II. Literature Review
2.2 Aggregation bias
The aggregation bias has been firstly point out in macroeconomic (Theil, 1954) and urban economics (Goodman, 1998). As one of the rapid growing services in the global economy, the contribution of tourism on economic growth has long been investigated in past literatures (West, 1993; Archer, 1995; Gunduz & Hatemi-J, 2005;
Kim et al., 2006). Among these literature, the problems of the aggregation bias may result in that relationship between tourism and economy with aggregate data may offer little guidance for policymakers when formulating tourism marketing and economic growth policies (see also Oh, 2005). This problem of aggregation bias has been shown to exist in housing market by Zietz, Zietz, and Sirmans (2008) and has been controlled of using disaggregate analysis by Corgel, Lane, and Walls (2013).
Most hotels have very different price and quality characteristics. The hotel industry in U.S. can be organized into six chain scale divisions as follows luxury, upper upscale, upscale, upper midscale, midscale, economy, independent (Corgel et al., 2013). According Taiwan Tourism Development Act, the hotel industry has been categorized into international tourism hotel, hotel, and host family by different room number, operation purpose, and facility. One important concern here is that the heterogenous nature of hotel by different service quality, physical location, and customer base will suspect resulting in aggregation bias.
A strategic location will decide the short or long term superior performance of the hotel in terms of revenue generation (Johns et al., 1997; Nicolau, 2002). The different hotel locations which attribute the characteristics of accessibility, urban development, public good and service and agglomeration effect (Yang, Wong, &
Wang, 2012) could present an essential factor that strongly influences a tourist’s hotel selection decision. Rivers, Toh, and Alaoui (1991) noted that the convenience of
17
location significantly affect tourists’ hotel selection. Lewis and Chambers (1989) showed that location was the most important factor influencing hotel selection for business tourists. However, some previous researches offered that leisure tourists also place high priority on locational advantage in selecting their hotels (Barsky & Labagh, 1992; Chu & Choi, 2000). Otherwise, Tsaur and Tzeng (1995) evidenced that hotel location factors, such as the convenience of transportation and parking, were among most important factors in assessing the service quality of hotel. Therefore, this paper employee the disaggregate data analysis with different location of Taiwanese international tourism hotel to offer more information to the hotel managers and policy makers. The aggregate data such as international visitor arrivals dataset does not accurately demonstrates that the impact nationality different on Taiwanese international tourism hotel markets.
This study try to examine the relationship among tourism expansion, economic variable, and hotel performance using the disaggregated hotel location and tourist arrivals to illustrate the aggregation bias problem and the advantage of a better explanation on the role of each tourism and economic variable on hotel performance in Taiwan. Therefore, the finding may provide more precise information to hotel manager when choosing the tourism marketing destination.
18
CHAPTER III
METHODOLOGY
According to pioneer research of economy, tourism, and hotel firm performance, this paper focuses on the two main research issues as shown in figure 5. We focus on the disaggregation investigation among the relationship of exchange rate, tourist arrival and hotel performance base on the difference of tourist nationality and hotel location.
19
3.1 Variables and pooled regression for estimated beta
Panel regression methodology can be estimated by two popular statistical models for the fixed effects method and the random effect method (Dimitrios, 2005). Baltagi (2005) and Hsiao (1986) states that panel regression methodology can reduce problems associated with multicollinearity and estimation bias by control heterogeneity and time-varying. Under the fixed-effect model we assume that there is common constant for all cross sections. That is, all differences of the constants are due to sampling error. By contrast, the random-effects model allows the constants to differ-that the constant are associated with random parameters underlying different population.
In order to understand the fixe effects model appropriation, the F-test is used to test the validity of the fixed effects method. The null hypothesis is that all the constants are the same, and the OLS method is more suitable than the fixed effect method. This hypothesis is tested by the F test, which is based on loss of goodness-of-fit. The Hausman specification test compares the fixed versus random effects is used to determine if the fixed or random effects should be appropriate.
Under the null hypothesis, the individual effects are uncorrelated with the other regressors in the model (Hausman, 1978). If null hypothesis is rejected, which means that individual effects are correlated to the other regressors and a random effect model produces biased estimators. Therefore, the fixed effect model is better than random effect model.
