Efficiency Analysis of the Australian Accommodation Industry

Abstract

This article measures and compares performance in the accommodation industry across Australian states and territories. The authors specify a Bayesian frontier model and estimate the model using panel data spanning the period 1998 to 2009. The results indicate that there are differences in efficiency across states and territories and also between sectors. Empirical results show that the hotel sector is the most efficient, followed by the guest house and motel sector and, finally, the serviced apartment sector. The Bayesian regression results indicate that the key determinants of efficiency in the accommodation industry are the international attractiveness of the state or territory, the share of large accommodation providers in the sector, and the prevailing economic conditions within the state or territory. The findings that the efficiency measures differ across regions and types of sector indicate that adopting a holistic approach to policy formulation and implementation may not be appropriate to ensure competitiveness of the accommodation industry in Australia.

Keywords

hotel management econometrics data envelopment analysis

Over the past decade or so, there has been a growing number of theoretical and empirical studies measuring and explaining technical efficiency across the accommodation industry. These studies have traditionally used different techniques for estimating technical efficiency, hereafter referred to simply as efficiency. Most of these analyses have been limited to simple ratios and partial productivity indicators (e.g. gross profit, salesor revenue, occupancy, or spending). Although these ratios provide some insights into efficiency in the accommodation industry, the approaches have been limited to examining efficiency in relation to specific areas of operation of the accommodation industry. For example, the widely used labor productivity measure only provides information on the efficiency of labor usage in firms. Arguably, the use of a single output measure for examining the overall performance of an industry may be misleading as it fails to capture efficiency measures in various sections of operations in the industry (for a review, see Coelli, Rao, O’Donnell, & Battese, 2005).

Given the limitations of the traditional single output measures of efficiency, there has been a shift in research focus toward the use of multiple output–input measures of efficiency. The most common multiple measures of efficiency have been the stochastic frontier (SF) and data envelopment analysis (DEA) methods. In contrast to simple ratios, both DEA and SF provide a more accurate and comprehensive measure of efficiency. In addition, the inclusion of multiple inputs and outputs ensures that different components of operations are captured in the estimation of efficiency (Assaf & Agbola, 2011a,b).

In the last three decades, a plethora of studies have employed variants of the SF and DEA methodologies in the analysis of efficiency across industries. Studies that have applied the SF and DEA methodologies in the hospitality industry include those by Anderson, Fish, Yi, and Michello (1999); Anderson, Randy, and Fok (1999); Barros (2005); Wang, Hung, and Shang (2006); Barros (2006); Barros and Santos (2006); Chen (2007); Barros and Dieke (2008); Assaf, Barros, and Josiassen (2010); and Assaf and Knežević (2011), among others. The application of the SF and DEA methodologies also extends to other sectors of the hospitality industry. Studies by Reynolds (2003) and Reynolds and Thompson (2007) examined the restaurant industry, whereas Barros and Dieke (2008) and Assaf et al. (2010) and Assaf and Agbola (2011a) examined the hotel sector, and Anderson, Lewis, and Parker (1999) examined the travel agency sector. These methodologies have also been extended to country-specific analyses and includes studies on the hospitality industry in the United Kingdom (Sigala, Airey, Jones, & Lockwood, 2004), the United States (Anderson, Fish, et al., 1999; Anderson, Lewis, et al., 1999; Anderson, Randy, et al., 1999), Portugal (Barros & Alves, 2004; Barros & Mascarenhas, 2005), and Spain (Pérez-Rodríguez & González, 2007), among others.

