Multi Input Intervention Model for Analyzing. the Impact of the Asian Crisis and Terrorist. Attacks on Tourist Arrivals in Bali

Transcription

1 Applied Mahemaical Sciences, Vol. 7, 203, no. 35, HIKARI Ld, hp://dx.doi.org/0.2988/ams Muli Inpu Inervenion Model for Analyzing he Impac of he Asian Crisis and Terroris Aacks on Touris Arrivals in Bali Sri Rezeki, Suharono 2 and Suyadi 3,3 Deparmen of Mahemaics Educaion Faculy of Teacher Training and Educaion Universias Islam Riau Perhenian Marpoyan, Pekanbaru, Indonesia 2 Deparmen of Saisics, Faculy of Mahemaics and Naural Sciences Insiu Teknologi Sepuluh Nopember Kampus ITS Sukolilo Surabaya 60, Indonesia sri_rezeki@yahoo.co.id Copyrigh 203 Sri Rezeki, Suharono and Suyadi. This is an open access aricle disribued under he Creaive Commons Aribuion License, which permis unresriced use, disribuion, and reproducion in any medium, provided he original work is properly cied. Absrac The purpose of his sudy is o develop muli inpu inervenion model for analyzing he impac of he Asian financial crisis and erroris aacks Bali on he number of ouris arrivals. This research focuses mainly on he developmen of a model ha could be used o explain he magniude and periodic impacs of he Asian financial crisis since July 997 and erroris aacks referring o he Bali bombings on Ocober 2 h 2002 and Ocober s 2005, respecively. Monhly daa comprising he number of ouris arrivals in Bali via Ngurah Rai airpor are used as he daa for his case sudy. The resuls show ha he Asian financial crisis and Bali bombings yield negaive impacs on he number of ouris arrivals o Bali. Generally, he Asian financial crisis did give a negaive permanen impac afer a 0 monh delay. The firs and second Bali bombings also yield negaive impacs which were direc and emporary effecs. However, he firs aack had longer period effec han he second, i.e. 8 and 5 monhs afer onse aacking, respecively. In addiion, his

2 676 Sri Rezeki, Suharono and Suyadi research also discusses how o assess he effec of an inervenion in ransformaion daa. Keywords: Asian crisis, Bali bombing, Touris arrivals, Inervenion model INTRODUCTION Inernaional ouris arrivals have been affeced due o disrupions caused by a range of evens ha may occur in he desinaion iself, in compeing desinaions, original markes, or hey may be remoe from eiher. In recen years, major disrupions ha have affeced he inernaional ouris arrivals include he Gulf War in 99, he Asian financial crisis in 997, he erroris aack on Sepember, 200 on he US, he SARS and avian flu in 2003 as well as naural disasers such as hurricanes, sunamis, and earhquakes [8, 9, 2, 4]. The ourism indusry in Indonesia is an imporan componen as well as a significan source of foreign exchange revenue for he Indonesian economy. However, ourism in his counry is also subjec o he effecs of naural and man-made disasers. Some of naural disasers ha have aken place are: he 26 December 2004 Aceh sunami, 27 May 2006 and 30 Sepember 2009 earhquakes in Yogyakara and Padang respecively on, and he bird flu epidemic in Besides hese naural disasers, man-made disasers caused by erroris aacks and bombings such as he incidens on 2 Ocober 2002 and Ocober 2005 in Bali, 5 Augus 2003a he Jakara s Marrio Hoel, 9 Sepember 2004 in he Ausralian embassy in Jakara and more recenly, he 7 July 2009 bomb aacks of JW Marrio and Riz-Carlon hoels in Jakara. Bali as an island is one of he primary desinaions in Indonesia for inernaional ouriss. Tourism is a vial source of employmen in Bali and people have migraed o he resors and hoels of Bali no only from elesewhere in he island, bu also from oher pars of Indonesia. The ourism indusry does no only generae direcly relaed employmen, bu also creaes a grea deal of indirec employmen opporuniies [3]. From 989 o 2009, he Balinese ourism indusry experienced hree major disasers: namely he Asian financial crisis of 997 and he erroris aacks of 2002 and Hichcock [7] gave a business review abou Asian financial crisis and ourism in Bali. More recenly, Pura and Hichcock [0] also made a review abou errorism and ourism in Bali and Souheas Asia. This paper developes he muli inpu inervenion model and apply i o examine he impac of he Asian financial crisis and erroris aacks on ourism in Bali as a way o esablish a beer undersanding of how hese changes and rends affec inernaional ourism. The paper is organised as follows: a brief