Chen (2010) employees the F-test and Hausman test to determine the appropriate model of the fixed or random effects. He found that fixed effect model should be used to performance panel regressions base on the hotel firm performance. Therefore, this paper uses the F-test and Hausman test to determine the appropriate model in the
20
panel dataset.
The hotel operation variable are collected from Taiwan Tourism Bureau report and divided into four categories: rooms、occupancy、 Occupancy rate、Average rate Revenue and Employee. Room is measured by the room numbers of hotel capacity.
Occupancy is the number of rooms occupied. Occupancy Rate presents the monthly number of rooms occupied divided by the monthly hotel capacity of the room numbers. Average rate is the monthly average hotel room price. Revenue is the monthly room department revenue income. Employee indicates the monthly hotel employee. Pioneer research of exploring the interactions between the hotel room rates and the number of international inbound tourists only consider the operation variable of room rate with limitation of higher frequency data, monthly data (Lee, 2010). This paper introduces additional variables in the empirical model to further investigate the relationship between the hotel operation and the number of international arrivals.
Total international tourist arrivals of hotel are also collected from Taiwan Tourism Bureau monthly report. This report is compiled from Taiwan Tourism Bureau completed by all international tourism hotels monthly in Taiwan. It can be used as a proxy for international tourism hotel demand. This variable includes the arrivals from North American, Japan, China, Australia, Asia, and Europe. The domestic tourists are also included in this paper.
21
The following pooling linear regression for operation variable of hotel
i
was developed for evaluating the influence of hotel international arrivals.
The models are classified into seven categories: Taipei, Kaohsiung, Taichung, Eastern, TaoChuMiao, Scenery, and Others by Taiwan Tourism Bureau. The dependent variable (operationit) is denoted as revenue, room rate, occupancy of hotel.
The explained variables include the numbers of domestic tourists, numbers of U.S.
tourists, numbers of Japanese tourists, numbers of Asian Tourists, numbers of European tourists, numbers of Australian tourists.
22
3.2. Data
The empirical data comes from the monthly operation report and tourist numbers of Taiwan international hotel published by Taiwan Tourism Bureau from 1999 January to 2013 October. To understand the relationship between operation variable and tourist variant during the whole period, the dataset used in this study with complete information contains 60 different international hotels across 178 months and 10680 observations were formulated a panel dataset.
Figure 6, 7, and 8 show that hotel performance as revenue, occupancy rate, and room rate present uptrend except the year 2003 of SARs event during the sample periods. However, we also find that hotel locations difference seemly lead to different hotel performance pattern during the sample periods.
Table 2 reports the sample firm-year mean and standard deviation of hotel important operation variables for the whole sample periods in the seven regions. The Hotel Classification system of Taiwan Tourism Bureau base on the geographical location is classified into Taipei, Kaohsiung, Taichung, TaoChuMiao, Eastern, Scenic, other regions.
Table 2 shows that the mean and STDV values of the hotel operation variables in different regions. The highest mean value of rooms is about 360 in Kaohsiung and 3 times more than lowest mean value (102) in others location. The mean of occupancy rate is highest (71%) in Taipei and lowest (45%) in others. The mean of average rate is higher in Taipei (3122) and scenic (3181) regions. The hotels in location of Taipei and Kaohsiung have obviously higher average revenue (23832632, 15915359) than other geographical regions. Finally, the mean of employee numbers are 344 and 342 in Taipei and Kaohsiung than other geographical location.
23
Figure 9, 10, 11, 12, 13, 14, and 15 illustrate the trends of domestic tourists and foreigners tourists with different country in different hotel locations. According to these figures, we can understand that the tourist behaviors with different nationality are very different in hotel location difference. More domestic tourists have accommodated in Scenery and Others, but foreigners have consumed more in main city areas. And Chinese tourist arrival numbers significantly increased after year 2005.
International tourist arrived numbers can be treated as travel demand then directly affect the hotel operation. The foreigner and domestic travel demand drivers and creates opportunities for profitability in hotel business model. Table 3 reports the mean and standard deviation of the international hotel visitor arrival in different geographical regions. For domestic tourists, highest mean of the arrival number is 5990 in the Scenic region, which imply that domestic tourist prefer to travel the eastern areas. For Chinese tourists, they mostly arrive hotel is in the eastern with mean value of 2820. Taipei region is most attractive location to stay a night, because the tourists from the North America, Japan, Asia, and Europe have their highest mean values in Taipei from the result of Table 3.