The objective of this article is to estimate and compare measures of efficiency of accommodation service providers across Australian states and territories. Despite a growing number of studies on efficiency in the accommodation industry abroad, studies on Australia have been limited. The few studies that have examined efficiency in the accommodation industry in Australia include those by Assaf and Agbola (2011a, 2011b). This present study aims to bridge the knowledge gap by empirically estimating and comparing measures of efficiency across the hotel sector, the guesthouse and motel sector, and the serviced apartment sector in Australian states and territories. This study makes four major contributions to the hospitality and tourism literature. First, the study focuses on the Australian accommodation industry whose efficiency measures have not been researched. In a changing global economy, characterized by growing uncertainty, an understanding of the efficiency in one of the growing industries in Australia is pertinent. Second, the study extends the analysis to examine the efficiency in three sectors of the accommodation industry in Australia. No previous study, to the authors’ knowledge, has examined efficiency across these sectors. Third, given that each state and territory has its own tourist board, an understanding of levels of efficiency across states and territories will provide strategic information in the formulation and implementation of policy in the accommodation industry. Last, this study models efficiency in the Australian accommodation industry using a Bayesian output distance function framework. The usefulness of this approach over the DEA approach is that, unlike the DEA, it accounts for random error in estimation and it is not sensitive to the choice of inputs and outputs (Coelli et al., 2005). Furthermore, the Bayesian output distance function allows for the inclusion of multiple outputs unlike the SF and DEA models, which can only incorporate one output in the estimation process. Finally, the Bayesian output distance function imposes theoretical regularity conditions of monotonicity and curvature on the function, thereby making the estimation more flexible than the Maximum Likelihood method often used in the estimation of SF and DEA models.

The rest of this article is organized as follows: The next section provides an overview of the Australian accommodation industry. Then we discuss the methodological framework employed in the estimation of efficiency in the Australian accommodation industry. The data and their sources are described next. The next section reports and discusses the efficiency estimates derived from the application of the Bayesian output distance function to annual data spanning the period 1998 to 2009. This section also reports the regression results of the determinants of efficiency across sectors in the accommodation industry. Finally, we conclude and draw some implications of the empirical findings for policy formulation and implementation in the Australian accommodation industry.

The Australian Accommodation Industry: an Overview

The Australian accommodation industry has made an important contribution to the country’s economic growth. During 2009-2010, the industry employed around 71,500 direct workers. It also generated a total income of AU$9.9 million. Total industry value added was AU$4.8 million, representing 0.5% of Australia’s gross domestic product (GDP). The operating profit before tax for the industry was AU$1.1 million and the operating profit margin was 11% (Australian Bureau of Statistics [ABS], 2010).

The accommodation industry is geographically spread across the Australian states and territories. Three states in the eastern part of Australia, namely, New South Wales (NSW), Victoria (VIC), and Queensland (QLD), account for nearly 76% of total income and attract a large proportion of the total labor employed in the industry. The state of NSW accounts for almost a third of total income (32.8%) and total employment (31.8%). The state of VIC accounts for 19.2% of total income and 19.6% of total employment. Finally, the state of QLD accounts for almost a quarter (23.8% or AU$2,348.4 million) of total income from the accommodation industry in Australia (IBISWorld, 2009). The NSW, VIC, and QLD economies, along with Western Australia (WA), represent the largest proportion of the Australian economy. For instance, the economy of NSW on its own was valued in 2008-2009 at AU$382 billion, representing around 33% of total GDP of the Australian economy. The other largest contributors in terms of GDP are VIC, QLD, WA, South Australia (SA), and Australian Capital Territory (ACT), in that order. Additionally, in terms of economic growth, WA, NSW, QLD, and ACT recorded the largest growth in the last few years (IBISWorld, 2009).

The three major tourism destinations in Australia are NSW, VIC, and QLD. This is because these states are home to some of the most attractive cities in the world, namely, Sydney, Melbourne, Brisbane, and Cairns. These cities have the greatest number of airline flights and are major entry and exit points into and out of Australia. Also important are tourism features and attractions in those states, such as the Sydney Opera House, Great Barrier Reef, Uluru, and Great Ocean Road, all of international significance. The states of NSW, VIC, QLD, and WA also have the largest percentage of large accommodation providers, which is again due to the high concentration of tourism activities in these states. In addition to having some of the largest hotel brands in the accommodation industry, NSW on its own has around 26% of large accommodation properties in Australia, followed by VIC with 22%, QLD with 19%, and finally, WA with 13%.

