SIMULATION MODELING OF FOREIGNERS

DEMAND FOR TOURIST SERVICES IN EGYPT
ASSOC. PROFESSOR PhD IVAN R. MARCHEVSKY,
HEAD ASSISTANT PROFESSOR RADOSLAV G. YORDANOV

І. Actuality of the problem
Tourism is one of the traditional, leading sectors in the economy of the Arab Republic of Egypt. Over the last years, sustainable growth has been observed both in the volume of tourist services demanded in the country and in the share of the tourist sector in the gross domestic product. According to data provided by the Ministry of Tourism, in 2007 tourism accounted for 11.3 % of the GDP. The sector provided 19.3 % of the revenue in foreign currency and 12.6 % of the employment in the country.
The government’s plans to increase investments in the sector further confirm the importance of tourism for the economy of the country. As a result of that policy, the number of tourists visiting Egypt is expected to reach 14 million annually by 2011, revenue from tourism should reach $12 billion per year, the number of people employed in the sector is expected to grow to 1.2 million, and according to the forecasts the country will have a 1.02% share of the global tourism market.
The accomplishment of this objective depends on two main factors – the capacity of the material base and changes in the size and structure of tourists flow by regions and on a global scale.
The improved development of the material base in tourism is the result of the priorities in the policy implemented by the government as well as the volume of investments made. Data provided by the Central Bank of Egypt show that in 2007 the tourist sector attracted 4% of total investments in the country and 13% of investments in the service sector, the investments in hotels
and restaurants having a 22% annual growth. These figures prove that the country has the necessary material capacity for tourism development.
Another major factor for the accomplishment of the established goals is the dynamics of global tourists flow. The Arab Republic of Egypt is a key player on the global tourism market. According to data provided by the World Tourist Organization (UNWTO), over the last five years the country has ranked among the World’s 25 Top Tourism Destinations, ranking 23rd in 2007. The main reason for that has been the steady trend towards an annual increase in the number of foreigners visiting the country (see Fig. 1).

Fig.1 Number of tourists that visited Egypt during the period 1999 - 2007

This sustainable growth in the number of tourists after 2001 is due to the following factors:
Recent liberalization of political regimes in many countries allowed a greater number of people to travel freely and to join the global flow of tourists.
Rising purchasing power of consumers from target markets – more than two-thirds of the tourists are people from European countries and the share of tourists from Eastern Europe has been increasing.
Expedient actions of the government of Egypt towards developing high-quality tourist products and promoting the country as a tourist destination.
Diversification of supply – alongside with traditional cultural and archaeological tourism (one-third of globally famous monuments being in Egypt) other varieties of tourism are being developed as well – recreation tourism, religious, therapeutic, sports, conference and eco-tourism, golf-tourism, desert tourism, yacht and submarine tourism, festival tourism, etc.
The analysis of the geographical structure of tourists visiting the Arab Republic of Egypt brings us to the conclusion that tourists from Europe have the biggest share in the flow of tourists to the country. Data provided by the Central Agency for Public Mobilization and Statistics (CAPMS) reveal that in 2007 the total number of tourists that visited Egypt was 9,756 people, the number of Europeans accounting for 75.6% of foreign tourists, 24.9% of them being people from East-European countries. Another interesting fact is the steady trend towards an increase in the number of East-European tourists and their rising share in the total number of people visiting Egypt (see Fig. 2). In 2007 alone, the number of tourists from Eastern Europe rose by 40.9% as compared with their number in 2006.

Fig.2 Structure of foreign tourists that visited Egypt during the period 2002-2007
At the same time, East-European countries and Bulgaria in particular, are among the countries where demand for tourist services is increasing the fastest at a steadiest rate. All these facts enable us to define the East-European market as a strategic one for tourism industry in Egypt.
All facts presented so far make us believe that making forecasts about the demand for tourist services by foreigners in Egypt could be a matter of real interest to the bodies engaged with tourism management in the country, as providing reliable forecasts is a starting point when trying to select an appropriate strategy for the development of the sector.

