You are on page 1of 18

PERFORMANCE BENCHMARKING FOR

TAIWAN'S INTERNATIONAL TOURIST HOTELS
CHYAN YANG*
Institute of Business and Management
National Chiao Tung University
No. 118, Section 1, Chung-Hsiao West Road, Taipei 100, Taiwan
e-mail: professoKyang@gmail.com

WEN-MIN LU
Department and Graduate Institute of Finance
National Defense University
No. 70, Section 2, Zhongyang North Road, Beitou, Taipei 112, Taiwan
e-mail: wenmin.tu@gmail.com

ABSTRACT
Jhis study proposes an alternative data envelopment analysis (DEA) method to explore the managerial performance, the input congestion, and the benchmarks of intemational tourist hotels (ITHs) for
a small open economy, Taiwan. Several empirical results are shown: (1) most ITHs operate at
decreasing returns to scale, indicating that ITHs are facing a highly competitive environment; (2) the
intemational chain ITHs are generally more efficient than independent-owned ones; (3) ITHs
Iocated in resort areas operate slightly better on average than ones located in metropolitan areas; (4)
ITHs that are close to CKS intemational airport operate slightly worse on average than ones far from
CKS intemational airport; (5) congestion analysis reveals that inefficient ITHs lack the ability to
integrate their resources, especially in the total area of the catering division and the number of guest
rooms; and finally, (6) the reference-share measure shows that efficient intemational chain ITHs are
able to more easily become benchmarks. The findings show that efficient intemational chain ITHs
are more competitive and they should provide examples of operating practice.
Keywords: Data envelopment analysis; Performance measurement; Input congestion; Referenceshare measure.

1. INTRODUCTION
This study proposes an alternative data envelopment analysis (DEA) method to examine the
managerial performance, input congestion, and the benchmarks of fifty-six Taiwanese intemational tourist hotels (ITHs) in 2002. The results of this study can provide Taiwanese ITHs' operations with insights into resource allocation and competitive advantage and help with strategic
decision-making, especially regarding operational styles under an intense competitive environment through high ITH density.
Taiwan's ITH industry is experiencing competitive pressure due to the rapid growth of new
ITHs, a decreasing number of tourists, deteriorating economic conditions, and inefficient management. Inefficient management is the key issue that top managers should pay attention to. In
other words, the total number of Taiwan's ITHs has increased from 44 to 62 within the period of
1985 to 2004, while the total number of ordinary tourist hotels has decreased from 79 in 1985 to
25 in 2004. Moreover, as a result of the Asia Financial Crisis in 1997, Taiwan's 921 earthquake
in 1999, the 911 terrorist act in 2001, the Second Persian Gulf War in 2003, and the SARS epidemic in 2003, the number of foreign tourists visiting Taiwan are decreasing. With the external

*Corresponding author.
Reed. Apr. 2005, Rev. Sept. 2005, Ace. May 2006

INFOR vol. 44, no. 3 Aug. 2006

229

with the Malniquist productivity index for the period 1999-2001. and an efficiency indicator (Wijeysinghe. eight four-star ITHs closed down in 1998. lodging index (Wassenaar and Stafford. Much of the research on the hotel industry often uses performance indicators. (2004) implemented DEA to measure 25 Taipei ITHs' performances under three operational styles of ITHs using operating data from 2000. 1994) which may explain why researchers are interested in using it to investigate the efficiency of converting multiple inputs into multiple outputs. 1975 and Coltman. (1999).. DEA is also a theory-based. Hwang and Chang (2003) utilized DEA and the Malmquist productivity index (Fare et al. The major reasoh is that they use a more comprehensive efficiency measure and are able to capture more inefficiency.. As suggested by Anderson et al.. Barros and Alves (2004) analyzed the efficiency of Portugal's publicowned hotel chain. 2002). By using the stochastic frontier approach (SFA). Anderson et al.. 1992) to measure the managerial performance of 45 Taiwanese hotels in 1998 and the efficiency change of them from 1994 to 1998. This study contradicts previous studies. the RevPar (revenue per available room) index (Ismail et al. 1986). Furthermore. 1978). 1993). Tsaur (2000) used DEA to measure the operating efficiency of 53 ITHs in Taiwan using operating data from 1996-1998. In comparison to the traditional approaches such as ratio analysis and regression analysis (Sherman. The finding shows that not all of Taipei's franchised or managed ITHs performed more efficient than the independent-owned ones. This study provides the owners of single properties with the ability to benchmark a manager's performance. To overcome the drawbacks mentioned above. The study reports that few hotels achieved total productivity improvements in that period. These studies mostly focus on single indicators such as cost-volume-profit (Fay et al. 1997. a revenue performance indicator (Baker and Riley. 2004). DEA has gained several more advantages. transparent. 1982). Anderson et al. Bell and Morey (1995) employed DEA to measure the relative efficiency of 31 travel departments in the United States. (2000) employed DEA to re-evaluate the managerial efficiency of 48 hotels using the data in Anderson et al. 1989). Enatur. which find the hotel industry to be nearly perfectly competitive and efficient. (1999). and obviously these techniques have not taken into account the mix and nature of services provided. To survive. the lodging industry's sales receipt information (Van Doren and Gustke. the concept of perishable asset revenue management to measure performance (Kimes. Although these accounting and financial indicators in terms of simple ratios provide important and useful infonnation for benchmarking a hotel's financial performance. and reproducible computational procedure. there are in fact many factors relative to hotel performance. 1991). One major advantage is that DEA has emerged as the leading method for efficiency evaluation in terms of both the number of research papers published and the number of applications to real world problems (Seiford. . geographically dispersed over the continental United States. 1971. DEA has been used to measure hotel performance over the last decade. measuring the relative efficiency of a hotel requires methods that are more sensitive than accounting and ratio measures and that can explicitly consider the mix of service outputs produced. Previous studies that used DEA to investigate the relative efficiency of the hotel industry are now described as follows. Gattoufi et al. Table 1 presents the characteristics of these main studies using DEA. The study reports that the market for lodging services seems to be operating efficiently in Taiwan.230 CHYAN YANG AND WEN-MIN LU reasons stated above and inefficient hotel management. DEA has many desirable features (Charens et al. Jaedicke et al. the island's ITHs need to identify the critical input/output factors to improve their operating efficiency and managerial performance. Effective management strategies are developed specifically to each of the six clusters of hotels. Morey and Dittman (1995) implemented DEA to probe the generalmanager performances of 54 owner-managed hotels of a nationally known chain. 1994). Chiang et al. (1999) evaluated the managerial efficiency of 48 hotels using operating data in the year 1997. This study shows that the entire industry can be partitioned into six clusters based on relative managerial efficiency and efficiency change.

