1 views

Uploaded by Bicăjanu Bianca

jhbh

- 19016055 MagzID Case
- Introduction to Forecasting Presentation
- BDMDM
- 2144 - Geography of Tourism Course Outline - Winter 2012
- Statistical Analyses of Multivariate Time Series Data With Application to Compacting Effects on Soil Chemical and Biological Properties in Forestry
- Box Jenkins
- 100.Defining a Museum
- Tourism Timeline
- Lecture on ARIMA Using Minitbab
- Study on the Diversity and Predictors of Internal and International Migration in Bangladesh a Panel Data Approach
- Zaur Imrani
- Domestic and International Tourism in a Globalized World
- DRTA Prediction Log
- An Introduction to Multivariate Statistics
- VUSSC Tour Guiding
- Policy Paper on Tourism
- Ma Thematic Neural Network
- Skill-Importance-in-Volleyball.pdf
- International Tourism Forum Programme "Development of wine tourism in the countries of the Danube River Basin"
- TASK Compass 5

You are on page 1of 13

**ISSN 1549-3644
**

© 2012 Science Publications

**Modeling Tourist Arrivals Using Time Series Analysis: Evidence From Australia
**

1

Gurudeo AnandTularam,

2

Victor Siew Howe Wong and 3Seyed AbdelhamidShobeiriNejad

1

Mathematics and Statistics, Science Environment Engineering and Technology (ENV),

2

Finance, School of Business, Finance and Economics

3

Science Environment Engineering and Technology (ENV),

Environment Future Centre, Griffith University, Australia

**Abstract: Australian tourism has a logistic trend as Butler’s model shows. The stagnation has not
**

been reached so opportunities exist to increase tourism. The logistic model predicts 7.2 million tourists

in 2015 but time series models of ARIMA and VAR improve the prediction and explain the data. The

ARIMA (2, 2, 2) fits well while the VAR lead to Granger causalities between the three data sets. A

regression model (R2 = 0.99) using Australian tourist arrival as a function of Europe and World

arrivals allowed to further understand the Granger causality. The ARIMA model predicts tourist

numbers to be approximately 6 million in 2015. The VAR technique allowed impulse response

analysis as well. A two-way causality between the tourist in Australia, Europe and World exists, while

impulse response indicated different effect patterns, where tourist arrivals increase in the first period

and declines in the second period but experience seasonal fluctuations in the third period. The

strongest causalities in were period 1 between World and Europe; period 2-a one-way causality from

Australia to World and period 3-a two-way causalities between Australia, Europe and World. The

impulse responses results were aligned with the Butler theory.

**Keywords: ARIMA, auto-regression, mathematical modeling, butler model, tourism modeling,
**

granger causality, impulse response, variance decomposition model, logistic

regression, time series

**INTRODUCTION encourage researches to investigate on number of tourist
**

arrivals and attempt to make a more accurate prediction

The number of tourists has significant impacts on for future planning.

global economy. These impacts can be counted as both The number of short-term international arrivals to

of negative and positive effects. The most important Australia has significantly grown over the last 6 decades

effect of tourism on economy can be known as number from about 60,000 to approximately 6 million arrivals

of changes on supply and demand chain in the per year presently. From 1956 to 2009 the Australian

destination which is the host of tourists. Tourists’ tourism industry has been growing steadily following the

demand or simply tourist consumption contributes to trend of worldwide tourism and increasing over time.

GDP, increasing the employment rate, making new However, in recent times the number of tourist arrivals

source of revenue for local people, private and public has starting slowing down and this may be just evident in

sectors and destination’s government and so on. For Fig. 1. However when compared with actual numbers

instance, the consumption international and domestic Australian tourism is significantly less than either Europe

tourism was approximately AUD$ 95.653 billion from or the World as shown in Fig. 2.

July 2010 to year-ended June 2011. It shows that the This Fig. 1 shows the trend of changes in tourist

tourism share of GDP and total employment rate were 2.5 arrivals to both Australia and the world. The above

and 4.5%, respectively. Moreover, the share of tourism model gives this chance to look at the trend of

industry in export was 8% 0f total Australian export in growth of both Australia and the world international

2010-11 (ABS, 2011). This significant impact is enough to arrivals at the same time.