As mentioned earlier, the focus of this article is to measure and compare the efficiency of three accommodation sectors, namely, the hotel sector, the guesthouse and motel sector, and the serviced apartment sector. Together, these sectors account for more than 70% of total accommodation establishments in Australia (ABS, 2010). The Australian accommodation industry is generally characterized by a high level of competition due to the existence of many accommodation providers that provide the same service at a cheaper price. Consumers are relatively price conscious and therefore will try to ensure that they receive value for money, and they have a choice of a large number of operators who compete to provide excellent service to their guests. With each state and territory having unique characteristics, particularly in terms of distribution of large accommodation properties as well as the concentration of international tourism, the comparison of efficiency measures would assist in formulating policies that are targeted at specific sectors of the accommodation industry across the states and territories. The identification of the highest performing state or territory can serve as benchmark for developing best practices that can be adopted by low-performing states and territories in Australia.

Method

Adopting the Bayesian output distance function, let represent the vector of outputs that can be produced using a vector of inputs Following the parameterization of Fare and Primont (1995), an output distance function can be expressed as follows:

D_{0}^{t} (x^{t}, y^{t}) = \min {θ : (y^{t} / θ) \in P^{t} (x^{t})},

where the functional form is nondecreasing, positive, and linearly homogenous and convex in outputs and nondecreasing and quasi-convex in inputs. The function takes the value of less than or equal to one if (x^t, y^t) belongs to the feasible production set, P^t(x^t) That is, D₀(x, y, t) ≤ 1 if (x^t, y^t) ∈ P^t(x). Moreover, D₀(x, y, t) = 1 if (x^t, y^t) belong to the “frontier” of the production possibility set.

To estimate efficiency, we assume that the distance function in (1) follows a translog functional form, which for the case of M outputs and P inputs can be expressed as follows:

\begin{array}{l} \ln D_{0} (x, y, t) = β_{0} + \sum_{m = 1}^{M} β_{m} \ln y_{m} + 0.5 \sum_{m = 1}^{M} \sum_{n = 1}^{M} β_{m n} \ln y_{m} \ln y_{n} + \sum_{p = 1}^{P} γ_{p} \ln x_{p} \\ + 0.5 \sum_{p = 1}^{P} \sum_{j = 1}^{P} γ_{p j} \ln x_{p} \ln x_{j} + \sum_{p = 1}^{P} \sum_{m = 1}^{M} δ_{p m} \ln x_{p} \ln y_{m} + ω_{t} t + \frac{1}{2} ω_{t t} t^{2} \\ + \sum_{p = 1}^{P} α_{x p} t \ln x_{p} + \sum_{m = 1}^{M} α_{y m} t \ln y_{m} \end{array}

where β₀, β_m, β_mm, γ_p, γ_pj, δ_pm, ω_t, ω_tt, α_xp and α_ym are unknown parameters to be estimated. If we replace –ln D(x, y, t) with a nonnegative term, u, this captures the effect of inefficiency, and a random error, v. The output distance function in the Model (2) can be rewritten as follows:

\begin{array}{l} - \ln (y_{m}) = β_{0} + \sum_{m = 1}^{M} β_{m} \ln (y_{m} / y_{M}) + 0.5 \sum_{m = 1}^{M - 1} \sum_{n = 1}^{M - 1} β_{m m} \ln (y_{m} / y_{M}) \ln (y_{n} / y_{M}) \\ + \sum_{p = 1}^{P} γ_{p} \ln x_{p} + 0.5 \sum_{p = 1}^{P} \sum_{j = 1}^{P} γ_{p j} \ln x_{p} \ln x_{j} + \sum_{p = 1}^{P} \sum_{m = 1}^{M - 1} δ_{p m} \ln x_{p} \ln (y_{m} / y_{M}) + ω_{t} t + \frac{1}{2} ω_{t t} t^{2} \\ + \sum_{p = 1}^{P} α_{x p} t \ln x_{p} + \sum_{m = 1}^{M - 1} α_{y m} t \ln (y_{m} / y_{M}) + u + v . \end{array}

Equation (3) has the same functional form as the traditional SF production models commonly used in the literature. Coelli et al. (2005) provide a detailed discussion of the similarities between the Bayesian output distance function and the SF functional form, hence not discussed here. As noted by Lovell, Richardson, Travers, and Wood (1994), the choice of the normalized output ensures homogeneity and symmetry conditions are satisfied. We therefore normalize the output distance function by the Mth output.¹ We also let the inefficiency term (u) in (3) varies with some exogenous variables (zs). This would provide further insights on the source of efficiency variation between the different states and territories.

