Tourism Management xxx (2012) 1e9

Contents lists available at SciVerse ScienceDirect

Tourism Management
journal homepage: www.elsevier.com/locate/tourman

Efficiency and its determinants in Portuguese hotels in the Algarve

Ricardo Oliveira, Maria Isabel Pedro*, Rui Cunha Marques
CEG-IST e Center for Management Studies, Av. Rovisco Pais, 1049 001 Lisbon, Portugal

h i g h l i g h t s

< We measure the efficiency of Portuguese hotels in the tourism region of the Algarve.
< Robust non-parametric methods are applied to compare the performance of hotels.
< The paper provides evidence of higher efficiency in hotels without golf courses.
< Results show that hotel star rating is not a relevant issue for their performance.

a r t i c l e i n f o a b s t r a c t

Article history: This paper uses non-parametric techniques to investigate and compare the efficiency of Portuguese
Received 23 December 2011 hotels in the Algarve, the comparison being made between those hotels possessing golf courses and
Accepted 22 June 2012 those that do not. The Algarve is one of the more important tourism locations in Portugal and a popular
destination for tourists from many countries and specifically attracts tourists seeking to play golf. By
Keywords: using data envelopment analysis (DEA) this study investigates the influence of star ratings, golf courses
Algarve
and location on hotel efficiency. Additionally the study uses the test devised by Carvalho and Marques
Efficiency
that takes into account the equality of the entire efficiency distribution. We conclude that star rating is
Golf courses
Hotels
not a significant determinant of efficiency but location and the existence of golf courses may have some
Statistical tests relevance. A major finding is that it is those hotels that do not possess golf courses that are the more
Star rating efficient.

2012 Elsevier Ltd. All rights reserved.

1. Introduction
The Algarve is a tourism region in Portugal that traditionally
attracts the major part of domestic Portuguese tourists as well as
tourists from other countries of Europe and the world. It is one of
the main tourism destinations for visitors from countries of
Northern and Central Europe, and it was ranked as the 10th Euro-
pean and the 20th world destination by the UNWTO in 2007. In
addition to its attractive beaches, the tourism offering in the
Algarve focuses especially on golf.
The golf product has undergone strong expansion in Portugal,
both in quantity and in reputation. In 2006 it accounted for 6.3% of
the reasons given by foreign tourists for visiting Portugal (PENT,
2007). The Algarve has more than 40% of the golf courses in the
country (31 out of 70) and enjoys high international rankings that
contribute to the area’s attraction as a tourist location for both the
* Corresponding author.
E-mail addresses: ricardo.d.oliveira@ist.utl.pt (R. Oliveira), ipedro@ist.utl.pt
(M.I. Pedro), rui.marques@ist.utl.pt (R.C. Marques).
domestic and international market. For example, the Algarve was
considered the ‘Best Worldwide Golf Destination’ several times in the
last decade by the International Association of Golf Tour Operators
(IAGTO), and was identified by the same organization as a golf
destination of first choice. Additionally, the Algarve golf courses
were also acclaimed by ‘Rheingolf Magazine’ and ‘Golf Digest’ that
ranked the San Lorenzo and Vilamoura Old Course among the 100
best golf courses in the world. The hotels with golf courses are
generally rated 5-star. There are 28 5-star hotels in the Algarve
(7.3% of the total) and of these 29% have golf courses.
Based upon its geographic and climatic features, the Algarve can
be further divided into the windward and leeward sub-regions, in
the west and east respectively. This has implications for the
patterns of weather, and from a golfer’s perspective, the pattern of
prevailing winds.
Given the importance of the Algarve, and the contribution made
by golf toward its destination image, this paper seeks to assess the
efficiency of the hotels by the use of data envelopment analysis
(DEA) and to assess to what degree association with a golf course
impacts upon a hotel’s operational efficiency. In doing so it builds
upon the work of Barros, Botti, Peynoch and Solonandrasana (2011)

0261-5177/$ e see front matter
2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.tourman.2012.06.009

