A financial ratio-based predicting model for hotel business failure

Zhai, Shan-Shan; Choi, Jeong-gil; Kwansa, Francis Article A financial ratio-based predicting model for hotel business failure Global Business & Finance Review (GBFR) Provided in Cooperation with: People & Global Business Association (P&GBA), Seoul Suggested Citation: Zhai, Shan-Shan; Choi, Jeong-gil; Kwansa, Francis (2015) : A financial ratiobased predicting model for hotel business failure, Global Business & Finance Review (GBFR), ISSN 2384-1648, People & Global Business Association (P&GBA), Seoul, Vol. 20, Iss. 1, pp. 71-86, https://doi.org/10.17549/gbfr.2015.20.1.71 This Version is available at: https://hdl.handle.net/10419/224324 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc/4.0/ . IntroductionⅠ A. Problem Statement Hospitality industry involves high operational and financial risk. According to Ole Skalpe (2003), among the accommodation and restaurant industries the variation in earnings is high and the sales variability is fairly low, whereas the financial leverage and operational leverage are high. Operational leverage measures the flexibility of the firm’s operational cost structure. Accommodation and restaurants have high levels of fixed costs (like rent, property tax and interest expense), so the operational Hotel Management Department, Kyung Hee University, Seoul, South† Korea, E-mail: [email protected] leverage is high in this industry. This industry is also capital intensive and often hotel properties have been financed with loan to value ratios in excess of 80%. These debt obligations require fixed interest charges and regular instalment payments which become quite problematic during recessionary periods. Thus, financial leverage makes the firm more sensitive to changes in the business environment. Using borrowed money can increase the shareholders’ return on investment, but high financial leverage also raises the risk of bankruptcy if they are unable to make payment on their debt. They may also be unable to find a new lender in the future. As many hotel owners experienced during the 2008 recession, properties that had been financed with bullet loans which had come due around 2007 and 2008 were suddenly finding there were no lenders interested in refinancing their loans, thereby causing the loans to go into default. Consequently, GLOBAL BUSINESS & FINANCE REVIEW, Volume. 20 Issue. 1 (SPRING 2015), 71-86 pISSN 1088-6931 / eISSN 2384-1648 Http://dx.doi.org/10.17549/gbfr.2015.20.1.71 ∣ 2015 Global Business and Finance Associationⓒ GLOBAL BUSINESS & FINANCE REVIEW www.gbfrjournal.org6) A Financial Ratio Based Predicting Model for Hotel Business Failur e ‐ Shan Shan Zhai‐a,JeongGilChoi‐b, and Francis Kwansac aGraduate student at Hotel Management Department, Kyung-Hee University, Seoul, South Korea bProfessor of Hotel Management at Hotel Management Department, Kyung-Hee University, Seoul, South Korea cAssociate Professor of Hotel Management at Hotel, Restaurant & Institutional Management Department, University of Delaware, Newark, DE, USA A BSTRACT The purposes of this study were to find financial ratios that uniquely characterize failed hotel firms, and develop a multiple discriminant model which can predict business failure in the Korean hotel industry. Nine financial ratios that classify 86 hotel firms into failed and non failed groups were identified. Of these nine ratios, two, including‐ debt ratio and fixed assets turnover ratio, were extracted and their prediction accuracy in terms of hit ratio was 91.9%. The model suggests that debt burdened firms with low fixed asset turnover ratio are more likely to fail.‐ It means that a prudent debt financing policy is necessary to avoid business failure and fixed assets must be used effectively in order to maintain a viable enterprise. Prediction models for business failure are not homogeneous across all countries. Hotel investors and creditors can benefit from the model in screening out failing firms and lowering their investment risk. Keywords: Korean hotel industry; Business failure; Multivariate Discriminant Analysis; Z score. ‐ GLOBAL BUSINESS & FINANCE REVIEW, Volume. 20 Issue. 