We use four inputs (the number of persons employed, number of establishments, bed spaces, and the number of rooms) and three outputs (the number of room nights occupied, average length of stay, and operational revenues) in our study. The selection of these input and output variables was driven by two criteria. First, they are the main source of labor expenditures, capital expenditures,² and operational revenues in a hotel. Second, they have been used in most related studies in the literature (Anderson, Fish, et al., 1999; Anderson, Lewis, et al., 1999; Anderson, Randy, et al., 1999; Assaf et al., 2010; Barros, 2006; Barros & Alves, 2004; Bell & Morey, 1995; Brown & Ragsdale, 2002; Chen, 2007; Hwang & Chang, 2003; Morey & Dittman, 1995; Reynolds, 2003; Shang, Hung, Lo, & Wang, 2008; Wang et al., 2006).

As mentioned, we also use in this study exogenous variables that were selected based on the extant literature and includes the international attractiveness of each state and territory, percentage of large accommodation providers in each state and territory, and economic growth of each state and territory (GDP; see Barros, 2006; Chen, 2007; Shang et al. 2008; Wang et al. 2006). It is important to emphasize that the focus of this article is on the industry and not on individual firms. As such, it is important that the exogenous variables reflect the accommodation industry characteristics of each Australian state and territory. To the authors’ knowledge, no previous study has incorporated macro-level variables in explaining efficiency in the accommodation industry in Australia. Clearly, an understanding of the macroeconomic determinants of efficiency at the industry level is important for the formulation and implementation of policies that has impacts on the accommodation industry (Barros & Dieke, 2008).

The international attractiveness of each state and territory is measured as the number of international visitors to a state or territory and expressed as a share of the total number of international visitors to Australia. In Australia, international visitors are a major segment accounting for about 18.0% of total visitor expenditures (domestic plus international). The literature provides evidences that international tourists usually stay longer and generate more spending in many tourism destinations (Rosenbaum & Spears, 2006; Thrane & Farstad, 2011). Thus, accommodation properties that are located in internationally attractive states are expected to generate higher revenue from the sale of rooms and other hotel services. Support for this claim also comes from Bernard and Jensen (1999) and Wagner (2005), who argue that firms that are able to sell their products to foreign customers are more productive than domestically oriented firms, mainly due to the extra revenues that they can generate. Accommodation properties that have a high percentage of international sales are also able to generate new directions from international tourists and from their experience on international markets. Indirectly, such conditions will also help them improve their performance and strengthen their competitive position in the domestic market (Assaf & Knežević, 2011).

Several studies have assessed the impact of size on the performance of firms. It is argued that large firms have the capacity to be more efficient, because they can use more specialized inputs, coordinate their resources better, and reap the advantages of economies of scale (Alvarez & Crespi, 2000). There are however some studies that argue that small firms can be more efficient, as they have flexible, nonhierarchical structures and do not operate in a monopolistic environment that generally creates a lower incentive to improve performance (e.g., Jovanovic, 1982). In the accommodation industry, studies assessing the impact of size have also come to different conclusions (Barros & Dieke, 2008; Brown & Ragsdale, 2002), especially when the results were tested in different countries. In fact, some studies argue that in the case of a developing economy, firms may be more profitable if they increase their size or join a larger group to make up for any external market failures (Ghemawat & Khanna, 1998; Khanna & Palepu, 2000).

With all these contradictory arguments, it is important to provide further supporting evidence about whether accommodation property size has an important impact on efficiency. As the focus here is on the state level, the impact of size can be assessed by taking the percentage share of large accommodation properties in each state and territory. It is expected that states and territories that have a higher percentage of large hotels are more efficient.

Data Sources and Description

Data on all the above input, output, and exogenous variables were collected separately for each of the following accommodation sectors analyzed in this study:

The hotel sector

The guest house and motel sector

The serviced apartment sector

The sample consists of data for each Australia state and territory (NSW, VIC, QLD, SA, WA, Tasmania [TAS], Northern Territory [NT], and ACT), making it possible to have an interstate comparison for each of the accommodation providers listed above. Our data are collected at the sector level and not at the firm level. This is again in line with the aim of this study, which seeks to analyze the efficiency of the various accommodation sectors across the different Australian states.