cn T -S la o^ C > ^ <" CO C ^ Cfl o 2 S ^ Tt o o .a E uu ''^ U Oi U u u UJ Q uu iJ5 <" Q Q II Q •a o -a ca irt — o\ PQ C - oi U •si ^ E ^^ t3 r ON E . a.otal reven ues g 1 sei•vices rom vario IU u o 2 2 ?> IU c« u B IU 00 O cfl . ^ 3 u 3 B U c/1 3 O s> otal rom room rever aciliities-s.5^ •s 00 ca B E « D — .=1 ^ cfl E 2 -5 2 2-a o u 2 . 2 ON < X C- U CQ .c 1 . OS IS -5 2 2-a o lu . B O S BS-S30IAJ31 1 3 o 1BJ31 I 3 13 IU o o (U T3 O < LU Q 2 u 2 u2 u > >> ca ca ca 3 c —'CScnTtio^-'CNcn IT) ' B U (U Stat > 5 <U CU 00 00 00 u 2 'E 13 .atisfa ion index ndex.

(0 •c 2 o o 2 "^ s o .> o (U u o 00 O •S 2 ca 2 | .° B B o. andC 8 ca ^ ^ eo B § o KCi u 1 &0 1 .S o •^ p K o -o (iii 1 ^-g 2_ 2 3 ca ca B 2 •S g jj u u E^ S u •3 2l^ •c <U CL C E E 8 I U I 3 E E a 3E c3 a T3 B XI . o 0- X •4^ 3 CU •= E'SU OQ Q a.2 2 • 3 ca 5. •a c c ca j:3 o ^ Tsai.

maintenance crews. 2001) means that there are increments in inputs which. section 4 concludes with the findings of this study. The input and output factors used ih this study are defined as follows. In this study the production approach is used to design the performance model. The first issue concerns the different operational characteristics.. The key ITH operating characteristics (i. and shopping facilities.nd relating expenses. 2. result in a decrease in output. • Number of employees: refers to the number of individual employees who are involved in the operation of ITHs. closeness to international airport or not) are needed to be figured out and to provide insight into what causes imperfectly competitive conditions for some ITHs. international chain or independent-owned. namely. this study applies the reference-share measure (Zhu. catering costs. The results will also aid operation managers in improving their ITHs by benchmarking their ITHs against similar ITHs. RESEARCH DESIGN 2. organizational activities refer to the conversion of inputs in various resources to output. This method not only detects congestion.1 Performance model The ITH industry provides guests with services such as accommodation. social activities. This paper is organized as follows.PERFORMANCE MEASUREMENT IN HOTEL SERVICE PROVISION 233 It should be noted that there are several important issues not being taken into account in the previous literature.and high-ranking executives. By definition. Accordingly.e. water and electricity fuel expenses. The design of the performance model and an introduction of the methodology are addressed in Section 2.. rent and so forth. • Number of guest rooms: refers to the amount of guest rooms that can be provided for rent by an ITH. Input factors • Total operating expenses: the items of operation expenses of international tourist hotels. 1984) model. An excessive amount of labor or capital input can be a major source of inefficiency. however. the unit of measurement is simply 'room'. entertainment. • Total area of catering division: refers to the total floor space used by the operational units of all the ITH's catering facilities. From a system perspective. including medium. and repairmen. Finally. The second issue is related to the so-called input congestion. catering. These efficient ITHs are worthy of further analysis to identify the real benchmark in each input/output. cooks. convention venues. but also determines the amount of congestion and simultaneously identifies factors responsible for causing congestion in an inefficient ITH. 2000) defining a ranking measure by combining the factor-specific measure and VRS (variable retums-to-scale) / BCC (Banker et al. they include salary a. metropolitan areas or resort areas. measured in units of thousand NT$. guest rooms and catering staff. The third issue concerns a situation whereby using the DEA model may produce many ITHs with a full efficient status denoted by unity. . without any subsequent adjustment being made for size or quality. maintenance and repair costs.. Output is a concrete measurement showing that an organization has reached its objectives. depreciation expenses. The empirical results and interpretations are provided in section 3. To discriminate between these efficient ITHs. measured in square feet. We will use herein a slack-based approach (Cooper et al. managerial performance. The various ITHs' characteristics are evaluated to determine their relationships to the ITH industry's efficiency. The problem of input congestion thus far is less discussed in the literature regarding the hotel industry. The performance model measures the managerial performance of ITHs in using four inputs to produce five outputs. 2001) to measure the input congestion. input congestion (Cooper et al. as a whole..