ІІ. Tourism demand evaluation model
In order to respond adequately to tourism demand, private entrepreneurs and/or government institutions providing tourist services need an efficient evaluation model. This, in practice, means to develop correct and precise evaluation criteria of the market potential.
A number of factors influencing tourism demand on a global scale have been enumerated in specialized literature. Many of them however are qualitative and do not allow quantitative evaluation (for example political risk, existing sights of interest, etc.). We therefore believe that when evaluating the potential of a market only factors that allow a relatively precise quantitative interpretation should be employed. An adequate evaluation can be achieved if the number of factors that influence market potential is reduced to three, namely: the number of tourists expected to visit the country during the year, the average number of nights per tourist, and the average daily expenditures per tourist. We would then apply the following evaluation model for tourism market potential based on our assumptions so far:
(1)

where:
is Market Potential
is Number of Tourists
is Number of Nights per Tourist
is Daily Expenditures per Tourist
The data we use to compute the model have been found on the web-sites of Egypt State Information Service. Table 1 gives the number of tourists (NT), the number of tourist nights (NTN), tourist revenues (TR), as well as their derivative indexes – number of nights per tourist (NNT) and revenues per tourist (RT). Some of the rows provide no values due to the incompleteness of the sources.
Table 1. Initial data for the tourist sector in Egypt
Number of tourists
(NT) Number of tourist nights
(NTN) Number of nights per tourist
(NNT) Tourism revenues
($) Revenues per tourist
($) Daily expenditures per tourist
($)
1. 2. 3. (2/1) 4. 5. (4/1) 6. (5/3)
1999 4797000 31002000 6,46 4000000000 833,85 129,02
2000 5506000 32788000 5,95 4300000000 780,97 131,15
2001 4648000 29813000 6,41 3400000000 731,50 114,04
2002 5191684 32662994 6,29 3422800000 659,29 104,79
2003 6044160 53129907 8,79
2004 8103611 81667918 10,08
2005 8607507 85171917 9,90 6797853309 789,76 79,81
2006 9082797 89304053 9,83 7192982456 791,93 80,54
2007 10714267 104496678 9,75 8200000000 765,33 78,47

Two approaches can be applied in order to determine the evaluation model – the deterministic and the stochastic (or simulation) approach. When applying the deterministic approach, each variable has an input of one value only and in result a single value is produced at the outcome of the model. According to the stochastic approach, on the other hand, it is impossible to determine the value of a random variable as it can fluctuate within a certain interval. The variable then could have an unlimited number of values within the limits of that interval. This, in turn, would result in just as large a number of values generated for the dependent variable at the outcome of the model. The crucial step when conducting a simulation analysis is to select an appropriate probabilistic distribution for each of the independent variables. When there are data for past periods available, formal criteria are applied to select the best fitting distribution (the Anderson-Darling criterion, the Kolmogorov-Smirnov criterion, and the -criterion). When historical data are not available, the people engaged with the analysis have to examine thoroughly even the smallest details that are related to the problem researched.
In a similar situation useful sources could be established standards, experts’ opinions, physical limits (boundaries) or limits generally accepted as normal, information from analogous product-market contexts, surveys conducted among target customers, theoretical treatments, empirical studies, verified assumptions, planned activities, etc.
The Monte Carlo method is most frequently applied when conducting simulation analyses. That method is an iterative procedure for evaluation of deterministic models where unknown variables and parameters of the model are represented by random probabilistic inputs instead of a single (most probable) input value. All probable combinations of inputs (or at least a sufficient number of them) are used in order to simulate all probable outputs (or at least a sufficient number of them). The result of iterative computation of the model with the Monte Carlo method is producing a range of values, each of them characterized by a certain degree of probability. The use of random values for independent variables and parameters with this model automatically turns it from a deterministic model into a stochastic one. The application of the Monte Carlo method (and all simulation methods in general) is appropriate when using a complex non-linear model which contains a large number of variables and parameters whose values are unknown and cannot be defined in another way. Data generated by the simulation are usually displayed as histograms by computing the probabilities for obtaining different values of the dependent variable (see Fig.3).
In order to evaluate the potential of Egypt tourism market for 2008 by a simulation, we should make several assumptions about the type of probabilistic distributions of variables included in the model with equation (1) – number of tourists (NT), number of nights per tourist (NNT), and daily expenditures per tourist (DET).
Simulation modeling of the NT variable. By applying regression analysis and quadratic trend model (y = a + bt + ct2) we project the time series Number of Tourists one step ahead. Thus we evaluate the number of expected tourists in 2008: NT2008 = 12 391 857 ( , α = 0,000). We consider it highly improbable that the real number of tourists in 2008 should remain outside the interval 12,391,857 2,000,000. A similar assumption allows us when modeling the variable Number of Tourists to employ the Beta-PERT distribution with a minimum value of 10,391,857 (what we obtain when we reduce 12,391,857 by 2,000,000), the most probable value being 12,391,857 and the maximum amounting to 14,391,857 (i.e. the sum of 12,391,857 and 2,000,000).
Simulation modeling of the NNT variable. Between 2003 and 2007, the variable Number of Nights per Tourist leveled off just slightly below 10. By applying descriptive analysis of this fragment of time series NNT (i.e. for NNT2003-2007) we obtain the results presented in Table 2.

