Modelling the economic impact of international tourism on the Chinese economy: a CGE analysis of the Beijing 2008 Olympics

Transcription

1 Please reference this paper as: Li, S.N.; Blake, A.; Cooper, C. (2011) "Modelling the economic impact of international tourism on the Chinese economy: A CGE analysis of the Beijing 2008 Olympics", Tourism Economics, 17(2): This paper is for individual use only. Modelling the economic impact of international tourism on the Chinese economy: a CGE analysis of the Beijing 2008 Olympics SHINA LI UK Centre for Events Management, Leeds Beckett University, Leeds, UK. E- mail: S.Li@leedsmet.ac.uk. ADAM BLAKE School of Services Management, Bournemouth University, Talbot Campus, Poole BH12 5BB, UK. CHRIS COOPER Business School, Oxford Brookes University, Oxford, UK. International inbound tourism to China has grown phenomenally since 1980 and the hosting of the Olympics in 2008 was an important milestone. This paper takes the first step in applying computable general equilibrium (CGE) modelling to forecasting the economic contribution of tourism generated by the Beijing Olympics. CGE modelling has been widely applied to different tourism issues in many countries. In China, it has been used in fields such as taxation and international trade. However, economic impact studies on China s tourism using CGE modelling have not been found. The paper includes two types of estimations: ex ante and ex post. The ex ante estimation was conducted before the Beijing Olympics and thus predicted the impact of international tourism based on historical data, such as previous literature and historical statistics. The ex post estimation was conducted several months after the Beijing Olympics and the estimation was based on up-to-date statistics published by the China National Tourism Administration. The economic impact generated from the two types of estimations is compared. It was found that, while the economic impact of international tourism was predicted to be positive in the ex ante estimation, this impact was analysed as negative in the ex post estimation.

2 Keywords: economic impact; international tourism; Beijing 2008 Olympics; CGE modelling; tourism policy

3

4 The data for international inbound tourism in China are impressive. Both tourism receipts and tourist arrivals have grown rapidly since the late 1990s. In 2007, the volume of inbound tourists reached 132 million, with a spend of US$41.9 billion (Shao, 2008), and the UN World Tourism Organization forecasts that China will become the largest inbound international tourism destination in 2015, with 100 million international arrivals. Developing inbound tourism is considered as one of the important elements in China s development policies. Given the value of international tourism to the Chinese economy, it is therefore important to assess the economic contribution of international tourism using an appropriate method, and it is also of significance in identifying the key factors affecting the contribution of international tourism to China s economy. The Olympics can bring overall positive tourism impacts to host economies during the Olympic year, according to existing papers such as NSW Treasury (1997), Madden (2002, 2006) and Blake (2005). It seemed that the Beijing 2008 Olympic Games would be a major opportunity for China to generate foreign exchange from inbound tourism during Positive effects of tourism brought by the Beijing Olympics were forecast by a number of studies which were conducted before the Olympics (BMBS, 2003; Zou, 2007; Zhang and Zhang, 2008) and it was predicted that these positive effects would reach their peak in the Olympic year (Li, 2008; Li and Blake, 2009). However, the data collected by the China National Tourism Administration published several months after the Beijing Olympics revealed that both the number of inbound tourists and their expenditures decreased in 2008 compared to This paper takes the first step to applying computable general equilibrium (CGE) modelling in analysing the economic impact of China s international inbound tourism. It contributes to the tourism literature by building a China tourism CGE model and estimating the economic consequences of tourism in the context of the Beijing 2008 Olympics. A static China tourism CGE model is built based on the 2002 China input and output table. The paper consists of two types of estimations: the ex ante and ex post estimations. The ex ante estimation was conducted before the Beijing Olympics and thus predicted the impact of international tourism based on historical data, such as the previous literature and historical statistics. The ex post estimation was conducted several months after the Beijing Olympics and the estimation was based on up-to-date statistics published by the China National Tourism Administration. The main objectives here are to explain how CGE modelling can be used in the impact of international tourism on China s economy and to explore the features of the impact of international tourism brought by the Beijing Olympics. Four research questions are developed. First, how can we build a China tourism CGE model for evaluating the economic impact of international tourism in China? Second, how can we predict a change in international tourism during the Beijing Olympic Games and what do the findings of this CGE model mean? Third, what are the differences between the tourism effects of the Beijing Olympics predicted before the Olympics and analysed after the Olympics? Finally, what are the tourism implications for developing international tourism in China? It is important to note that a change in international tourism spending is different from the economic impact of international tourism (Li and Blake,