24
CHAPTER IV
EMPIRICAL RESULTS
In this sub-section, there are two main issued are discussed base on the empirical model. This second section considers the relationship between tourist arrivals and hotel operation with different disaggregated analysis in two categories. The first category consists of the investigation between domestic and foreigner tourist arrivals and hotel operations in section 4.1. The second category consists of the investigation between the Chinese tourist arrival and hotel operations in section 4.2.
25
4.1 Nationality effect on the hotel operation
From Table 4 to Table 6, the empirical results show that the impact of different tourist nationality effects on the international operation variables with different locations. The time series include 178 monthly observations from 1999 January to 2013 October. The cross section data is collected by different areas which are defined by Taiwan Tourism Bureau. The dependent variables are classified into four categories: Revenue, Rate, Occupancy, and Employee. The interpretations of the empirical results are as follows:
Revenue. Table 4 illustrates the empirical results of nationality effects on the hotel revenue with difference location in Taiwan. Table 4 shows the estimated coefficients of different nationality tourists on the room sector revenue. Most explanatory variable coefficients are significant positive at the 10% level. The positive value for coefficients indicates that foreigner visitors significantly increase the hotel revenues.
From Table 4, we can observe that the relationship between the nationality and the hotel revenue would vary by different hotel location. In Taipei area, the coefficient of European tourists is significantly largest (2976.1630) than other coefficients of tourist origins. It implies that European tourist arrivals could increase more hotel revenue than other foreigner and domestic tourists in Taipei location. This empirical result is also shown in the hotel location of Kaohsiung, TaoChuMiao, Scenery and Others area. However, the domestic tourists have little impact on the hotel revenue in Taipei. In Kaohsiung area, the highest coefficient is the variable of U.S. tourist arrivals with 1610.3390. The Asian tourist arrivals would bring highest impact on the hotel revenue in Taichung.
For domestic tourists, the coefficient (1466.1100) in scenic area shows the most
26
significant economic impact. The Japanese customers with second highest coefficient value have positive influence on the hotel revenue in the Eastern location.
Therefore, the positive relationships between the hotel revenue and numbers of international customers are significantly influenced by location difference. Compare to the difference of the location, the Australian tourists has most little influence coefficient on the hotel revenue such as in Kaohsiung, Taichung, Eastern, Eastern, Scenery, and Others. For the comparison of nationality, the most insignificant coefficients is in Eastern, which only domestic, USA and Japanese tourist have positive impact on the hotel revenue. The hotel revenues are most sensitive to the European tourists in most locations.
Rate. The average room rate level would dependent on the hotel location according to Table 2. We can realize that Taiwan tourism present expansion in recently years and good and service price level have increased until now. Table 5 reports the results for the tourist nationality effect on the room rate level in different hotel locations. The significant positive coefficients present the positive relationship between the customer numbers and the room daily rate.
The coefficients of different nationality only in the location of TaoChuMiao are all significantly positive. It implied that the foreigner and domestic tourists push the hotel room price in over past 13 years in TaoChuMiao. Thus, the coefficients of Japanese and American tourists are significant positive with room rate in Taipei.
However, in Kaohsiung, different foreigner tourist arrivals such as Japanese, Asian, and European have significant positive relationship with hotel room rate. Meanwhile, in the scenic area, the coefficient of domestic tourists (0.0870), American tourists (0.1389), Asian tourists (0.5528), and Austrian tourists (0.4483) are higher than other areas. There are only the domestic tourists and European tourists with positive impact
27
on the hotel price. Therefore, hotels with different location have different room rate growth pattern which have been affected by different foreigner and domestic tourists in our sample period. The hotel room rates are most sensitive to the European tourists in most locations.
Occupancy. Hotel performance has positive relationship with occupancy (Corgel et al., 2013). Table 6 illustrates the relationship between the room occupancy and tourist arrival numbers. The positive coefficient indicates the positive relationship between tourist arrivals and the hotel occupancy rate.
The numbers of domestic customers have significantly positive relationship with room occupancy in some locations, and have significantly negative relationship with room occupancy in other locations. The coefficients of European tourists presents negative coefficient in Taipei, Taichung, Taiwan eastern areas, the location of TaoChuMiao, and others areas. Which imply the European tourists decrease as the the hotel occupancy rate increase in these regions. Domestic tourists and foreigner tourists such as American, Japanese, and Asian have more strong positive impacts on the hotel occupancy rates than other locations. For European tourists and Australian tourists, they have more storing positive impacts on the hotel occupancy rates in each Scenery, and Others regions.
