Urbanization effects on hotel performance: A case study in China

This paper studied the impacts of urbanization regarding the performance of hotel in Guangdong through the study of panel data from 1994 to 2014. Urbanization was measured by economic, geographical landscape, population, and social cultural dimensions. This research suggested that urbanization generally improves hotel performance, but with different effects. Therefore, stakeholders, such as hospitality managers, could provide corresponding governance strategies according to the characteristics of the city and province in China. Subjects: Tourism, Hospitality and Events; Urban Studies; Economics, Finance, Business & Industry


ABOUT THE AUTHORS
Jian Ming Luo is an assistant professor in the Faculty of International Tourism and Management at City University of Macau. He has extensive international and multinational tourism working experience in China, Hong Kong, Macau, and Canada. Apart from involvement in the tourism industry, he also has research and teaching experience. His teaching and research interests focus on Economic/Social/Environmental/Cultural Impact of Tourism, Socially Responsible Investing, and Consumer behavior.
Chi Fung Lam worked in several hospitality and tourism industries in a few different countries including Hong Kong and Canada. His research area focuses on tourism demand analysis and economic impacts. His recent publications appeared in the highly prestigious international academic journals.

PUBLIC INTEREST STATEMENT
Urbanization and tourism are important Chinese development strategies. Indeed, China is probably one of the countries in the world with the fastest urbanization rate. The role of the Belt and Road initiative further elevates this "world processing factory" into a "world economic nexus" of the twenty-first century. Urbanization has played a facilitating role on the tourism and hotel development. There is a need for a macroeconomic perspective in studying hotel management. This study tests the macroeconomic factors that matter to hotel performance. The result of this study shows the contributions of macroeconomics variables toward microeconomics variables are significant. The results show that urbanization is an important factor in hotel performance in China.
province to represent China. The open door policy was initially implemented in Guangdong. The unexpected growth of oversea tourists in Guangdong brought unprecedent opportunities to Guangdong hotel industry and created a huge demand on the lodging industry. Guangdong was one of the most popular destinations for international tourists (Zhang, Luo, Xiao, & Denizci Guillet, 2013). Guangdong contained five major airports, an efficient network of railways and highways that connects tourists to many of China's main tourist destinations. Therefore, this study used Guangdong to act as a representative sample to investigate the hotel performance in China.
Ever since the open door policy, the growth of hotel in Guangdong has raised by the increasing amount of tourists. Guangdong was the top destination in China among oversea tourists (SBG, 1995(SBG, -2017. There were over thirty-four million visitors visited Guangdong in 2016. Over twenty percent of the visitors were oversea tourists and the remaining were domestic tourists (China National Tourism Administration [CNTA], 1995[CNTA], -2016. Within the hotel sector, Guangdong contained 16,693 hotels, including 983,102 hotel rooms and 58.8% of room occupancy rate (see Figure 1).
Between the nineteenth and twentieth centuries, urbanization was a notable social appearance. Economically, Bradshaw and Fraser (1989) argued that modernization and industrialization would increase the proportion of urban population and hence increase the labor force in urban regions. Ciccone and Hall (1996) and Becker (2007) argued that the increase in urban population and labor forces created economics of scales on both the producers and customers. Urbanization hence contributed to create a tourism center (Kastarlak, 1971). Shen (2000) claimed that close to ten percent of the population in urban used hotels to accommodate temporarily. The increase of urbanization in China would hence increase the demand for hotel accommodation, regardless business travelers or leisure travelers. Zhang et al. (2013) discussed that urbanization would be the reference tool on hotel investment. According to Charles and Anderson (2016), urbanization was not only a good indicator to the size of market, but also the type of marketing strategies firms need to adopt. Urbanization was a crucial phenomenon internationally during the last two centuries. However, this issue did not appear in hospitality literature frequently. Urbanization would bring certain benefits to the society, but at the same time, it would bring social and political problems. The goal was benefit and cost control (Spence, Annez, & Buckley, 2009). Some Chinese scholars made some systematic studies in this area, however, many questions were unresolved (Qiu, 2007). The research question is how the macroeconomic factors influences on hotel operation. Particularly, this study tried to examines the link between urbanization and hotel performance. There were three objectives of this study. The first objective was to investigate the relationship between urbanization and hotel performance. The second objective was to evaluate the factors in urbanization and how they affect hotel performance. The third objective was to provide suggestions and recommendations to hotel management.  Child (1972) defined organization performance as the actual output or results of an organization as measured against its intended outputs. Organizational performance could be further divided into several areas of outcome measurements. The first was financial performance, which included revenue, and return on assets and investment. The second was the product market performance, which included sales and market share. The last was returned to shareholders, which included the total return of a share and the value added (Richard, Devinney, Yip, & Johnson, 2009). Performance measurement was important and necessary for hotel general managers (Phillips, 1999). Financial results simply reflected the performance of underlying operations (Chin, Barney, & O'Sullivan, 1995). Key Performance Indicators (KPIs) or ratios were based on the key drivers or the business and reflected the performance and the progress of the business. Particularly within the hospitality and tourism industry, managers tended to use KPIs to measure performance (Ireland, 2014). The most common indicators were Occupancy rate, Average Daily Rate and Room Revenue per room (Damonte, Rompf, Domke, & Bahl, 1997;Enz, Canina, & Walsh, 2001). Occupancy rate measured the efficiency of room sold. This rate was the ratio between room night being sold and the room night available for sale (American Hotel & Motel Association, 1996).