5 2009). The former is the model input and the latter is the model output. When the economic impact of tourism brought by the Beijing Olympics was modelled in the ex ante estimation, the model input was predicted based on historical data and the practices of previous Olympics. When this impact is analysed in the ex post estimation, the model input is estimated according to statistics by the China National Tourism Administration. The model input is considered as a demand shock to the CGE model and then the model output can be generated. In this paper, the model input is the difference between international tourism spending with and without holding the Beijing Olympics and it is not the total spending of international tourism during the Olympic year. In addition, the model input included only spending by international tourists whose main purpose of visiting was just the Beijing Olympics and no other purpose, such as visiting friends and relatives. In this case, it is difficult to calculate accurately the model input in studies on the economic impact of tourism generated by holding an event. CGE modelling of the economic impact of international tourism on national economies Existing studies of the impact of international tourism on the Chinese economy have used input output (I O) modelling (Yan and Wall, 2001; Oosterhaven and Fan, 2006). In terms of the methodology used in the two studies, there are crucial limitations in the assumptions of I O models, which may have led to unreliable results (Dwyer et al, 2000). For example, the assumptions of CGE models allow wages and prices to change flexibly and also place constraints on factors such as labour, capital and land, while I O modelling assumes fixed wages and prices and no constraints on factors. CGE modelling has been described as a paradigm shift from I O modelling in tourism economic impact evaluation (Dwyer et al, 2004a). The existing literature has applied CGE models in tourism impact studies, which cover a series of tourism issues such as a decrease or increase in tourism demand, shocks and tourism crises (for example, the foot-and-mouth disease and the 9/11 terrorist attacks), tourism taxation, environment, globalization and special events, for example, festivals, the World Cup and the Olympic Games. These studies relate to a variety of areas, that is both developed (such as the USA and the UK) and developing (Mauritius and Indonesia) economies, and both countries (Spain and Australia) and regions (Hawaii and New South Wales in Australia). However, existing literature on the economic impact of tourism in China has not yet been found. The economic impact of international tourism during the year of the Olympic Games is one of the most significant components of the effects of the Olympic Games on national economies (Blake, 2005). Other main components of the economic effects of the Olympic Games include operation expenditures by the Organizing Committee of the Olympic Games (OCOG), tourism legacies before and after the Olympic Games and investment in infrastructure and in Olympic venues and related facilities (Li and Blake, 2009). Although the total economic impact of the Olympics should include all these effects, given that the aim of this paper is to model the impact of international tourism, only

6 international tourism impacts generated by the Beijing Olympics will be analysed. The existing literature shows that the number of international tourists and tourism spending increased due to holding the Olympics, including the 1984 Los Angeles Olympics (Economic Research Associates, 1984), the 1992 Barcelona Olympics (Brunet, 1995; Dong, 2004), the 1996 Atlanta Olympics (Humphreys and Plummer, 1995), the 2000 Sydney Olympics (NSW Treasury, 1997; Haynes, 2001; Madden, 2002, 2006; Giesecke and Madden, 2007), the 2004 Athens Olympics (Papanikos, 1999) and the 2012 London Olympics (Blake, 2005). For example, Madden (2002) predicted that during the Sydney Olympic Games year the increased international tourism spending would be around AUS$400 million. Blake (2005) estimated there would be a growth of 364 million flowing into the economy from international tourism expenditures in the London 2012 Olympics. Although Madden (2002) and Blake (2005) applied CGE modelling to estimate the total economic impact of the Olympics, they did not analyse particularly the impact of international tourism brought by the Olympics. A China CGE model of the Beijing 2008 Olympic Games: assumptions and simulations Forecasting tourism impacts using a CGE model The model is calibrated to the China 2002 input and output (I O) table formulated by the China National Statistical Administration (2006a), which is the most recent data source. This table of 2002 prices has been updated to 2005 prices. It includes 122 industries, total export and both rural and urban households. Rural and urban households are aggregated into one representative household. For the sake of modelling tourism impact, total exports in the 2002 China I O table are separated into tourism exports and non-tourism exports using data on tourism expenditures in 2005 (Appendix 1). Tourism exports were formed based on foreign exchange earnings of international tourism listed in Appendix 1, and non-tourism exports were obtained by subtracting tourism exports from the original total exports. Demand by international tourists is a Cobb Douglas function of individual product demands (Figure 1a), with demand for the Cobb Douglas aggregate product being a downward sloping function of the aggregate tourism price (Figure 1b). The price elasticity of international tourism demand is estimated as 1.37, according to recent research conducted by Song and Fei (2006). Figure 1a can also be formulated in the following function: p T = n p i i where p T is the aggregate international tourism price; is a shift parameter; p i are individual product prices; i i = 1. Figure 1b can be expressed in the following function:

7 p T p T C D P(1) p(2) p(n) q T (a) (b) Figure 1. Demand by international tourists. q T = Q T pt e where q T is the demand quantity by international tourists; Q T is the benchmark demand quantity by international tourists; e is the exchange rate; and is the price elasticity of international tourism demand, which is 1.37 in the model. Model assumptions and model closure This China CGE model is a single-country static model, which captures the short-run impact of the Beijing Olympic Games in There are different economic agents in the economy, 122 industries, representative households, government and two factors, labour and capital. On the demand side, utility maximization subject to resource constraints leads to demand functions. On the supply side, profit maximization subject to resource constraints leads to supply functions. Prices, as one set of the most significant endogenous variables, are adjusted to make all excess demands equal to zero, which means that every market clears simultaneously. The Leontief function, Cobb Douglas function, the constant elasticity of substitution (CES) function and the constant elasticity of transformation (CET) function are applied to production and demand functions (Appendix 2). This model includes hundreds of functions and the software used to solve the model is the General Algebraic Modelling System (GAMS), which is a high-level modelling system used for solving mathematical programming problems. In particular, a subsystem of GAMS, called Mathematical Programming System for General Equilibrium Analysis (MPSGE), will be applied in this research to estimate the economic effects of the Beijing Olympics. A nesting structure can be used to illustrate a model structure. A three-level nesting structure of production is introduced in the model (Figure 2). 1 Each level shows an elasticity of substitution. The first level reveals supply of each product is a CES function of domestic supply and imports using the elasticity of. Production of domestic outputs is a CET function of exports and supply using the elasticity of. Production of domestic products uses a fixed coefficient

8 Exports Domestic output Supply Imports Zero Intermediate inputs Zero Value added input(1) input (2) input(n) factor(1) factor(2) factor(n) Figure 2. Three-level nesting structure of production. of zero between value-added and intermediate inputs, and between different intermediate inputs in levels two and three. In another group of the third level, the CES function with elasticity of is used between factors. The Armington elasticity of substitution between imports and domestic goods, the elasticity of transformation elasticity, and the elasticity of substitution between factors of production are taken from the Global Trade Analysis Project (Hertel, 1997) database and are adapted to the industries used in this study (Appendix 4). The government receives taxes on production and income, and consumes public goods. The remainder of government revenue after purchasing public goods is transferred to the representative household. Foreign investment is assumed to be fixed, which leads to a fixed current account balance. The representative household receives revenues through offering factors, labour and capital, to produce domestic and export goods, and it receives government transfers. The representative household purchases both domestic and import goods, pays taxes and saves additional revenues. International visitors are treated as a separate group of consumers who purchase national goods and services. The Armington assumption is used in the model, which specifies that domestic goods and import goods cannot be substituted perfectly by each other. Francois and Reinert (1997) point out that the Armington assumption can be used to solve two difficulties in the model. First, to assume that domestic goods and imported goods are not perfect substitutes, it is admitted that not only inter-industry trade but also intra-industry trade is possible in the economy. Second, the difficulty that the model cannot support a number of goods that exceed the number of factors can be solved (Francois and Reinert, 1997, p 7). In a small country assumption, domestic prices are determined by world prices, which are exogenous variables, and this small country is a price taker (Dervis et al, 1982). China is a big country in terms of territory and population, but its prices in most industries still could not affect world prices. Thus, a

9 small country assumption is employed, which indicates that domestic prices are determined by world prices. Sensitivity analysis is used to test the reliability and confidence of CGE models. The prediction of the Olympic economic impact was carried out in the ex ante estimation before the Olympic Games were held, which may put constraints on the availability of data. The main assumptions and parameters in the model, such as elasticities, may also introduce uncertainties into the model. A key elasticity in the model is the price elasticity of international tourism demand. In order to test the degree of effects that this elasticity may bring to the results, the base elasticity ( 1.37) is halved to and doubled to 2.74 in two sensitivity tests. Sensitivity analysis will also be applied to three groups of main elasticities Armington elasticity, output transformation elasticity and factor substitution elasticity used in the model by doubling one elasticity and maintaining the other two unchanged. For example, Armington elasticity is doubled, while output transformation elasticity and factor substitution elasticity are kept the same as the base case. Prediction of a change in international tourism brought by the Beijing Olympics (model input) Ex ante estimation This section estimates the model input, which is the demand shock to the CGE model. In this research, the demand shock is the international tourism spending brought by the Beijing Olympics. Both the number of international tourists arriving in China and their expenditures increased markedly from 1996 to According to the growth trend of China s tourism and the experience of previous Olympics, the ex ante prediction forecast that international tourism receipts would increase during the Olympic year. The increased international tourism expenditures brought by the staging of the Beijing Olympic Games were uncertain because the event had not occurred when the research was conducted. We try to address this uncertainty by introducing three scenarios (low, central and high) which assume different international visitor expenditures per visitor per day. According to the China National Tourism Administration, foreign visitor spending per visitor per day was US$136.85, US$ and US$ in 2000, 2001 and 2002, respectively. Based on the trend of the increase in foreign visitor spending per day per visitor, the 2008 figure was estimated to be US$150, which was included in the low scenario. Foreign visitor spending per day per visitor in the high scenario was taken from an analysis by the Beijing K&D Consulting Company, which predicted that a foreign visitor would spend US$300 per day during the Beijing Olympics. The central scenario assumed an average of the low and high scenario, which was US$225 per foreign visitor per day (Table 1). The expenditures per visitor per day of other types of international visitors, including athletes, officials, media visitors, sponsor visitors and the Olympic Family, were a certain proportion of that of foreign tourists calculated according to the estimation in previous studies (Airola and Craig, 2000; Blake, 2005). The Beijing Tourism Administration forecast that the number of foreign tourists attributable to