Corresponding Author: Gurudeo AnandTularam, Senior Lecturer in Mathematics and Statistics in School of Environment,

Environment Future Centre, Griffith University, Australia

348

. & Stat. J. 8 (3): 348-360. 1: Short-Term International visitors Arrivals 1956-2010 compared to the World (Based on ABS. 2: Low levels of tourism in Australia compared to Europe and the World 349 . 2012 Fig. Math. 2011) Fig.

thus not tourism are costly and require a long-term approach and listed now as a priority for potential tourists in that the planning in order to develop relevant infrastructures rate of growth of the tourist numbers gradually (Song and Turner. In addition. representing an exponential from an origin to a destination. by aggregation of the concept rate m (as before). 2006). & Stat. This framework is used to (Nejad and Tularam. For tourism facilities in both public and private areas become easier comparison in the next graph the trend of well developed while in the fourth consolidation stage. It is then important for our models Exploration. Butler terms of infrastructure and like. as the world’s tourism cycle. involvement. tourists number continue to increase but the destination Australian investments on projects facilitating now becomes well known and visited by many. 2: to the destination. When t0 = 0 the model may usually given by the total number of tourist arrivals be written as xt = x0mt. The model is given as: the Tourist Area Cycle Theory (Nejad and Tularam.. 1 βt = Tularam and Illahee. Development. where the negative effects model tourist arrivals. 2010). The involvement stage is a time period in which tourist numbers increase mainly due to X xt = (2) an increased awareness of the destination as a tourist base. X development. and discrepancy of tourist numbers at previous tourist arrivals can be linked in both Area Cycle as well time t as a ratio of maximum tourist numbers X. namely. both public and private organizations budgeting and monetary positions (Li et al. Consolidation to accurately predict arrival numbers at it is critical for and Stagnation over time. The first stage of exploration tourism is not recognized The solution to the differential equation developed as an economic activity in that only a few people travel above can be given by Eq. exploration. consolidation and stagnation (Fig. organizations often rely on mathematical models (Song The evolution of tourist area’s stages including: and Turner. 3). 2006). 2010).. A Variance decomposition 1 + e − m( t − a ) model (VAR) can be used to study Granger Causality and impulse response for furthering understanding of A represents time at half the carrying capacity and tourism numbers worldwide. Similar analyses have been done in other areas (Tularam. Involvement. In fact. The VAR model allows an X being the maximum tourist capacity. In the first part of this study a and leveling or stagnation period are both captured by mathematical a logistic model is developed based on the model. a stagnation stage is reached of tourism to Australian economy is also well accepted and to cater for the arrivals much planning and when potentially all tourists know the destination well development is indeed necessary for future needs in including the facilities on offer. examination of the nature of integration of the tourism This model demonstrates that the tourist numbers market when compared with Europe and the World growth at time t+1 is proportional to constant growth tourism numbers. the initial value tourist number is common variable to measure tourism demand and greater than zero (x0 >0). However in this study an integrated time series xt = βt ARIMA model is considered together with a linear X multivariate Vector Auto-Regression (VAR) model is also for a more detailed study of drivers of Australian Where: tourism in terms of Europe and the World. the number of before). X − xt The growth is proportional to Eq. A more comprehensive model is developed by Theory (TACT) is a conceptual framework proposed by combining Malthus’ law and the Verhulst assumption Butler (1980) for modeling. In order to prepare for (1980) argued two possibilities after stagnation stage the future. tourist number at last time t (as of TACT and the study in this study. 2005). 2010. 2012 This Fig. companies and is reached. The contribution and impact declines until finally. 1: X Literature review: Butler (1980) theoretically studied the conceptual cycle of tourist evolution and identified X − xt xɺ t = m x t (1) five stages. 8 (3): 348-360. namely. Australian governments. 1 + e − m(t − a ) 350 . The Tourist Area Cycle growth. 2 shows the trend of changes in By the start of the development stage. the destination’s international arrivals to both Europe and the world. Mathematical model:The initial tourist model may be According to Song and Li (2008). 2010). J. Math. tourist arrivals or written as xt = x0 em (t-t0) where x is positive and when t tourist population at a particular time is the most = 0 given t0 = 0. rejuvenation or decline. the international arrival is shown.

3). Math. & Stat.594. 1 + e − m( t − a ) X prediction is 5. 2010) The constant parameter ‘a’ and the maximum the logistic model and the actual result for 2010 at capacity X are both assumed to be fixed for a updated in this study is 5. given range of 5. 2012 Fig. the authors have updated the data and general given that cyclones. 4: Fitted growth rate and observed data (Nejad and Tularam.400 (within the 95% particularly destination. 8 (3): 348-360.500 arrivals using within 95% ran based on logistic model. 2010). reasonably good in 2010 paper.878. As shown. The new model in this study 1 xt includes the “actual arrivals” in 2010 and the new βt = the model may be written as = βt . Equation 2 (Fig. however.692.065) where β t is proportion (percentage) of tourist at time t to suggesting the theoretical model predictions are the capacity (Nejad and Tularam. 351 .. 3: The evolution of a tourist area (Adapted from Nejad and Tularam. floods and financial crises recalculated the predictions based on the model in have all impacted Queensland in this time period.594. 2010) Fig. It was noted that the Nejad and This Table 1 shows the prediction of tourist arrivals Tularam (2010) prediction was 5.500). Based on the somewhat overestimates. J.824.932.886-5.709 (5.