Please cite this article in press as: Oliveira, R., et al., Efficiency and its determinants in Portuguese hotels in the Algarve, Tourism Management
(2012), http://dx.doi.org/10.1016/j.tourman.2012.06.009
2 R. Oliveira et al. / Tourism Management xxx (2012) 1e9
and their study of Portuguese hotels, by comparing the efficiency of
4-star and 5-star hotels and also those with and without golf
courses. Furthermore, the efficiency of the hotels of the two sub-
regions is also compared. DEA has numerous advantages over
other (parametric) methods (e.g. stochastic frontier analysis e SFA)
since it does not require any a priori assumption of a function to
represent the production or cost frontier, neither does require as
many data assumptions as parametric methods with, for example,
respect to normality of distribution, independence of variables or
assumptions relating to error terms (Kumbhakar & Lovell, 2000). A
sample of 84 hotels for the years 2005e2007 is used in this
investigation.
This paper makes several contributions to the literature. The
first concerns the novelty of investigating the determinants of
efficiency using non-parametric methods, particularly the influ-
ence of star ratings and the existence of golf courses. Second, robust
methods are used to compare the statistical distributions of effi-
ciency and a new statistical test is applied. Finally, it is the first time
that the efficiency of hotels specific to the Algarve has been
analyzed and this still remains (thus far) as one of the few studies of
this nature in Portugal.
The remainder of the paper is organized as follows. Section 2
presents a literature review on hotel efficiency. Section 3 intro-
duces the methodologies adopted and proposed for this
research. Section 4 describes the case study on efficiency esti-
mation and its determinants, comprising 84 Portuguese hotels in
the Algarve and, finally, and Section 5 presents the concluding
remarks.
2. Literature review
2.1. Efficiency and tourism
Hotels must compete globally to attract customers (Tarim,
Dener, & Tarim, 2000), yet in doing so are still required to achieve
desirable rates of return. This issue alone would be enough to place
efficiency and its measurement at the top of the agenda for
managers. Yet studies of this kind are a relatively new area of
research in the field of tourism. As Davies (1999, p. 296) notes, “. it
is surprising that the literature of IO (industrial economics) has
little to say about this area of economic activity.”
According to several authors (Anderson, Fok, & Scott, 2000;
Barros & Dieke, 2008; Barros, Peypoch, & Solonandrasana, 2009;
Chen, 2007), in the past, performance evaluation was restricted
to performance indicators. However, as these are partial
measures of productivity, and do not take into account either the
multiple variables of the hotel industry, or the interactions
between them, it is suggested that these measures have not been
wholly effective. Recently, methods such as SFA or DEA have
included multiple inputs and outputs of the hotel industry in
their analysis and as a result they have produced more
comprehensive, accurate and understandable performance
measures.
Several studies in the literature apply DEA to determine the
efficiency of hotels. For example, Hwang and Chang (2003) eval-
uated performance as an important part of management control,
and studied efficiency measures in the hotel industry in 45 inter-
national hotels in Taiwan. Anderson, Fish, Xia, and Michello (1999)
discussed the difficulties of measuring the efficiency of hotels.
They concluded that efficiency in the management of resorts
(hotels with leisure segments) was greater than that of business
hotels that primarily provided solely accommodation, and addi-
tionally that efficiency in the management of international hotels
that chiefly dealt with foreign tourists was greater than that of
hotels whose core clients were domestic tourists. They also
Please cite this article in press as: Oliveira, R., et al., Efficiency and its determinants in Portuguese hotels in the Algarve, Tourism Management
(2012), http://dx.doi.org/10.1016/j.tourman.2012.06.009
suggested that efficiency in the management of hotels that are part
of a hotel chain is greater than that of independent local hotels.
Rodriguez and Gonzalez (2007) studied the efficiency of hotels
located in Gran Canaria (Spain) over a period of 10 years,
concluding that inefficiency was common with serious conse-
quences in terms of costs. The authors also concluded that econ-
omies of scale in this industry mainly depend on the size of firms.
According to Hwang and Chang (2003), efficiency may be related
either to differences in the type of tourists or to the style of
management. In turn, Barros, Dieke, and Santos (2009) studied the
efficiency of hotels in Luanda (Angola), suggesting that hotels that
adopt strategic approaches get better results and, therefore, are
more efficient than more short-term oriented hotels. Other
studies, such as Barros, Botti, Peypoch, Robinot et al. (2011), have
focused on the competitiveness of the destination. Finally, Asalos
and Trandafir (2009) suggest that the heterogeneity and
complexity of the tourism product actually demands a wider scope
of coverage and evaluation of efficiency when compared with
other industries. They conclude that economic efficiency in
tourism is very complex because it is the result of a wide range of
activities including domestic tourism, international tourism, food,
services, and sale of advertising and transportation, among others.
Additionally tourism activity comprises two simultaneous aspects
of efficiency that interact with each other, namely economic and
social efficiency.
2.2. Efficiency of 4- and 5-star rated hotels
The star rating of hotels aims to harmonize international stan-
dards in order to differentiate and improve hotel facilities and
services. However, there is no universally accepted and adopted
rating system. The World Hotel Rating Agency, headquartered in
Belgium, is the international institution most often used to classify
hotels. Having a greater number of stars suggests more luxury, and
better services, food and beverages, entertainment, panoramic
views and a variety of rooms of different sizes. Additional
requirements, such as spa centres and fitness facilities, as well as
location, are also commonly considered in the definition of a stan-
dard. Nevertheless, a greater number of stars do not necessarily
correspond to greater efficiency and greater profitability.
Using room segmentations and price guide as a model, Smith
(1999) and Ismail, Dalbot, and Mills (2002) ranked each hotel in
one of five price segments of the market: luxury (upper scale),
upscale, mid-price, economic and budget, and examined the port-
folio of each multi-brand hotel company. These studies suggested
that a higher star rating permitted higher occupancy levels and
higher levels of quality.
Regarding star rating performance, Ray and Phillips (2005)
suggest that the greater the number of stars, the better the
performance. They suggest that this may be related to the fact that
the property is associated with foreign partners who bring better
and more experienced management techniques, modern technol-
ogies and (foreign) customers who spend more. Assaf and Agbola
(2011) suggest a positive relationship between the number of
stars and efficiency. Khataei, Farsin, and Mousavi (2008), in a study
of luxury hotels in Tehran, Iran, observed that 5-star hotels are
more efficient than 4-star hotels.
However, Chen (2007) states that there are greater efficiency
differences between 4-star hotels and yet nevertheless as a group
they perform better than 5-star hotels. Tarim et al. (2000) also
argue that 4-star hotels are technically more efficient (72%) than 5-
star hotels (52%) and suggest that the major difference between 4-
and 5-star hotels is the fact that 5-star hotels provided less satis-
factory services for their clients (who are more demanding with
higher expectations) and less profit.
 R. Oliveira et al. / Tourism Management xxx (2012) 1e9
2.3. Efficiency and golf courses
As noted in the introduction, tourism in the Algarve has
particularly focused on the competitive advantages that arise
from the nature resource-based products of ‘sun and beach’, but to
these attributes have added higher value-added segments, such as
golf.
According to Correia, Barros, and Silvestre (2007), golf was
identified as one of the ways of addressing the issues of seasonality,
especially in Mediterranean destinations, and they concluded that
golf courses are a profitable business both for tourism development
and for the golf activity itself. Malpezzi (1999), using the golf
market in the United States, estimated the need for new courses
from econometric models based on population and number of
holes. This study seems to confirm the findings of Martinez (1992),
which point to the growing popularity of golf and golf based
tourism, explained by an ageing population and increased life
expectancy, all of which have consequences for tourism. Catalão
(2001) addresses golf from the perspective of the offer, identi-
fying its importance in terms of planning. The author refers to the
impact of golf and concludes that sustainability is the fundamental
axis for the quality of such sports tourism. Horwath Consulting
(1992) and Pinheiro (1994) note the advantages of golf in the
economy and its importance in a strategy for seasonal adjustment.
Nature and favorable weather conditions were considered by
Correia et al. (2007) as factors motivating golfers to keep coming
back to a given location. In this case, the climate is a clear “pull
factor” (Lundtorp, Rassing, & Wanhill, 1999) for visitors returning to
the Algarve. Stynes, Sun, and Talhelm (2000) studied golf tourists in
Michigan and assessed the impact of individual golf course devel-
( )
 X
n X
n
b
q ðx; yÞ ¼ min q > 0 y  g y ; qx  g x ; g  0; i ¼ 1; :::; n
CRS i i i i i
i¼1 i¼1
opments on a local area. Tezcan (2004) argues that golf tourism in
Greece is an essential component of, and a priority for Greek
tourism policies. Although the development of golf tourism will not
solve all the problems of Greek tourism, it can certainly contribute
to the objectives of extending the tourist season and lead to high
levels of activity. Finally, Priestley (2006) suggests that the process
of the development of golf as a tourism product has a profound
impact on the location in which it takes place and can cause
conflicts and even imbalances where such policies are imple-
mented due to the resources required to sustain gold courses.
2.4. Efficiency and location
Many features have been shown to impact on the efficiency of
hotels and golf courses as indicated above. One factor specific to
both golf and hotels is the obvious one of location, and Guizzardi
and Bernini (2010), in a study on Italian tourism, suggest that the
location of the hotels has a positive correlation with technical
efficiency, particularly for those in cities by the beach or in cities of
recognized cultural importance.
Moliner, Claver, and Molina (2011) argue that location is a factor
that might explain performance, revenues and occupancy rates,
and suggest that location can determine 31.56% of the variance in
revenue performance. One implication is that hotels need to
conserve resources in order to keep the destination attractive
Please cite this article in press as: Oliveira, R., et al., Efficiency and its determinants in Portuguese hotels in the Algarve, Tourism Management
(2012), http://dx.doi.org/10.1016/j.tourman.2012.06.009
3
(Gomezelj & Mihalic, 2008). Attractiveness of a location however
goes beyond the impact of natural features and Chou, Hsu, and
Chen (2008) observe that location selection involves the provi-
sion of an overall distribution blueprint for the region, and that
traffic and transportation conditions are also very important. For
his part Coltman (1989) states that synergy can be achieved if
facilities such as commercial areas, convention centres, and airports
are taken into account (Coltman, 1989).
This review therefore indicates a number of key variables that
therefore need to be incorporated into any modeling of hotel
performance, and thus the next part of the paper reviews issues
pertaining to the research methodology.
3. Methodology
3.1. The DEA method
As noted previously, DEA possesses numerous advantages over
parametric methods (e.g. SFA), since it does not require any a priori
assumption of a function to represent the production frontier
(Gonçalves, Liang, Peypoch, & Sbai, 2012). Generally on emay state
that non-parametric methods use mathematical programming
techniques for the estimation of the production technology or
frontier (see, for example, Fried, Lovell, & Schmidt, 2008 for an
overview of the benchmarking techniques and their advantages).
The DEA method was developed by Charnes, Cooper, and
Rhodes (1978) who, by extending the ideas of Farrell (1957),
proposed the calculation of the efficiency of observations from the
following problem of linear programming, for an input orientation,
assuming constant returns to scale (CRS):
(1)
where x˛<pþ are the inputs used to produce the outputs y˛<qþ and q
the DEA-CRS efficiency scores for a given observation.
This non-parametric methodology creates a production fron-
tier formed by observations that represent the best practice,
thereby enabling the estimation of efficiency scores for each
observation computed from the distance to that frontier. The
DEA efficiency scores are estimated assuming that the tech-
nology of an observation i (i ¼ 1,.,n) is characterized by the
production set J: J ¼ fðx; yÞ˛<pþq þ jx can produce yg: In addi-
tion, the DEA model assumes the properties of free disposability
of inputs and outputs and the convexity of production in set J
(Charnes et al., 1978). The observations located on the frontier
are considered efficient and have efficiencies equal to one, while
the observations that lie below the frontier (for an input orien-
tation) are inefficient, displaying efficiencies less than one. The
production frontier can be estimated by considering both
constant returns to scale (CRS) and variable returns to scale
(VRS) technologies. The CRS approach assumes that all obser-
vations operate at an optimum level while the VRS approach
compares observations with a similar scale, resulting in effi-
ciencies that consider the differences in scale (Banker, Charnes, &
Cooper, 1984). The CRS DEA efficiency scores are computed from
the previous linear program while the VRS DEA efficiency scores
Pn
(q ) are obtained by adding the constraint i ¼ 1 gi ¼ 1 to
CRS
expression Eq. (1).
4 R. Oliveira et al. / Tourism Management xxx (2012) 1e9

