UBR, Volume 10, Number 1.

2010 ISSN: 1555-1296

P I I Ed- on: BUlsir,e!:~sEnviron et S uthe!ast

INTERNATIONAL JOURNAL
OF BUSINESS RESEARCHTM

ponsors
Jesse H. Jones School of Business School of Business
Texas Southern University, Houston, Texas, USA Hampton University, Hampton, Virginia, USA
http://www.tsu.edu bttp:/lwww.hamptonu.edu

la ~)n~ r:: i S
OeanIDr; Phapruke Ussahawanitchaklt DeanIProfessor Sid Howard Credle
Professor! Dr. C. B. Clalbome ProfessorlDr. YlShnuprasad Nagadevara

A Publication of the
Academy of International Business and Economics
PromaIIng GIobII ComfIetiIIveMII ,.
A1BE ORG
 AN EMPIRICAL APPROACH TO THE EVALUATION OF INTERNATIONAL TOURISTS'
EXPENDITURES IN THAILAND USING THE ARFIMA-FIGARCH MODEL

Kanchana Chokethaworn, Chiang Mai University, Chiang Mai, Thailand

Jittaporn Sriboonjit, Thammasat University, Bangkok Thailand
Chukiat Chaiboonsri, Chiang Mai University, Chiang Mai, Thailand
I Prasert Chaitip, Chiang Mai University, Chiang Mai, Thailand
I

I
~STRACT

~iJdieS of international tourists' expenditures have used a variety of forecasting procedures to evaluate
turism economic phenomena. Forecasting methods have developed utilizing ARFIMA(p,d,q)-
rGARCH(p,d,q) approach, primarily because the forecasting concept alone is sufficient explanatory
wer to cover many important tourism economic phenomena, particularly those which possess an
olutionary statistical structure. The secondary data were used to produce forecasts of international
rists' expenditures based on the ARFIMA(1,-0.672, 1)-FIGARCH(1,-0: 180, 1) method. The international
rists' expenditures in Thailand for the period of 2009-2010 may experience some decline. The public-
vete pettnersnip should plan to provide a remarkable chance to convey simultaneously a mixture of
rts of the industry and to establish major tourism course of action initiatives to set in motion to lay the
ndation at the educational level to assist all stake holders even if possible to initiate some of those
asures necessary.

ywords: Thailand, international tourists' expenditures, tourism market, forecasting procedures

INTRODUCTION

e study has suggested a new approach for investigating the structure of the international tourists'
eipts as a benchmark to assess the overall importance of tourism in Thailand. Sizeable international
rists' receipts can be a good indicator of the role tourism plays in an economy in terms of both Gross
mestic Product and foreign exchange generation.

ESEARCH OBJECTIVE

secondary data were used to produce forecasts of international tourists' expenditures in Thailand
ed on the ARFIMA(p,d,q)-FIGARCH(p,d,a) method for the study period of 2009-201 O.

HE SCOPE OF STUDY.

selected countries were the countries that have impact on the international tourism industry of
iland (Source of Data: Immigration Bureau, Police Department). This paper focuses on forecasting a
le variable approach based on 2000-2008 international tourists' expenditures in Thailand. Also, the
IMA-FIGARCH model was tested to forecast the expenditures in Thailand during the study period of
9-2010. However, this model has never previously been used for forecasting the international tourists'
enditures in Thailand.

General Model of ARFIMA

MA models as discussed by Box and Jenkins (1976) are frequently used for seasonal time series. A
eral multiplicative seasonal ARIMA model for a time series Z t can be written

------ (1J)

where
B = the backshift operator (8 z: - Z t-1)
S = the seasonal per:cc
0(B) = (1- 018 - - .e:E:s ::6 non-seasonal AR operator
<P(BS) = (1- <P18s - - :1::::;. .s tr e seasonal AR operator

INTERNATIONAL JOURNAL OF E;.. 5 1',::5 S :;.::5 :;t.RCH, Volume 10, Number1, 2009 141
 e
8(B) = (1- 6~B - .. . - S:B ) is the non-seasonal moving average (MA) opera.;:
p(B) = (1- p, BS - .. . - 2: 8,",5) is the seasonal moving average (MA) operator
(1_B)d(1_B5) = non-seasc+a : -'-'5:e cing of order d and seasonal differencing of order L:

ARFIMA models were proposed by Granger a~d Joyeux (1980). After that, Hosking (1981) also prc::::
this method to fit long-memory data. An a .:oregressive fractionally integrated moving-average (AR:
process is ARFIMA(p,d,q) model as well. as it can be written by: (see equation 14E ).