The data series for each of the inputs and outputs is available on a quarterly basis, and is also seasonally adjusted.³ Data on the variables used in this study were collected directly from the ABS and Tourism Australia database. Specifically, the study uses the tourism accommodation statistics from the ABS, which contains the results from an ongoing quarterly survey of the Australian accommodation industry. The data series is reported for each state/territory in Australia and by tourism regions as defined by the respective state/territory tourism commissions. Data are available on a quarterly basis from 1998 to 2009. Table 1 provides some summary statistics of the data.

Table 1

Descriptive Statistics of the Data

Statistic	Number of Persons Employed	Number of Establishments	Bed Spaces	Number of Rooms	Operational Revenues	Average Length of Stay (Days)	Number of Nights Occupied
Hotel sector
Mean	69,965	801	210,341	80,441	776,906,000	2	4,899,000
SD	1808.30	38.91	7613.36	3052.87	144,615,000	0.09	414,000
Min	66,547	733	197,513	74,302	525,795,000	2	4,109,000
Max	73,695	863	224,994	86,489	1,042,393,000	3	5,643,000
Guest house and motel sector
Mean	29,357	2,443	247,998	85,939	441,033,000	2	4,339,000
SD	790.51	39.62	2804.83	1354.62	73,425,000	0.03	265,000
Min	27,707	2,382	243,578	83,558	331,210,000	2	3,843,000
Max	31,342	2,491	254,578	87,902	563,367,000	2	4,808,000
Serviced apartment sector
Mean	13,929	830	143,991	44,456	380,421,000	3	2,665,000
SD	1498.78	122.76	22,204	7,573	119,804,000	0.21	541,000
Min	11,400	626	110,381	32,437	178,063,000	3	1,647,000
Max	15,572	975	172,221	54,736	576,553,000	4	3,467,000

Results and Discussion

Measuring Efficiency in the Accommodation Industry

The Bayesian estimation of the model involved 60,000 draws after discarding the first 20,000 as a burn-in. For the purpose of comparison, we tested whether a fixed effect model provides a better fit than the random effect model. We compared the two models using the deviance information criterion (DIC) test (Spiegelhalter, Best, Carlin, & van der Linde, 2002). A model with a smaller DIC is considered to be a better fit. The results indicate that the random effect model has a DIC of −110.12 compared with a DIC of −70.51 for the fixed effect model. We conclude that the random effect model is superior and hence proceed to estimate the Bayesian model within a random effect framework.

Given that the accommodation industry consists of three sectors, namely, the hotel sector, the guest house and motel sector, and the serviced apartment sector, we proceed to estimate the Bayesian posterior regression estimates for these three sectors. Table 2 reports the posterior estimates of the model. The results indicate that, across the three sectors, the models perform well with most coefficients being correctly signed and statistically significant.

