lag operators </li></ul><ul><li>the long run relationship is given by equation (3) </li></ul><ul><li>the long run coefficients from the ARDL are obtained as </li></ul>
  22. 22. The ARDL estimation technique <ul><li>like the other single-equation cointegration approaches, the ARDL is valid only when there is a unique cointegration vector </li></ul><ul><li>Pesaran et al. (1996) suggest a three step procedure of testing for the existence of a unique cointegration vector </li></ul><ul><li>first, estimate the error correction model (ECM) in equation (5) where the lag length (p) is such that the error term is not serially correlated </li></ul><ul><li>second, test H0: with the critical values provided in Pesaran et al. (2001) </li></ul>
  23. 23. The ARDL estimation technique <ul><ul><li>if F<FL no cointegration; if F>LU cointegration exists; if FL<F<FU the test is inconclusive and the order of integration of the underlying variable must be known to proceed further </li></ul></ul><ul><li>third, re-estimate the ECM in equation (5) several times with each of the explanatory variables (Z, S, U and O) as dependent variables and then test for the joint significance of the lagged level coefficients as in the second step </li></ul><ul><ul><li>t he number of significant F-statistics shows the number of cointegrating vectors and for us to proceed with estimating the ARDL model given in equation (1) we require that only one F-statistic be significant </li></ul></ul>
  24. 24. The bounds testing approach <ul><li>at the 5% level of significance, there is only one cointegrating vector corresponding to the Visits equation </li></ul><ul><li>given the unique cointegration vector, estimation can then proceed using the ARDL model given in equation (1) </li></ul>Dependant Variable Visits Zimprice Saprice Oilprice Unemploy F-Statistic 4.2627** 2.5086 2.0255 0.60253 2.2149 **Significant at 5% level of significance The critical values for case of unrestricted intercept and no trend for k=5 and T=80 are Lower Bound I(0)= 2.878; Upper Bound I(1)=4.015 using Narayan (2005) critical values
  25. 25. Estimation of the ARDL <ul><li>in estimating equation (1), we had two specifications capturing the September 11 terror attacks differently </li></ul><ul><ul><li>the first specification includes a dummy for the period from September 2001 onwards to capture a permanent effect on international holiday visits </li></ul></ul><ul><ul><li>the second specification includes a dummy for the immediate period following the terror attacks i.e. September to December 2001 to capture a temporary effect on international holiday visits </li></ul></ul><ul><li>based on the AIC, the ARDL(1,1,0,0,0) and ARDL(1,1,1,0,0) were selected and the LM test shows no serial correlation </li></ul><ul><li>the estimated long run coefficients are computed using the formula in equation (4) </li></ul>
  26. 26. Results (Model 1) (Model 2) ARDL(1,1,0,0,0) Estimates Long-run Elasticities ARDL(1,1,1,0,0) Estimates Long-run Elasticities Variable Coefficient Std Error Coefficient Std Error Coefficient Std Error Coefficient Std Error Visits(-1) 0.2183** 0.0848 0.1854** 0.0807 Zimprice -0.6056*** 0.2167 -0.1481 0.1677 -0.5159** 0.2109 -0.1452 0.1453 Zimprice (-1) 0.4898** 0.2361 0.3977* 0.2263 Saprice 0.2896 0.2488 0.3704 0.3182 -1.0045 0.8295 0.03998 0.3016 Saprice(-1) - - 1.037 0.7905 Unemploy 0.04098 0.0987 0.0524 0.1255 -0.1123* 0.0663 -0.1378* 0.0794 Oilprice -0.2620 0.1822 -0.3351 0.2337 -0.4454*** 0.1659 -0.5467*** 0.1968 Constant 9.023*** 1.2184 11.542*** 0.9814 11.2843*** 1.3315 13.8518*** 0.6984 Eclipse 1.7210*** 0.3417 2.2015*** 0.5202 1.8117*** 0.3184 2.2239*** 0.8778 Malaika 1.0186*** 0.2509 1.303*** 0.3307 0.8145*** 0.2419 0.9998*** 0.3083 Political 0.04115 0.1572 0.05264 0.2012 0.0727 0.1511 0.0893 0.1855 Nov -0.0371 0.1287 -0.04745 0.1646 -0.0155 0.1237 -0.0190 0.1517 Dec 0.7178*** 0.1321 0.9182*** 0.2036 0.6874*** 0.1248 0.8439*** 0.1851 Sept11 - - -0.7359*** 0.2165 -0.9033*** 0.2519 Sept01 -0.3625* 0.2089 -0.4637* 0.2526 - - Adjusted R2 0.5679 0.6035 F –stat 15.2205[p-value 0.000] 11.8891[p-value 0.000] Serial Correlation LM Test 7.6646[p-value 0.811] 7.2973[p-value 0.8371] Heteroscedasticity LM Test 0.96432[p-value 0.326] 1.5920[p-value 0.207] ***Significant at 1% level of significance (L.O.S); **Significant at 5% LOS; *Significant at 10% L.O.S
