Beruflich Dokumente
Kultur Dokumente
Chukiat Chaiboonsri
Received a scholarship from the Indian Government study Ph.D. (Economics)
at Bangalore University from 2005-2010.
chukiat1973@yahoo.com
N. Rangaswamy Ph.D.
Professor & Chairman , Department of Economics, Bangalore University, Bangalore
Economics Department
Bangalore University
Authors:
Chukiat Chaiboonsri
Parsert Chaitip
N Rangaswomy
.
Best Wishes,
Voratas Kachitvichyanukul
APIEMS 2006 Conference Program Chair
Preface
Chukiat Chaiboonsri
1/06/2006
Contents
Page
Preface A
Contents B
Tables C
Abstract 1
1. Introduction 2
2. Research Aim and Objective 3
3. Study area in this research 3
4. The methodology and research Framework
4.1 The back ground concept of International Tourism Demand Model 4
4.2 Unit-Root Tests 6
4.2.1 DF-Test, ADF Test (1979) 6
4.2.2 Phillips-Perron Test (PP-Test:1987,1988) 7
4.2.3 The KPSS-Test (1992) 7
4.2.4 The DF-GLS Test (1996) 9
4.2.5 The ERS Point Optimal Test 9
4.2.6 The Ng and Perron (NP-test:2001) 11
4.3 Cointegration and Vector Error Correction Mechanism 12
5. The results of the research
5.1 The results of the Unit-Root Test 16
5.2 The results of the analysis of Modeling International 17
Tourism Demand in Thailand
5.2.1 The results of the analysis of Modeling 17
International Tourism Demand in
Thailand as in long-run from VAR Model
5.2.2 The results of the analysis of Modeling International 19
Tourism Demand in Thailand as in long-run
from VECM Model
5.2.3 The results of the analysis of Modeling International 21
Tourism Demand in Thailand as in Short-term dynamics
base on VECM Model
6. The conclusions of research and policy recommendations 23
Bibliography 26
Page
Chukiat Chaiboonsri,
Candidate in the Indian Government Ph.D. (Economics) program
at Bangalore University, 2005-2010.
chukiat1973@yahoo.com
N. Rangaswamy Ph.D.
Professor & Chairman , Department of Economics, Bangalore University, Bangalore
Abstract
-------------------------------------
*
This paper has been presented by me namely APIEMS 2006 Conference at AIT Bangkok , Thailand
Year GDP
(million baht)
1997 3,073,615
1998 2,749,684
1999 2,871,980
2000 3,008,662
2001 3,072,925
(Source: National Economic and social Development Board of Thailand)
Defined
And assume that (+ Yt), (-TCt), (- Pt) and explain that when income at
time t is increasing then the demand for international tourism is increasing
together. When a measure of transportation costs from the origin to destination
country at time t is increasing then the demand for international tourism is
decreasing. And when a measure of tourism price of goods and services is
increasing then the demand for international tourism is decreasing. And the
equation (1A) can be expressed in log-linear (or logarithmic) [equation number
(2A)] .
where
Where
When added term (ßi Σmi=1∆Yt-i) in equation (4B) then t-statistics value of
α before Yt-1 to be change as well as all t-statistics value of them to be change
too. So ADF-Test corrects for higher order serial correlation by adding lagged
differenced terms on the right-hand side. The hypothesis test for unit root in
time series data by ADF-Test method as for the DF-test method and same
conclusion about time series data are stationary or non-stationary.
where
And the PP-Test (tpp) has a t-statistic is computed as equation (8B) as well
as where tb , sb are the t-statistics and standard error of (β t) received from
regress in equation (5B) and s* is the standard error received from regress in
same equation.
where
The asymptotic distribution of the PP-Test (tpp) is the same as the ADF-
Test. And the hypotheses to be tested are follow up:
Yt = Xt + ε t ------- (9B)
where
X t : Xt = a 0 + b0 t + ε t [ intercept and trend ]
X t : Xt = a 0 + ε t [ intercept ]
ε t : is a stationary random error
Yt : is data test stationary or non-stationary
Where
L = the lag operator
α* = 1+c*/ T , c* = -7 : in the model with drift, c* = 13.5 : in
the linear trend case
z* = (1,t)
if a 0 > Critical values for the DF-GLS Test for a model with linear
trend (Elliott et al.1996) and rejected H0 or rejected null hypothesis as well as
conclusion that time series data is stationary or I(d) = I(0) .
if a 0 < Critical values for the DF-GLS Test for a model with linear
trend (Elliott et al.1996) and accepted H0 or accepted null hypothesis as well as
conclusion that time series data is non-stationary or I(d) = I(d) .
where
The P(T) statistics was used in ERS Point Optimal Test for unit root test in time
series data and show it below that : (See equation 12B)
where
and where
The null hypothesis of ERS Point Optimal Test for unit root test in time
series data can show below that :
if P(T) statistics > Critical values for ERS test statistic are
computed by interpolating the simulation result provided by ERS (1996,table
1,p.825) for T = {50, 100, 200, ∞} then accepted H0 : α = 1 : [ time
series data is non-stationary ] and said that time series data is non-stationary.