3 Muli inpu inervenion model 677 lieraure review abou he impac of he differen crisis and errorism on ourism based on a ime series approach, daa descripion and he modeling mehod, resuls based on he model, evaluaions of he he impac due o he inervenions, conclusions and recommendaion. MATERIAL AND METHOD The relaionship beween ourism and errorism or poliical insabiliy has been invesigaed exensively by many researchers since 980s. Sonmez [3] did a comprehensive lieraure review focusing on he relaionship beween hese phenomena during 8 years, i.e. from Generally, researchers have relied on wo main approaches for evaluaing he effec of crisis or errorism o he ourism indusry by focussing on he impac of hese evens on micro-ouris preferences using individual ouris daa or focussing on esimaing he aggregae effecs using ime series daa. The firs approach could be seen in Yechiam, Barron and Erev [5] who analyzed he personal experience in conribuing o he differen paerns of response o rare erroris aacks. Arana and Leon [] sudied he shor-run impacs of he Sepember aacks in New York on ouris preferences for compeing desinaions in he Medierrnean and he Canary Islands, and Riichainuwa and Chakrabory [2] who sudied abou perceived ravel risks regarding errorism and disease in Thailand. This research focusses on he second approach ha uilizes a ime series analysis for assessing he impac of crisis or errorism on ourism. Enders and Sandler [5] and Enders, Sandler and Parise [6] were among he firs researchers who used ime series approach for analyzing he negaive impacs of errorism on ourism revenues in Spain and oher Eruopean counries. The research can be used o suppor subsiuion effecs beween hese counries as a resul of he ouris s goal of minimizing he risk of facing a error aack. Similar resuls were also found by Drakos and Kuan [4] who sudied he regional effecs of errorism on ourism in hree medierranean counries. MULTI INPUT INTERVENTION MODEL The muli inpu inervenion model is [] bi ( ) ( ) k s B B q B i Y i X i d ( B) ( B)( B) a., ri p Eq. () shows ha here are k evens ha have affeced he ime series daase. As an illusraion, consider a muli inpu inervenion wih wo evens, namely ()

4 Y Y* 678 Sri Rezeki, Suharono and Suyadi a pulse funcion occurring a T 40 wih ( b, 2, 0) s r which is followed by a sep funcion a T 2 60 wih b, s, r ), hus 2 ( ( 0 B) B ( ) q B Y [( 0 B 2 B ) B ] P, 2,. S a d (2) B ( B)( B) The impac is Y * P 0 P P S p ( 0 ) S2, 2 ( 0 ) S2,,, 2 2, , which can also be wrien as Y * 0, 0,, 2, 0, 0 2 m i i 2 2 m i2 i2 2, T T T T T T k T m, 2 and k 4 m. (3) The illusraion of Eq. (3) and is impac are represened in Figure where 25, 0, 5, 5, 4, and T=40 T2=60 0 T=40 T2= (a) (b) Fig. : (a) Simulaion of he Inervenion Model, (b) Inervenion effec of Muli Inpu Inervenion where Pulse Funcion b, s 2, r 0) occurred a 40 ( ( 2, s2, r2 and followed by he Sep Funcion b ) a 60 The firs inervenion ha affeced he daa a 4, wih a magniude of 25. The pulse funcion inervenion had an effec ha lased for 3 periods beyond T 40 wih magniude effecs of 0 and 5 on he second and hird afer he inervenion, respecively. Afer ha, he effec of his pulse inervenion will be equal o zero. The second inervenion began a T This sep inervenion was deeced a 6 and is impac was 5. From 62 o 65 he impacs of his sep inervenion were 26.5, 32.25, 36.5, and 37.3, respecively. I is noed ha he impac did no increase beyond 38.

5 Muli inpu inervenion model 679 PROCEDURE FOR BUILDING MULTI-INPUT INTERVENTION MODEL * Rezeki, Suharono and Suyadi [] showed ha he inervenion response or Y is easily formulaed using he response values char for deermining he order of inervenion model using b, s, and r. The inervenion response denoed as * Y is basically residual or error which is he difference beween he acual daa and he ARIMA model forecass based on he daa before he inervenion. A complee procedure of he inervenion model building can be used o evaluae hese k inervenion funcions a ime T, T 2,,T k as according o he following procedures. Procedure. Dividing he Daase ino k+ Pars Par daase is he daa before he firs inervenion wih n 0 as he ime periods, i.e.,2,, T. Denoed asy 0. Par 2 daase is he daa from he firs inervenion unil jus before he second inervenion based on n ime periods, i.e. T, T, T 2,, T2. Denoed asy. Par k+ daase is daa from he k h inervenion unil he end of daa analysis based on as many as n k ime periods, i.e. Tk, Tk, Tk,, n. Denoed as Y. k Procedure 2. Modeling of he Firs Inervenion a. Sep Use he ARIMA model building for ime series daa before he firs inervenion occurs ( ), so we have Y 0 q ( B) p ( B )( Y0 B ) Forecasing of Daa 2 ( Y ) using he ARIMA model. In his sep, we ge he forecas daa, i.e. Y ˆ, Yˆ,..., Yˆ. T T T n b. Sep 2 * Calculae he response values of he firs inervenion or Y. These values are he residuals of he daa from T, T, T 2,..., T2. This is based on he forecasing mehod of he ARIMA model as shown in he firs sep. This sep produces response values of he firs inervenion based on * * * Y, Y,..., Y. T T T2 d a.