4: 2012 5.745 6.160.612 6. q) model can be written using 2010 5.152).883 5. 2013 5. − ϕ p Lp ) (4) This Fig.890.871.id(0. 2) contains 2 autoregressive (p) parameters and 2 moving average (q) parameters Table 2: New logistic model predictions of tourist arrivals to Australia computed for the series differenced twice. 4) fits the data well with an R. Europe and • Stagnation (2008-dependent to the world’s rate) World.412 5.656 6. where a structural break The theoretical model’s 95% range predictions due could be identified (financial crises).068. Math.136. d.237. J. 2010). Tularam and Illahee. 6: model: The common time series analysis may be conducted using autoregressive moving average models 1 b 12 y t b 1 0 c 1 1 c 1 2 y t − 1 ε yt (6) b = + + known as ARIMA (p.325.144 (0. Period 2 (1988-2001) consisted of the This Table 2 shows the prediction of tourist arrivals September 11.130 ∆ d Yt (1 − ϕ1 L − ϕ 2 L2 − . z t = b 20 − b 21 y t + c 21 y t − 1 + c 22 z t − 1 + ε zt Model of tourism evolution: with εit ~ i.400 autoregressive parameters (p). The updated data confirms the stages as follows: cause y2. y2 then y1 is said to terms of the Butler model and identified the year for Granger-cause y2. the structure • Involvement (1973-74 to 1983-84) allows Granger Causality tests to be conducted that may • Development (1983-84 to 2003-04) indicate whether there is one or two-way Granger • Consolidation (2003-04 to 2008) Causality between tourist arrivals in Australia.214 6. As A more in depth analysis of the tourism data is an example.075. ARIMA (p. In particular. 2010).943.932.85). In general Year Lower bound Point estimate Upper bound form. in October 1987.757.157 6. or group of variables.137-0.012.827.082.801 5. a general form of a two-variable VAR (1. The structure of the VAR model is such that it provides information about a variable’s 6453205 xt = −0.d(0.199 differencing (d) and moving average parameters (q).824.εz) = 0 in matrix form ARIMA-autoregressive integrated moving average the above can be written as Eq. or group of variables.755 5.205 6..040. A 95% confidence interval shows a good fit. & Stat.878. (5) 2010.709 5.996. The notion of Granger Causality does not imply true Causality in that it only implies forecasting ability • Exploration (1973 to 1974) (Roca and Tularam.159) with the international arrivals growth rate of 0.935..179 6. otherwise it is said to fail to Granger- each stage. for example.646 For example ARIMA (2.129-6.. 2. Tularam et al..890.598 6.695 5. 2012 Table 1: Predictions of tourist arrivals to Australia by Nejad and autoregressive and moving average parameters. Tularam (2010) ARIMA models explicitly include differencing Year Lower bound Point estimate Upper bound 2010 5.007.991.189.683 p q 2013 5.. 2001 attacks in the US and Period 3 (2002- within 95% rang for-squared 0. 1994) including 21 1 z t b 20 c 21 c 22 z t − 1 ε zt 352 .144054 ( t −1993.948.991. 2012. 2010). 2009) was the Global Financial Crisis (GFC) in 2008.799 (Tularam. Variance decomposition-granger causality and impulse Using the turning point as the half of the carrying capacity response analyses: The VAR modeling process allows occurring around in the middle of 1993 (1993.581.. The calculated carrying capacity u = The usual error term .886 5. − θ p L )u t 2 q regression international arrivals to Australia.080. y1 is found to be helpful for predicting another Nejad and Tularam (2010) analyzed their results in variable.023 2014 5.930 intercept form or with lag notation (L) as follows Eq. 2012).85) (3) forecasting ability for other variables (Roca et al. 5: ARIMA and VAR methods.065 2011 5.884.σ2) for this model is 6. d.440 5.949. 1) conducted in this study in terms of time series using with k = 2 are shown in Eq.281 6.998.817.996.022 2014 6.991. The three types of parameters are: the 2011 5.580 5. 3 for international Australian tourist arrivals Period 1 (1950-1987) consisted of the stock market crash are shown in Table 2. σε2i ) and cov (εy.117.205 (6. similar to those used in y t = b 10 − b 12 z t + c 11 y t − 1 + c 12 z t − 1 + ε yt other studies (Roca and Tularam.042. 1+ e Granger (1969) argued that if a variable.344 5.ui~i. the more specific examinations of the relationships between mathematical model can now be written as: the three data series.177 6. To study impacts over different periods time series data was divided into sub-periods.946...453. q) (Box et al.348 6. using Eq. the number of 2012 5.042.i. 4 shows the model results of logistic = (1 − θ1 L − θ 2 L − .900 6. Where: The logistic model (Fig. 8 (3): 348-360.622 Yt = ∑ϕ Y i =1 i t −i + u t + ∑ θ ju t − j j=1 2015 6. Y = The time series data squared value of 0.