The choice of input and output variables used in the performance model can be traced to the literature (see Table 1) and the hotel operating reports are published by Taiwan Tourism Bureau (TTB). • Average room rate: refers to the ratio between the income from guest rooms and the actual number of guest rooms. these input/output factors are excluded from our performance model. the total area of the catering department appears as an input measurement in Tsaur (2000) and Hwang and Chang (2003). attached operating income. stores. For instance. the average production value per employee in the catering division appears as an output measurement in Tsaur (2000). catering services. measured in units of thousand NT$.105 15.844 334 312 1.684 866 5.224.789 242 161 784 24. The performances of the ITHs are accessed based on the data obtained for the year 2002.550. The 56 ITHs of various sizes and geographical dispersement are selected since they are officially ranked as being either four or five 'plums'.201 515.340 15 1.250 Average production value of catering 237.857 142.039 Output Variables Total operating revenues (>'|) 529.S. 2. measured in NT$/room. Dev.329.569 1.) Number of guest rooms (x^) Total area of catering division (^•4) Maximum Mean Std. Similarly. Input/output data are reported as the total number throughout the year and can be found in The Operating Report of . Note that in Taiwan the highest rating of an ITH with five 'plums' is equivalent to five 'stars' in the U.118 11 2. Information on the service-satisfaction index and the physical-facilities-satisfaction index are not available.597.810 Average occupancy rate (y2) 60 Average room rate (y^) 2.393.643 26 50 48 2. Table 2 presents descriptive statistics for our dataset.172. laundry.727 83 Output factors • Total operating revenues: the operational revenue of ITHs includes the income from guest rooms.044.917 537.CHYAN YANG AND WEN-MIN LU 234 Table 2: Descriptive statistics for the 56 ITHs in Taiwan Input Variables Total operating expenses {x^) Number of employees (jC2. The data are extracted from the annual report of the TTB. and service fees. Minimum 491.197 division (36 square feet) (y^) 989 873 3.091.382 434. • Average production value of the catering division (per 36 square feet): refers to the ratio between total revenues from the catering division and the total floor space of catering division. Each of these ITH is treated as a decision making unit (DMU) in the DEA analysis.504. measured in NT$/individual.379.194 572. measured in NT$/(per 36 square feet). • Average occupancy rate: refers to the ratio between the actual number of guestrooms let and those available to be let.379 3.726 Average production value per employee in the catering division (y^) 1.942.354 263.180. Therefore. • Average production value per employee in the catering division: refers to the ratio between total revenues from the catering division and the number of employees in the catering division.2 Sampte and Data This study investigates 56 ITHs in Taiwan based on the ITHs' operation data shown in the period 2002.

78 0.21 0..31 0.00 0.00 0. 2. and outputs j .92 0.03 y\ yi ^3 ^4 >'5 1.56 0. This report is commonly deemed as valid. we arrive at the PTE in each ITH. Notice that all the correlation coefficients are positive.43 0.61 0.00 0. these inputs and outputs hold 'isotonicity' relations.35 0.61 1. reliable.PERFORMANCE MEASUREMENT IN HOTEL SERVICE PROVISION 235 Table 3: Correlation coefficients among input variables and output variables Input variables Output variables -^3 Xo •«4 y\ yi y^ >'4 1.. respectively.3 DEA Modets Assume that the objective of each ITH is to minimize its inputs.24 0.45 0. keeping the output level constant in the BCC model. or 56 > 3(4 + 5) = 27.00 0.45 0. e is non-Archimedean infinitesimal. Xj. Table 3 shows the correlation matrix of inputs x..00 0. Cooper et al.22 0. and available to the public. Consequently.. where n is the number of ITH\ tn and . sY.j* > 0.75 0.47 0. The PTE of the target ITHg is defined as PTE = G^.56 1.74 0.00 International Tourist Hotel in Taiwan (2003) published by the TTB. (2001) suggested that the number of ITHs should be at least triple the number of inputs and outputs considered.27 1. The pure technical efficiency (PTE) of the target ITHo (o = 1.t.52 0. which is larger than triple the number of inputs (4)/outputs (5).97 0.52 1. .57 1.68 0.00 0. n) can be computed as a solution to the following linear programming (LP) problem: S.00 0. By varying the index 'o' over all ITHs. the developed DEA model should hold a high construct validity in this study.62 0.73 0.45 1. the newest published document. and thus these variables are justified to be included in the model. Let Xij and yrj be the amount of the ith input consumed and the amount of the rth output produced by the jth ITH. (1) 9o. V/ and r.55 0.12 0.60 1.14 0.00 0.? are the number of inputs and outputs.56 0. If PTE = 1 and all input and .53 0. Therefore.28 0.50 0. In this study the number of ITHs is 56. respectively.90 0.

one can easily estimate the returns to scale (RTS) by the TE and PTE scores and ^ J = . 2001) is defined as a congestion measure to capture input congestion and identifies its sources and amounts by the BCC model in Eq."=\ ^j = 1 is dropped from Eq." A slack-based approach (Cooper et al. If PTE is smaller than 1. . . if TE ^ PTE.. employees. then the target ITHo is scale inefficient and there is potential input savings through the adjustment of its operational scale. If Y.. The scale efficiency (SE) for the target ITHo is then obtained as: SE = TE/PTE. (1). Input congestion for TTHg can be computed as a solution to the following linear programming (LP) problem (Cooper et al. Whether the scale inefficient ITHg should be either downsizing or expanding depends on its current operating scale. then it is operating at the constant returns to scale size. (2) The SE represents the proportion of inputs that can be further reduced after pure technical inefficiency is eliminated if scale adjustments are possible.. The amount of congestion in each input for ITHg can then be determined by the difference between each pair of sj~* and 8t*.ni. . This rnethod not only detects congestion. then the technology is said to exhibit constant returns to scale (CRS). "Input congestion" was first defined in Cooper et al. This study measures the congestion of operating expenses. following the result of Zhu and Shen (1995).. where 5^* are optimal values in Eq. That is. (1). and the area of the catering division for the managerial performance model. (3) Xj > 0. then CRS prevails. Notice that the inequality in Eq. then Xl"=i ^7 < 1 indicates IRS (increasing retums-to-scale) and ^ J ^ i Xj > 1 indicates DRS (decreasing retums-to-scale). if TE = PTE. It has a value of less than or equal to one. s^ =s-*-5t*>0. and s** are obtained from Eq. i= 1 . (1). . That is. (2001) as "increasing in one or more inputs associated with decreasing in one or more outputs. The input congestion is next measured. The technical efficiency (TE) of the target ITHg is defined as TE = Xo under the input-oriented CRS model (Chames et al. guest rooms. s~ and s*. . then ITHg is technically inefficient."* =Xio. otherwise. i^l.o . (4) . If the target UHg has a value equal to one. s~*. If SE is less than one.. 1=1 s. where 0*. but also determines the amount of congestion and simultaneously identifies factors responsible for congestion. The objective in Eq.8t = e*x. n XXij . 1978). are equal to zero. To determine the current operating region for scale inefficient ITHg. (1) and all variables are constrained to be nonnegative. (1). 2001).. . S-* > 8f. (3)..236 CHYAN YANG AND WEN-MIN LU output slacks. m constraints is reversed from the usual form exhibited in Eq. . (3) is to maximize the sum of the input slacks with the additional constraint s^* > 5^ limiting each slack to the maximum value obtained in the preceding solution to Eq.?.. .t. (3) for the inputs implied in the first / = l . m. then the ITHg is technically efficient. Ij in any optimal solution to the CRS model.