3.2. Comparing groups of efficiency hotels, the test is performed in accordance with the following steps
(Carvalho & Marques, 2011):
To identify the most efficient groups of hotels and therefore the
best practices, the equality of the distributions of DEA efficiency 1. Compute the estimated DEA efficiency scores b q DEA of each hotel
scores was assessed for the 5-star hotels against the distributions of for the whole sample of 5- and 4-star hotels from the following
DEA efficiency scores for the 4-star hotels (doing the same for the linear program; for example, for this situation it considers
existence or otherwise of a golf course and the windward or an output orientation and a VRS technology (Banker et al., 1984),
leeward hotel location), using for this a statistical test suggested by (
Carvalho and Marques (2011). This test is an improvement of the b Xn X
n X
n
q DEA ðx; yÞ ¼ max q gi yi  qy; gi xi  x; gi ¼ 1; gi
statistical test provided in Simar and Zelenyuk (2006), often i¼1 i¼1 i¼1
regarded as one of the more powerful tests for this form of )
calculation.  0; i ¼ 1; :::; n
To assess whether certain groups of observations are more
efficient than others, the literature on non-parametric efficiency (6)
measurement proposes the application of various statistical tests.
where x˛<pþ are the inputs used to produce the outputs y˛<qþ ;
However, most of them assess the equality of the location param-
eters of the efficiency scores distribution (such as the mean and the
2. Smooth the original estimates of the efficiency scores DEA
median) of various groups (e.g. KruskaleWallis test) rather than the
equal to 1, according to the following expression, obtaining
equality of the entire efficiency distribution. In addition, many of  n1þn2
*
these tests have proved, at times, to be insufficiently powerful, b
q DEAi :
especially when samples are small or have high dimensionality. i¼1
The statistical test that is applied aims to test whether two 8
<b
samples of DEA efficiency n1
scores that n2are identically, independently b
q
*
¼
q DEA þ ε if bq DEA ¼ 1 (7)
distributed (i.i.d.) fb
q 1i gi¼1 and fbq 2i gi¼1 are drawn from the same DEA b
: q DEA if bq DEA s1
distribution, and it consists of testing the following null hypothesis:
n o
H0 : f ðxÞ ¼ gðxÞ for almost all x (2) where ε ¼ uniform n2=ðpþqþ1Þ ; qt  1 and qt is the 2.5th