1(SPRING 2015), 71-86 72 the need for financial risk management and planning is paramount in the hospitality industry due to the capital structure, the vulnerability to business cycles, and the competitive nature of the industry. Korean firms in the hospitality industry have been facing great challenges in recent years. A recent study reported that more than 70% of hospitality firms fail within the first 5 years of operation (Choi, 2008). A report by National Statistical Office showed that about 124, 299 new lodging and restaurant firms, which accounted for 20.88% of the whole industry, were established in 2011, but about 127, 443 firms which accounted for 22.07% of the industry went bankrupt or were closed in the same year. The hospitality industry has the most business failures of all businesses in the retail industry (Kim, 2011). Rapid expansion, accompanied by fierce competition and poor market conditions, have been blamed for a significant number of hospitality firms going out of business (Gu, 2002). The overall market saturation in the hospitality sector has not only led to the dreadful operating performance of individual firms (Park, Choi, & Ahn, 2008) but also intensified the high competition among these firms. The Bank of Korea in 2008 reported that the intensity of service industry competition in Korea is higher than in Japan or the United States. The report found that in Korea, there were 0.9 hotel firms per 1,000 populations compared to 0.5 firms in Japan and 0.2 in the U.S. In addition, the financial performance of firms in this industry has not been good recently. According to the Ministry of Culture, Sports and Tourism (2009), the tourist hotels are experiencing significant difficulties, because of low occupancy rates. The current occupancy rates dropped around 10% compared with 10 years ago. Firms operating in this highly competitive environment are vulnerable to failure especially under poor market conditions. The current credit crisis and the extended slowdown of the nation’s economy are likely to make market conditions even tougher in the near future and may lead to more hospitality firms’ failure (Youn & Gu, 2010a). Previous research on bankruptcy has shown that not all firms fail in an unforeseeable manner (Altman, 1984). The problems that lead to the bankruptcy of a business seldom arise overnight. Warning signals preceding failure may emerge much earlier than the actual failure. Therefore, the troublesome signs of a firm can be used to predict the business failure before it actually occurs (Gu, 2002). A number of studies have used prediction models to identify warning signs of business failure. Many studies (Platt, 1989; Wight, 1985; Rutledge, 1985) suggest that prediction models for business failure are not homogeneous across all industries, and different prediction models are the result of different characteristics that are unique to a specific industry. Bettinger (1981) suggested that it is necessary to consider that a business failure prediction model should reflect the unique nature of a given industry. Most failure prediction studies have been conducted using U.S. firms and only a handful of studies have developed failure prediction models for Korean business firms (Youn & Gu, 2010a). This study fills the gap by developing an appropriate and reliable business failure prediction model for the Korean hotel industry. The effective business failure prediction models will be useful to managers, stockholders, investors, creditors and employees. Reliable models for predicting business failures will enable managers to take steps to avoid its occurrence. It will be useful to stockholders by providing them a signal and allowing them to take preventive actions, including portfolio diversification, and shorten the length of time in which losses are incurred. Hospitality investors and creditors could use the model to screen out failing firms and lower their investment risk (Youn & Gu, 2010b). Furthermore, such pre warning signals will assist in‐ making good career planning decisions. The specific purposes of this study are summarized below. B. Purpose of the Study The objectives of this study were as follows: (1) To identify financial ratios, representing the financial characteristics of failed hotel firms in contrast with those of non failed hotel firms.‐ (2) To derive a discriminant function to separate the failed hotel firms from the non failed hotel firms.‐ (3)Todevelopaclassificationmatrixtoassessthe prediction accuracy of the discriminant model. (4) To provide guidance that can be used for strategic decision making in hotel business and provide recommendations to failed hotel firms. Shan Shan Zhai, Jeong Gil Choi, and Francis Kwansa ‐‐ 73 . Literature ReviewⅡ A. Business Failure and Bankruptcy Business failure is the condition in which a firm cannot pay its lenders, preferred stock shareholders and suppliers (Lin, Yeh, & Lee, 2011). Net income has been widely used to define business failure. Schwartz & Menon (1985) used the reporting of net loss or negative net income as an indication that a firm was having financial difficulty. A firm was considered to have entered the failure process if it had an initial net operating loss following at least three consecutive years of profitability, defined as net income greater than zero. Therefore negative net income is considered by researchers as one of the early warning signs of failure. Altman (1993) provided three generic terms to represent business failure: ‘economic failure,’ ‘technical insolvency,’ and ‘bankruptcy’. According to Altman (1993) economic