when positive. (6) jeE Here. (1996) report the application for output-specific efficiency measures which are derived from the radial component and non-zero slacks. Lewin et al.specific measures are still multi-factor performance measures. Here. . s. r= i. (5) jeE Y^hyrj>yrd. represent the congesting amounts in each of the J = 1. j€ E.. (4) are then called input congestion slacks. These factor. x/ > 0.. (1). the reference-share measure (Zhu. (6) determine the maximum potential decrease of an input and increase of an output while keeping other inputs and outputs at current levels. r ^ q. f > yrd... the factor-specific {kth input-specific and qth output-specific) measure comes via the following two linear programming problems and the existing BCC model's best practice frontier..t. i^k.PERFORMANCE MEASUREMENT IN HOTEL SERVICE PROVISION 237 where if in Eq. jeE E ^ = i' jeE e^.. /^jXij <Xid. The factor-specific measures in Eq. (5) and Eq. 2000) is defined as a ranking measure by combining the factor-specific measure in Eqs.. d &N. The kth input-specific DEA model can be written as follows: e^' = Min e j . while 5t* > 0 represent the corresponding technical inefficiency components.t. since all related factors are considered in a single model. To identify the inputs/outputs that are most important or to distinguish those efficient ITHs which can be treated as benchmarks. s. E and A^ respectively represent the index sets for the efficient and inefficient ITHs identified by Eq. m inputs. d£N.s. (1982) and Torgersen et al.. (5) and (6) and the BCC model in Eq. (1). for a particular inefficient ITHa. The qth output-specific DEA model can be written as follows: G^* = Max (j)^. These if values.

EMPIRICAL RESULTS AND ANALYSIS 3. (5).0.^^ '^ ^^O' clear from Eq. Approximately 54% of ITHs need to reduce their inputs if they are to become efficient. Using a Z-test.969). the qth output-specific reference-share measure for each efficient ITH. (7) and Eq. Mean = 0. the more important an efficient ITH is in benchmarking. (6).)xkd/ E d d£N . On the basis of Eq.^*)xiai and [1 . The result reveals that the overall technical inefficiencies of Taiwanese ITHs are primarily due to the pure technical inefficiencies rather than the scale inefficiencies. Mean . j G E. is d€N deN where Xj* and (^f are optimal values in Eq.{\/^'^*)]yc. j e E.876) and scale efficiency (SE.67% less inputs.1 Managerial Performance Based on the controllable aspect from a manager's point of view. The technical efficiency (TE. it is found that. Nearly 6 1 % of the ITHs operate at decreasing returns to scale (DRS). The rest of the ITHs operate at increasing returns to scale (IRS). (8) that an efficient ITH which does not act as a referent ITH for any inefficient ITH will have a zero reference-share measure. Terms A' and W are weighted optimal lambda values across all the inefficient ITHs. and then ITHs can be subject to improving their scale efficiencies. Therefore.jeE ^j .1. approximately 30% of the ITHs are constant returns to scale (CRS). Mean = 0. we reject the null hypothesis that the sample mean is one at the 5% level of significance. The reference-share Aj^ (or YIJ) depends on the values of Xf and 0^* (or Xf* and if''/). The rest of the ITHs are regarded as efficient. The total number of Taiwan's ITHs increased from 43 to 62 (growth rate = 31%) within the period of . This result implies that ITHs are facing a highly competitive environment in Taiwan. This implies that the number of ITHs is approaching market saturation. the kth input-specific. (6).238- CHYAN YANG AND WEN-MIN LU On the basis of Eq. ITHs can produce the same level of measured output with 12. on average. The weights. and the nature of returns to scale (RTS) is reproduced in Table 4. respectively.95*)^M and J [1 I deN are normalized and therefore we have Y^jeE ^j = 1 ^"^ Y. 3. From Table 4. the reference-share here measures the contribution that an efficient ITH makes to the potential input (output) improvement in inefficient ITHs. the performance model in this study is run under the assumption of input minimization (also known as input orientation). (5). This study further investigates the status of returns to scale for ITHs. (1 .d characterize the potential decrease on the kth input and increase on the qth output. is j ~ / I j ^ d > deN where Xj* and 0^* are optimal values in Eq. This also suggests that managers should focus firstly on removing the technical inefficiency of ITHs.848) is decomposed into pure technical efficiency (PTE. holding the current input ratios constant. reference-share measure for each efficient ITH. As regards to the pure technical efficiency (PTE). Note that (1 . This indicates that overall ITHs still have room for improving their pure technical efficiencies.Q''. The larger the reference-share measure is.