Hotel performance
This study employed occupancy rate as an indicator to measure hotel performance. The occupancy rate was a crucial indicator to measure hotel performance. Positive links between occupancy rate and hotel revenue was established by various researchers (Barden, 1998;Jeffrey & Barden, 2000;Norkett, 1985). Furthermore, it was also a common management objective, regardless explicitly or implicitly, for many hotels (Barden, 1998). Jeffrey and Barden (2000) also stressed the importance of such measures allowed hotel managers to compare occupancy rate with other hotels and hence to monitor hotel performance on a reliable and regular basis.

Urbanization and hotel performance
External environment and entrepreneurial process were inseparable (Fogel, 2001). External environments could provide a threat to business. Several researchers demonstrated the lack of competition, the lack of management practice, lack of resources, government regulations, economic policies and the recession of the economy would impose a threat to business (Buick, 2003;Christie & Crompton, 2001;Garcia & Velasco, 2002). There was a clear relationship between urbanization and economic growth (Burgess & Venables, 2004;Kasarda & Crenshaw, 1991;Landes, 1969;Williamson, 1987). However, the discussion of urbanization was absent from economic analysis of growth and development.
Since the hospitality industry was part of the service industry, hotel performance was also affected by the urbanization. Urbanization was one of the key factors to China's structural changes in terms of economic composition and employment (Huang & Bouis, 1996). Recently, many researchers use the four connotations, population, economy, social cultural, and geographical landscape, to measure urbanization (Chen, Lu, & Zha, 2010;Fang & Yao, 2006;Luo & Lam, 2016;Zhang et al., 2013). Luo and Lam (2016) studied the relationship between urbanization and hotel development. They examined the critical factors of urbanization on hotel development in China via qualitative semi-structured interviews. Population, economic, geographical landscape, and social cultural dimensions was identified as major connotations of urbanization impact on hotel development. Luo, Qiu, Goh, & Wang (2016) and Luo, Qiu, & Lam (2016) showed that urbanization has a significant relationship with tourism development in Guangdong. Furthermore, both studies showed that the impact was uneven across regions and segments of tourism and specifically, there was a significant relationship between domestic tourism development and urbanization but an insignificant relationship between foreign tourism development and urbanization. Within hotel sector, urbanization showed different effects on hotel development in China (Zhang et al., 2013).