  27. 27. Analysis of results <ul><li>in both specifications, taste formation, domestic price, the festive season, the solar eclipse, Miss Malaika and September 11 terror attacks are significant </li></ul><ul><li>furthermore, changes in income and transport costs are significant in model 2, which has higher explanatory power </li></ul><ul><li>doubling current visits increases next year visits by 18.5% </li></ul><ul><ul><li>there is need to improve the tourist experience for a larger proportion of visitors to return or recommend the country to others </li></ul></ul><ul><ul><li>high quality of infrastructure and tourist facilities is important for attracting future visitors </li></ul></ul><ul><ul><li>Kester (2003) observes that deficiencies in facilities, lack of image and poor perceptions are among the major obstacles to arrivals in Africa </li></ul></ul>
  28. 28. Analysis of results <ul><li>the immediate impact of solar eclipse and Miss Malaika was to increase average monthly holiday visits by 512% & 126% </li></ul><ul><li>the immediate effect of a 10% domestic price increase is a decline in international holiday visits of 5.16% </li></ul><ul><ul><li>since the long run price elasticities are insignificant, holiday visits to Zimbabwe are insensitive to domestic and foreign prices; a result consistent with a wanderlust destination (e.g. the Victoria Falls) </li></ul></ul><ul><li>transport costs and changes in income have significant long run elasticities of -0.55% and 13.8% respectively </li></ul><ul><li>there is evidence of seasonality in holiday visits to Zimbabwe; </li></ul><ul><ul><li>in recent times, this could be capturing Zimbabweans in the diaspora visiting friends and relatives during the festive season every December </li></ul></ul>
  29. 29. Analysis of results <ul><li>the effect of the September 11 attacks was to reduce holiday visits to Zimbabwe both in the short run and long run </li></ul><ul><ul><li>the September 11 attacks dummy also happens to capture a period of increased political instability in the country thereby making it difficult to separate out the effects of these two instability phenomena </li></ul></ul><ul><li>the error correction term is significant and has the correct sign </li></ul><ul><li>model 1 estimates that 78.2% correction in the deviation from the long run equilibrium is made in the first month while model 2 puts the correction at 81.7% </li></ul><ul><li>the results confirm the existence of a long run relationship and indicate a fairly quick speed adjustment </li></ul>
  30. 30. Conclusion <ul><li>this paper investigated the determinants of international tourism demand for Zimbabwe </li></ul><ul><li>taste formation, transport costs, income, events such as the solar eclipse, Miss Malaika and 9/11 attacks are significant </li></ul><ul><li>this should guide the priority areas for policy intervention </li></ul><ul><ul><li>enhance the quality of services to tourists in order to reinforce taste formation to attract more international tourists to Zimbabwe </li></ul></ul><ul><ul><li>improve tourism infrastructure in order to reduce travel costs </li></ul></ul><ul><ul><li>promote pleasant events to shape international tourism demand for the country </li></ul></ul><ul><li>the results confirm the existence of a long-run equilibrium and indicate a fairly quick speed of adjustment </li></ul>