808 1.000 0.717 0.848 0.989 0.985 1.649 0.924 0.060 1.000 1. CRS denotes constant returns to scale.522 1.000 1.594 1.845 0.918 0.858 1.918 0.834 1.669 0.789 0.000 1.826 0.000 1.000 1.000 1.248 1.000 1.000 0. MA: metropolitan area.916 0.000 1. Close Close Close Close Close Close Close Close Close Close Close Close Close Close Close Close Close Close Close Close Close Close Far Far Far Far Far Far Far Far Far Far Far Far Far Far Far Far Far Far Far Far Far Far Far Far Close Far Far Close Far Far Far Far Note: TE = PTE X SE.000 0.000 1.889 0.119 1.000 0.550 1.681 0.654 0.000 0.000 1. RTS: IRS denotes increasing returns to scale. DRS denotes decreasing returns to scale. CKS: Closeness to Chiang Kai-Shek (CKS) International Airport (1 hour driving).750 0.994 0.978 0.000 1.388 1.647 0.941 1.000 1.800 1.000 1. I: independently owned and operated.839 0.000 1.664 0.000 1.908 0.000 1.PERFORMANCE MEASUREMENT IN HOTEL SERVICE PROVISION Table 4: Code HOI H02 H03 H04 H05 H06 H07 H08 H09 HIO Hll HI2 H13 H14 H15 H16 H17 H18 H19 H20 H21 H22 H23 H24 H25 H26 H27 H28 H29 H30 H31 H32 H33 H34 H35 H36 H37 H38 H39 H40 H41 H42 H43 H44 H45 H46 H47 H48 H49 H50 H51 H52 H53 H54 H55 H56 Mean ITH Name Grand Hotel Taipei Ambassodor Hotel Magnolia Crown Hotel Imperial Hotel Taipei Gloria Prince Hotel Emperor Hotel Hotel Riverview Taipei Caesar Park Hotel Taipei Gold China Hotel Brother Hotel Santos Hotel Ritz Taipei Hotel United Hotel Sheraton Taipei Hotel Taipei Fortuna Hotel Holiday Inn Asiaworld Plaza Hotel Royal Taipei Howard Plaza Ho"tel Taipei Rebar Crowne Plaza Hotel Grand Hyatt Taipei Grand Formosa Regent Taipei Sherwood Taipei Far Eastem Plaza Hotel The Westin Taipei Hotel Kingdom Hotel Holiday Garden Summit Hotel Linden Hotel Kaohsiung Grand Hi-Lai Hotel Howard Plaza Hotel Kaohsiung Splendor Kaohsiung Park Hotel Hotel National Taichung Plaza International Hotel Evergreen Laurel Hotel Taichung Howard Plaza Hotel Taichung Splendor Taichung Astor Hotel Hualine Marshal Hotel China Trust Hotel Hualien Parkview Hotel China Yangmingshan Hotel Grand Hotel Kaohsiung Caesar Park Hotel Kenting Hotel Royal Chihpen Grand Formosa Taroko Howard Beach Resort Kenting The Hibiscus Resort Taoyuan Holiday Hotel Hotel Nanhwa Hotel Tainan The Westin Resort Hotel Royal Hsinchu Ambassador Hsinchu Formosan Naruwan Hotel Tayih Landis 239 Effficiency scores of the 56 ITHs TE PTE SE 0.000 1.000 1.905 1.962 1.995 0.304 0.385 1.000 1.337 1.963 0.977 0.966 1.000 1.080 1.138 0. .000 1.000 0.989 0.863 1.000 1.000 1.649 0.938 1.923 0.000 1.000 1.000 0.662 0.729 0.968 1.269 1.000 1.974 1.000 1.000 1.370 1.912 0.956 0.682 0.652 0.960 1.996 0.707 1.232 1.962 0.000 1.768 1.923 0.000 0.206 1.950 0.214 1.569 1.786 1.000 1.515 0.965 0.686 0.999 0.000 0.000 1.094 1.000 1. 0.982 0.000 1.983 0.912 0.000 0.000 1.245 RTS MS Location DRS DRS DRS DRS DRS CRS CRS DRS DRS DRS DRS DRS CRS DRS DRS IRS DRS DRS DRS CRS CRS CRS CRS CRS IRS DRS DRS DRS DRS DRS IRS CRS DRS DRS DRS DRS DRS IRS DRS CRS DRS CRS CRS CRS CRS CRS DRS IRS DRS CRS CRS DRS DRS DRS DRS DRS I C I I MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA RA RA RA RA RA RA RA RA RA RA RA MA MA MA MA MA MA MA MA cI I 1 I I I I I ( ( ( C C ( C 1 C C C 1 1 1 1 C C C 1 C CKS Close Close .332 1.000 1.889 0.223 1.796 0.708 0.808 0.000 0.000 1.112 1.995 0.000 1.759 0.734 0.856 1.000 0.994 0.000 0. management contract.738 1.072 1.856 1.093 1.646 0.000 1.765 1.914 0.000 1.926 0.000 1.926 .985 1.884 0.983 0.918 0.550 1:000 0.055 1.995 0.500 1.799 0.512 0.685 0.000 0.862 1.385 0.000 0.000 1.000 0.000 0.000 1.878 1.803 0.051 1.646 0.000 0.761 0.876 0.918 0. Location: RA: resort area.000 1.935 0.847 1.671 0.000 1.955 0.969 1.000 1.988 0.000 1.713 1.720 1.908 0.399 2.496 1.877 0.000 0.665 0.000 0.000 1. C: international chain including franchise.803 0.789 0.713 0.999 0.000 0.000 0.932 0.737 0.000 0.711 0.000 1. Management style (MS).743 1.981 1.854 0.000 0.000 1.000 0.000 1.000 1.000 0.287 1.397 1.792 1. and membership.