Variables and empirical measurement
This study investigated the relationship between urbanization and hotel performance in China. Hotel performance was measured by the hotel occupancy rate. Based on China's urbanization characteristics, the theoretical research of urbanization, the features of the Guangdong province and available secondary data, urbanization could be operationalized by following four dimensions. In this study, urbanization was measured by geographical landscape, economic, population, and social cultural dimensions, these determinants are used to measure the level of urbanization in China (Chen et al., 2010;Fang & Yao, 2006;Luo, 2016;Luo & Lam, 2016;Luo, Qiu, Goh, et al., 2016;Luo, Qiu, & Lam, 2016;Zhang et al., 2013).
Economic dimension was measured by the Share of service industry in GDP (SSGDP), and GDP per capita (GDPpc) (Bradshaw & Fraser, 1989;Chen et al., 2010;Ehrlich & Holdren, 1971;Fang & Yao, 2006;Liu, 2004;Luo & Lam, 2016;Luo, Qiu, Goh, et al., 2016;Luo, Qiu, & Lam, 2016;Xu & Watada, 2008). Generally speaking, GDPpc was regarded as the average income from a macroeconomic perspective. Economists normally started cross-national comparisons of urbanization with local and per capita income or its closely related variable, GDP. A positive relationship between urbanization and income was observed, i.e. the higher the urbanization rate, the higher the income. To achieve a high and sustainable growth, manufacturing and services industry were the leading force. The productivity increase in the agricultural sectors increased the labor force available to the manufacturing industry in the urban area (Spence et al., 2009). Since the main purpose of this study was to relate urbanization and tourism activities, which was part of the service industry, the SSGDP was chosen because this variable represented the relative importance of the service sector in the cities or provinces (Luo, Qiu, & Lam, 2016).
Social cultural dimension was measured by number of Hospital beds (NHB) (Chen et al., 2010;Fang & Yao, 2006;Liu, 2004;Luo, 2016;Luo, Qiu, Goh, et al., 2016;Luo, Qiu, & Lam, 2016;Zhang et al., 2013). The variable is a proxy of the social welfare structure of the lifestyle in the cities. Annemans, Van Audenhove, Vermolen, and Heylighen (2011) confirmed the hospital bed is a social dimension and it is used to study the relationship between hospital bed and the hospital building. Health was the basic objective of development. Moreover, health was a prerequisite for the increase in productivity and was seen as vital components of growth and development (Todaro & Smith, 2012). As population increased, the number of people who demand for health care would increase and hence increased the demand for NHB.
Geographical landscape was measured by the Area of garden and green (AGG) (Chen et al., 2010;Fang & Yao, 2006;Liu, 2004;Luo & Lam, 2016;Luo, Qiu, Goh, et al., 2016;Luo, Qiu, & Lam, 2016). According to China City Statistical Yearbook (DUSS)  the AGG is the vertical projection area in the city for vegetation or planted. Theoretically, land mosaic was a result of planning and an effective spatial arrangement (Forman, 2008). Feagan and Ripmeester (2001) confirmed green space is the geographic identities at the level of the lawn. As the city develops, when the urbanization rate increases, the demand for land for industrialization would increase. Hence, the AGG would decrease (Forman, 2008). Therefore, this variable reflected the impact of urbanization.

Research method
This study employed a quantitative method. Statistical analysis was used to test the relationship between hotel performance and urbanization. In the regression analysis, dependent variable which is hotel performance was measured by the hotel occupancy rate. According to the literature review, urbanization was measured by social cultural, economic, population, and geographical landscape dimensions as independent variables. Secondary data were collected on an annual basis from 1994 to 2014 in Guangdong. These data were generated from (1) China Statistical Yearbook (NBSC, 1995(NBSC, -2016, (2) China City Statistical Yearbook (DUSS, 1995(DUSS, -2015, (3) The Yearbook of China Tourism Statistics (CNTA, 1995(CNTA, -2016, and (4) Guangdong Statistical Yearbook (SBG, 1995(SBG, -2017. The proposed model of this study was as below: 1. HOR i,t = α 0 +β 1 GDPpc i,t + β 2 SSGDP i,t +β 3 NAPP i,t + β 4 NHB i,t + β 6 AGG i,t + ε i,t

Dependent variables:
HOR i,t = Hotel occupancy rate in the i th city at time t, ln transformed.