10 Table 1. Estimation of international visitor spending during the Beijing 2008 Olympic Games (central scenario ex ante). Type of Visitor Days per Spend per Spend per Spending, international visitors numbers visitor day per visitor total visitor (US$) (US$) (US$ million) Foreign tourists 250, , Athletes, foreign 13, , Officials, foreign 2, ,050 5 Media visitors, foreign 18, , Sponsor visitors, foreign 6, , Olympic family, foreign 4, , Total 294, staging the Olympics would increase by 250,000. The estimation of other types of international visitor number and days per visitor was adapted from the estimates of the Houston bidding for the 2012 Olympics (Airola and Craig, 2000) and of the London bidding (Blake, 2005). The increased international visitor spending due to holding the Beijing Olympics was projected to be US$601 million in the low scenario, US$899 million in the central scenario and US$1,199 million in the high scenario. Ex post estimation The actual tourist number and expenditures decreased rather than increased in Figure 3 displays the rate of percentage growth of China s international tourism over the previous 12 months in 2007 and The growth rates of both arrivals and receipts have been decreasing since February 2008, and from April 2008, the rates have been negative. Several factors may be responsible for the decline of China s international tourism arrivals and receipts in Major events/crises occurred in 2008, including the snow crisis (January and February), the Tibet riots and disturbances in the torch relay (March), an earthquake (May), turbulence in the world economy (from the second half of 2008) such as the global financial crisis, high oil prices and rising inflation, and government restrictions on visas due to holding the Beijing Olympics (from the end of May). Visa rules were tightened during the Beijing Olympics in order to enhance security at the Games. Restrictions included suspending some multiple entry visas and requiring more visa application documents, such as hotel booking and airline tickets. The new visa regulations were implemented according to the practice of previous Olympics. There is a delay in the response of any change in international tourism to an event/crisis, as it takes time for tourists and tour operators to reschedule or cancel a tour (Blake et al, 2003). The decline in the growth rate of China s international tourism in the first half of 2008 compared to 2007 was probably due mainly to the snow crisis, the Tibet riots and the torch relay, while the decrease after September, when the Paralympics finished, was likely caused by

11 Jan-07 Mar-07 May-07 Jul-07 Sep-07 Nov-07 Jan-08 Mar-08 May-08 Jul-08 Sep-08 Nov-08 Jan-09 Percentage growth over previous 12 months Arrivals Receipts 25 Figure 3. China s international tourism arrivals and receipts. Source: UNWTO and China National Tourism Administration ( the global financial crisis. It seems that both the Beijing Olympics (mainly visa restrictions) and turbulence in the world economy were mainly responsible for the decrease in international tourism during the four months June, July, August and September. In order to capture the change in inbound tourism receipts caused by the Beijing Olympics only, this paper separates the impacts of the turndown in the world economy from the impacts of the Beijing Olympics during the four months and the method of the separation used in this paper is developed from that used in Blake et al (2003). The change in international tourism receipts brought about by the Beijing Olympics ( r) is r = A E F where A is the actual international tourism receipts in June, July, August and September in 2008 (with the impact of both the Beijing Olympics and the turbulence in the world economy); E is the estimated receipts if the Beijing Olympics and turbulence in the world economy had not occurred in June, July, August and September 2008; and F is the change in tourism receipts brought by the turbulence in the world economy in June, July, August and September 2008 only. The data for actual international tourism receipts (row A) shown in Table 2 can be obtained from the China National Tourism Administration. The data for tourism receipts if the Beijing Olympics and the turbulence in the world economy had not occurred (row E) can be estimated based on the growth rate in China in 2007, which is illustrated in Figure 3. The monthly average growth rate in 2007 compared to the corresponding months in 2006 was 12%, and

12 Table 2. Estimation of international visitor spending (US$ million) during the Beijing 2008 Olympic Games (central scenario ex post). Calculation Jun-08 Jul-08 Aug-08 Sep-08 Total International tourism receipts (real data) A 3,084 3,256 3,198 3,142 12,680 Estimates (without the Olympics and turbulence in the world economy) E 3,490 3,736 3,806 3,577 14,608 Estimates (impact of turbulence in the world economy) F Change (impact of the Olympics) r ,238 this average rate during the four months (June, July, August and September) in 2007 was slightly lower at 11.2%. This ex post estimation takes the average of the two percentages, which gives 11.6%. The tourism receipts if the Beijing Olympics and the turbulence in the world economy had not occurred in 2008 (row E) can be estimated by multiplying 11.6% by actual tourism receipts in The difference between rows A and E (A E) gives the change in tourism receipts brought by both the Olympics and the turbulence in the world economy. In order to capture the impact of the Beijing Olympics only, the impact of turbulence in the world economy (F) needs to be subtracted from (A E), which is A E F. The percentage change in the growth rate of China inbound tourism spending brought by the turbulence in the world economy is estimated using the information of the growth rate of world inbound tourism arrivals, as data on the rate of world inbound tourism spending in 2008 are not available. The United Nations World Tourism Organization (UNWTO) evaluated that turbulence in the world economy was the major factor which caused a decrease in world inbound tourism in 2008 (UNWTO, 2009). In Table 3, the average growth rate of the world inbound tourism in the first half of 2008 is calculated to be 5%, which represents a growth rate when the turbulence in the world economy had not yet happened. If we subtract this average growth rate (5%) from the real percentage change in June, July, August and September, we obtain the change in growth rate brought by the impact of turbulence in the world economy for the four months. The average rate ( 5.275%) of the four months is used to estimate the change in China s inbound tourism caused by turbulence in the world economy in row F (Table 2) through multiplying 5.275% by the real data of China s inbound tourism receipts in Finally, the change in international tourism receipts brought by the Beijing Olympics ( r) can be calculated using the formula (A E F), and thus a total decrease of US$1,238 million in inbound visitor spending brought by holding the Beijing Olympics is estimated.