especially for resort areas. the ITHs are consequently facing greater competition. and closeness to CKS Characteristics Management Type Location Closeness to CKS Independent-owned (Mean) International chain (Mean) Metropolitan area (Mean) Resort area (Mean) Far (Mean) Close (Mean) Number of ITHs Mean 34 22 45 11 30 26 0. It is discovered that the international chain ITHs are more efficient on average than independentowned ones.856 0. As a result. To determine whether differences exist in various operating characteristics including management type (either international chain or independent-owned). some influential events are provided. location (either metropolitan area or resort area). These rural ITHs benefited from the annual government subsidiary policy to civil service employees. Due to Taiwan's 921 earthquake in 1999 and the 911 terrorist act in 2001. a sudden drop of foreign tourists (-24.. A non-parametric statistical analysis is presented in Table 5.05 level.168 *Statistically significant at 0.849 Mann-Whitney test(p-vatue) 0.314 O. a concept from the areas of transportation and agriculture. Moreover. To help the ITHs located in resort areas.000 annually. location. reductions in the usage of a proper subset of inputs may generate an increase in one or more outputs.000 government employees including military personnel. 1996) for unknown distribution scores. have achieved managerial know-how in efficiently operating an ITH.907 0. The result shows that gradually independent-owned ITHs. 2003). 3. This finding is consistent with the fmding by Hwang and Chang (2003). In order to explain the result of the other two characteristics. With existing inputs as sunk costs or fixed costs to business operations. the number of foreign tourists visiting Taiwan started decreasing. like international chain ITHs.240 CHYAN YANG AND WEN-MIN LU Table 5: Non-parametric statistical analysis of management style. After projecting an inefficient . because using a Mann-Whitney test shows no significant difference at the 5% level.2 Measuring Input Congestion Input congestion. and closeness to international airport (either close or far) for managerial efficiency.932 0.862 0.tlO 0. The majority of the ITHs far from CKS international airport are less influenced by the decreasing number of foreign tourists. The result reveals that ITHs located in resort areas regained their advantage in 2002 since the government provided subsidies to government employees in 2001. a non-parametric statistical analysis (Mann-Whitney test) is used (Brockett et al. this created a somewhat oversupply situation. 1989-2003.907 0.5%) in 2002 saw each ITH on average take up less foreigners. refers to the situation that. It is estimated that there are nearly 570. They will try to minimize their losses by serving tourists whenever they can so as to cover their variable costs. too many ITHs competing for a limited number of foreign tourists made most ITHs operate at DRS and they exhibited diseconomies of scale. when holding the usage of other inputs constant. Government subsidies provided support to government employees for domestic tours of up to seven days with a ceiling of NT$ 16. The performances of ITHs near the CKS international airport were mainly influenced by the decreasing steam of foreign visitors after Taiwan's 921 earthquake in 1999 and the 911 terrorist act in 2001. However. the gap of efficiency score in management type is less than the previous study (Hwang and Chang. the government provided government subsidies and discount native tour packages to government employees in 2001. Table 5 presents that ITHs located in resort areas operate slightly better on average than ones located in metropolitan areas.

Table 6 presents a summary of inputs' congestion after radial technical inefficiency is removed. especially in the total area of the catering division and the number of guest rooms. a specific efficient DMU to a frontier. internet service in guest rooms. . on average. One may want to know the importance of each efficient DMU by measuring the extent of inefficiencies of other inefficient DMUs. We therefore suggest that ITHs' managers may take some actions to improve the performance of the room departments and catering divisions. it is possible that the most efficient DMU detected by Andersen and Petersen's (1993) method may never appear in the reference set for inefficient DMUs.616 45 86 392 13. various themed restaurants. 24 ITHs could reduce the number of employees by 45 persons. (2001) to capture input congestion and identify its sources and amounts.3 Analysis of Best-Practice-Frontier ITHs For an efficient 'decision making unit (DMU/ITH in our case). a lack of discrimination and inappropriate weighting schemes. as constructed by the remaining DMUs.59 25. Li and Reeves (1999) presented the Multiple Criteria Data Envelopment Analysis (MCDEA) which focuses on solving two key problems. the input congestion measure is calculated in order to provide information about the eifect on output improvement through further individual input reduction.84 Input factors Total operating expenses (A:. Number of employees (ATJ) Number of guest rooms x^) Total area of catering division (x^) ITH onto the frontier by a proportional (radial) input decrease. Almost 40% of ITHs underutilize the employees and the total area of the catering division. The result denotes that inefficient ITHs lack the ability to integrate their resources.61% to 25.61 12.9% and 25. it may attract more foreign guests to visit Taiwan and hence increase the occupancy rate of ITHs and the production value of catering divisions.PERFORMANCE MEASUREMENT IN HOTEL SERVICE PROVISION 241 Table 6: Descriptive and summary statistics for inputs congestion Number of ITHs with slack Mean Total slack as percent of total inputs 5 24 14 22 76. In Li and Reeves' (1999) and Tone's (2002) methods. 3. The most efficient (or important) DMU is the one holding the largest radial distance. Holding the level of ITHs' operations constant.84) than other input factors. To help the Taiwanese hotel industry. A way to accomplish such a task is to count the number of times a particular efficient DMU is referred to by other DMUs (Smith and Mayston. 1987). Their excessive use of inputs accounts for about 13. 14 ITHs could reduce the number of guest rooms by 86 rooms.616. are larger (20.84% of total inputs.988. In this section we use a slack-based approach following Cooper et al. In doing so. and special smart deals. Tone (2002) proposed a super-efficiency model by using a slacks-based measure (SBM). and 22 ITHs could reduce the total area of the catering division by 392 (36 square feet). However. Andersen and Petersen (1993) ranked efficient DMUs by measuring the radial distance form. The input congestion slacks. the contribution which an efficient DMU makes to the potential input (output) improvement in inefficient DMUs cannot be measured. five ITHs could reduce their use of total operating expense by NT$76. the role it plays to be benchmarked by other inefficient DMUs is also important. namely.39 20. the total area of the catering division and number of guest rooms. booking through the internet. such as the quality of room service. the government can play an important role by organizing more international conferences and business exhibitions in Taiwan as well as providing better tourism environments and various tour packages.988.) .