failure means that an asset’s return on investment is significantly and continually less than the return on similar investments; whereas technical insolvency refers to a condition in which a firm cannot meet its current obligations, signifying a lack of liquidity. On the other hand, bankruptcy is a more severe condition in which a business enterprise, unable to meet its debt obligations, petitions a court for relief from its obligations and requests either a reorganization of its debts or liquidation of its assets. The three business failure terminologies suggest that a continuum exists in business failure. Most firms may not suddenly face technical insolvency or bankruptcy. Usually they tend to move from economic failure to technical insolvency, and lastly to bankruptcy. B. Previous Business Failure Prediction Studies In 1996, Dimitras, Zonakis, and Zopounidis, extensively reviewed a total of 47 published journal articles that developed failure prediction models between 1932 and 1994. Comparing the literature on business failure Dimitras et al. (1996) found that the primary failure prediction methods employed were discriminant analysis and logistic regression analysis. Beaver (1966) first used profitability, liquidity, and solvency ratios to predict a firm’s failure. Beaver’s findings showed that univariate ratio analysis could be useful in predicting business failure. After this, statistical linear models began to be applied to the problem of corporate bankruptcy prediction. Following this, Altman (1968) first developed a multivariate bankruptcy model and tested it empirically. He computed an individual firm’s discriminant score using a set of financial and economic ratios. A fairly well known problem emerged‐ with multiple discriminant analysis, the need to have the variance covariance matrices of the predictors to be‐ the same for both groups, and to avoid this problem the logistic regression analysis was chosen to predict business failure. Ohlson (1980) introduced the logistic regression with a sigmoid function to the multiple discriminant analysis along with relaxed assumptions for the model. Deakin (1972) developed a prediction model with 14 ratios. He concluded that comparing statistical techniques for failure prediction using financial data as early as three years prior to the event, discriminant analysis could predict business failure best and achieved a high accuracy. Ravi and Ravi (2007) presented a comprehensive review of the work done, during 1968 2005, in the application‐ of statistical techniques to solve the bankruptcy prediction problem faced by banks and firms. The significant difference between the studies by Dimitras et al. (1996) and Ravi and Ravi was the methodologies used in developing the models for failure prediction. For the model development in the earlier studies, discriminant analysis and logistic regression analysis were the preferred methods. However in the recent studies, artificial intelligence techniques, such as neural networks, seem to dominate more (You & Gu, 2010b). C. Previous failure prediction studies in hospitality industry There are some researchers who have investigated industry specific business failures. For the hospitality‐ industry, there are a few documented business failure prediction studies. Olsen, Bellas, and Kish (1983) used graph analysis of financial ratios instead of sophisticated models. The drawback of the study is its lack of sophisticated statistical analysis but it is easy to apply. Others have also attempted GLOBAL BUSINESS & FINANCE REVIEW, Volume. 20 Issue. 1(SPRING 2015), 71-86 74 to predict bankruptcy using event analysis. Kwansa and Parsa (1991) presented a study of business failure in restaurant companies. Their study identified events in the bankruptcy process that characterized restaurant companies that have filed for bankruptcy under Chapter 11. Although it does not discriminate between failing and non failing firms with a quantitative model, it‐ compares the two groups based on the characteristics common to failing firms. Cho’s study in 1994 investigated business failure in the hospitality industry and developed logit models for predicting restaurant and hotel failures. Gu (2002) analyzed bankruptcy in the restaurant industry using multiple discriminant analysis (MDA). The limitation associated with the MDA model is the assumption of multivariate normal distribution of independent variables. Kim and Gu (2006a) developed logit models for predicting bankruptcy in the hospitality industry. A major contribution of their study was that they attempted to predict firm bankruptcy 2 years in advance. Increasing the prediction time frame provides firm with more time to correct their errors and thereby reduce the risk of ultimate failure. Using the same data set from Gu’s (2002) earlier study, Kim and Gu (2006b) developed a logistic regression model for