percentile of the empirical distribution of b

q DEA > 1;
where f ð,Þ is the probability density function (PDF) of b
q 1 and gð,Þ is
the PDF of bq2 .
3. Apply a bootstrap method which consists of randomly drawing
The statistical test has a statistic:  n1þn2
*
pffiffiffiffiffiffiffiffiffiffiffiffiffi  n1 and n2 observations with replacement from b q DEAi to
Tnb ¼ n1 n2 h Inb  Cn;b b sb (3)  n1  n2
i¼1
* *
obtain b
q and b
q , respectively;
1i 2i
where i¼1 i¼1
 n1  n2
8 * *
<   Xn1 X
n1 h  i 4. Estimate the statistic of the test Tnb using b q 1i and bq 2i
 n1  n2
K b q 1i  bq 1j h
i¼1 i¼1
In ¼ 1 hn21 
b
: instead of bq 1i and b q 2i respectively, and estimate the
i¼1 j¼1 i¼1  i¼1
  X
n2 X
n2 h  i
bandwidth h ¼ min h1 ; h2 , where h*1 and h*2 are obtained
* * *
þ1 hn22  K b q 2i  bq 2j h (4)
i¼1 j¼1  n1  n2
9 * *
 i= using Silverman’s rule (1986) applied to b
q 1i and b
q 2i ;
  X
n1 X
n2 h
b b i¼1 i¼1
2 hn1 n2  K q 1i  q 2j h
;
i¼1 j¼1
5. Repeat steps 3 and 4 B times to obtain B bootstrap estimates of
the statistics of the test.
and n1 and n2 represent the sample size of the groups 1 and 2
respectively, Kð,Þ a kernel function, h the bandwidth, and b q 1 and bq 2 The null hypothesis of equal distributions will be rejected if the
the estimated DEA efficiency scores of the groups 1 and 2. pffiffiffiffiffiffi p-value is below the significance level (a), being determined by the
And Cn;b ¼ Kð0Þ=h  ð1=n1 þ 1=n2 Þ, where Kð0Þ ¼ 1= 2p in
pffiffiffiffiffiffi following expression:
the case of Gaussian kernel, that is, KðuÞ ¼ eu =2 = 2p.
2

And finally: X
B n o
b
p ¼ I Tnb * > Tnb B (8)
8
<   X n1 X
n1 h  i2 b¼1
sb 2b ¼ h n21  K b q 1i  bq 1j h
: where Tnb * are the bootstrap estimates of the statistics of the test
i¼1 j¼1
and Tnb are the original test statistics, obtained from the DEA effi-
  X n2 X
n2 h  i 2
ciency scores bq DEA estimated in step 1.
þ h n22  K b q 2i  bq 2j h (5)
i¼1 j¼1
9 4. Case study
  Xn1 X
n2 h  i2 =
þ 2h n1 n2  K b q 1i  bq 2j h
; 4.1. Sample
i¼1 j¼1

Consider, for example, that group 1 corresponds to the group of The sample of this study comprises 4- and 5-star hotels oper-
5-star hotels and that group 2 is associated with the group of 4-star ating in the Algarve in the period 2005e2007 inclusive. All the

Please cite this article in press as: Oliveira, R., et al., Efficiency and its determinants in Portuguese hotels in the Algarve, Tourism Management
(2012), http://dx.doi.org/10.1016/j.tourman.2012.06.009
 R. Oliveira et al. / Tourism Management xxx (2012) 1e9 5

Table 1 Table 4
Sample and number of hotels and hotel groups in the Algarve. Average efficiencies taking into account the determinants of efficiency.