13 Table 3. Growth rate (%) of world inbound tourism arrivals Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Average Growth rate of arrivals real data) (average for the first half of 2008) Difference between with and without turbulence in the world economy The international tourism impact of the Beijing Olympics (model output) The economic impact of international tourism brought by holding the Beijing Olympics generated from ex ante (three scenarios) and ex post estimations are discussed at the whole economy and industry level. Table 4 shows the main results of the impact of international tourism on the Chinese economy due to holding the Beijing Olympics. Economic welfare, which shows the impact on households, is measured by the equivalent variation (EV in row A). Blake (2000) points out that a welfare change (EV) is equivalent to a real GDP change. The change in international tourism demand (row B) is the demand shock (or the model input) estimated in previous sections (Tables 1 and 2). The change in real tourism consumption (row C) is different from row B in that the former takes the change in prices into account while the latter does not. The change in the price of foreign tourism consumption is shown in row D. The ratio of change in welfare to the change in real tourism consumption is a key ratio (A/C), which reveals whether low and high scenarios differ from the central scenario in the ex ante estimation. In the ex ante estimation, the increase in welfare, which was US$177 million in the central scenario (row A), would be brought to households by holding the Beijing 2008 Olympic Games. The change in real tourism consumption (US$862 million in the central scenario in row C) is less than the change in international tourism demand (US$899 million in row B). This occurred because there was a 0.09 increase in the price of foreign tourism consumption (row D), which offset real tourism spending. Every unit change in real tourism consumption would change welfare by (row A/C). In the low and high scenarios, it was projected that households would gain welfare equivalent to US$118 million and US$236 million, which was stimulated by an increase of US$576 million and US$1,151 million in real tourism consumption, respectively. Change in EV per change in real tourism consumption (0.205) remains the same in the three scenarios. Sensitivity tests are conducted by changing key elasticities used in the model: price elasticity of international tourism demand, Armington, output and factor substitution elasticity. The

14 Table 4. Macro-economic impact of the Beijing Olympic Games. Calculation Ex ante Ex post Low Central High Change in EV (US$ million) A Change in international tourism demand (US$ million) B ,199 1,238 Change in real tourism consumption (US$ million) C ,151 1,188 Change in price of foreign tourism consumption (%) D Change in tourism expenditure (US$ million) Change in EV per change in international tourism demand E A/B , , Change in EV per change in real tourism consumption A/C Change in EV per change in tourism expenditure A/E results indicate that the change of elasticities will not change the model output qualitatively (Appendix 5). The ex post estimation shows that there would be a welfare loss of US$297 million brought by a US$1,238 million decrease in international tourism demand. This means that every US$100 decrease in tourism demand would cause US$25 decrease in welfare (row A/C). When the economic impact of the Beijing Olympics was modelled in ex ante estimates, model input was generated from the practice of previous Olympics, previous studies and the trend of development of China s international tourism. The practice and trend of China s tourism indicate two important points: the previous Olympics always brought a tourism boom, especially during the Olympic year; the percentage change of international tourism arrivals in China and tourism revenues has been increasing during the recent decade. Therefore, it was predicted that the Beijing Olympics would bring an increase of US$899 million in international tourism demand. By employing a China tourism CGE model, it can be modelled that an increase of US$899 million in tourism demand would increase welfare by US$177 million. However, real statistics collected by the China National Tourism Administration revealed that the number of international tourist arrivals and their expenditures decreased in The ex post prediction of model input was based on the real statistics and, as a result, the Beijing Olympics caused a welfare loss. Table 5 shows the economic impact of international tourism brought by holding the Beijing Olympics at industry level. For the purpose of presenting the results, 122 industries were aggregated into 13 industries (Appendix 6). In the ex ante estimation, holding the Beijing Olympics was predicted to provide a stimulus to tourism-related industries through an increase in international tourism demand. As a result of the expansion of international tourism, outputs of tourism-related industries increased. For example, outputs of air