Table 2

Posterior Estimates of the Distance Frontier Model

Variable	Coefficient	Hotel Sector Estimates	Guest House and Motel Sector Estimates	Serviced Apartment Sector Estimates
Intercept	β₀	0.0112 (0.0033)	0.0156 (0.0015)	0.0182 (0.0016)
(ln y₁)	β₁	0.1956 (0.0412)	0.1798 (0.0365)	0.1698 (0.0193)
(ln y₁)(ln y₁)	β₁₁	0.0049 (0.0019)	0.0045 (0.0017)	0.0052 (0.0022)
(ln y₃)	β₃	0.0027 (0.0015)	0.0031 (0.0016)	0.0052 (0.0018)
(ln y₃)	β₃₃	0.0032 (0.0013)	0.0036 (0.0014)	0.0042 (0.0014)
(ln y₁)(ln y₃)	β₁₃	0.0014 (0.0011)	0.0014 (0.0014)	0.0018 (0.0017)
ln x₁	γ₁	−0.3516 (0.0434)	−0.3432 (0.0411)	−0.3634 (0.0436)
ln x₂	γ₂	−0.6356 (0.1455)	−0.6329 (0.1394)	−0.8521 (0.1458)
ln x₃	γ₃	−0.0714 (0.0135)	−0.0538 (0.0129)	−0.0582 (0.0124)
ln x₄	γ₄	−0.0266 (0.0018)	−0.0238 (0.0018)	−0.0321 (0.0022)
(ln x₁)²	γ₁₁	−0.0165 (0.0145)	−0.0159 (0.0141)	−0.0231 (0.0147)
(ln x₂)²	γ₂₂	−0.3155 (0.1365)	−0.3097 (0.1329)	0.4523 (0.1342)
(ln x₃)²	γ₃₃	0.0085 (0.0019)	0.0069 (0.0020)	0.0058 (0.0021)
(ln x₄)²	γ₄₄	0.0041 (0.0015)	0.0039 (0.0014)	0.0142 (0.0017)
(ln x₁)(ln x₂)	γ₁₂	0.1446 (0.0275)	0.1555 (0.0286)	0.1672 (0.0235)
(ln x₁)(ln x₃)	γ₁₃	−0.0126 (0.0016)	−0.0118 (0.0015)	−0.0158 (0.0018)
(ln x₁)(ln x₄)	γ₁₄	0.0145 (0.0011)	0.0139 (0.0011)	0.0142 (0.0014)
(ln x₂)(ln x₃)	γ₂₃	0.0311 (0.0126)	0.0309 (0.0133)	0.0422 (0.0126)
(ln x₂)(ln x₄)	γ₂₄	−0.0436 (0.0054)	−0.0419 (0.0050)	−0.0567 (0.0057)
(ln x₃)(ln x₄)	γ₃₄	0.1285 (0.0103)	0.1251 (0.0100)	0.1452 (0.0105)
(ln x₁)(ln y₁)	δ₁₁	0.0065 (0.0135)	0.0061 (0.0131)	0.0069 (0.0135)
(ln x₂)(ln y₁)	δ₂₁	0.0017 (0.0005)	0.0017 (0.0005)	0.0022 (0.0005)
(ln x₃)(ln y₁)	δ₃₁	0.0032 (0.0013)	0.0030 (0.0011)	0.0252 (0.0018)
(ln x₄)(ln y₁)	δ₄₁	0.0023 (0.0011)	0.0021 (0.0009)	0.0035 (0.0013)
(ln x₁)(ln y₃)	δ₁₃	−0.0054 (0.0032)	−0.0052 (0.0032)	−0.0057 (0.0034)
(ln x₂)(ln y₃)	δ₂₃	0.0056 (0.0022)	0.0048 (0.0018)	0.0052 (0.0022)
(ln x₃)(ln y₃)	δ₃₁	0.0423 (0.0014)	0.0415 (0.0012)	0.0378 (0.0015)
(ln x₄)(ln y₃)	δ₄₁	0.0482 (0.0198)	0.0388 (0.0222)	0.0482 (0.0192)

Table 3 reports the efficiency estimates across the three sectors in the accommodation industry. These results have been averaged over different years under analysis. The last row represents the overall average. We can see that the hotel sector seems to be the best performer for most years and has the highest average, operating at 82%. Between the other two sectors, the guest house and motel sector has the higher overall average efficiency. The difference in efficiency between the serviced apartment sector and the hotel sector is around 5% and between the guest house and motel sector and the hotel sector only 2%.

Table 3

Efficiency Comparison Between the Various Accommodation Sectors

Period	Hotel Sector	Guest House and Motel Sector	Serviced Apartment Sector
1998	0.7572	0.7462	0.7438
1999	0.7720	0.7453	0.7437
2000	0.7719	0.7324	0.7245
2001	0.7601	0.7393	0.7272
2002	0.7633	0.7721	0.7383
2003	0.7824	0.7820	0.7448
2004	0.8230	0.8028	0.7700
2005	0.8636	0.8117	0.7963
2006	0.8600	0.8504	0.7810
2007	0.8840	0.8592	0.8118
2008	0.8598	0.8456	0.8137
2009	0.8802	0.8330	0.7952
Average	0.8148	0.7933	0.7658

Table 4 reports the results for each of the different accommodation sectors across the various states and territories. For the hotel sector, each column in this table represents a different state and territory. The last row represents the average efficiency score for each state and territory over the period of this study. It is clear that the state of NSW has the most efficient hotel sector followed by VIC and QLD. The average efficiency for NSW is nearly 88%, which indicates that the hotel sector in this state is around 12% from achieving maximum efficiency. Out of the different states, SA and TAS have the lowest efficiency on average. The NT has the lowest efficiency in the sample. From Table 4, it is also clear that all states and territories generally achieved an increase in efficiency between 1998 and 2009. Those that achieved the highest efficiency increase are SA, WA, and TAS, with WA improving the most. However, it is important to note that between 2007 and 2009 several states experienced an efficiency decline, probably due to the impact of the economic crisis.