006(World) + εt. we can construct an impulse graph for ∆ 2 Yt (1 − ϕ1L − ϕ2 L2 ) = (1 − θ1L − θ2 L2 )u t (8) each variable as the response to a certain shock. it is predictions made up to year 2015.as fitted imposed.130 and MAE = 0. 7 shows the 95% error bound. tourist numbers. q = 2 and d = 2 Eq. the y-equation. BIC another. J. customary to set the size of the shock equal to its This Table 4 shows the prediction of tourist arrivals standard deviation.809-5. another analysis was undertaken using the multiple Where: regression method.h = ( σii ) −1/ 2 ( e′ ∑ x e ) . For our VAR (1.2) model • The impulse response of z in response to a shock in parameters with significances ME = 0.. the z-equation. Error bounds are how.353. RMSE = the y-equation.810 constant terms may be ignored: Yt = ψ (L) xt If (5. Thus. arrivals based on the ARIMA model. The 80 and 95% bounds are shown Figure for the difference in the timing of the effects. 7. Every structural shock affects every other variable. meaningful interpretations of the initial using the multivariate analysis.0036.27 and BIC = -46.076 The forecast using the ARIMA (2.692. The equation is expressed as: Australia = - generalized method is invariant to re-ordering of the 2. Finally. 8: countries. 1) example. the reaction to a shock of unit size. 5. interest may be in: The results of the Residuals. Europe and the World.2) is given for The impulse response functions will all have the 80 and 95% accuracy limits for the next ten years in same general shape and if the system is stable. There will be a set. It is noted that Granger Causality simply refers to (marginal) RESULTS predictive content. εz estimated coefficients for the ARIMA mode are present • The impulse response of y in response to a shock in in Table 3 suggesting a close fit with data. Figure 6 shows the ARIMA predicted number of Consider now the moving average representation of Australian arrivals up to 2015. 7: MR-multivariate analysis: To further cross check the nature of the predictions and examine the kind of Ψ ij. 2012 The test of the joint hypothesis that none of the z’s impact response of each variable to shocks to any is a useful predictor. is called a Granger Causality test.811) compares well to the actual of E ( x t x′t ) = ∑ x such that shocks are contemporaneously 5.613. The impulse response functions traces the (AIC = -57. 5 shows that the model satisfies all • The impulse response of z in response to a shock in required tests for a suitable model for tourist data. Australian tourist arrivals at a particular time is highlighted in darker grey shows 80% and lighter grey likely to respond to changes in Europe and World shows 95% error bound.445. Math. then the generalized impulse response tourism numbers. Impulse response This Fig.59 + 0. 6 shows the forecast of Australian tourist analysis provides useful information for example. y. j i (7) relationship between Australian.400 even when both a flood and a second wave of the financial crises placed rather sudden shocks to correlated. h = The horizon This Table 5 shows parameters for multiple regression analysis. *** indicates 1% level of The impulse responses computed using the significance.028(Europe) + -0.285. 353 . & Stat.9) suggesting an excellent effect of a shock to one country on to the other fitted model: with p = 2. ACF and Lung-Box • The impulse response of y in response to a shock in tests for the ARIMA analysis is shown in Fig. VAR (m) model where the 2010 prediction is a much a better: 5. εy This Table 3 shows the ARIMA (2. Since orthogonality is not This Fig. the Table 4 and the 2010 actual value fits rather well in the impulse responses will all approach zero. εy 0. variables in the VAR.2. 2) model that fitted the Australian tourist duration and impact of tourism in one country to arrival data with a high degree of it and low AIC. The significance of the tourist modeling and prediction analysis can be strengthened ARIMA: A code in R was used to obtain a timer series further using impulse response analysis to observe ARIMA (2. The impulse response then shows to Australia within 80 and 95% accuracy range.2. it is noted that the the multiple-equation. function of Yi to a unit (one standard deviation) shock in Xj is given by Eq. 2. 8 (3): 348-360. The the z-equation. The results of a Multivariate linear σii = The ith diagonal element of ∑x Regression (MR) model based on Australian ei = A selection vector with the ith element equal to dependency on Europe and the World highlighted in one and all other elements equal to zero Table 5 and Fig. above and beyond lagged values of other variable may be determined. εz This Fig.873.

Math. 6: Forecast for Australian tourist arrivals (80 and 95% error bound) 354 . 8 (3): 348-360. & Stat. 2012 Fig. 5: ARIMA residuals and Ljung-Box statistic Fig.. J.