because tourism has not only become one of the largest sources of income for Taiwan. important points not touched. Although the reference-share measures give a different ranking list in terms of input/output factors measured. Although the international tourism industry's efficiency has been widely discussed in the previous literature and the DEA technique is frequently used to explore this topic. H21. H22. We can now identify the inputs/outputs that are most important or distinguish those ITHs which can be treated as benchmarks. The results of this study can provide Taiwanese ITHs' operations with insights into resource allocation and competitive advantage and help with strate- . In Table 7 the reference-share measures are reported for the performance model.242 CHYAN YANG AND WEN-MIN LU In the current study the reference-share measure (Zhu. and economies of scale. This section gives the ranking list of the performance model for all those efficient ITHs. while for other input/output factors Grand Formosa Regent Taipei (H21) is still efficient. Hotel Royal Taipei (HI 7) plays a leading role in terms of average occupancy rate (yj) given the current levels of other inputs/outputs. From the perspective of research methods. 2000) defines a ranking measure by using the factor-specific measure and the BCC model. 32 referenceshare measures are greater than 10%. CONCLUDING REMARKS While service industries have become increasingly important in the global economy in recent years. efficient international chain ITHs (HI7. there are still some . Grand Formosa Regent Taipei (H21). Those ITHs which have a reference-share measure of zero are self-evaluators in Table 7. with ranking correlation coefficients ranging from 0. which is a particular technically efficient ITH. There are 26 pure technical efficient ITHs in the performance model. Of the total 234 reference-share measures. the result of this analysis is robust. This result is quite reasonable since international chain ITHs have a better reputation. Even if these ITHs are efficient. Grand Formosa Regent Taipei (H21) is therefore an important benchmark as the above factors are concerned.83 at the 5% level of significance. China Yangmingshan Hotel (H42) has a leading role in terms of number of guest rooms (xj) and average room rate iy^).) and total operating revenues (}'|). but also has an effective means to stimulate global economic development. This paper therefore aims to explore the efficiency. Grand Formosa Taroko (H46) plays a leading role in terms of the number of employees (xj) and average production value per employee in the catering division iy^). The ranking lists are all very similar. with the ranking in parenthesis and ordered by the average ranking of the efficient ITHs. has the biggest reference-share in total operating expenditure (x. intemet marketing. the input congestion. From the perspective of a research topic. The problem of input congestion thus far is less discussed in the literature of the hotel industry. they are revealed as being too different in the input/output space to be either a reference to other units or to be referenced.60 to 0. Sherwood Taipei (H22) leads in terms of the total area of the catering division (X4) given the current levels of other inputs/outputs. given the current levels of other inputs/outputs. China Trust Hotel Hualien (H40) has a leading role in terms of average production value of the catering division (y^) given the current levels of other inputs/outputs. H42. and the benchmarks of the ITHs from a more complete viewpoint. Therefore. an efficient reservation system. The findings show that the international chain ITHs are more competitive and they should provide examples of operating practice. the ranking list shows a clear and stable indication of the ITHs that may be pointed out as benchmarks to be referred by others. the problem that many units are easily calculated as being efficient in DEA is usually ignored. In summary. the various ITHs' characteristics are rarely taken as the research target. and H46) are frequently referenced and efficient independentowned ITHs can hardly become benchmarks. As for the number of employees ix2) and average production value per employee in the catering division iy^) are concerned. given the current levels of other inputs/outputs. a brand image. but not in the leading place. 4. the issue of the international tourism industries is especially valuable for a small open economy like Taiwan.

. ^^ . ^ ^ ' ^ ^ O V O O O O Q Q O ^ ^ * ^ ^ ^ *'• ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ '^**' •s fN fN (N C .—^ ^ . ^ f N ^ ^ ^—' f N f N CN r ^ '~^ CN f N f N r ^ C^ f N .^ ^. ^ : '^^ ^ * " ' . • " > " ' ^ ' ^ .( N CN fNONONrfONcnONONOOoOOvOQOOOOOO^O — »o — oo — \ q r n i n p p r n o q S p P p p p p l n O SHddd^dddddddd I O 00 S •5S g I o . p p | p p p p p p ^ddddcJd — d d d d d d d CD CO 2 cc ^ ^ ' ^ ^ .2 • >ninvor-^r^odcidc\CT\d(NrnrnTtTt'u-i»/^i^r^ododovovd—^—^ r^^r^^G'^^-^o6cicit^'zic^-^o^cioci-^^<6cic><D — f*^(Noor-oo«o \Ovfnn-jqpppo^ — ooooo >r) m in y-i i n i n i — Tt i o ^ ^ 00 2 ^ K (N fN fN <N 3 (0 CO CU . 7 S-1^ j ^ j j .

L. Boston: Kluwer Academic Publishers.W. the overall technical inefficiencies of Taiwanese ITHs are primarily due to pure technical inefficiencies rather than scale inefficiencies. American Business Review 18(1). Thirdly.. New perspectives on productivity in hotels: some advances and new directions.D. and quality of employees. Our study is in terms of highly aggregated measures of outputs and inputs. Fourthly. International Advances in Economic Research 10(3). Seiford. 11. M. A. 215-225.C. most ITHs are operating at decreasing returns to scale (DRS).. This finding is consistent with the finding by Hwang and Chang (2003). and Wang. European Journal of Operational Research 2(6).P.. and Seiford. M. A few notes of caution are in order here. A further investigation would be the examination of performance over time (panel data) by using the Malmquist productivity change index techniques. F. Fish. Brockett.M. Anderson. Data envelopment analysis: theory. J. Chames. and Golany.. 1261-1264. REFERENCES 1. Inc. Chames.. (2001). Management Science 42. 5. Tsai.. and Petersen. and Riley.S. 8. W. W. 297-311. Chames. F. Management Science 30(9). Thrall. and Alves. Barros. Secondly. L. Productivity in the tourism industry. 45-57. It is also hoped that the models and methods implemented in this study can bring about other related research to a variety of industries. congestion analysis denotes that inefficient ITHs lack the ability to integrate their resources.244 CHYAN YANG AND WEN-MIN LU gic decision-making. and Scott.. M. 429-444. Chiang. Cooper. International Journal of Hospitality Management 13(4).. and Zhu.E... W... Measuring the efficiency of decision making units. For instance.W. 3 1 ^ 7 . (2000). Li.W. 4. Xia. consumer satisfaction.W. K. M..I.. Hotel industry efficiency: an advanced linear programtning examination. Some models for estimating technical and scale inefficiencies in data envelopment analysis. Lewin. Banker.M. and Tone.. the international chain ITHs are generally more efficient than independent-owned ones. W. 712-715. Journal of Productivity Analysis 16.L. (1996). L. applications.M. E. Firstly. There are important qualitative dimensions of outputs that are not taken into account. Sixthly. (2000). C. 3. Sensitivity and stability analysis in DEA: some recent developments. M. these qualitative dimensions can be quality of services. R. Cooper. 1078-1092. ITHs located in resort areas operate slightly better on average than ITHs located in metropolitan areas. Cooper W. R. B. R. 466-472. especially in the total area of the catering division and the number of guest rooms.I. Data envelopment analysis: a comprehensive text with models. J. L. P. indicating that ITHs are facing a highly competitive environment in Taiwan.. R. Using rank statistics for determining programmatic efficiency differences in data envelopment analysis. Hospitality management accounting. The findings show that the international chain ITHs are more competitive and they should provide examples of operating practice. P. Baker. 13. A procedure for ranking efficient units in data envelopment analysis. A. 7. Measuring efficiency in the hotel industry: a stochastic frontier approach. R. references and DEA-solver software. Managemerit Science 39(10). Y. The findings can briefly be concluded as follows.H. (2004).. (1978). The primary reason is that ITHs located in resort areas regained their advantage in year 2002. Our findings can serve as a guideline in the tourism industry for coping with issues relating to ITHs. This also suggests that managers should focus firstly on removing the pure technical inefficiency of ITHs. Hospitality Management 18. (1993). (1994)... Our basic methodology will still remain valid. Boston: Kluwer Academic Publishers. It would be desirable to treat these outputs explicitly in the models used herein. (1994).W. Annals of Tourism Research 31(3). 9. (1978). N. Boston: CBI Publishing Co. international chain ITHs are able to more easily become benchmarks. Anderson. (1984). and Michello.M. (1999). S. Such an approach would allow a dynamic view of the multidimensional performance of ITHs.P. methodology and application. and Cooper. Tone. since year 2001 government provided subsidies to government employees. Coltman. Fifthly. Cooper. especially regarding operational styles under intense competition caused by high ITH density. among those efficient ITHs. Seiford. and Rhodes.M.. 6. 12. A. 2.M. 10. Fok. A. ITHs that are close to CKS international airport operate slightly worse on average than ones far from CKS international airport. W.. however. Anderseti. and then ITHs can be subject to improving their scale efficiencies. K. (2004). A DEA evaluation of Taipei hotels. 40-48.