Stars Hotel Hotels in Total of hotels in % of the Observations 4 Stars 5 Stars With golf Without golf Windward Leeward
groups (no.) sample (no.) the Algarve (no.) sample (no.) Input orientation
5 13 20 28 71.43 39 TE 0.518 0.520 0.525 0.517 0.520 0.517
4 15 36 141 25.53 45 PTE 0.711 0.750 0.628 0.762 0.727 0.733
Total 28 56 169 33.14 84 SE 0.763 0.715 0.862 0.700 0.747 0.725
Output orientation
TE 2.041 2.005 1.934 2.055 2.023 2.026
PTE 1.699 1.646 1.800 1.632 1.676 1.672
Table 2 SE 1.261 1.284 1.109 1.326 1.277 1.257
Statistical characterization of the variables.

Variable Average Standard deviation Minimum Maximum

Rooms (no.) 377 260.7 96 1151 The sample corresponds to 71.43% of the total number of 5-star
Employees (no.) 205 131.9 38 561 hotels, and 25.53% of the 4-star hotels, so it can be considered
F&B capacity (no.) 399 247.9 110 900
Other costs (V) 10,660,899 9,816,586 691,057 54,868,846
representative. The total number of observations is equal to the total
Total revenue (V) 11,134,053 10,160,298 954,369 71,023,628 number of hotel groups of the sample (28) multiplied by the
number of years studied (3) and, therefore, equals 84 (Table 1).

4.2. The model

Table 3
Inefficiency of Algarve hotels using input and output orientations (2005e2007).
The model adopted 4 inputs and 1 output. As inputs, we
Input orientation Output orientation considered the physical characteristics associated with the hotels,
CRS: TE VRS: PTE SE CRS: TE VRS: PTE SE particularly the number of rooms, number of employees, the F&B
Average 0.519 0.729 0.741 1.923 1.534 1.201 (food & beverage) capacity and the ‘other costs’. In this last variable
Maximum 1.000 1.000 1.000 1.000 1.000 1.000 we considered all costs, other than the costs of staff, depreciation
Minimum 0.319 0.444 0.389 3.135 2.786 2.309 and the interest paid. The output considered was the total revenue.
Standard 0.125 0.188 0.177 0.439 0.477 0.355
deviation
Other variables could be relevant, such as those related to the
Median 0.495 0.684 0.768 2.020 1.776 2.015 quality of service provided; however, there were no data about
Efficient hotels 2 12 2 2 9 2 them, assuming that star rating is not a wholly satisfactory proxy
Efficient hotels in 2 1 for such a variable. Table 2 presents the main statistical parameters
all three years
of the variables.
The choice of the variables used as inputs and outputs was
determined by the past studies of Morey and Dittman (1995), Johns,
necessary data were collected from the SABI and AHETA (Associa- Howcroft, and Drake (1997), Anderson et al. (2000), Brown and
tion of Hotels and Tourist Enterprises of Algarve) databases. Ragsdale (2002), Barros (2004), Shang, Hung, Lo, and Wang
The total number of hotels that were star rated in the Algarve in (2008), and Barros and Dieke (2008) and the information made
2005, 2006 and 2007, were respectively 371, 379 and 382. There available by the hotels.
was no change in the number of hotels or in the number of hotel The models adopted were input and output oriented and used
chains in the 3 years studied as far as 5- and 4-star hotels are both CRS and VRS technologies. We believe that, for the tourism
concerned. The total number of 5-star hotels was 28 and the total industry, an output orientation might be more relevant, but in the
number of 4-star hotels was 141. particular case of hotels the scores achieved with the input orien-
The sample used comprises 13 hotel groups owning 20 5-star tation are also important since they are a measure of competi-
hotels, and 15 hotel groups owning 36 4-star hotels. Table 1 tiveness (lower costs can lead to cheaper room prices). Therefore,
summarizes the information collected. we use both orientations in order to obtain more robust results and
Of these 56 hotels in the sample, 7 have golf courses, of which 3 a higher degree of confidence.
are 5-star and 4 are 4-star. Regarding their location, 38 are in the For small numbers of observations Banker, Charnes, Cooper,
west (10 are 5-star hotels of which 2 have golf courses, and 18 are 4- Swartz, and Thomas (1989) propose the rule that in the selection
star hotels, of which 3 have golf courses) and 18 are in the east (10 of variables the total number of inputs and outputs should be less
are 5-star hotels with 1 offering golf, and 8 are 4-star hotels with 1 than a third of the number of observations, which caveat was
offering golf). adhered to in this case.

Fig. 1. Inefficiency intervals for the Algarve hotels.

Please cite this article in press as: Oliveira, R., et al., Efficiency and its determinants in Portuguese hotels in the Algarve, Tourism Management
(2012), http://dx.doi.org/10.1016/j.tourman.2012.06.009
Table 5
Results of test of the equality of efficiency distributions of various groups of hotels.

T CRS VRS

Average P value Tnb Bandwidth (h) Conclusions Average P value Tnb Bandwidth (h) Conclusions
efficiency efficiency
Output 4-Star 0.518 0.85 1.48 0.1916 P 0.642 0.7975 1.326 0.236 P
orientation 5-Star 0.520 0.664
With golf 0.525 0.08 0.8964 0.1327 P 0.590 0 7.2328 0.1217 O
Without golf 0.517 0.673
Windward 0.520 0.0250 2.8991 0.1544 O 0.650 0.265 0.307 0.2136 P
Leeward 0.517 0.657
Input 4-Star 0.518 0.6975 1.231 0.0489 P 0.711 0.2775 0.34 0.0931 P
orientation 5-Star 0.520 0.750
With golf 0.525 0.065 1.3832 0.0381 P 0.628 0 6.303 0.0504 O
Without golf 0.517 0.762
Windward 0.520 0.0175 2.8119 0.0386 P 0.727 0.9825 2 0.0871 P
Leeward 0.517 0.733

P: Accept the hypothesis of equal distributions of efficiency to confidence level of 95%.

O: Reject the hypothesis of equal distributions of efficiency to confidence level of 95%.

