Microsoft Word 2007 ichrie conference Proceedings Final-Final 06-06-07. doc

Data and Variables 
The balance sheet and income statement data for the hotel industry (NAICS 721110) between 1990 and 2004 
are retrieved from the COMPUTAT database. The segregated revenue data used for constructing internationalization 
proxy was collected from the COMPUSTAT database’s Industry Segment and Geographic Segment tapes. These 
two sets of data are merged based on company-year. After deleting observations with missing data and outliers, 
ninety observations from 14 international hotel companies (hereafter, international hotels) and 180 observations 
from 36 domestic hotel companies (hereafter, domestic hotels) are retained for analysis. The international hotels 
mean those who made revenues from overseas in this study. 
Many existing studies (Berger & Ofek, 1995) have adopted a dummy variable to surrogate the diversification 
status. The use of dummy variable, especially in a cross-sectional data, would inevitably encompass a wide range of 
characteristics that can not be controlled in the model. Using the geographical segmentation revenue data, this study 
was able to measure internationalization as a ratio of non-U.S. revenue to total revenue to test the non-linear 
relationships between internationalization, leverage, and performance. A higher ratio indicates a firm receives a 
larger portion of the revenue from overseas. The long-term debt ratio (long-term debt/total assets) was used as a 
dependent variable to measure leverage behavior in this study because lodging firms usually possess a high 
percentage of long-term debt to finance fixed assets such as real estate (Arbel, 1990). Moreover, the use of long-
term debt, instead of short-term debt, could provide more specific information about the financial leverage behavior 
of lodging firms because the long-term debt is usually a preferred funding source for growth in the lodging industry, 
whereas short-term debt is used primarily for current assets and cash shortages (Hovakimian et al., 2001). 
Following extant performance studies (Berger & Patti, 2006; Bettis, 1981; Chatterjee & Wernerfelt, 1991; 
Hitt et al., 1997), return on assets (ROA) was adopted as a proxy for firm performance. Return on equity (ROE) was 
not chosen because it ignores the impact of other forms of resource invested and could artificially inflated by high 
leverage, which is confounded with the performance impact of capital structure (Simerly & Li, 2000). Extant 
literature suggests that the size of a firm may influence its leverage (Barclay & Smith, 1995), decision-making 
capabilities, and performance (Simerly & Li, 2000). Therefore, the natural logarithm of sales (Hitt et al., 1997; 
Singh & Nejadmalayeri, 2004) was adopted as the proxy for firm size. Risk is another controlling variable common 
in all three equations because it could influence manager’s choice on the level of diversification (Chatterjee & 
Wernerfelt, 1991) and leverage (Singh & Nejadmalayeri, 2004). Moreover, profitability could be just a tradeoff 
between risk and expected return (Berger & Patti, 2006). Therefore, in this study a standard deviation of 3 years of 
EBIT (Berger & Patti, 2006) was adopted as the proxy for risk. Growth opportunities were found to have significant 
influences hospitality firms’ leverage and profitability (Chathoth & Olsen, 2007). Following Tang and Jang (2007), 
the market-to-book ratio is adopted in this study to capture the relative value of the growth opportunities viewed by 
the market. Since PP&E usually serves as inflation-resistant collateral for loans, hotels with high level of PP&E 
carry a higher debt capacity. Therefore, asset structure measured as a ratio of PP&E to total assets was included in 
equation 2 to control for internationalization’s influence on leverage. Stimpert and Duhaime (1997) also argued that 
international firms are able to build new strategic assets faster and more cheaply than their competitors by 

---