20. Cornell Hotel and Restaurant Administration Quarterly 43(5). the worst of times: difference in hotel performance following 9/11. Multi-factor performance measure model with an application to Fortune 500 companies.W. T. R.T. European Journal of Operational Research 143. Zhu. 3 2 ^ 1 . C. and Cook. L. (1987). P. C. Journal of Productivity Analysis 1. and Dittman. International Journal of Management Science 20. Iowa. and Reeves. Omega. S.. A. Jossey-Bass Inc.B. Best of times. 21. Publishers.PERFORMANCE MEASUREMENT IN HOTEL SERVICE PROVISION 245 14.. 32-33. Hwang. and Reisman. Li. Cooper. Smith. B. Journal of Productivity Analysis 3(1).L. Rappaport (ed.). A discussion of testing DMUs. Taiwan Tourism Bureau (2003).S. (2002). Seiford. 569-586. Tourism Management 24. William C.J.. 14-19. and Gustke.. 16. Annals of Tourism Research 9(4). W.K. (1999). and Roos. CS. J.. Managing productivity of health care organizations in R. lodging industry. B. A slacks-based measure of super-efficiency in data envelopment analysis. NJ. Oral. and Kittelsen. Prentice-Hall.. Englewood. The lodging index: an economic indicator for the hotel/motel industry.M.A. Enz. H. Using data envelopment analysis to measure hotel managerial efficiency change in Taiwan.N. (2004). and Stafford. and Canina. 29. The operating efficiency of international tourist hotels in Taiwan. 379-398. Annals of Operation Research 73. Breakeven occupancy for a hotel operation. Evaluating the administrative efficiency of courts. Taipei. 34.C. F0rsund. B. 393-438. San Francisco. (2002). A. (1971).R. Wijeysingle. Using RevPar to analyze lodging-segment variability.. 227242. S. (1989). Data envelopment analysis literature: bibliography update (1951-2001).H. Morey. A. S. J. M. Socio-Economic Planning Sciences 38.. Rhoads. Li. A bibliography for data envelopment analysis (1978-1996). (2000). S. 85-101. 543-563. and Thrall. Omega. (1982). 31. Cornell hotel restaurant and administration quarterly 36(5). M. S. Cornell Hotel & Restaurant Administration Quarterly 30(3). Dalor. M. 401-11. . Journal of Travel Research 30(1). Morey. Gu. Management Accounting 71(2). H. J. Tone K. 22. 29-37. 25.. 32.. 22-32. The basics of yield management. G. R.A. (1982).Y. Cliffs. Asia Pacific Journal ofTourism Research 6(1). Fay. R. 28. (2000). Evaluating a hotel gm's performance: a case in benchmarking. (1975). R.D.159-229. Productivity Changes in Swedish Pharmacies 1980-1989: a non-parametric malmquist approach. (2003).).M. Deng. 26.C. (1997). Taiwan Tourism Bureau. (1995). 35. Jaedicke. The operating report of international tourist hotels in Taiwan 2002. 27. International Journal of Management Science 10. 507-517. 36. Slack-adjusted efficiency measures and ranking of efficient units. Bubuque. (1996). Gattoufi. and Chang. 30-35. 73-80. Fare. 33.C. Ismail.H.' returns to scale. In A. European Journal of Operational Research 115.A.E. (1992). European Journal of Operational Research 81. K. Grosskopf. 15. 105-124. Wassenaar. (2001). A multiple criteria approach to data envelopment analysis. 18. Tsaur. Cornell Hotel and Restaurant Administration Quarterly 43. Socio-Economic Planning Science 35. 81-21. Sherman.R.Y. Using DEA to improve the management of congestion in Chinese industries (1981-1997). (1991). Z. Information for Decision-Making-Quantative and Behavioural Dimensions (2nd Edition). T. Spatial analysis of the U. D. 357-369. (2002). (1993).. and Rosenblatt. R. 19..S. 24.. F. Lewin. D. and Shen.D. S. 37. Zhu. A. R. P. E. Managerial Accounting for Hospitality Service Industries.. 23. R. and Robichek. and Mayston. 590-596. R. Van Doren. Measuring efficiency in the public sector. Silkman (eds. Lindgren. X. Torgersen. Brown Publishers.. European Journal of Operational Research 123(1). S.C. 30. Kimes. and Mills. 17. Cost-volume-profit analysis under conditions of uncertainty. J.. L. (1986). L. (1995).