Independent variables:
GDPpc i,t = GDP per capita of the i th city at time t, measured in ln transformed RMB. SSGDP i,t = Share of service industry in GDP in the i th city at time t, measured in ln transformed.
NAPP i,t= Non-agricultural population proportion in the i th city at time t, measured in ln transformed.
NHB i,t = Number of hospital beds in the i th city at time t, measured in ln transformed unit. AGG i,t = Area of garden and green in the i th city at time t, measured in ln transformed ha.

Error term:
ε i,t = "ε i,t " is a residual term when city "i" and year "t".
This study used descriptive statistics and regression analysis to investigate the relationship between hotel performance and urbanization. This study appointed four different estimation methods: pooled Ordinary Least Square (OLS), Feasible Generalized Least Squares (FGLS), Random effect and Fixed effect regression to study urbanization effects. The panel data analysis method was used to test the relationship between urbanization and hotel performance. Cross-sectional data could explain the relative urbanization level within the year. Time series data could explain the evolution of hotel development throughout the years. Panel data methods were now popularly adopted by urbanization researchers and were considered to be the most appropriate approach for the present research. The advantages of panel data include: (1) the ability to resolve the heterogeneity problem in the micro basis; (2) increase variability and hence reduces the multicollinearity problem; (3) the ability to study the dynamic adjustment more precisely (Kennedy, 2003).

Descriptive statistics results
The descriptive statistics of the variables used in this study were listed in Table 1. Since there are twenty-one cities in Guangdong and this study covered 20 years of observation, there were 441 observations collected. The average occupancy rate (HOR) throughout the years was around 56%, while the average share of service industry was around 37%. The average of GDPpc was 24,389 RMB. The same estimate for NAPP was around 41%. The average number of NHB was around 9122 units and for the AGG was 10,143 ha. One should handle these data with extreme care, since there were pooled data. The correlation between HOR and all other variables were positive and significant.

Panel data regressions results
Four different methods of regression would be used to investigate the relationship between HOR and other variables. The first was pooled OLS; the second was Feasible Generalized Least Square (FGLS); the third and forth were Fixed effect and Random effect respectively. Since our data-set was composed of a short period of observation (20 years) and a large number of cities, according to Baltagi (2008), it was common to employ pooled OLS, which meant the slope coefficients were the same across cities. Hence, this was our main regression model and the result was shown in Table 2. Most of the slope coefficients in this model were significant, except for AGG and NHB. The R 2 for this model was 0.343 (with adjusted R 2 0.107). Variance inflation factor was used to investigate to significance of the multi-collinearity problem and the results were less than the recommended value (VIF < 10) suggested by Neter, Kutner, Nachtsheim, and Wasserman (1996). Due to the time -series nature of the data, a Durbin-Watson test was used to study the significance of autocorrelation. The values of the test statistics suggested that there were no significant autocorrelation problem (D-W statistically less than 2) (Chatterjee & Hadi, 2006).
The issue of multicollinearity tended to inflate the variable of the slope coefficient. This would reduce the individual significance, but the independent variables, as a whole, could be significantly explaining the dependent variable (Kennedy, 2003). Feasible Generalized Least Square had the ability to transform the heteroskedastic nature under an unbalanced panel (Baltagi, 2008), hence, the second regression model, FGLS, was conducted and the result was shown in Table 3. The results of  Tables 2 and 3 are similar. The signs of the slope coefficient were close to identical between Tables   Table 2

. Pooled OLS regression statistics in Guangdong province
Notes: GDPpc is GDP per capita in each city; SSGDP is the share of service industry in GDP in each city; NAPP is nonagricultural population proportion in each city; NHB is number of hospital beds in each city; AGG is the area of garden and green in each city. Critical values for the Durbin-Watson test at 5% significance level are 0.69146 and 2.16189.
These values are calculated dependent on T (length of the balanced panel-time periods the individuals were surveyed) and K (number of repressors') (Savin & White, 1977). As the test statistic lies between lower and higher bound, we do not reject null hypothesis of zero autocorrelation in the residuals. *Correlation is significant at the 0.01 level (2-tailed).