Table 2. Descriptives for the row NNT2003-2007





8,79 10,08 9,67 0,51

When these data are available, several probabilistic distributions could be employed for simulation modeling the NNT variable: a continuous uniform distribution with a minimum value of 8,79 and a maximum of 10,08; a triangular or Beta-PERT distribution – with a minimum, a most probable and a maximum value of 8,79, 9,67 and 10,08 respectively; and a normal distribution with a mean value of 9.67 and a standard deviation of 0.51. As the number of cases in the time series is not sufficient for fitting empirical and theoretical distributions by applying formal criteria (the minimum number of cases observed should be at least fifteen), an expert decision must be made about the type of distribution. We select to work with normal distribution as it is typical for many socio-economic processes and phenomena.
Simulation modeling of the DET variable. We interpolate the missing values for the years 2003 and 2004 in the variable Daily Expenditures per Tourist by a cubic trend model (DET2003 = 96.09; DET2004 = 87.07; ; α = 0.004). Thus the time series DET1999-2007 has the following descriptives:

Table 3. Descriptives for the DET1999-2007 row





78 131 100,11 20,81

Formally, it would be possible to apply normal distribution here (as we have the mean value and the standard deviation). It would be inappropriate to do so, however, since the variable DET1999-2007 has explicitly expressed trend. It would be inappropriate to put a mean value in a simulation modeling of the DET variable for 2008, provided that a similar level was current in the middle of the examined period (from 1999 to 2007). That period has a steady trend of decreasing daily expenditures per tourist. That decrease obviously stopped when a level of $79 was reached. When taking into account absolute changes over the past years we come to the conclusion that the actual value of DET2008 could be expected to lie somewhere in the interval between $75 and $83.
We therefore consider it more appropriate to apply a Beta-PERT distribution with a minimum value of 75, most probable value of 79, and a maximum value of 83.
In the simulation analysis we use the Latin Hypercube Sampling procedure which provides us with a more balanced uniformly presented sampling as compared to the Monte Carlo Sampling procedure. After conducting 100,000 simulation iterations (100,000 replications), we obtain 100,000 values for the dependent variable. The mean value is 9,466,332,639. This is the evaluation (in $) of the market potential for tourist services in Egypt for the year 2008. Figure 3 shows the main findings of the simulation analysis:


Fig.3. Results of the simulation modeling

In addition to providing us with an estimated mean value, the Monte Carlo method also enables us to define the probability for obtaining a specific result. For example, the probability that the market potential would exceed $9 billion is 71.21%; there is a 25.15% probability that it would exceed $10 billion. The probability that the market potential would be more than $11 billion is only 2.7%. There is a 46.01 % probability that the market potential would be somewhere in the interval between $ 9 billion and $10 billion, etc.
It is not our intention to consider the issue of validation of the employed method or the reliability of the computed forecast evaluation in this paper. Our goal is to present these summarized results from the application of the simulation method of Monte Carlo. This method could also be useful for sensitivity analysis (i.e. evaluation of the influence of independent variables on the dependent one), optimization of target functions, etc.
The main conclusions from the tourism sector evaluation methodics presented in this paper and the results from its testing could be summarized as follows:
In the first place, the simulation can be useful approach for evaluation of tourist services demand in presence of uncertainty.
Second, by applying simulation modeling of market demand for tourism services we could eliminate some of the difficulties due to insufficient or unavailable information about past developments or the influence of random factors.
Third, the simulation approach could ensure a balance between variability and predictability in evaluation of tourist services demand.
In the fourth place, variability in the evaluation of the market potential enables the bodies engaged with tourism management to develop various strategies for tourism development.

Rate the article
Bad	1	2	3	4	5	6	7	8	9	10	Very good
Result : 100%							Participated in the vote : 1