Tourists Reaction Calculation Ethiopia Economy

While per capita income and population are supposed to proxy the generating potential of
the origin countries, CPI ratio, exchange rate ratio, and distance are meant to capture the
degree of costliness of visiting Ethiopia for tourists. Distance from Addis to capital cities of
the sending countries is an important variable that proxies cost of travel, importance of
nearness for cultural similarities (and willingness to move), business and transit travels.
Note
that price differential doesn’t account for travel costs as the price differential may be the
same between Ethiopia and Kenya and between Ethiopia and USA but the travel cost is
notably different. Urbanization rate, length of road network and the number of internet users
in Ethiopia are thought to account for changes in Ethiopia’s power of attraction for
international tourists while the lagged value of tourist arrivals is used to capture mouth-tomouth advertisement and the already existing potential and image of Ethiopia as a tourist
attraction to each country. African dummy is included to capture the effect of the presence of
African Union and United Nations Economic Commission for Africa in Addis Ababa on tourist
arrivals in Ethiopia.
Using these moment conditions, Arellano and Bond (1991) propose a two-step GMM
estimator where the error terms are assumed to be both independent and homoskedastic,
across countries and over time in the first step and such assumptions are relaxed in the
second step where the residuals obtained in the first step are used to construct a consistent
estimate of the variance-covariance matrix. This GMM estimator is generally called the
difference GMM estimator.
However, Blundell and Bond (1998) show that when the lagged dependent and explanatory
variables are nearly a random walk, lagged levels of these variables are weak instruments
for the regression equation in differences. Instrument weakness influences the asymptotic
and small sample performance of the difference estimator. In addition, Beck et al. (2000)
note that differencing may decrease the signal-to-noise ratio, thereby exacerbating
measurement errors.
Arellano and Bover (1995) describe how, if the original equation in levels is added to the
system, additional instruments can be brought to increase efficiency. In this equation,
variables in levels are instrumented with suitable lags of their own first differences.
Unfortunately, additional assumptions are required as the country specific effect appears
again in the system through the equation in levels. For the differences to be appropriate
instruments, we assume that there is no correlation between the differences of these
variables and the country specific effect.
The additional moment conditions for the second part of the system (the regression in levels)
are:
E[ (yit−s − yit−s−1)× (ηi + ε it ) ]= 0 for s =1 (5)
E[ (X it−s − X it−s−1)× (ηi + ε it ) ]= 0 for s =1 (6)
We use the moment conditions in 3, 4, 5 and 6 and employ a two-step GMM procedure to
generate consistent and efficient parameter estimates.
It is clear that consistency of the GMM estimator depends on the validity of the instruments.
Arellano and Bond (1991), Arellano and Bover (1995) and Blundell and Bond (1998) suggest
two specification tests. The first test, Arellano-Bond test of autocorrelation, examines the
hypothesis that the error term ε it is not serially correlated. Here, we test whether the
differenced error term is second order serially correlated . The second suggested test is the
Sargan test of over identifying restrictions, which tests the overall validity of the instruments
by analyzing the sample analog of the moment conditions used in the estimation process.
12
However, the Sargan statistic, which is the minimized value of the one-step GMM criterion
function, is not robust to heteroskedasticity or autocorrelation (see Roodman, 2006). Thus,
we use another statistic, the Hansen J statistic, which is the minimized value of the two-step
GMM criterion function, and is robust. Finally, the software and the package used for our
dynamic panel estimation are Stata 9.2 and xtabond2 of Roodman (2006) respectively.
Following the review of the literature on most frequently used determinants of tourist flows by
Crouch (1994) and Lim (1997) and applications by Naudé & Saayman (2004), the following
empirical model is set forth to be tested.
it
t t t
it it it it i it t it
Year Year Year
Urban Road Internet Africa Year Year Year
TA TA PCI EXR DIST CPI Kenya POP
β β β ε
β β β β β β β
β β β β β β β
+ + +
+ + + + + + +
= + + + + + + +
−
4 5 6
1 2 3
15 15 16
8 9 10 11 12 13 14
1 1 2 3 4 5 6 7
(7)
: Where TAit is the number of tourist arrivals from country i in year t ;TAit−1 is the number of
tourist arrivals from country i in yeart −1; PCI it is the per capita income of the sending
country i in year t ; EXRit is the exchange rate between the currencies of Ethiopia and
origin countryi in year t ; DISTi represents an air distance from the capital of the origin
country i to Addis Ababa; CPIit stands for the ratio of Consumers’ Price Indices (CPIs) of
Ethiopia and the origin country i in year t ; Kenyat represents the ratio of CPIs of Ethiopia
and Kenya in year t ; POP it stands for the total population of the sending country i in year t;
Urban t, Roadt, Internet t
respectively represent the urbanization rate, the length of road
network in Kilometers and number of internet users in Ethiopia at time t; Africa and Year
denote dummy variables for the sending countries being African and six years respectively
and ε it is the error term.