Math. & Stat. 2006). Europe numbers. Figure 8 other factors. The R-squared value was fear of economic down turn. 8: Histogram and QQ plot of MR model The multivariate analysis shows a significant Over years (1978-2000) there has been a steady model that the model highlights the Australian growth in the Australian tourist numbers but during dependency on the Europe and World numbers 2001-2003 the number of tourists fell down due to the (Tularam and Keeler.982 and the adjusted R-squared 0.. 8 (3): 348-360. 9 and 10 show the closeness of fit of the Europe and the world international movement is multivariate model-close to normality of the histogram interesting. SARS. 2012 Fig. Iraq war among 0. The number of arrivals compared to These Fig. may have been the initiator of the industry.9807. J. It is not 355 . 7: 95% error bound on model fit using MR Fig. The global financial crisis lowered further shows the QQ plot demonstrating that the numbers in 2007-2009 but the rate of reduction in Australian data fit is in line with the assumptions of a Australia was less than the world and Europe and linear multivariate regression method. Europe has a long history of tourism and and linear QQ plot of MR model shows a good fit.

12 shows the results of Granger causality Australia is causing the World at 5% (medium) level of estimated based on the VAR model in a visual diagram. J. Period 1 (1950-1987) consists of the stock decrease/increase for the first 5 years and gradually market crash in October 1987. Australian increasing the responses are plotted over 100 periods/years. and from 2000 to 2007. The arrivals was growing from 1956 to 1985 at an average rate of 0. 1964.33 Mp/yr. 2001 attacks in the US. 1956. the market started years. The Crisis (GFC. Math. times that of the earlier period rate. the tourist destination having about 50% of the world’s Causality was both ways for Europe and the World and international arrivals. Table 6 shows the world’s international arrivals In the second period. Australia decreased by a quarter in functions found. the Granger Causality tests tourism for Europe and the World. The effects of both the cyclone Yasi and In the first period (1950-1987) may be characterized by Queensland floods do not appear in 2010 numbers Exploration and Involvement stages. 2 and 3. 11. & Stat. 356 . followed by the World and Europe.5 Olympics in 2000 located in Sydney. responses generally show signs of fluctuations every The flow and strength of Causality or interaction 5 years and it indicated a season cycle and the between the countries is summarized in Fig. stronger for the European continents and tourists are This Table 6 shows the proportion of arrivals to travelling to the European countries during 1950 to Australia to the World. when Australia times the world’s growth rate. Figure 13 shows the impulse responses the next period. a 10% (low) level of significance from Europe to This Table 5 shows the proportion of arrivals to Australia. arrivals in Australia and World during 1950-2010. however. while the results of this is a two-way Causality for all the world was growing in the same period at an average countries at 1% (high) level of significance. 1% (high) level of significance. In the first period (1950-1987). 1987 because there were 8 Olympics events in the This table shows average growth rate for World.5 times from the to one country on to the other countries. recovering. Australia experienced an Australia to the World. while the World was increasing at 1. all countries are Granger causing period characterized by Consolidation and the tourist arrivals between one country and the other at Stagnation stages of the Butler theory.5 Mp/yr. attracted millions of tourists. responses show a sharp peak in tourist number after 25 Granger causality: The VAR modeling process years for Australia. In particular. the economy is growing Europe to the World. Such dynamic first to the second period. As noted Europe and the World. similarly. In the final period. where the because of a lag effect. 25. 1968. economies are globalised with more The pattern of Australian tourist numbers over time information coming from media and news. Period 2 (1988-2001) declining afterwards. 1972. This is possibly due to the average growth rate of around 9%.065 Mp/yr while the world was growing at Impulse response analysis: The impulse response an average rate of 28. Europe and Australia during 1956-2009. Clearly.3% while Europe has a rate around performing better and there is one-way Causality from 5. In the last period (2002- and Period 3 (2002-2009) refers to the Global Financial 2009). 11 shows the number of international (weak) level of significance from Australia to the World. This is around 1. tourist can is similar to the world growth as shown in Fig. The responses generally earlier. In this period. In the full period. for Australia at 10% (weak) level of significance to Australia was growing at 0. Australian rate increased 6. The rate reached a maximum around 1990 and continued to responses are generally the same pattern among the grow but at a lower rate until the 2007 financial crisis. easily search for places to visit from the internet with Further analysis of Fig.9%. increase for 70-90 years and declines afterwards. a significant analysis analyzed the duration and impact of tourism in growth has occurred in Australia in terms of actual one country to another by tracing the effect of a shock numbers. 8 (3): 348-360. Australia indicate if there is two-way Granger Causality between responses show an increase of tourists coming from the tourist in Australia. 2012 surprising then that Europe is the world’s leading significance. Australia was growing at 0..26 Mp/yr and the world was Europe and the World.1 for the full period. countries where the number of tourists fall up to 40-50 While there is a dip after 2007.04 million persons per year (Mp/yr). 12. showed interesting relationships between the three data The second period (1988-2001) shows a decline in series. 1960. when the world increased by 3 relationships are captured in the impulse response times respectively. This is the whole study period. period 1. In the responses completes after 20-25 years. European countries (1952. the responses show mixed signals. however. In the same period. 1976 and 1984). 2007-2008). 11 shows that Australian increased attention using marketing campaigns. This period is characterized by consists of the September 11. The Causality is at 10% This Fig. the Australian economy is average is around 6. only This Fig. the Development stage. except rate of 8 Mp/yr. in the period 1985 to 2000. Europe and World.