restaurant bankruptcy prediction and compared its predictive ability with that of the MDA. In comparison with Gu’s (2002) MDA model, the out of sample testing‐‐ results showed that the two models are equally effective in predicting restaurant bankruptcy. More recent studies have used the artificial neural network as the method to predict business failure. Youn and Gu (2010b) developed an artificial neural network (ANNs) bankruptcy prediction model for the U.S. restaurant industry and compared its performance with a logistic regression model estimated from the same data set. The findings showed that the logistic model was not inferior to the ANNs model in terms of prediction accuracy. Researchers also attempted to compare the traditional qualitative method and artificial neural network method. Youn and Gu (2010a) developed logistic regression and artificial neural network (ANN) models to predict business failure for Korean lodging firms. . MethodologyⅢ A. Research Model and Hypothesis Altman (1968) first developed a multivariate bankruptcy model and tested it empirically. Using MDA, Altman established his bankruptcy predictive model with five financial ratios from an initial list of 22 variables. Altman et al. (1977) extended Altman’s work in 1968 and expanded his original MDA model to include seven ratios. The new model could correctly predict over 70% of the bankrupt firms 5 years in advance. In order to analyze business failure in the Korean hotel industry, this study used the multiple discriminant model to discriminate the failed hotel firms from non failed firms.‐ Discriminant analysis is an appropriate statistical technique, which involves the construction of a classification model based on sample data and the function, which is then used to classify an object into one of several groups. MDA establishes a classification model in which independent variables (financial ratios) are used to discriminate between failed versus non failed‐ firms (Gu, 2002). The discriminant model is specified as follows: Z=W 1X1+W 2X2+W 3X3+ ... + WnXn where Z = Discriminant score Wi = Discriminant weight for variable i Xi = Independent variable i Based upon the above literature review, the following hypotheses were developed to guide this study: Hypothesis 1: Certain financial ratios can differentiate between failed hotel firms and non failed hotel firms.‐ Hypothesis 2: The MDA discriminant Z score for Korean‐ hotel firms are significantly different for failed and non‐ failed firms, thus comparing the Z score of individual firms‐ with the critical cutting score value can separate the failed group from the non failed group.‐ Shan Shan Zhai, Jeong Gil Choi, and Francis Kwansa ‐‐ 75 Table 1. Sam p le hotel firms Year Financially failed firms Total assets Non failed firms‐Total assets 1999 2001‐Kaya Co, Ltd. 7,102,986 Sentro 7,897,484 2007 2008‐Kabo Hotel 9,049,640 Juro International 9,526,918 2002 2005‐Grand Hotel 8,892,057 Raemian Tourist Development 8,944,865 1999 2000‐Hotel Green Villa Cheju 17,119,464 Siheung Tourist Hotel Co., Ltd. 17,123,128 1999 2007‐Naksan Development Co., Ltd. 29,031,157 Seoul Garden Co., Ltd. 29,198,652 1999 2000‐Hotel NamTaepyung Yang, Ltd. 13,442,337 Samwha Development, Co, Ltd. 12,832,832 2007 2008‐Neo Campus 21 10,094,114 Sky Sea Resort‐10,098,610 1999 2001‐New Crown Tourist Hotel 11,694,838 YuSong Hot Spring Development 11,817,477 2002 2006‐New Prince Hotel 9,561,551 New Regent Hotel 9,274,246 2004 2006‐Hotel Amigo 14,718,130 Ocean Valley Co., Ltd. 732,056 1999 2000‐Dae Deok Crystal Co., Ltd. 16,326,799 Mibong Co., Ltd. 16,765,068 2001 Duck San Spa Hotel 6,694,407 Tovice Leisure Industry Co., Ltd. 8,547,349 2001 2002‐Dong Nam Sea World Tourist Hotel 20,480,355 Nam Young Co., Ltd. 20,415,956 2007 2008‐Dong Busan Tourism Hotel 6,363,840 River Ville Inc. 7,832,682 2005 DML 31,468,260 Sunrise Leisure Group Corp. 31,424,887 1999 2001‐Midas Hotel Co., Ltd. 18,760,433 Sunshine Co., Ltd. 18,096,804 2005 Muju TL land Co., Ltd. 6,180,091 Sung Dong Leisure Co. 8,742,675 2002 2005‐Moonhwa Tourist Hotel 7,562,374 Sejong Wedding.Co.Ltd 9,856,228 2004 2007‐Suwon Tourist Hotel 24,919,617 Teachi world Jeju Hotel 24,357,664 2007 Swiss Condominium 6,608,251 IB Hotel Co., Ltd. 8,381,680 2004 2005‐S.H Leisure Tourist. Co., Ltd. 7,069,570 You Eal. Co,Ltd. 8,001,740 2007 Airport Condotel Co., Ltd. 6,980,967 Daewoo Songdo Hotel Co., Ltd. 8,004,826 1999 2006‐Olympia Suite Co, Ltd. 41,246,654 Hotel Prima 40,198,802 1999 2002‐Woo Ju Co., Ltd. 17,714,800 Hotel Kukie 18,315,818 2001 2007‐Grace Hotel 11,864,148 Jiri Mountain Spa Land, Co, Ltd 11,751,752 2006 2008‐ILYeon Investment 10,446,335 Hotel Park Business 10,287,130 1999 2002‐Core Hotel.Co., Ltd. 15,528,960 Yeonjen Development Co., Ltd. 15,612,653 1999 2003‐Jung Won Hotel 9,488,511 Royal Kingdom Hotel 9,854,167 2004 2007‐ChoSun Tourist Hotel 7,719,082 Shin Wha Tourism Development 8,320,430 1999 2000‐Centeral [is this the right word here, or is it Central?] Heights Development 11,961,712 Sun San Terminae Co., Ltd. 12,520,264 2001 2003‐Crown Tourist Hotel 7,320,527 Jooyoung Twenty One Co., Ltd.