6 6720 Sri Rezeki, Suharono and Suyadi Deermine b, s, r based on he firs inervenion by using he plo of * * * response values Y T, Y T,..., YT 2 and a confidence inerval of widh, i.e. 3 ˆ a0, where ˆ a is Roo Mean Square Error (RMSE) from he 0 previous ARIMA model. These inervals are based on he deerminaion of conrol char bounds during he saisical qualiy conrol for deecing oulier observaions. c. Sep 3 Esimae he parameer and es he significance for he firs inervenion model. Conduc a diagnosic check o examine he residual assumpion inclusive of whie noise and normaliy disribuion. In his sep, he firs inpu inervenion model is s ( B) B Y ( B) r b X, q ( B) ( B)( B) p d a. (4) Procedure 3. Modeling of he m h Inervenion Model, where m = 2, 3,..., k. a. Sep Forecas Daa m ( Y ) based on he m h inervenion model. In his m sep, we will obain he forecased values from he m h inervenion model Y ˆ, Yˆ,..., Yˆ. Tm Tm Tm nm b. Sep 2 Calculae he m h inervenion responses Y ) which is he residual of ( * m he daa for T, T,, T. This is based on he forecasing of he m m m h ( m ) inervenion model. These response values are denoed as Y * Tm, Y,..., Y * * Tm Tm h Idenify b m, sm, rm from he m inervenion model from he plo of * * * response values Y Y,..., Y, and he confidence inerval of c. Sep 3 widh 3 ˆ. a m T, m Tm Tm h Esimae he parameer and conduc a significance es for he m inervenion model. Conduc a diagnosic check o examine he residual assumpion inclusive of whie noise and normaliy disribuion. The resul of his sep is.

7 Touris Arrivals (Thousands) Muli inpu inervenion model 672 b j ( ) ( ) m s B B q B j Y j X j d ( B) ( B)( B) a., rj p (5) h This procedure is done ieraively unil he las ( k ) inervenion. As a resul of hese seps, evenually we would obain he following muli inpu inervenion model b j ( ) ( ) k s B B q B j Y j X j d ( B) ( B)( B) a., rj p DATA The number of ouris arrivals in Bali from uary 989 unil December 2009 is used in his sudy. The daa are 252 monhly records of he arrivals. Figure 2 illusraes he daa in a ime series plo. 250 July '97 Oc '02 Oc ' Monh Year Fig. 2: Monhly ouris arrivals o Bali from uary 989 December 2009 During his period, here were hree inervenions which may have affeced he number of ouris arrival in Bali. These inervenions are he Asian financial crisis which occurred from July 997 unil December 2009 [9] and he Bali bombings which occurred in Ocober 2, 2002 and Ocober, In his analysis, he Asian financial crisis is he sep funcion inervenion variable, whereas he Bali bombings are he pulse funcions. From he graph, we could see ha he daa did no decline drammaically according o he period when he Asian financial crisis occured which was in July 997. Wih ha excepion, he graph shows ha he number of ouris arrivals dropped direcly and dramaically during he firs and second Bali bombings.