Math. J. 8 (3): 348-360.. 2012 Fig. 11: World and Australia international arrivals 1950-2010 357 . 10: Percentage of share of Australia to the World (1956-2010) Fig. & Stat. 9: Percentage of share of Europe to the World (1950-2010) Fig.

The actual value is well within the levels respectively.900-6. ** more accurately predicts for the year 2010 with a value of and * indicate significance at the 1.954. All test statistics are Chi.2 million arrivals in 2015 but a more realistic The Table 7 shows the Granger causality results 6.2) Australia = Australia(-1) + 6.260.866 with the following equations.863.810 5.019 World = World(-1) + 13.5009Australia(-1) + ε 2010 5. That is.168.805-6.8739** Australia = Australia(-1) + 9. The null 6. 8 (3): 348-360.009*** response analyses related to VAR modeling process Table 6: Average growth rate for World.934-6. 2) model predicts using data up to and including 2009 square with the p-values listed in brackets.475.244 5.409.011 5. & Stat.866-6.016.157.854.3815 5.157 This Fig.353. For the interest of saving to 6.863.237.179 and 95% range of arrivals may be given as estimated based on the VAR model. ***.764. 12: Significant linkages between tourist in Australia..809-5.3 model based on Butler.157 for 2015.012 5.336 2012 5. 2. a 95% range The following regression estimates was estimated of arrival numbers predicted lie within the range 5.295.790 0. the Granger Causality Equations for AR2 -0.012 in 2015.618.9444Australia(-1) + ε 2013 5.569. The Granger and impulse World -0.0 modified Butler model predicts growth predicts around 7.9945** For example.080.294 5.7392Europe(-1) + ε 2014 5.2. J.873.468. The ARIMA (2. cause the column variable.613.527 World = World(-1) + 7.78. The models were calibrated Europe 5.4615** MA1 -1.731 5.499-6. 95% range of 5.111.686-6.462.130 using the updated theoretical hypothesis is that the row variable does not Granger logistic model presented in this study.448.3101World(-1) + ε Year Point Low80% .445.9 using Australian tourist arrival data (1956-2010) and the Australia 9.168. we only show the full period equations as Coefficients Coefficients follows (ε-N (0.863.high95% Europe = Europe(-1) + 10.233.7195** the Full Period: MA2 0.938.234.679.3670Europe(-1) + ε Table 4: Forecast for the next 6 years based on ARIMA(2.658.093-6. Concluding comments: In this study a number of Constant -2.030-6. 2012 Fig.219.930 5.184 0*** models of tourist arrivals were developed and analyzed Europe 0.100. Math.285.006 -2.501 5.525 5.417.016. Europe and Australia (1956 was cross checked and justified with the use of a 2009) multivariate linear model and ARIMA model was Growth rate (%) shown to be in line with and better than the theoretical World 6.304 2015 6. Table 5: MR model coefficients for the tourist arrivals to Australia Model Coefficients T Sig.028 6.521-5.1)): AR1 0. 358 .590 -10.2942World(-1) + ε 2011 5.902 5.high80% Low95% .160.86606. Europe and World Table 3: Estimated parameters of the ARIMA space.800 5.443.257.811 Europe = Europe(-1) + 17. 5 and 10 percent 6.403.662-6. 13 shows the impulse responses analysis based on the VAR model.804-5.556 0*** using time series analyses.510-6.

359 . The logistic model relationship between the tourist numbers recorded for derived from theory was compared with the time series Europe. theoretical model was not as accurate but performed Australia tourist arrivals have been growing since 1974 well and can be used a benchmark for checking and have strong influences from tourist arrival to performance of other models. & Stat. Europe and World . 8 (3): 348-360. period 1.. 13: Impulse response for Australia. 2 and 3 The results of the analysis show: There is a close Europe and indeed the World. Math. 2012 Fig. The when compared to the other numbers of tourists.full period. J. Australia and the World even though data based ARIMA model and the ARIMA performed Australia’s tourist arrives are very low in actual terms the better in terms of the prediction for 2010.