if P(T) statistics < Critical values for ERS test statistic are computed
by interpolating the simulation result provided by ERS (1996,table 1,p.825) for
T = {50, 100, 200, ∞} then accepted H1 : α = a*: [ time series data is
stationary ] and said that time series data is stationary (perception : the ERS –
Test was used to test unit root for time series data have big simple size at least
more than 50 observations).
where
c* = { -7 if xt = 1 or z* = 1, -13.5 if xt = (1,t) or
z* = (1,t) }
f0 = ΣT-1j =-(T- 1) γ* (j) . k(j/t),
j = the j-th sample autocorvariance of the ε t
t = a truncation lag in the covariance weighting
γ (j) = ΣTt = j+1(ε tε t- j)/ T, T = the number of observation
*
or
f0 = kernel- based sum-of-covariance estimator, and
autoregressive spectral density estimators
The null hypothesis of Ng and Perron(2001) Test for unit root test in time
series data can show below that :
But if equation (1C) has three variables (Y, W, Z) then these are three
possible long run relationships then can show equation numbers (2C), (3C) and
(4C).
where
This way of specifying the system contains information on both the short-
and long-run adjustment to changes in Zt (∆Zt) and rewriting (6C) as :
(equation number (7C)).
Which gives the n eigenvalues λ^1> λ^2 >…….> λ^n and the
corresponding eigenvectors ν^ = (ν^1 >,…….,> ν^n). Those r elements in ν^
which determine the linear combinations of stationary relationships can be
denoted β ^ =(ν^1 >,…….,> ν^r), that is, these are the cointegration vectors. The
VECM model has been develop by Hendry (1995) and he used the Johansen-
Juselius (JJ) methodology to study long-run relationship among M1, the price
level, output and interest rate in Canada. The VECM modelling procedure form
can be written and it begins by defining an unrestricted vector autoregression
(VAR) involving up to k-lags as well as can show below that : (See equation
(12C))
where
where
The simply of VECM model can rewrite based on concept of this research
follow by equation (14C) and this equation described below.
where
The simply of VECM model can rewrite based on concept of this research
follow by equation (16C) again and this equation described below.
∆ln(Expdt) ∆ln(Expdt)
∆ln(D1t) ∆ln(D1t)
∆ln(GDPt) ∆ln(GDPt)
∆ln(POt) = + Γ(L) ∆ln(POt) DZt + αβ′ Xt-1 + ε1 ----- (16C)
∆ln(RPt) ∆ln(RPt)
∆ln(RERt) ∆ln(RERt)
∆(SDRt) ∆(SDRt)
where
And the simply of VECM model in short–term dynamics was used in this
research can be rewrite as equation (17C) as well as descried below.
∆ln(Expdt) ∆ln(Expdt)
∆ln(D1t) ∆ln(D1t)
∆ln(GDPt) ∆ln(GDPt)
∆ln(POt) = + Γ(L) ∆ln(POt) + DZt + λα + ε1 ----- (17C)
∆ln(RPt) ∆ln(RPt)
∆ln(RERt) ∆ln(RERt)
∆(SDRt) ∆(SDRt)
where
From : computed
The results imply that when D(lnRER(-1)) of Thailand with the America
dollar increase by 1% then American tourist spending on goods and services in
Thailand increases by 1.23%. The results of short-term dynamics indicate that
D(lnRER(-2)) has a positive impact on international tourist spending on goods
and services in Thailand. The results imply that when D(lnRER(-2)) of
Thailand with the America dollar increase by 1% then American tourist
spending on goods and services in Thailand increase by 2.51%. And finally, the
results of short-term dynamics indicate that D(SDR(-1)) has a positive impact
on international tourist spending on goods and services in Thailand. The results
imply that when the D(SDR(-1)) of Thailand with America dollar increases by
1% then American spending on goods and services in Thailand increases by
0.06%. Furthermore, the results of short-term dynamics indicate that D(SDR(-
2)) has a positive impact on international tourist spending on goods and
services in Thailand. The results imply that when D(SDR(-1)) of Thailand with
the pound sterling increases by 1% then English tourist spending on goods and
services in Thailand increases by 1.10%.
From : computed
Diagnostic Test Malaysia China England Germany France America Canada
Autocorrelation
Test 40.47 35.13 35.81 40.37 37.63 38.33 33.42
LM(12) (0.27) (0.50) (0.92) (0.28) (0.39) (0.36) (0.56)
(P-value)
Normality Test
JB -Test 3.04 3.63 3.38 2.45 2.70 3.03 3.26
(P-value) (0.21) (0.16) (0.18) (0.24) (0.25) (0.21) (0.19)
Hetero
White –Test 554.56 542.50 - 552.38 555.03 554.25 550.97
(P-value) (0.39) (0.53) (0.41) (0.38) (0.39) (0.43)