‐7,353,196 1999 2007‐Paradise Hotel Dogo 11,331,325 Hotel Sorak Park 11,235,573 1999 2001‐Hotel Honey Croun 6,059,242 Woo Young Development 8,513,337 2007 2008‐Hanwori World Resort 83,941,832 YS Investment Corp. 81,564,240 1999 2000‐Hotel Songdo Beach 185,431,776 Seoul Lake Side Co., Ltd. 178,428,154 1999 2000‐Daeeuysaneop. Co., Ltd. 56,420,811 Sejong Investment & Development Corporation 55,574,352 2001 2007‐Dawn Beach Co., Ltd. 25,272,432 Paik Nam Tourist Co., Ltd 27,159,867 2006 2007‐Seoul Leisure Tourist Hotel 9,343,932 Prado Hotel 9,559,708 1999 2000‐Shinsung Resort Co., Ltd. 15,377,326 Hando Tour Co., Ltd. 15,669,970 1999 2002‐Shinhan Development Company 19,402,300 Hotel Paragon 19,438,615 1999 2001‐JinYang Enterprise Co., Ltd. 14,715,089 Deoku Hot Spring Hotel 15,335,306 1999 2003‐Jin Won Tourist Co., Ltd. 10,364,224 Yong Chang Ind. Co., Ltd. 10,512,604 2002 2003‐TaeJin Tour Co., Ltd. 8,357,856 Hotel Taegu Co., Ltd. 8,259,930 Source: Korean financial supervisory service database and Korean companies information database. GLOBAL BUSINESS & FINANCE REVIEW, Volume. 20 Issue. 1(SPRING 2015), 71-86 76 B. Sample Depending on Act on External Audit Ltd, when a firm is over a certain size (current assets seven billion won) should receive an audit obligatorily from a CPA (certified public accountant). If a firm that should undergo an external audit does not, it means there is a big problem in management (Huo, 2010). In this paper, the corporations that formerly underwent external audits, but didn’t get an external audit for 3 years, were defined as failed firms. The samples gathered for this study, were gathered from Korean Financial Supervisory Service Database (DART) and Korean Companies Information Database (KOCO info). By searching the lodging category, 86 failed hotel firms’ list was abstracted from the Korean companies’ information database. Firms with incomplete or unavailable financial information were removed from the list. In order to compare these firms under the same conditions, firms with different fiscal years were eliminated. Firms that were in the positive for last year’s net income to net sales ratio were taken away. If the last year’s net income to net sales ratio is positive, it’s hard to know whether firms failed or not. Firms which did not receive an audit and did not exceed three years were excluded. In this study, only hotel firms that did not get an external audit more than 3 years were defined as failed firms. After excluding firms that did not meet these conditions, 43 failed hotel firms remained. During an audit from a CPA (certified public accountant), there are financial audit reports including the financial statements of firms. Financial statements of the 43 failed hotel firms were extracted from their financial audit reports found in the Korean Financial Supervisory Service Database and the Korean Companies Information Database. As the established year and failed year are different for every firm, the total years of available financial statements are different. For example, table 1 shows that Paradise Hotel Dogo had available financial data for 9 years, while Duck San Spa Hotel’s financial data was only for 1 year. In order to compare them, the average financial data was used in this study. It has been a common practice in business failure prediction studies to use one to one match of failure and‐‐ non failure cases (Gu, 2002). Paired sampling can be‐ matched according to industry classification and size measured by either assets or sales. Therefore, this study also adopted paired sampling to develop the MDA model. Table 1 lists the failed firms paired with their control firms. There are 86 sample firms including 43 failed hotel firms and 43 non failed firms. To match the failed firms,‐ average assets of each non failed firm was calculated within‐ the same year period. After this was done, firms with the most similar asset sizes were selected. Forty three‐ non failed hotel firms with the similar sizes in terms of‐ assets were selected to comprise the control sample. After coding the sample data, it was found that the extreme values can heavily influence the statistical analysis. This study, using the Winsorising method cleaned the data. Winsorising or Winsorization is the transformation of statistics by limiting extreme values in the statistical data to reduce the effect of possibly spurious outliers. The distribution of many statistics can be heavily influenced by outliers. A typical strategy is to set all outliers to a specific percentile of the data. C. Classifying Variables Predicting business failure is an important management science problem and its goal is to differentiate