Table 4

Average efficiency comparison of the hotel sector across the various states

Period	NSW	VIC	QLD	SA	WA	TAS	NT	ACT
1998	0.8263	0.8430	0.7889	0.7167	0.7275	0.6668	0.6654	0.8131
1999	0.8479	0.8387	0.8274	0.7237	0.7212	0.6566	0.7310	0.8291
2000	0.8858	0.8602	0.8283	0.7130	0.7258	0.6350	0.7668	0.7605
2001	0.8409	0.8332	0.8333	0.7078	0.7018	0.6133	0.7558	0.7943
2002	0.8651	0.8026	0.8320	0.7200	0.7285	0.6270	0.7344	0.7970
2003	0.8853	0.8380	0.8370	0.7040	0.7970	0.6711	0.6551	0.8716
2004	0.9283	0.8863	0.8678	0.7278	0.8153	0.7208	0.7463	0.8910
2005	0.9069	0.9185	0.8943	0.7973	0.8890	0.7993	0.7718	0.9320
2006	0.8994	0.9246	0.8993	0.8463	0.8931	0.6944	0.7941	0.9286
2007	0.9020	0.9329	0.9084	0.8412	0.9317	0.8874	0.7957	0.8730
2008	0.8744	0.9014	0.8622	0.8405	0.9137	0.7718	0.7795	0.8351
2009	0.8724	0.8730	0.8571	0.8429	0.9190	0.7825	0.7233	0.8910
Average	0.8779	0.8710	0.8530	0.7651	0.8245	0.7555	0.7299	0.8514

Note: NSW = New South Wales; VIC = Victoria; QLD = Queensland; SA = South Australia; WA = Western Australia; TAS = Tasmania; NT = Northern Territory; ACT = Australian Capital Territory.

Table 5 reports the efficiency estimates for the guest house and motel sector across the different states. Again, they are averaged over different time periods. NSW has the highest average efficiency score on this sector. This is followed by WA, ACT, QLD, and VIC, in that order. For NSW, the state is now around 11% from achieving full efficiency. SA and the NT, which have the lowest efficiency on the guest house and motel sector, are around 32% from achieving full efficiency. The average efficiency scores indicate most states and territories have achieved an increase in efficiency between 1998 and 2009. Those that have achieved the highest increase in efficiency are QLD, WA, and ACT, with WA again having the highest average efficiency. On the other hand, SA and NT experienced the lowest increase in efficiency and were generally the least efficient across all the years under analysis.

Table 5

Average Efficiency Comparison of the Guest House and Motel Sector Across the Various States

Period	NSW	VIC	QLD	SA	WA	TAS	NT	ACT
1998	0.8705	0.7496	0.7312	0.7108	0.7231	0.6793	0.6631	0.8131
1999	0.9262	0.7513	0.7600	0.7230	0.7738	0.7189	0.6991	0.7969
2000	0.9035	0.7320	0.7245	0.7051	0.7487	0.6763	0.6921	0.8038
2001	0.8568	0.7273	0.7628	0.6855	0.7969	0.6950	0.6930	0.7999
2002	0.8719	0.7426	0.7708	0.6988	0.7920	0.6933	0.6817	0.7953
2003	0.8843	0.7341	0.7903	0.6678	0.8218	0.6780	0.6758	0.8260
2004	0.8815	0.7464	0.8280	0.7018	0.8543	0.7075	0.7596	0.8405
2005	0.9065	0.8046	0.8398	0.6843	0.8478	0.8900	0.6514	0.8060
2006	0.9052	0.7856	0.8413	0.7121	0.8335	0.7107	0.6106	0.8491
2007	0.9202	0.8262	0.8425	0.7593	0.8948	0.6638	0.6859	0.9018
2008	0.9149	0.8219	0.8207	0.6916	0.9100	0.7111	0.7373	0.9020
2009	0.8936	0.7780	0.8167	0.7276	0.8838	0.7109	0.6817	0.8895
Average	0.8938	0.7666	0.7941	0.7115	0.8384	0.7118	0.6859	0.8353

Note: NSW = New South Wales; VIC = Victoria; QLD = Queensland; SA = South Australia; WA = Western Australia; TAS = Tasmania; NT = Northern Territory; ACT = Australian Capital Territory.