0(~ )Ll dy \ = 0 + 8(~) \

with 0(~) = 1- 01~ - O2[32_ - 0p~P

and 8(~) = 1- 8 1(~) - 8 2(~)2 - - 8 q(f3)q
where
o = constant term
8(~) = moving-average operator at order q
t = error term of equation 14E
0~~) = the autoregressive operator at order p
Ll Y\ = differencing operator at order d of time series data y \

For d within (0, O.5), the ARFIMA process is said to exhibit long memory or long range p:: - .
dependence
For d within (-0.5, D), the process exhibits intermediate memory or long range ne; = >,.
dependence
For d within [0.5, 1) the process is mean reverting and there is no long run impact to future vs _-=-
of the process
The process is short memory for d=O corresponding to a standard ARMA process

3.2 General Model of FIGARCH

The-simplest GARCH model is the GARCH(1,1) model: (see equation number 4H)

2 2 2
at = Uo + Ul ~ t-l + Ala \-1

Now the variance of the error term has three components: a constant, last period's volatility (the ,:-:; = -
term), and last period's variance (the GARCH term). In general, it could have any number of ARCH :~-.
and any number of GARCH term and the GARCH (p,q) model refers to the following equation for -
( see equation 5G ) .

at 2 _
- Uo
+ U 1~
2
\-1
+ 0"
+ U P~
2
\-p
+' ""la 2
t-l
+ +'"" qa2 t-q

Baillie, et al. (1996) proposed the fractional integrated GARCH (FIGARCH) model to determine :-1
memory in return volatility. The FIGARCH(p,d,m) process is defined as follows: (equation :1y)

( 1- L)d 4> (L)2\ = W + [1 - ~(L)l u \

where u t = 2\ - a2\ and also the FIGARCH model derived from standard GARCH model with frac. ;' ~
difference operator, ( 1- L)d The FIGARCH(p,d,q) model is transformed standard GARCH wher : :
and IGARCH model when d = 1.

3.3 AIC (Akaike, 1973) and BIC (Bayesian Information Criterion)

Table 1 shows forecasting methods based on ARFIMA-FIGARCH models for forecasting mterna: :-.,;
tourists' expenditures in Thailand period of 2009 to 2010 based on based on AIC criterion ar: ::
criterion.

INTERNATIONAL ~: _ =<. " .!._:= 3 ....SINESS RESEARCH, Volume 10, Number 1, 2009 142

$
 TABLE 1: ACCURACY COMPARISON IN SAMPLE FOR DIFFERENT FORECASTING MODELS
BASED ON AIC CRITERION AND BIC CRITERION

Models of foreca~tip
ARFIMA(1 .d, 1)-FIGARCH(1,d,1)
1 7.100 28.848
d of ARFIMA = -0.672, d of FIGARCH = -0.180
ARFIMA(1,d,1 )-FIGARCH(1,d,2)
2 9.102 33.568
d of ARFIMA = -0.114, d of FIGARCH = -0.076
ARFIMA(1,d,2)-FIGARCH(1,d,1 )
3 9.066 33.532
d of ARFIMA = -0.680, d of FIGARCH = -0.140
ARFIMA(2,d,2)-FIGARCH(1,d,1 )
4 11.026 38.211
d of ARFIMA = -0.366, d of FIGARCH = -0.163
From: computed

'om table 1, the best model to forecast international tourists' expenditures in Thailand during the
~oecified period is ARFIMA(1,-0.672,1 )-FIGARCH(1,-0.180,1) and the value of Akaike Criteria(AIC) from
is model is 7.10. Also the value of BIC from this model is 28.848. This model is the best model among
ese models because the values of both AIC and BIC are less than other models (Torre, Didier and
emoine, 2007). Consequently, the ARFIMA(1,d,1 )-FIGARCH(1,d,1) model was chosen for selection as
e best model for forecasting international tourists' expenditures in Thailand for this period (see more
etails in Table 2 and Figure 1.).