Table 1. Descriptive statistics and correlations in Guangdong province
Notes: HOR is the hotel occupancy rate in each city; GDPpc is GDP per capita in each city; SSGDP is the share of service industry in GDP in each city; NAPP is non-agricultural population proportion in each city; NHB is number of hospital beds in each city; AGG is the area of garden and green in each city.
*Correlation is significant at the 0.01 level (2-tailed). 2 and 3. However, one should notice that the significance level of slope coefficients improved and the variables that were insignificant in Table 2, i.e. NHB and AGG, were significant. The Wald χ 2 value was 20,256.74 that indicated the degrees of freedom of the χ 2 distribution used to test the Wald χ 2 statistic at least one of the predictors' regression coefficient was not equal to zero (Baltagi, 2008).

Variables
This study performed the Hausman test specification for Random Effect and Fixed Effect, Prob>χ 2 = 0.1764 > 0.05, the test recommended the Random Effect model (Greene, 2003). Based on the result of the Random Effect regression, occupancy rate in Guangdong could be explained by all the independent variables except Hospital Bed, i.e. the slope coefficient was not significant. Within Random Effect result, the growth of hotel occupancy rate in Guangdong was significantly related to the levels of GDPpc, SSGDP, NAPP and AGG in each city. The overall R 2 for this model was 0.252. This suggests that the model could explain around 25.2 percent of the variation of the dependent variable. Individual coefficient of each variable represented the marginal effect of the independent variables on the dependent variables. All three models, pooled OLS, FGLS, and Random Effects suggested that urbanization affected the hotel occupancy rate. The results confirmed that occupancy rate was affected by economics, population, social, cultural, and geographical landscape (See Table 4).

Table 3. Cross-sectional time-series FGLS regression statistics in Guangdong province
Notes: The standard errors of FGLS regression was normalize by N standard errors. GDPpc is GDP per capita in each city; SSGDP is the share of service industry in GDP in each city; NAPP is non-agricultural population proportion in each city; NHB is number of hospital beds in each city; AGG is the area of garden and green in each city.

Variables
Coefficients Sig.  Table 4

. Random-effect (within) regression statistics at city level in Guangdong province
Notes: The standard errors of fixed-effect regression was conventional standard errors. GDPpc is GDP per capita in each city; SSGDP is the share of service industry in GDP in each city; NAPP is non-agricultural population proportion in each city; NHB is number of hospital beds in each city; AGG is the area of garden and green in each city. Σ_u is the standard error of the unobserved effects that are unchanged over time. Σ_e is the total sum of squared error. ρ is fraction of variance due to the unobserved effects.