15 Table 5. Industry-level impact in the central scenario. Central scenario Change in tourism exports Change in output (US$ million) (%) Ex ante Ex post Ex ante Ex post Primary industry Secondary industry Railway transport Road transport Water transport Air transport Communications Accommodation Catering Tourism Residential services Recreation Other services transport, recreation and accommodation were predicted to increase by 0.98, 0.60 and 0.43, respectively. Air passenger transport was the largest tourism exporter, with an increase of US$118.4 million. However, the growth of international tourism during the Beijing Olympics brought crowding-out effects on secondary industries. This allocation consequence captured by CGE models has also been observed in previous studies. The secondary industry experienced a percentage reduction in output (0.046). Sensitivity analysis of the industry level also shows that changing elasticities will not change the model output qualitatively (Appendix 7). On the contrary, the Beijing Olympics brought a decrease in tourism exports in all industries, and the largest decrease of US$248 million can be seen in air transport in the ex post estimation. This decline was attributable to a decline in tourism expenditure in the Olympic year, which has been discussed previously. Tourism-related industries also experienced a decline in percentage change in output, with the largest decreases in air transport (1.324), recreation (0.815) and accommodation (0.594). It seems that the industry affected most adversely is air transport. Rather than being crowded out, as shown in the ex ante estimation, there was an increase of in percentage change in output in secondary industry in the ex post estimation. Conclusions The main contribution of this paper is that it is the first time that CGE modelling has been applied to China s tourism issues. It models the economic impact of international tourism brought by holding the Beijing Olympics. This economic impact generated in the ex ante estimation is compared with that generated in the ex post estimation. It was discovered that while the economic

16 impact of international tourism was predicted to be positive in the ex ante estimation, this impact was analysed to be negative in the ex post estimation. This was due mainly to a different model input. The model input, which was a change in international tourism revenue during the Olympic year in this research, was predicted as positive in the ex ante estimation, but as negative in the ex post estimation. This was different from previous Olympics, which brought a tourism boom to the host country in the Olympic year. There were several factors which caused a decrease in international tourism arrivals and expenditure. New rules on tightening visas was one of the most crucial factors. There was a debate on the impact of new visa rules on China s international tourism. One viewpoint was that the new rules made it harder to obtain a visa and thus reduced international tourism arrivals during the Olympics. Another viewpoint supported the visa rules, as Olympic security was the highest priority. This paper approves of the second view. If the Beijing Olympics went smoothly, it would be more likely to create a good image of China as a tourism destination. More international tourists would travel to China after the Olympics, which would enhance tourism legacies. This would have been an ideal scenario if the global financial crisis had not occurred. However, when China s international tourism market was facing this crisis, even the tourism legacies brought by the Beijing Olympics could not stimulate international tourism arrivals automatically. Therefore, it is important for the China National Tourism Administration and different tourism-related firms to promote China with the help of the Olympic legacy. Tour operators could design a variety of Olympic-themed tours, including visits to Olympic venues, to attend events or to visit the Olympic museums. For example, the main Olympic stadium (the Bird Nest) was covered with Christmas decorations for tourists to photograph, and concerts have been held in the National Aquatics Centre since September In addition, there is a need to upgrade tourism facilities further, improve tourism services after the Olympic Games and hold other events (such as the 60th Anniversary of China s National Day in 2009 and the Shanghai 2010 World Expo). The difference between the economic impact of international tourism generated in the ex ante and ex post estimations implies that tourism is vulnerable to external changes. When the impact was forecast before the Olympics, the positive aspect of hosting the Beijing Olympics, such as changing the image of the host country and attracting international visitors, was considered. However, when the impact was analysed after the Olympics, it seemed that the negative aspect, that is restrictions on visa rules due to Olympic security needs, counteracted and even exceeded the positive aspect. This makes forecasting of tourism demand or tourism impact, using either CGE modelling or other methods such as econometrics, more difficult. One possible solution could be the application of the scenario method. This paper has employed this method through estimating international visitor spending in high and low scenarios. However, the low scenario did not consider the situation of the change of tourism revenue becoming negative, as according to the practice of previous Olympics and the trend of the development of China s international tourism, there was no implication during the Beijing Olympics of a possible decline. A suggestion can be drawn from the comparisons of the impacts predicted