The hotel occupancy is most sensitive to the Asian tourists in most locations such as Taichung, TaoChuMiao, and Others. The result of Table 6 demonstrates that individual hotel employee increase only has significant positive relationship with specific tourist nationality. This relationship would change with different hotel locations. The hotel managers in different location should project the service quality planning on their target tourist market based on our result.
28
4.2 China effect on the hotel operation
According to the report of UNWTO Tourism Highlights, 2014, China ranked the first-large in international tourism expenditure in 2012 from the ranked 7th in 2000.
The spending by outbound Chinese tourists extended its lead further, which increased expenditure in 2013 by a massive US$ 27 billion to a record US$ 129 billion. This oversee spending amount is almost ten time rather than the amount in 2000. It is boosted by disposable incomes surge, foreign travel permission and currency appreciation.
The Taiwan authority has released the regulation of Chinese people traveling to Taiwan in 2002, January. But until 2009, the obviously increasing numbers of China tourists have shown in international tourism hospital report of Taiwan Tourism Bureau.
Therefore, the increasing China tourists should have significant impact on the hotel operation performance.
From Table 7 to Table 9, we use China arrival number as independent variable to test the impact of china tourist numbers on each international hotel operation variables of different location in Taiwan by the pool regression with time series and cross section data. The time series include 58 monthly observations from 2009 January to 2013 October. The cross section data is collected by different areas which are defined by Taiwan Tourism Bureau. The dependent variables are classified into four categories: Revenue, Rate, Occupancy, and Employee. The interpretations of the empirical results are as follows:
Revenue. Table 7 illustrates the relationship between the Chinese tourist arrivals and the hotel revenue by difference hotel location in Taiwan. The estimated coefficients in Table 7 are significant positive at the 10% level but only in Kaohsiung presents insignificant on the room department revenue. The positive value for
29
coefficients indicates that more Chinese visitors would significantly increases the hotel revenues. The highest coefficient is 2203 of the hotel in other area. The second high coefficient number is 1488 in Taipei. The results illustrate that Chinese tourists have higher contributed spending in the hotels of other area and Taipei. But the hotels in Kaohsiung have little economy benefit from the Chinese tourist arrivals. Compare to Table 5, Chinese tourist as European travelers have significant impact on the hotel revenue located inOther area.
Rate. Table 8 reports the relationship between the Chinese tourist arrivals and the hotel room rates in different areas. The coefficients are only significant negative in Taipei, Kaohsiung, and TaoChuMiao. The room rate coefficient in Kaohsiung is negative 0.0934 and larger than others. It implies the relationship between Chinese arrivals and room rate is different to the previous section result of other foreigner tourists and presents the Chinese tourists choose the hotels with lower room price especially in Kaohsiung. This result shows that more and more Chinese tourists stay at normal hotel than international tourism hotels as the hotel room rate has increased continually.
Occupancy. Table 9 illustrates the relationship between the room occupancy and Chinese tourist arrivals. The positive coefficient of different areas all indicates that Chinese tourist arrivals have positive relationship with the occupancy of hotels. The occupancy of hotels in Others location which include Tainan, Keelung, Penghu, and Yilan counties are highly sensitive to Chinese tourist arrivals due the largest coefficient 0.0074% in Table 11 as the Asian tourist arrival results in Table 7. This result implies that Chinese tourists could increase room occupancy of hotels in all locations.
30
CHAPTER V
CONCLUSION
This paper incorporates the disaggregation analysis into panel regression model to investigate the relationship between the tourist arrivals from different country and hotel operation in different location.
The results of past research focus on the aggregated data of tourist arrivals and hotel operation variables but neglect the difference of the hotel location and arrival nationality. It only offer less information for hotel managers to understand that which country tourist is import target market to their business operation variables with different hotel locations. Meanwhile, the sample sizes in this paper are more and broader than previous researches which only consider the listing firms (the firms trading in the stock exchange).
31
5.1. Foreigner and domestic tourist arrivals
On the second part of our results, we explore how the references of international
On the second part of our results, we explore how the references of international