8 6722 Sri Rezeki, Suharono and Suyadi RESULTS All he resuls and models repored in his sudy were esimaed using Saisical Analysis Sysem (SAS) and he graphs were produced by MINITAB. The following secions will ouline he resuls of he pre-inervenion model using he Box-Jenkins procedure and he firs, second and hird inervenion models. Pre-inervenion Model Resuls The Box-Jenkins procedure [2] was uilized for his research which included he idenificaion, parameer esimaion, diagnosic checking, and forecasing o find he bes ARIMA model before he firs inervenion, i.e. he Asian financial crisis since July 997. The resuls of he idenificaion sep, parameer esimaion, parameer significance es, and diagnosic checking show ha wo models, i.e. ARIMA(0,,)(0,,) 2 and ARIMA(0,,)(,,0) 2, are appropriae as a means for forecasing he monhly ouris arrivals in Bali before Asian financial crisis. The comparison of he mean square errors (MSE) showed ha ARIMA(0,,)(0,,) 2 yielded less MSE han ARIMA(0,,)(,,0) 2. Thus, he bes ARIMA model for daa before he firs inervenion is ARIMA (0,,)(0,,) 2, i.e. 2 ( B)( B ). (6) ( B)( B ) ln Y a 2 The Firs Inervenion Model Resuls This secion presen he resuls of he inervenion model by illusraing he impac of he firs sep funcion inervenion, namely he Asian financial crisis from July 997 unil December 999 or a he ime 03,04,, 32. Is funcion could be wrien as S, 0,, 02 03,04,,32. The firs sep in his modeling is o deermine he order b, s, and r for he firs sep funcion inervenion model. This is done o deermine he order of he inervenion model and o explain he decrease in he number of ouris arrivals in Bali due o he Asian financial crisis. A residual char is as in Figure 3 is used o show his sep.

9 Residual Muli inpu inervenion model July '97 May ' T-0 T-5 T T+5 T+0 Time T+5 T+20 T+25 T+30 Fig. 3: Response Values on he Number of Touris Arrivals in Bali (in Naural Log) afer he Firs Inervenion and Prior o he Second Inervenion The resuls of esimaion parameer and diagnosic checking seps show ha he bes model of he firs sep funcion inervenion and prior o he second pulse funcion inervenion is 2 ( B)( B ) ln Y S, S, S, 3 a. 2 (7) ( B)( B ) Based on he model presened in Eq. (7), he inerpreaion of he impac of he Asian financial crisis is no direc as i was delayed for 0 monhs or i sared only on May 998 when here were rios, killings, and desrucion of commercial disrics in Java, paricularly he ani-chinese senimen rios in Jakara. Resuls from he Second Inervenion Model Afer modeling he firs inervenion based on he inervenion model due o he Asian financial crisis, anoher analysis of he second pulse funcion inervenion was conduced. This was based on he Ocober 2, 2002 Bali bombing a which is equaed wih T 66. Thus, he pulse funcion is wrien as P2, 0,, The firs sep in his analysis is o deermine he order of he second inervenion model. By applying similar procedure as he firs inervenion model, he resul show ha he bes inervenion model for he number of ouris arrivals in Bali afer he second pulse funcion inervenion and prior o he hird pulse funcion inervenion is ln Y S, S, S, P2, P 2,

10 Touris Arrivals (Ln ransformaion) 6724 Sri Rezeki, Suharono and Suyadi P2, P2, P 2, P2, P2, P2, 6 2 ( B)( B ) P 2, 8 a. 2 ( B)( B ) (8) Resuls from he Third Inervenion Model The final analysis of he hird pulse inervenion funcion based on he second Bali bombing which ook place on Ocober, 2005 is equaed wih T 202. So, he pulse funcion in his inervenion could be wrien as P3, 0,, By using he same procedure as previous inervenion, he resuls show ha he final muli inpu inervenion model for he number of ouris arrivals in Bali afer he hird pulse inervenion funcion can be wrien as ln Y S, S, P2,.2580P2, P P2, 3 2, P 2, P2, P2, P2, P 2, P3, P 3, P3, P3, P3, P3, 5 ( B)( B ( B)( B 2 ) 2 ) a, (9) where S, he sep funcion of he Asian financial crisis, P 2, is he pulse funcion of he firs Bali bombing, and P 3, is he second Bali bombing. The effec from he reconsrucion and he forecas of he final inervenion model for ransformaion daa (naural log daa) are presened in Figure Variable ln Y Pre-Inervenion s Inervenion 2nd Inervenion 3rd Inervenion July '97 Oc '02 Oc '05 ' Monh Year Fig. 4: Effec Reconsrucion and Forecass of he Firs, Second and Third Inervenion models (a Transformaion Daa)