17 before and analysed after the Olympics: a wide range of scenarios, including even the worst and best situations, can be applied to studies of forecasting of tourism demand and impact. The prospect of applying CGE modelling to China s tourism studies is promising. Further research can extend attempts to build dynamic or multiregional models (for Beijing, the rest of China or different regions of China), estimate the long-run impact (before, during and after the Olympics), run simulations using different tourism policy scenarios and include other elasticities in the sensitivity analysis. This paper has discussed the tourism impact during the Olympic year, but it does not model the tourism legacies brought by holding the Beijing Olympics, since tourism legacies are a long-run impact. In order to capture the total economic impact of tourism due to holding the Beijing Olympics, an analysis should be conducted in five to ten years. Endnotes 1. Figure 2 shows a basic structure of this model and Appendix 3 provides more detail. 2. Except in 2003, when tourism was affected negatively by the outbreak of severe acute respiratory syndrome (SARS). References Airola, J., and Craig, S. (2000), The Projected Economic Impact on Houston of Hosting the 2012 Summer Olympic Games, Department of Economics, University of Houston, Houston, TX. Blake, A. (2000), The Economic Impact of Tourism in Spain, TTRI Discussion Paper No 2000/2. Blake, A. (2005), The Economic Impact of the Olympics, TTRI Discussion Paper No 2005/5, Nottingham University Business School, Nottingham. Blake, A., and Sinclair, M.T. (2003), Tourism crisis management US response to September 11, Annals of Tourism Research, Vol 30, No 4, pp Blake, A., Sinclair, M.T., and Sugiyarto, G. (2003), Quantifying the impact of foot and mouth disease on tourism and the UK economy, Tourism Economics, Vol 9, No 4, pp BMBS [Beijing Municipal Bureau of Statistics] (2003), A report of the impact of the Olympic expenditures on Beijing s economic development: an application of input output techniques [in Chinese], in Liu, Q., ed, Study on Beijing Olympic Economy, Beijing Press, Beijing, pp Brian, R., and Copeland, R. (1991), Tourism, welfare and de-industrialization in a small open economy, Economica, Vol 58, pp Brunet, F. (1995), An economic analysis of the Barcelona 92 Olympic Games: resources, financing and impact, in Miquel, M.D., and Botella, M., eds, The Keys to Success, Autonomous University of Barcelona, Barcelona, pp China National Statistical Administration (2006a), The China 2002 Input Output Table, China Statistics Press, Beijing. China National Statistical Administration (2006b), The China Statistical Yearbook 2006, China Statistics Press, Beijing. Dervis, K., De Melo, J., and Robinson, S. (1982), General Equilibrium Models for Development Policy: A World Bank Research Publication, Cambridge University Press, Cambridge. Dong, J. (2004), Economic Effects of Olympic Games on Host City, Economic and Scientific Publishing Co, Beijing. Dwyer, L., Forsyth, P., Madden, J., and Spurr, R. (2000), Economic impacts of inbound tourism under different assumptions regarding the macroeconomy, Current Issues in Tourism, Vol 3, No 4, pp Dwyer, L., Forsyth, P., and Spurr, R. (2004a), Evaluating tourism s economic effects: new and old approaches, Tourism Management, Vol 25 pp Dwyer, L., Forsyth, P., and Spurr, R. (2004b), Estimating the impacts of special events on an economy, Journal of Travel Research, Vol 43, pp Economic Research Associates (1984), Community Economic Impact of the 1984 Olympic Games in Los Angeles and Southern California, Los Angeles Olympic Organizing Committee: Los Angeles, quoted

18 in Kasimati, E. (2003), Economic aspects of the Summer Olympics: a review of related research, International Journal of Tourism Research, Vol 5, pp Francois, J.F., and Reinert, K.A. (1997), Applied Methods for Trade Policy Analysis: A Handbook, Cambridge University Press, Cambridge. Giesecke, J., and Madden, J. (2007), The Sydney Olympics, Seven Years On: An Ex Post Dynamic CGE Assessment, CoPS General Paper No G-168, Centre of Policy Studies, Monash University. Haynes, J. (2001), Socio-economic Impact of the Sydney 2000 Olympic Games, The 2001 Seminar of the International Chair in Olympism, Barcelona. Hertel, T.W., ed (1997), Global Trade Analysis: Modelling and Applications, Cambridge University Press, Cambridge. Humphreys, J.M., and Plummer, M.K. (1995), The Economic Impact of Hosting the 1996 Summer Olympics, The University of Georgia and IRE Advisors, Georgia. Li, S.N. (2008), What did Sydney Olympics tell Beijing? selected interviews with personages in politics, business and academics relating to the Sydney Olympics, in Olympic Economic Research Institute, eds, 2007 Report of the Beijing Olympic Economy, Beijing Publishing House, Beijing. Li, S.N., and Blake, A. (2009), Estimating Olympic related investment and expenditure, International Journal of Tourism Research, Vol 11, pp Madden, A.R. (2002), The economic consequences of the Sydney Olympics: the CREA/Arthur Andersen Study, Current Issues in Tourism, Vol 5, No 1, pp Madden, A.R. (2006), Economic and fiscal impacts of mega sporting events: a general equilibrium assessment, Public Finance and Management, Vol 6, No 3, pp NSW Treasury (1997), Economic impact of the Sydney Olympic Games, Research and Information Paper, Office of Financial Management, New South Wales. Oosterhaven, J., and Fan, T. (2006), Impact of international tourism on the Chinese economy, International Journal of Tourism Research, Vol 8, pp Papanikos, G.T. (1999), Tourism Impact of the 2004 Olympic Games, Tourism Research Institute, Athens [in Greek], quoted in Kasimati, E. (2003), Economic aspects of the Summer Olympics: a review of related research, International Journal of Tourism Research, Vol 5, pp Shao, Q.W. (2008), Speech of Shao, Qi Wei at China s Tourism Conference, Tourism Research [in Chinese], Vol 1, Song, H., and Fei, B. (2006), Modelling and forecasting international tourist arrivals to mainland China, China Tourism Research, Vol 3, No 1, pp UNWTO (2009), UNWTO World Tourism Barometer, Vol 7, No 1, World Tourism Organization, Madrid ( accessed 21 March 2009). Yan, M., and Wall, G. (2001), Economic perspectives on tourism in China, Tourism and Hospitality Research, Vol 3, pp Zhang, W., and Zhang, X. (2008), Research on the effect on tourism and the countermeasures for the Beijing Olympic Games in the year of 2008, Economy and Management, Vol 22, No 2, pp Zou, T.Q. (2007), The Tourism Impact of Olympic Games The Impellent Effects of the Beijing Olympics on Chinese Tourism Industry, Social Sciences Academic Press, Beijing.