Discussion and conclusion
This study discovers the relationship between urbanization and hotel performance, particularly occupancy rate. The continuously improving urbanization in the next century contributes abundant opportunities for rapid growth in China. This can be achieved by utilizing existing resources, improvement on high value manufacture products and services, and the increase reliance on domestic consumer demand. The hospitality industry, one of the fastest growing industry in China, will be one of the industry that benefit most from these opportunities.
First, the results of all three regressions show a huge degree of consistency. At the provincial level, the coefficient on GDPpc, Share of service industry and NAPP, are significant even without the correction on standard error (Table 2), while the remaining slope efficient become significant after the standard errors are adjusted based on the FGLS method ( Table 3). The signs of the coefficient are consistent with the general understanding. The economic dimensions, GDPpc and SSGDP, are positively related to occupancy rate. As the income of individuals in the city increase and as the service industry becomes relatively more important, the demand of hotel accommodation will increase accordingly. Furthermore, as the number of people moving the urban area increase, the demand for hotel accommodation will increase. For Hospital bed, the explanation for the negative sign can be of two folds. First, in the long run and from the perspective of medical tourism competition, the development of medical tourism will cause tourists to substitute hospital bed with hotel accommodation and hence, the increase of hospital bed will lead to a decrease of occupancy rate. Second, since Hospital bed is a measure of the social welfare structure of the lifestyle in the cities. The increase of hospital bed would increase tourists' concern on the living conditions of the destination provinces. Therefore, this would lead to a decrease of occupancy rate. For AGG, since this study uses the starrated hotel occupancy rate as dependent variables, the occupancy rate for hostel or motel are not included. Hence, as the AGG increases, the geographical landscape of the area improves, then the tourists might feel comfortable to substitute star-rated hotels to other non-star-rated accommodation facilities. Therefore, the increase of AGG will lead to a decrease of occupancy rate.
Second, the results of this study show the impact of urbanization toward hotel performance is significant. After correcting the standard error of the slope coefficient, the slope coefficients for all independent variables are significant (Please see Table 3). However, the explanatory power of the regressions (Tables 2 and 4) are relatively low. Zhang et al. (2013) showed that the explanatory power of similar independent variables toward hotel growth rate is around 90 percent, while the explanatory power of this study is well below 50. The reason for such a low explanatory power can be as follows: First, the dependent variables in Zhang et al. (2013) is the number of hotels in Guangdong. To a certain degree, this is the amount of capital, or investment in the hospitality industry. This is a macroeconomic data. Second, the dependent variable in this study is occupancy rate. This is a variable that reflects the performance of individual hotels. Hence, this is a microeconomic data. Third, the independent variables that measures urbanizations in both studies are GDPpc, share of service industry, NAPP, NHB and AGG. These are all macroeconomics variables. Therefore, the result of this study shows the contributions of macroeconomics variables toward microeconomics variables are significant, i.e. the slope coefficients are significant, while these microeconomics variables can be fully explained by these marcoeconomics data.
This study provides two contributions. From the academic perspective, this study enhances the literature by identifying the effect of urbanization toward one of the key hotel performance indicators, occupancy rate. In particular, this study shows that GDPpc, SSGDP, NAPP will increase occupancy rate while the Hospital bed and AGG will decrease occupancy rate. From the hotel management perspective, the occupancy rate is an important indicator of hotel performance. However, this study shows that this indicator will be affected by the urbanization level of the corresponding area. Hence, when top management is trying to evaluate hotel performance, one should take the degree of urbanization of the area into account. For the policy-makers, this paper would help them restructure the city with understanding the effect of urbanization to hotel performance. Hence, it would be useful to improve the hotel business environment. To conclude, urbanization is an important factor in hotel performance in China. The results were consistent with previous literature, where the degree of urbanization would influence the marketing strategies of companies (Charles & Anderson, 2016;Steenkamp, 2001). This will have certain implications for the hotel operators in designing their hotel marketing strategies. For example, from the hotel management perspective, since for the same level of investment into the same category of hotels in different cities will have different impacts to the occupancy rate, to make cross city occupancy rate evaluations, hotel managers should take the degree of urbanization of each city into consideration.

Research limitations and future research
This research has a number of limitations that deserve further investigations. First, the quantitative research of this study uses secondary data. Some data, such as the number of star-rated hotel before 1994, are missing. Therefore, this study cannot provide a longer historical panel study. Furthermore, other factors, such as population density and the education demographic, are not included. Moreover, this study uses hotel occupancy rate to measure hotel performance. However, different hotel performance indicators might show different results. For future research, by using performance variables such as ADR, ROI, etc., the macroeconomic factors that matter to hotel performance can be established. Furthermore, city characteristics (city size, city intensity, city labor structure and so on) may also affect the hotel performance. Finally, some factors which may affect the hotel performance, such as policy and governance are not included in the research. Future studies may include these variables to further examine the relationship between urbanization and hotel performance. It may bring more contributions to hotel industry and tourism industry.