11 Muli inpu inervenion model 6725 CONCLUSION Sudying he impac due o unexpeced disrupions such as he Asian financial crisis and erroris aacks on ourism is imporan for forecasers, planners, invesors and operaors. This paper provides an analysis of he impac of hree inervenions, namely he Asian financial crisis and he wo Bali bombings on he number of ouris arrivals in Bali. The resuls show ha hese hree inervenions have significanly conribued o he decrease in he number of ouris arrivals. The Asian financial crisis ha occurred from July 997 o December 999 did no direcly affec he decrease of ouris arrivals in Bali bu i did cause a delayed reacion afer 0 monhs when he impac was only fel in May 998. The impac ook place due o he rios, killings, and desrucion of commercial disrics in Java, especially in Jakara. The decrease of ouris arrivals in Bali was due o hese crises and he number of ouriss in May, June and July 998 were 3399, 42822, and respecively. On he oher hand, he firs and second Bali bombing had affeced direcly he decrease of ouris arrivals. The firs aack had a negaive impac unil 8 monhs afer he aack whereas he second one had an impac for only 5 monhs afer he aack. Moreover, he reduced numbers of ouris arrivals due o he firs and second aacks were and 2699 ouriss respecively. We recommend ha he Indonesian governmen refer o he valuable daa as he impac of hese harrowing experiences and he afer effec responses o hese crises in Indonesia as well as oher counries. I is imporan ha ourism be used as a reference for crisis and disaser managemen in he fuure. There is a need for proper evaluaions, pre-warnings and responses o disasers as hese are imporan seps ha may reduce he impac from such urmoil. This paper has shown us he means o learn from hisory so ha we know how o brace ourselves in preparaion for unforeseen fuure disasers in. To recapiulae, his research shows ha he inerpreaion of an inervenion model for ransformaion daa could no be done direcly based on esimaed model parameers. Furher research is needed o undersand he precise impac of he inervenions on oher forms of daa ransformaion. ACKNOWLEDGMENT This research was suppored by Research Hibah Bersaing 202 of Universias Islam Riau Pekanbaru.

12 6726 Sri Rezeki, Suharono and Suyadi REFERENCES []. J.E. Arana and C.J. Leon, The impac of errorism on ourism demand. Annals of Tourism Research, 35(2) (2008), [2]. G.E.P. Box, G.M. Jenkins and G.C. Reinsel, Time Series Analysis: Forecasing and Conrol. (4h ed.), New Jersey: John Wiley & Sons, Inc., (Chaper 3, pp ), [3]. J. Cukier, Tourism Employmen in Bali: Trends and Implicaions. In R. Buler and T. Hinch (eds), Tourism and Indigenous People (pp ). London: Inernaional Thomson Business Press, 996. [4]. K. Drakos and A. Kuan, Regional effecs of errorism on ourism in hree Medierranean counries. Journal of Conflic Resoluion, 47 (2003), [5]. W. Enders and T. Sandler, Causaliy beween ransnaional errorism and ourism: The case of Spain. Terrorism, 4 (99), [6]. W. Enders, T. Sandler and G. Parise, An economeric analysis of he impac of errorism on ourism. Kylos, 45 (992), [7]. M. Hichcock, Tourism and oal crisis in Indonesia: The case of Bali. Asia Pacific Business Review, 8(2) (200), [8]. H.I. Kuo, C.C. Chen, W.C. Tseng, L.F. Ju and B.W. Huang, Assessing impacs of SARS and Avian Flu on inernaional ourism demand o Asia. Tourism Managemen, 29 (2008), [9]. B. Prideaux, E. Laws and B. Faulkner, Evens in Indonesia: exploring he limis o formal ourism rends forecasing mehods in complex crisis siuaions. Tourism Managemen, 24 (2003), [0]. I.N.D. Pura and M. Hichcock, Terrorism and Tourism in Bali and Souheas Asia. In Hichcock, Michael, King, Vicor T. and Parnwell, Michael (ed.), Tourism in Souheas Asia: Challenges and New Direcions (pp ). Copenhagen, Denmark: NIAS Press, []. S.Rezeki, Suharono and Suyadi, Applicaion of muliple sep funcions inervenion model for evaluaing he impac of crisis and new policy of gambling prohibiion o he ourism in Baam. Research repor of Hibah Bersaing scheme, Universias Islam Riau, Indonesia, 202. [2]. B.N. Riichainuwa and G. Chakrabory, Perceived ravel risks regarding errorism and disease: The case of Thailand. Tourism Managemen, 30 (2009), [3]. S. Sonmez, Tourism, errorism, and poliical insabiliy. Annals of Tourism Research, 25 (998),

13 Muli inpu inervenion model 6727 [4]. Y.S. Wang, The impac of crisis evens and macroeconomic aciviy on Taiwan s inernaional inbound ourism demand. Tourism Managemen, 30 (2009), [5]. E. Yechiam, G. Barron and I. Erev, The role of personal experience in conribuing o differen paerns of response o rare erroris aacks. Journal of Conflic Resoluion, 49 (2005), Received: Ocober 26, 203