7392 (0.A. 3.5974 (0. 2010. Tourism demand modeling from different regions tend to drive new short term and forecasting. ARIMA and VAR socially responsible investment markets worldwide models all helped in the development of a better integrated.F. Tularam. 37: 424-438. Time series Statistics. An econometric models and cross-spectral methods.5009 (0.. 2012. Impact ABS. Stat. 23: 281-301. Plann.5190) World 0.0000)*** 38. Res.0800)* Europe 1.0002)*** - Period 1 Australia .0652 (0. J. ----------------Europe---------------. 360 .9824 (0.2010. Forsyth. Account. Granger. 1.. -----------------World----------------- Full Period Australia . J.2942 (0. The nature of influence 441. Wong and G. arrivals around to Australia. Stud.0052)*** .0698)* World 1. 26: 633-642. analysis of rainfall and temperature interactions in Box.S.1080/00036846. H. Investigating causal relations by Tularam. S. Australia)..2638) Europe 3. H.0022)*** 6.0012)*** . 0. Math.. and G.0766)* Europe 7.) Edward Elgar. Wong and E. Applied Econom. G. G.. Australian Bureau of Tularam.3814 (0. G. 2008. Keeler. E.3101 (0. 1980.. 8 (3): 348-360. 2010-11.2085) 4.441 factors is better exemplified by the impulse response Roca.1000 (0.H. pp: 592.H. G. Relationship between El Nino Butler. R.4396) 1. DOI: possible based in this structure. The validity of the 10. The study of REFERENCES coastal groundwater depth and salinity variation using time-series analysis.4205) 0. DOI: 10. 6: Series Analysis: Forecasting and Control. 2006. Prentice Hall.0077)*** World 24..A. Cheltenham. Reinsel. V. further showed that the Butler’s stages can be verified 44: 2935-2944. Tourism Demand particularly useful for new investors and developers as Forecasting.. Sustain. 1969. 6: 431- predictive ability of models. 2006.0508)* ..7974 (0. Li. the differential equation. Jenkins and G. Detailed analyses not only showed the nature water flow? An econometric analysis of the global of significant interactions but the impulse response price integration of water stocks.431.W.5237 (0. ISBN-10: 0130607746. G.. Dwyer. Illahee. and M. J. Canadian Geographer.8158 (0..A.2307/1912791 Regional Urban Plann. J.P.. and L.3042) .1591) 3. Modeling multiple regression and further analyses of the VAR in tourist arrivals in destination countries: An term Granger Causality to provide the justification and application to Australian.7292 (0. Tourism Manage.. departments.1177/0047287505276594 models can be further clarified by the examination of a Nejad. Tularam.0000)*** - The interactions studied with the VAR model Li. Turner. 0. tourist numbers Song.S. Int. SARS.W. The concept of a tourist area cycle southern oscillation index and rainfall of evolution: Implications for management of (Queensland.. G. 2010. and H. 10.H.A. Such information is Song.6489 (0. 9.1108/10309611011092600 suggesting that in different periods. J.5817 (0.0120)** Europe 10.A. and G. 44: 82-99.M.A.0659)* 3. 2010. C. 3. 5: 378-391. resources. Are Finally. Roca. H. 2005.3670 (0.. DOI: volatility using EGARCH approach. In: International Handbook on the well as longer term planners and government Economics of Tourism.. 2011. Which way does analyses. Witt.D. and P.568403 breaks such as financial crises. & Stat. E.A.3844/jmssp. 1994. L.7111) - Period 3 Australia .0334)** - Period 2 Australia . 2012 Table 7: Granger causality results --------------Australia-------------. (Ed. DOI: understanding the overall picture of the tourism story 10. Dev. Time coastal catchments.J.4159 (0.. Mathematics Statistics.0048)*** 13. V. Math.9444 (0.2488 (0. 7. satellite account. Recent allowed more in depth analyses based as the Granger developments in econometric modeling and Causality and impulse response calculations were forecasting.0000)*** World 7. Travel Res. Tularam.2011. Roca.A. 2010. 2010.1372 (0. Environ.S. 17. 372-380. Song and S. 24: 5-12. Tularam. G.0557 (0. and G. J. Australian national accounts: Tourism Assessment Rev. Tularam. 3rd Edn. analysis on Australian superannuation funds Econometrica.2832 (0.1351 (0.2888 (0. DOI: when the data is studied using important structural 10. 29: 203-220.