Fig. 2. Distributions of the efficiencies in case of output orientation.

Please cite this article in press as: Oliveira, R., et al., Efficiency and its determinants in Portuguese hotels in the Algarve, Tourism Management
(2012), http://dx.doi.org/10.1016/j.tourman.2012.06.009
 R. Oliveira et al. / Tourism Management xxx (2012) 1e9 7

4.3. Results
Table 3 presents the results obtained and the statistical features
of the inefficiency of the Algarve hotels considering the input and
output orientations and the years 2005e2007.
The hotels show huge levels of inefficiency. For example, under
a CRS assumption the average efficiency (TE) is 1.923 and with the
VRS approach the pure technical efficiency (PTE) is 1.534 for output
orientation and 0.519 and 0.729 respectively for input efficiency.
This means that for the input-orientated model we can reduce the
inputs rooms (no.), staff (no.), F&B places (no.) and other costs (V) by
27.1% and maintain the same level of outputs (revenues), or we can
expand the outputs by 53.4% (revenues) for the same level of inputs
(rooms, staff, F&B places and other costs) for the output-orientated
model under VRS assumptions. The levels of scale inefficiency are
also meaningful, although greater when the input orientation is
adopted (at 0.741 and 1.201 inefficiency levels for input and output
orientation). Most results show increasing returns to scale (IRS)
technology. The number of efficient hotels is two for the CRS model
and 12 and 9 respectively for the VRS model with input and output
orientation. There are only two efficient hotels for the three years
using input orientation and one efficient hotel for output orienta-
tion. The minimum efficiency levels found were 0.319 and 3.135 (TE)
and 0.729 and 1.534 (PTE) for the input and output orientation
models and CRS and VRS approaches respectively.
Fig. 1 illustrates the intervals of inefficiency for the Algarve
hotels for the input and output orientations.
efficiency averages when location was taken into account. This
exogenous variable does not seem to be relevant to efficiency.
Performing a test to assess the equality of the efficiency
distributions of various groups of hotels, we found that the results
are identical for both input and output orientation. However, the
results are different if we consider either CRS or VRS technologies.
Table 5 presents the results and their analysis. Figs. 2 and 3
illustrate the results for the input and output orientations. From
these figures we can draw and visualize information about which
groups of hotels are more efficient when an output orientation is
assumed. When the CRS technology was considered, we found
statistically significant differences only among the hotels of
the windward and leeward Algarve; it seems that a majority of
the leeward hotels (observing the distribution of the inverse of
the efficiencies in Fig. 2) are more efficient than most hotels in the
windward region, although the although average efficiencies are
higher for the hotels in the windward, implying therefore
a greater variance in efficiencies for windward hotels. On the
other hand, if VRS technology is adopted, there are statistically
significant differences only between hotels with and without golf.
Those without golf are the most efficient. These results possess
interest, especially concerning the variable of location. As we

4.3.1. Analysis of the results

Analyzing the scores obtained in more detail reveals several
interesting results. The first is that inefficiency decreases between
2005 and 2007, and it is suggested this is a consequence of
increased competition, either locally of globally. The prominent
reason for both orientations (input and output) was increasing
returns to scale (IRS), meaning that big is better. This was found in
26 hotels using the input orientation and 23 using output orien-
tation. To identify outliers, we applied the concepts of supereffi-
ciency (Andersen & Petersen, 1993) and peer count (Charnes,
Cooper, Golany, Seiford, & Stutz, 1985) and we noted that
although there are some extreme observations, these do not
significantly influence their peers and are probably best practices.
We removed only one hotel (considered an outlier) because it
appeared to be overly influencing the results.
Table 4 presents the results for the hotel inefficiency ratings
taking into account their star ratings, the existence of golf courses
and their locations (windward or leeward) considering both input
and output orientations.
As shown in the table, the 5-star hotels seem on average to be
slightly more efficient than the 4-star hotels when using both the
input and output orientation models and CRS and VRS approaches.
However, concerning scale efficiency, the 4-star hotels are closer to
the optimal scale when compared to the 5-star hotels. This might
mean that the 5-star hotels should increase their size further to
improve their efficiency (and to have a higher return on the
investment made). As far as the presence of golf is concerned,
hotels with golf courses are slightly more efficient on average than
those without golf for the CRS model and substantially less efficient
when the VRS model is adopted. The results are consistent for both
orientations (input and output). The differences are more
substantial for the efficiencies computed with the VRS technologies
and for scale efficiencies. Concerning the latter, the result is also
related to the fact that 5-star hotels are normally larger and have
IRS, and simultaneously it is necessary to recover the high levels of
investment made in these amenities as well as the huge operating Fig. 3. Distributions of the inverse of the efficiencies (input-oriented distance func-
and maintenance costs. Finally, no differences were found in the tion) in case of input orientation.