firms with a high probability of failure in the future from non failed‐ firms. In other words, the business failure prediction is to build a model to forecast the moment of distress so that the firm’s economic agents may make a suitable decision. This information is basically given by financial ratios and additional information (e.g. activity, company, size, etc.) should also be taken into account. In this study, liquidity ratios, stability ratios, profitability ratios, activity ratios and growth ratios were used to predict business failure. Liquidity ratios can show a firm’s ability to meet short term obligations (Gu, 2002). A company’s ability‐ to turn short term assets into cash to cover debts is of‐ the utmost importance when creditors are seeking payment. Bankruptcy analysts frequently use the liquidity ratios to determine whether a company will be able to continue as going concern. Stability analysis is that in a certain period of time, in order to keep business activities smooth, ensuring that the various assets, liabilities, capital, and balanced financial structure are well organized is‐ important. Profitability ratios are used to assess a business’s ability to generate earnings as compared to its expenses and other relevant costs incurred during Shan Shan Zhai, Jeong Gil Choi, and Francis Kwansa ‐‐ 77 a specific period of time. Unprofitable firms with cumulative losses are likely to end up with negative net worth and eventually head for bankruptcy. Activity ratios measure a firm’s ability to convert different accounts within their balance sheets into cash or sales. Activity ratios are used to measure how effectively the company’s assets were being used and how quickly the company’s assets are actually being used in generating sales. Growth ratios can help to understand the growth potential of companies from the past to the present. Growth ratios can also be used as an indicator for forecasting the future (Choi, 2009). Previous bankruptcy prediction studies used financial ratios measuring liquidity, stability, profitability, activity and growth as variables for MDA models in the hospitality industry (Kim, 2006). Based on the ratios commonly used in previous studies of business failure prediction, this study selected 17 ratios, representing liquidity, stability, profitability, activity and growth ratios, as candidate variables for estimating an MDA model. This study collected these financial ratios from Korean Financial Supervisory Service Database and Korean Companies’ Information Database. The financial data collected for non failed firms were from the same years‐ as those compiled for failed firms. Table 2 shows the variables to be included in the model. . ResultsⅣ A. Tests of Hypothesis Hypothesis 1: Certain financial ratios can differentiate between failed hotel firms and non failed hotel firms.‐ This study utilized the Wilcoxon sum rank test to test if there are certain financial ratios that can differentiate failed firms from those of non failed firms. In general,‐ T test is an effective method to verify the differences‐ between the two groups. However, when extreme values are included in the dataset, that can decrease the effectiveness of the T test. On the contrary, a non parametric‐‐ test is not affected by extreme values in the dataset, so it provides a more stable test. A basic assumption of T test‐ is all variables should follow normal distribution, by contrast Wilcoxon test does not require such assumption of normality (Kim, 2005). The Wilcoxon Sum Rank Test showed that, at the 0.05 significant level, the two groups are distinctly different based on 9 ratios. These are current ratio, quick ratio, debt ratio, ratio of net income to net sales, total ordinary profit rate, normal profit to net worth, return on equity, fixed assets turnover and growth rate of total assets. These ratios are candidates to be included in the discriminant function model. Table 2. Classif y in g Variables Categories Financial Ratios Liquidity Ratios X1: current ratio (current assets/current liabilities) X2: quick ratio ( quick assets/current liabilities) Stability Ratios X3: debt ratio (total debt/total assets) X4: fixed assets to long term capital ratio (fixed assets/ long term capital)‐‐ Profitability Ratios X5: ratio of net income to net sales X6: total ordinary profit rate X7: normal profit to net worth X8: return on equity (net income/shareholder’s equity) Activity Ratios X9: total asset turnover ratio (total revenue/total assets) X10: inventory turnover ratio (total sales/average inventory) X11: fixed assets turnover ratio (total sales/ average fixed assets) X12: receivables turnover ratio (net credit sales/average accounts receivable) Growth Ratios X13: growth rate of total assets X14: growth rate of ordinary income X15: growth rate of net income X16: growth rate of stockholder’s equity X17: growth rate of sales GLOBAL BUSINESS & FINANCE REVIEW, Volume. 