FIGURE 1: GRAPHICAL PRESENTATION OF FORECASTING INTERNATIONAL TOURISTS'

EXPENDITURES IN THAILAND DURING 2009 BASED ON ARFIMA-FIGARCH

(Unit: Million Baht)

70,000.00

60,000.00

~~~~ ........ - --.-- - - - -- - __ ._------_ -

40,000.00 -r- 2009(Actual)

__ 2009(Forecastj
30,000.00
-rk-MAE

10,ooo,ool-----/~~~=::::~~t~:::=:::::~-----------------
. K ~~~ ~~ ,/ >(

January February March April May

From: computed

INTERNATIONALJOURNAL OF :: ~ s " =S S ::=S=.A,RCH, Volume 10, Number 1, 2009 143

TABLE 2: FORECAST INTERNATIONAL TOURISTS' EXPENDITURES THAILAND DURING
2010 BASED ON ARFIMA(1,-0.67,1)-FIGARCH(1,- 0.18,1)

January 51,289.33 50,396.32 893.01 1 .:.

February 46,069.96 59,045.36 12,975.40 28.' :
March 50,094.28 61,707.85 11,613.57 23.' :
April 43,935.01 58,964.93 15,029.92 34.:;:'
May 37,400.20 60,547.96 23,147.76 61.: ~
June 57,433.63
July 49,819.87
August 47,754.59
September 47,030.48
October 51,924.80
November 47,489.96
December 36,220.29
'"
Total 228,788.79 628,336.03 12,731.93 29.S.!

January 44,533.06
February 43,785.36
March 47,070.06
April 51,546.47
May 46,299.59
June 50,342.92
July 44,152.45
From: cc:": ,':~:
(*MAE: Mean Absolute Error, **MAPE(%): Mean Absolute Percentage :--

4. CONCLUSIONS

For the period of 2009 to 2010 the best ARFIMA-FIGARCH model indicated the ARFIMA(1 ,,0:- ~
FIGARCH(1 ,-0. 180,1) model because this model has a value of Akaike Criteria(AIC) = 7.100 c': "
value of BIC = 28.848. The values of both AIC and BIC were much lower than other models. Hence
model has been selected to be the efficient model to forecast the international tourists' expendi., ':-:
Thailand for this period (see more details at Torre, Oidier and Lemoine, 2007). The ARFIMA(1,-C -:'
FIGARCH(1 ,-0.18,1) model predicts that in 2009 the expenditures of international tourists in Thails -: "
be 628,336.03 Million Baht (see more information in table 2 and figure 1). Moreover, the value c: '. ::...,=-
Absolute Error (MAE) is 12,731.93 in the period of January-May, 2009. Also the value of Mean A,es: -:
Percentage Error (MAPE(%)) is 29.84 % in the same period (see more information in table 2 anc :; .:
1). The international tourists' expenditures in Thailand for the period of 2009-2010 may expe-..: ..
some decline. The public-private partnership should plan to provide a remarkable chance to :: - .0

simultaneously a mixture of parts of the industry and to establish major tourism course of action ini: ~ '-:
to set in motion to lay the foundation at the educational level to assist all stakeholders even if poss : : .
initiate some of those measures necessary.

REFERENCES:

Akaike, Hirotugu., "A new look at the statistical model identification". IEEE Transactions on Au~: - ~
Control, vol. 19 (6),1996,716-723.
Amato, J D., "Risk aversion and risk premia in the COS market", BIS Quarterlv Review, 2005, 55-6-
Armstrong, J.S. and Collopy, - Speculations about seasonal factors, November .: : -.J
http://hops.wharton.upenn.eduifo,e:2s: ". I

INTERNATIONAL JCLR~":L C,,: 3.S1NESS RESEARCH, Volume 10, Number 1, 2009 144
 -lIie, RT., Bollerslev, T. and ""~(S:32- _="2:::::-2 ,j j,:::;;"a:::;: ;::;-::;"a ZS::: autoreqressive
iditional heteroskedasticity", Journa,'of=s::-::-=:-':3 vc. 7t" "leSS, 2<:
: I, W., and Trimbur, T., "Seasonal hetercskeoas: C~, in time series: r,::je[r-,; est.rnation, and testing",
::rking paper. 2005. .

:::erslev, T., Gibson, M. and Zhou, H., "Dynamic estimation of volatility risk premium and investor risk
.ersion from option-implied and realized volatilities", working paper, 2005.
ex, G. E. P., and Jenkins, G. M., Time Series Analysis: Forecasting and Control (rev. ed.), Holden-Day,
an Francisco, 1976.
-aitip, P. and Chaiboonsri, C., "Forecasting with X-12-ARIMA and ARFIMA: International Tourist Arrivals
India", Annals of the University of Petrosani. Economics, VoI.IX, 2009.
-J, Fong-Un, "A fractionally integrated autoregressive moving average approach to forecasting tourism
e.nand", Tourism Management, Vol. 29, 2008, 79-88.
cornik, Jurgen A. and Oorns, Marius, " A Package for Estimating, Forecasting and Simulating Arfima
,cdels: Arfima package 1.0 for Ox", The Netherlands, 1999.
-echtling, D.C., Practical Tourism Forecasting, Butterworth-Heinemann, Oxford, 1996.
ai, P., and Vause, N., "Measuring investors' risk appetite", Bank of England Working Paper Series, No.
:3,2005.
-anger, C. W. J., and Joyeux, R, "An introduction to long-memory time series models and fractional
"Terencing", Journal of Time Series Analysis, Vol. 1, 1980, 15-39. '

.crkrnaz, Turhan, Cevik, Emrah Ismail and Ozatac, Nesrin, "Testing for Long Memory in ISE Using
~FIMA-FIGARCH Model and Structural Break Test", International Research Journal of Finance and
::onomics, Vol. 26, 2009,186-191.
ewis, C. D., Industrial and business forecasting methods, Butterworths, London, 1982.
'acauley, F.R, The smoothing of time series, National Bureau of Economic Research, 1930.
'akridakis, S., Wheelwright, S.C. and Hyndman, R J., Forecasting Methods and Applications. Third
:ition, John Wiley and Sons, 1998.
"Ier, Don M. and Willams, Dan, "Shrinkage Estimators for Damping X-12-ARIMA Seasonal", discussion
aoer, Virginia Commonwealth University, USA, 2003.
-oietti, T., "Seasonal specific structural Time Series" Studies in Nonlinear Dynamics & Econometrics,
:J 8 (2), 2004.
:-omas M, Trimbur., "Seasonal heteroskedasticity in Census Bureau construction series", Statistical
~search Division U.S. Census Bureau Washin ton DC, 20233-9100., 2006.
crre, Didier, Delignieres, D. and Lemoine, L., "Detection of long-range dependence and estimation of
rsctal exponents through ARFIMA modeling", British Journal of Mathematical and Statistical Psychology,
JI. 60, 2007, 85-106.

::urism Council of Thailand (TCT), November 2009; www.thailandtourismcouncil.org/home.php.

UTHOR PROFILES:
.anchana Chokethaworn Associate Professor, Faculty of Economics, Chiang Mai University, Chiang
.'3i,Thailand.
ttaporn Sriboonjit. Assistant Professor of Faculty of Commerce and Accountancy, Thammasat
-iversity, Bangkok, Thailand.
hukiat Chaiboonsri. Ph.D. Candidate. F2CU:Y of Economics, Bangalore University, Bangalore, India.
r. Prasert Chaitip earned his Ph.D. a: 1'/:ss:ss:p; State University, USA in 1989,Associate Professor,
acuity of Economics, Chiang Mai Universi., ::=;;"[2"= Mai, Thailand.

INTERNATIONAL JOURN . !..L:F := ~S',:SS :;:::5EARCH, Volume 10, Number 1, 2009 145