Please cite this article in press as: Oliveira, R., et al., Efficiency and its determinants in Portuguese hotels in the Algarve, Tourism Management
(2012), http://dx.doi.org/10.1016/j.tourman.2012.06.009
8 R. Oliveira et al. / Tourism Management xxx (2012) 1e9
based our analysis only on the average statistical parameter, our
results (and conclusions) are subject to bias. Fig. 2 is clear in this
respect, highlighting similarities of efficiency averages concerning
the location but indicating quite different efficiency distributions
between groups.
5. Concluding remarks
This paper measures the performance of a sample of hotels in the
Portuguese region of the Algarve. It was noted that the Algarve is
one of the more important European tourism regions, one reason
being its reputation for high quality golf courses. By applying the
non-parametric technique of DEA, we estimated the efficiency of
a sample of 84 hotels considering different models (CRS and VRS)
and orientations (input and output). Differing levels of inefficiency
were identified under differing assumptions under models that,
unlike previous studies, incorporated variables such as the star
rating of a hotel and whether it was lined with a golf course. An
additional variable was the importance of location since the region
studied (Algarve) has two distinct and well-defined areas (wind-
ward and leeward). Additionally the testing included the testing
suggested by Carvalho and Marques (2011), a procedure that boot-
straps data to compare the distributions of efficiencies. It was found
that, under certain circumstances, that star ratings do not matter as
far as efficiency is concerned but that location and the existence of
golf courses might be relevant. However, it is to be noted that
differences in test results emerged whether one was using CRS or
VRS, and the results are complicated by the possibility of some bias
in the sample. Nonetheless one clear result does emerge, and that is
that hotels without golf courses are more efficient.
Given a past policy to encourage golf based tourism this finding
poses some problems for future public policies, as there is little
doubt that the high reputation of the Algarve for golf is a factor in
its tourism industry. It raises issues that go beyond the data dis-
cussed and analyzed in this paper, which is the relationship
between the policies and concerns of the public sector with its wish
and responsibilities for developing regional income and employ-
ment, and the concerns of a private sector that requires a return on
investment and equally a return for private sector shareholders.
This is not a debate unique to hotels on the Algarve, but it does raise
a further issue in the application of DEA, and that is to what degree
might the technique be applied to the intangibles inherent in other
measures of company efficiency and policy that are inherent in
corporate socially responsible policies.
References
Andersen, P., & Petersen, N. (1993). A procedure for ranking efficient units in data
envelopment analysis. Management Science, 39(10), 1261e1264.
Anderson, R., Fish, M., Xia, Y., & Michello, F. (1999). Measuring efficiency in the hotel
industry: a stochastic frontier approach. International Journal of Hospitality
Management, 18(1), 45e57.
Anderson, R., Fok, R., & Scott, J. (2000). Hotel industry efficiency: an advanced
programming examination. American Business Review, 18(1), 40e48.
Asalos, N., & Trandafir, A. (2009). Conceptual issues and influencing factors of
profitability in tourism. In The international conference on economics and
administration. Romania: University of Bucharest.
Assaf, A., & Agbola, F. (2011). Modelling the performance of Australian hotels: a DEA
double bootstrap approach. Tourism Economics, 17(1), 73e89.
Banker, R., Charnes, A., & Cooper, W. (1984). Some models for the estimation of
technical and scale inefficiencies in data envelopment analysis. Management
Science, 30(9), 1078e1092.
Banker, R., Charnes, A., Cooper, W., Swartz, J., & Thomas, D. (1989). An introduction
to data envelopment analysis with some of its models and their uses. Research
in Government and Nonprofit Accounting, 5, 125e163.
Barros, C. (2004). A stochastic cost frontier in the Portuguese hotel industry. Tourism
Economics, 10(2), 177e192.
Barros, C., Botti, L., Peypoch, N., Robinot, E., Solonandrasana, B., & Assaf, G. (2011).
Performance of French destinations: tourism attraction perspectives. Tourism
Management, 32(1), 141e146.
Please cite this article in press as: Oliveira, R., et al., Efficiency and its determinants in Portuguese hotels in the Algarve, Tourism Management
(2012), http://dx.doi.org/10.1016/j.tourman.2012.06.009
Barros, C., Botti, L., Peypoch, N., & Solonandrasana, B. (2011). Managerial effi-
ciency and hospitality industry: the Portuguese case. Applied Economics,
43(22), 2895e2905.
Barros, C., & Dieke, P. (2008). Technical efficiency of African hotels. International
Journal of Hospitality Management, 27(3), 438e447.
Barros, C., Dieke, P., & Santos, C. (2009). Heterogeneous technical efficiency of hotels
in Luanda, Angola. Tourism Economics, 16(1), 137e151.
Barros, C., Peypoch, N., & Solonandrasana, B. (2009). Efficiency and productivity
growth in hotel industry. International Journal of Tourism Research, 11(4),
389e402.
Brown, J., & Ragsdale, C. (2002). The competitive market efficiency of hotel brands:
an application of data envelopment analysis. Journal of Hospitality & Tourism
Research, 26(4), 332e360.
Carvalho, P., & Marques, R. (2011). On evaluating efficiency of groups of observations.
Working Paper, Lisbon: IST.
Catalão, C. (2001). Golf tourism in the Algarve. Master thesis. Surrey: University of
Surrey.
Charnes, A., Cooper, W., Golany, B., Seiford, L., & Stutz, J. (1985). Foundations of data
envelopment analysis for Pareto-Koopmans efficient empirical production
functions. Journal of Econometrics, 30(1), 91e107.
Charnes, A., Cooper, W., & Rhodes, E. (1978). Measuring the efficiency of decision
making units. European Journal of Operational Research, 2(6), 429e441.
Chen, C. (2007). Applying stochastic frontier approach to measure hotel managerial
efficiency in Taiwan. Tourism Management, 28(3), 696e702.
Chou, T., Hsu, C., & Chen, M. (2008). A fuzzy multi-criteria decision model for
international tourist hotels location selection. International Journal of Hospitality
Management, 27(2), 293e301.
Coltman, M. (1989). Introduction to travel and tourism: An international approach.
Hoboken, New Jersey: John Wiley & Sons.
Correia, A., Barros, C., & Silvestre, A. (2007). Golf tourism repeat choice behaviour in
the Algarve: a mixed logit approach. Tourism Economics, 13(1), 111e127.
Davies, B. (1999). Industrial organization: the UK hotel sector. Annals of Tourism
Research, 26(2), 294e311.
Farrell, M. (1957). The measurement of productive efficiency. Journal of the Royal
Statistical Society, 120(3), 253e281.
Fried, H., Lovell, C., & Schmidt, S. (2008). The measurement of productive efficiency
and productivity growth. London: Oxford University Press.
Gomezelj, O., & Mihali c, T. (2008). Destination competitiveness e applying different
models, the case of Slovenia. Tourism Management, 29(2), 294e307.
Gonçalves, O., Liang, B., Peypoch, N., & Sbai, S. (2012). Technical efficiency
measurement and inverse B-convexity: Moroccan travel agencies. Tourism
Economics, 18(3), 597e606.
Guizzardi, A., & Bernini, C. (2010). Internal and locational factors affecting hotel
industry efficiency: evidence from Italian business corporations. Tourism
Economics, 16(4), 883e913.
Horwath Consulting. (1992). Estudo sobre o golfe em Portugal. Lisbon: Horwath
Consulting.
Hwang, S., & Chang, T. (2003). Using data envelopment analysis to measure
hotel managerial efficiency change in Taiwan. Tourism Management, 24(4),
357e369.
Ismail, J., Dalbot, M., & Mills, J. (2002). Using RevPAR to analyze lodging
segment variability. Cornell Hotel and Restaurant Administration Quarterly, 43(6),
73e80.
Johns, N., Howcroft, B., & Drake, L. (1997). The use of data envelopment analysis to
monitor hotel productivity. Progress in Tourism and Hospitality Research, 3(2),
119e127.
Khataei, M., Farsin, M., & Mousavi, A. (2008). Measuring the efficiency of selected
hotels in Tehran: a DEA approach. The Economic Research, 8(2), 1e24.
Kumbhakar, S., & Lovell, C. (2000). Stochastic frontier analysis. Cambridge (UK), New
York: Cambridge University Press.
Lundtorp, S., Rassing, C., & Wanhill, S. (1999). The off-season is ‘no-season’: the case
of the Danish island of Bornholm. Tourism Economics, 5(1), 49e68.
Malpezzi, S. (1999). The market for golf. Madison: University of Wisconsin.
Martinez, E. (1992). La economía del golf. Su evolución en el mundo y en España.
Estudios Turísticos, 114, 19e40.
Moliner, J., Claver, E., & Molina, J. (2011). Efectos empresa, grupo estratégico y
localización en el sector hotelero español. Cuadernos de Economía y Dirección de
la Empresa, 14(2), 123e138.
Morey, R., & Dittman, D. (1995). Evaluating a hotel GM’s performance: a case study in
benchmarking. Cornell Hotel and Restaurant Administration Quarterly, 36(5), 30e35.
PENT e Plano Estratégico Nacional do Turismo. (2007). Ministério de Economia e da
Inovação. Lisbon: Turismo de Portugal.
Pinheiro, S. (1994). Estudo sobre o golf em Portugal. Faro: Escola Superior Gestão
Hotelaria e Turismo, University of Algarve.
Priestley, G. (2006). Planning implications of golf tourism. Tourism and Hospitality
Research, 6(3), 170e178.
Ray, P., & Phillips, P. (2005). Performance comparisons of hotels in China. Interna-
tional Journal of Hospitality Management, 24(1), 57e73.
Rodriguez, J., & Gonzalez, E. (2007). Cost efficiency of the lodging industry in
the tourist destination of Gran Canaria (Spain). Tourism Management, 28(4),
993e1005.
Shang, J., Hung, W., Lo, C., & Wang, F. (2008). Ecommerce and hotel performance:
three-stage DEA analysis. The Service Industries Journal, 28(4), 529e540.
Silverman, B. (1986). Density estimation for statistics and data analysis. London:
Chapman and Hall.
 R. Oliveira et al. / Tourism Management xxx (2012) 1e9 9