20 Issue. 1(SPRING 2015), 71-86 78 Hypothesis 2: The MDA discriminant Z score for Korean‐ hotel firms are significantly different for failed and non‐ failed firms, and thus, comparing the Z score of individual‐ firms with the critical cutting score value, can separate the failed group from the non failed group.‐ A basic assumption of MDA is the multivariate normality of the classifying variables. According to Jackson (1983), in the case of a single discriminator X, the assumption is that the X variable in each group follows normal distribution. With more than one discriminator, the assumption is that the discriminators follow multivariate normal distribution. Shapiro Wilk test and Kolmogorov‐‐ Smimova test were used to test the normality of the 17 candidate variables. The test failed to reject the null hypothesis, which predicts that variables follow a normal distribution versus the alternative hypothesis that variable distribution is different from a normal distribution. The test results showed that 17 financial ratios, all followed normal distributions and the null hypothesis failed to be rejected at the 0.05 significance level, except total ordinary profit ratio in group 2. The assumption of equal covariance or dispersion matrices is usually tested by statistical programs. The most common test is Box’s M. In this study Box’s M significance is 0.000. Between each group for testing the equality of covariance matrices, Box's M statistic is presented. Box's M significant probability is less than 0.05 to 0.000 at 5% significance level. It means the null hypothesis that equality covariance is rejected. The discriminant analysis itself does not mean anything. But in reality, equality of the covariance matrix does not violate the assumptions extremely, and especially if the sample size is large enough, it does not cause any problems (Choi, 2000). Hair (2010) stated that the significance of the differences in the covariance matrices of the two groups was 0.0320. Even though the significance was less than 0.05, the sensitivity of the test to factors other than just covariance differences (e.g. normality of the variables and increasing sample size), make this an acceptable level. There are two computational methods that can be utilized to derive a discriminant function: the simultaneous method and stepwise method. Simultaneous estimation involves computing the discriminant function so that all of the independent variables are considered concurrently regardless of the discriminating power of each. Step wise‐ estimation is an alternative to the simultaneous approach; it involves entering the independent variables into the discriminant function one at a time on the basis of their discriminant power (Hair, Black, Babin, & Anderson, 2010). In this study the SPSS program was used to estimate the MDA model and two procedures (discriminant stepwise procedure and simultaneous procedure) were used for variable selection. This enables the comparison of the two procedures for classification accuracy. In the step wise‐ procedure, the variables in the discriminant model were entered one by one based on a cut off F value for pre‐ specified statistical significance level set at the 0.05 level. The discriminant stepwise procedure selected two variables from the 9 candidates for the model which could best discriminate the failed hotel firms from the non failed‐ hotel firms. The constant and the coefficients of selected variables are presented in Table 3. In the simultaneous procedure, all of the independent variables were considered concurrently, the 9 discriminant variables were used for the model which was better in discriminating the failed hotel firms from the non failed‐ hotel firms. The constant and the coefficients of selected variables are presented in Table 4. The model computed Z score of the sample companies,‐ Table 3 .Canonical discriminant function coefficients ( ste p wise p rocedure ) ‐ Function 1 X1 Debt ratio 0.19‐ X2 Fixed assets turnover ratio 0.04 (constant) 1.25 Z1= -0.19 X1+0.04X 2+1.25 X1= Debt ratio X2= Fixed assets turnover ratio Shan Shan Zhai, Jeong Gil Choi, and Francis Kwansa ‐‐ 85 23, 589 607.