- 19016055 MagzID CaseUploaded byisaac
- Introduction to Forecasting PresentationUploaded byMax Greco
- BDMDMUploaded bySaksham Goyal
- 2144 - Geography of Tourism Course Outline - Winter 2012Uploaded bynick_goel440
- Statistical Analyses of Multivariate Time Series Data With Application to Compacting Effects on Soil Chemical and Biological Properties in ForestryUploaded byStuart Fenech
- Box JenkinsUploaded bysiva_koti10
- 100.Defining a MuseumUploaded byPatty Sia
- Tourism TimelineUploaded byEvelyn Melendez
- Lecture on ARIMA Using MinitbabUploaded byEd
- Study on the Diversity and Predictors of Internal and International Migration in Bangladesh a Panel Data ApproachUploaded byIJSTR Research Publication
- Zaur ImraniUploaded byjjh
- Domestic and International Tourism in a Globalized WorldUploaded byMohit Sharma
- DRTA Prediction LogUploaded byAnonymous Env0Wed
- An Introduction to Multivariate StatisticsUploaded byKhampa Ngawang Phuntsho
- VUSSC Tour GuidingUploaded byale
- Policy Paper on TourismUploaded byShaing Shaing U
- Ma Thematic Neural NetworkUploaded bypeyman_qz_co
- Skill-Importance-in-Volleyball.pdfUploaded bycamry_ng
- International Tourism Forum Programme "Development of wine tourism in the countries of the Danube River Basin"Uploaded byVeronika Tomescu Raetsky
- TASK Compass 5Uploaded byPresa Rodrigo Gómez, "La Boca"

- ALBALACT Raport Amplasament 2016Uploaded byBicăjanu Bianca
- curs_BP_IDUploaded byGabriel Coltan
- 7.1_ Inventory Reduction Strategies 2018Uploaded byBicăjanu Bianca
- 7.1_ Inventory Reduction Strategies 2018Uploaded byBicăjanu Bianca
- modele fundamentaleUploaded byBicăjanu Bianca
- Exemple Probleme de Rezolvat Pentru Pregatirea Examenului La BCOUploaded byAnndu Andu
- 2013Uploaded byBicăjanu Bianca
- Script Proiect Analiza DatelorUploaded byBicăjanu Bianca
- ConsumatorUploaded byBicăjanu Bianca
- Tehnologii Web RaspunsuriUploaded byjohn
- curs_BP_IDUploaded byGabriel Coltan
- Dedu - Probabilitati culegere probleme (2).pdfUploaded byAndreea Oancea
- 2014_barem.pdfUploaded byBicăjanu Bianca
- 2. Stancu, S. - Microeconomie.pdfUploaded byBicăjanu Bianca
- 2013Uploaded byBicăjanu Bianca
- Game Theory in Ecomomics and BeyondUploaded byBicăjanu Bianca
- Antreprenoriat_RECOMANDARIUploaded byBicăjanu Bianca
- 5Uploaded byBicăjanu Bianca

- Kalman FilterUploaded byJamiu Ariremako
- ADL 07 Quantitative Techniques in Management AM4Uploaded bysolvedcare
- bayesGaussUploaded byYogesh Nijsure
- A Meta-Analysis of the Relationship Between MarketUploaded byHadi Jamshidi
- Analyzing Relationships Between VariablesUploaded bylogan159
- Enhanced Data Analysis Using SAS ODS Graphics and Statistical GraphicsUploaded byarijitroy
- 1. pendahuluan.pptUploaded bysidajateng
- GM533 Chapter 6 Study GuideUploaded byAngel Menchaca
- StatUploaded byAdriancisneros Pel
- Gentle Introduction to MCMCUploaded byjakub_gramol
- MathsWatch Essential Questions SAMPLEUploaded byChloëBooth
- Statistics ECA ReportUploaded byMuhamad Fauzi
- Mae 546 Lecture 16Uploaded byRajmohan Asokan
- Adv Algo TradingUploaded byVijay Kumar
- Mit18 05s14 Exam2 SolUploaded byErwin Dave M. Dahao
- MultilevelPRI WorkshopUploaded byAndré Luiz Lima
- Reability ODIUploaded byAnonymous mWAaDZmlJw
- Lecture EstimationUploaded byVishnu Venugopal
- Basic Statistics.pdfUploaded byUmair Adnan
- Applied Non Parametric RegressionUploaded byArun Kumar Nagar
- ANCOVA_R.pdfUploaded byLuis Luengo Machuca
- Long Term Memory in Stock Market ReturnsUploaded byboletodon
- Metrics10 Violating AssumptionsUploaded byJohn Lukas
- Ch18 AnswersUploaded byMichaelKiprono
- Econ275 (Stanford).pdfUploaded byInvest
- ExChapter6.pptUploaded byJacob Huston
- Normal Distribution CopyUploaded byBogdan Tanasoiu
- Project Scheduling -- Probabilistic PERTUploaded bySanjana Ganesh
- chap_13_2Uploaded byapi-19746504
- ap_statistics_midterm_review_12-13.rtfUploaded byshanelbbdo