Simar, L., & Zelenyuk, V. (2006). On testing equality of distributions of technical Maria Isabel Pedro is an Assistant Professor of Project
efficiency scores. Econometric Reviews, 25(4), 497e522. Management and Financial Accounting at Instituto
Smith, R. (1999). The US lodging industry today. Cornell Hotel and Restaurant Superior Técnico (IST) from the Technical University of
Administration Quarterly, 40(1), 18e25. Lisbon and a research member of the Center for
Stynes, D., Sun, Y., & Talhelm, D. (2000). Michigan golf tourists e Economic impacts. Management Studies (CEG-IST). Her areas of interest
East Lansing, Michigan: Department of Park, Recreation and Tourism Resources, comprise economics and the study of financial reports.
Michigan State University.
Tarim, S., Dener, H., & Tarim, S. (2000). Efficiency measurement in the hotel
industry: output factor constrained DEA application. Anatolia: International
Journal of Tourism and Hospitality Research, 11(2), 111e123.
Tezcan, B. (2004). Developing alternative models of tourism in Turkey. Master thesis.
Graduate School of Social Sciences, Middle East Technical University, Ankara,
Turkey.

Ricardo Lisboa Oliveira is an Assistant Professor at
Instituto Superior Manuel Teixeira Gomes (Lusófona
University), Portimão, Portugal and is also a research
member of the Center for Management Studies (CEG-IST)
of Instituto Superior Técnico (IST) from the Technical
University of Lisbon. His research interests are all related
to tourism.
Rui Cunha Marques is an Associate Professor of Decision
and Information at Instituto Superior Técnico (IST) from
the Technical University of Lisbon, a research member of
the Center for Management Studies (CEG-IST) and
a Senior Research Associate of Public Utility Research
Center (PURC) of the University of Florida. His areas of
interest comprise the regulation and governance of
public services as well as performance measurement and
benchmarking of organizations.

Please cite this article in press as: Oliveira, R., et al., Efficiency and its determinants in Portuguese hotels in the Algarve, Tourism Management
(2012), http://dx.doi.org/10.1016/j.tourman.2012.06.009