– Bettinger, C., (1981). Bankruptcy prediction as a tool for commercial lenders. Journal of Commercial Bank Lending, 63,1828. ‐ Beaver, W. G., (1966). Financial ratios as predictors of failure. Empirical Research in Accounting: Selected Studies. Journal of Accounting Research,5(Suppl.): 71 102. – Choi, H., (2008). More than 70% of lodging and restaurants firms fail within the first 5years. Seoul Economy. Retrieved from http://economy.hankooki.com/lpage/economy/200805/ e2008052218234270060.htm Choi, J. G., (2009). The art of hotel business accounting,21 century publishing. Choi, J. S., (2000). Modern statistical analysis using SPSS Ver 10, Bokdu Publishing. Cho, M., (1994). Predicting business failure in the hospitality industry: an application of logit model. Unpublished doctoral dissertation, Virginia Polytechnic Institute and State University. Deakin, E. B., (1972). A discriminant analysis of predictors of business failure. Journal of Accounting Research Spring, 10, 167 179. – Dimitras, A. I., Zanakis, S. H., & Zopounidis, C. (1996). A survey of business failures with an emphasis on prediction methods and industrial applications. European Journal of Operational Research, 90(3), 487 513. – Gu, Z., (2002). Analyzing bankruptcy in the restaurant industry: A multiple discriminant model. International Journal of Hospitality Management, 21(1), 25 42. – Huo, Y. H., Sul, H. K., & Jae, B. (2010). Empirical study to Develop the distress prediction model in tourist hotel industry. Tourism and Leisure Research, 22(6), 253 270. ‐ Hair,J.F.,Black,W.C.,Babin,B.J.,&Anderson,R.E.,(2010). Multivariate data analysis. Upper Saddle River, NJ : Prentice Hall. Jackson, B. B. (1983). Multivariate Data Analysis: An Introduction. Richard D. Irwin, Inc, Homewood. Kim, S. J., (2005). Comparing distress prediction models to the hotel corporate structure: Based on predictive powers.Korea Journal of Tourism Sciences, 28(4), 9 26. ‐ Kim, S. Y., (2011). Cost conscious SVM NN Hybrid Model for‐‐ the Hotel Bankruptcy Prediction. Korea Journal of Tourism Sciences, 35(8), 101 125. ‐ Kim, S. Y., (2006). Prediction of bankruptcy in the hotel industry: A multivariate discriminant analysis model. Journal of Hotel Administration, 15(1), 103 120. ‐ Kim, H., & Gu, Z. (2006). Predicting restaurant bankruptcy: A logit model in comparison with a discriminant model. Journal of Hospitality & Tourism Research, 30(4), 474 493. ‐ Kwansa, F. A., & Parsa, H. G. (1991). Business failure analysis: an events approach. Hospitality Research Journal (Proceedings of the 1991 Annual Conference of the Council on Hotel, Restaurant and Institutional Education), 23 34. – Lin, F., Yeh, C., & Lee, M. (2011). The use of hybrid manifold learning and support vector machines in the prediction of business failure. Knowledge Based Systems ‐ ,24(1), 95 101.‐ Mandelker, G. & Rhee, S.G. (1984). The impact of the degrees of operating and financial leverage on systematic risk of common stock. Journal of Financial and Quantitative Analysis, 19,4557. ‐ Ministry of Culture, Sports and Tourism, (2009). Annual Report of Tourism Trends. National Statistical Office, (2011). Analysis of 2004 2009 ‐ establishment and bankrupt business. Retrieved from http://kostat.go.kr/portal/korea/kor_nw/2/1/index.board?bmo de=read&aSeq=245574 Olsen, M., Bellas, C., & Kish, L.V. (1983). Improving the prediction of restaurant failure through ratio analysis. International Journal of Hospitality Management, 2, 187 193. – Ohlson, J.A., (1980). Financial Ratios and the Probabilistic Prediction of Bankruptcy. Journal of Accounting Research, 18(1), 109 131. ‐ Park, D., Choi, H., & Ahn, K. (2008). KIS industry outlook: Hotels. Retrieved from http://kisrating.com/report/industry_ outlook/IO20080131 08.pdf ‐ Platt, H. D. (1989). The determinants of inter industry failure.‐ Journal of economic and business, 41, 107 126.‐ Ravi Kumar, P., & Ravi, V. (2007). Bankruptcy prediction in banks and firms via statistical and intelligent techniques – Areview.European Journal of Operational Research, 180(1), 128. ‐ Rutledge, J. (1985). Restructuring: unload those assets. Chief Executive, 33,3638. ‐ Schwartz, K., & K. Menon. (1985). Auditor switches by failed firms. The Accountinq Review, 60, 248 261. ‐ Skalpe, O. (2003). Hotel and restaurants are the risks rewarded?‐ Evidence from Norway. Journal of Tourism Management, 24, 623 634. ‐ Van Horne, J. (1977). Financial management and policy. Englewood Cliffs, NJ: Prentice Hall. ‐ Wight, C. (1985). Business failures: early diagnosis and remedies. The Australian Account, 55, 30 39. ‐ Youn, H., & Gu, Z. (2010a). Predicting Korean lodging firm failures: An artificial neural network model along with a logistic regression model. International Journal of Hospitality Management. 29(1), 120 127. ‐ Youn, H., & Gu, Z., (2010b). Predicting US restaurant firm failures: the artificial neural network model versus logistic regression model. Tourism and Hospitality Research, 10(3), 171 187. ‐ Wikipedia. (2012a). Retrieved from http://en.wikipedia.org/wiki/ Business_failure Wikipedia. (2012b). Retrieved from http://en.wikipedia.org/wiki/ Bankruptcy Wikipedia. (2012c). Retrieved from http://en.wikipedia.org/wiki/ Winsorising Investopedia. (2012a). Retrieved from http://www.investopedia. com/terms/l/liquidityratios.asp Investopedia. (2012b). Retrieved from http://www.investopedia. com/terms/a/activityratio.asp Investopedia. (2012c). Retrieved from http://www.investopedia. com/terms/d/debtratio.asp Investopedia. (2012d). Retrieved from http://www.investopedia. com/terms/f/fixed asset turnover.asp ‐‐