19 Appendix 1 Table A1. Foreign exchange earnings of international tourism and its composition in Item Value (US$ million) Percentage Total 29, Long-distance transportation 8, Civil aviation 5, Railway Highway Waterway Visiting 1, Accommodation 3, Food and beverages 2, Shopping 6, Entertainment 1, Post and communication services Local transportation 1, Other services 3, Source: China National Statistical Administration (2006b, Table 19-2). Appendix 2 The main technologies applied in the model are the Leontief, the Cobb Douglas (C D), the constant elasticity of substitution (CES) and the constant elasticity of transformation (CET) technology. The following shows the basic forms of different functions including the Leontief, the C D, the CES and the CET functions and their input demand and price functions. A basic form of the Leontief function is defined as: Q j = min X 1,j, X 2,j, X 3,j,..., X i,j (A1) 1,j 2,j 3,j i,j Input demand and price functions for Leontief technology can be generated as: x i,j = q j (A2) p j = i,j p i i (A3) where Q j is output of good j, X i,j is the use of input good i in the production of good j, i,j is the fixed coefficient for input i in the production of j, i,j is the share of value of i in j, the lower case of variables are corresponding variables compared to benchmark values. For example

20 x i,j = X i,j/ X. i,j A basic form of the C D output (or utility) function is defined as: Q j = A X i i i Input demand and price functions for C D technology can be generated: (A4) p j x i,j = q j p i (A5) p j = (p i ) i,j i where = 1 and A is a shift parameter, which are constant and determined by i i technology. A general CES production function is defined as: i ( 1)/ /( 1) (A6) Q j = A[ i,j X i,j ] (A7) Input demand and price functions for CES technology can be generated: p j x i,j = q j p i p j = ( i,j p i 1 ) 1 / (1 i i where i,j is the share parameter and is the elasticity of substitution between i. A general form of CET function is defined as: ( 1)/ /( 1) (A8) (A9) X j = A[ i,j Q i,j ] (A10) Input demand and price functions for CET technology can be generated: p j q i,j = x j p i i 1 1 / (1 (A11) p j = ( i,j p i ) (A12) where the quantities Q i,j are a series of the output goods produced using the same input good X j, with an elasticity of transformation. The mathematical equations used in building CGE model for China are adapted from the above functions. Appendix 3 The use of nested functions in CGE models is convenient as it shows different elasticity parameters and the relationship of the inputs and outputs of functions. The model for China is built based on the following two nesting structures.

21 Armington goods/domestic sales CES: SUBA j Imported goods Domestic goods Production for exports CET: Total inputs Leontief: 0 Value added Intermediate inputs CES: SUBV j Leontief: 0 Factors Sector-specific factors Intermediate good 1 Intermediate good n Figure A1. Five-level nesting structure. Figure A1 displays a four-level nesting structure (from top to bottom): (1) The first level shows that Armington goods (or domestic sales) are CES functions of imported goods and domestic goods. The elasticity of substitution between imported goods and domestic goods is SUBA j. (2) The second level shows that total inputs are CET functions of the production of domestic goods and production for exports. (3) The third level indicates that total inputs are also Leontief functions of value added and intermediate inputs. (4) The fourth level includes two forms of functions. First, value added is a CES function of inputs of factors and sector specific factors. Second, intermediate inputs are Leontief functions of each individual intermediate good. Figure A2 is a two-level nesting structure, which specifies the functions of welfare. The first level indicates that household welfare is a C D function of household consumption and savings and the second level shows that savings are a C D function of investment in each individual good. Functions for market prices and quantities of output and input in individual nests will be developed for production, value added, Armington goods (or domestic sales), imports, exports, factor supply, savings and investment, household welfare, household income and government revenues, respectively. These functions are developed from the basic functions of CES, Leontief, C D and CET shown in Appendix 2.

22 Household welfare C D: Household consumption Savings C D: Investment in good 1 Investment in good n Figure A2. Two-level nesting structure. Appendix 4 Table A2. Elasticities used in the model. (Armington (output transformation (factor substitution elasticity) elasticity) elasticity) Primary industry a Secondary industry Railway transport Road transport Water transport Air transport Communications Accommodation Catering Tourism Residential services Recreation Other services Note: a A range is provided where more than one elasticity is used in different industries.