The efficiency results from the serviced apartment sector are expressed in Table 6. The ACT, NSW, QLD, and VIC are again the best performers. VIC, in particular, seems to enjoy the highest efficiency, followed by ACT and QLD. On the other hand, SA, TAS, and the NT seem to be the worst performers. A comparison between the different years indicates that most states and territories have achieved an increase in efficiency between 1998 and 2009. Those that have achieved the highest increase in efficiency are QLD, VIC, and WA, with QLD achieving the highest efficiency. The most efficient states in 2009 were the ACT, QLD, VIC, and NSW.

Table 6

Average Efficiency Comparison of the Serviced Apartment Sector Across the Various States

Period	NSW	VIC	QLD	SA	WA	TAS	NT	ACT
1998	0.7777	0.7782	0.7306	0.8173	0.7310	0.6978	0.7143	0.7829
1999	0.7866	0.8261	0.7236	0.6986	0.7859	0.7377	0.7544	0.7891
2000	0.7530	0.8487	0.7628	0.6639	0.7803	0.7162	0.7377	0.7464
2001	0.7227	0.8326	0.7948	0.6981	0.7734	0.6781	0.7275	0.7870
2002	0.8033	0.8206	0.8233	0.7453	0.7875	0.6896	0.7437	0.8235
2003	0.8408	0.8318	0.8362	0.6995	0.7620	0.7615	0.7113	0.8330
2004	0.8454	0.8286	0.8669	0.7643	0.7835	0.7164	0.7661	0.8609
2005	0.8443	0.8720	0.8593	0.7657	0.7987	0.7229	0.7771	0.8633
2006	0.8348	0.8612	0.8566	0.8026	0.8374	0.9178	0.8582	0.8549
2007	0.8514	0.8981	0.8598	0.8172	0.8769	0.8926	0.7939	0.8934
2008	0.8483	0.8904	0.8565	0.7997	0.8653	0.7639	0.7606	0.8804
2009	0.8326	0.8608	0.8449	0.8182	0.8222	0.7804	0.7437	0.8610
Average	0.8117	0.8457	0.8179	0.7575	0.8003	0.7562	0.7673	0.8313

Note: NSW = New South Wales; VIC = Victoria; QLD = Queensland; SA = South Australia; WA = Western Australia; TAS = Tasmania; NT = Northern Territory; ACT = Australian Capital Territory.

Determinants of Efficiency in the Australian Accommodation Industry

Table 7 reports the Bayesian regression results of the determinants of inefficiency across sectors in the accommodation industry. We run three separate regressions for each of the accommodation sectors under analysis. Given that we specify an inefficiency model, the coefficients of the variables in the model imply that an increase in these explanatory variables is associated with a decrease in inefficiency in production. The signs of the coefficients of the explanatory variables are as expected. The empirical results seem to favor those states and territories that have a higher international percentage of international tourists, have a higher percentage of large hotels, and enjoy better economic conditions. This provides a strong justification for NSW, VIC, and QLD having consistently the highest efficiency scores across the various accommodation sectors. These states, for instance, have the largest percentage share of international visitors due to their importance as major international attractions for visitors into Australia. They also have the biggest share of larger accommodation establishments in terms of number of rooms.

Table 7

Regression Results (Dependent Variable: Inefficiency)

Variable	Posterior Mean	Standard Deviation
Hotel sector
Constant	1.2111	0.0325
International attractiveness	−0.2752	0.0351
Percentage of large accommodation properties	−0.3511	0.0116
GDP	−0.0028	0.0010
Guest house and motel sector
Constant	1.4523	0.0566
International attractiveness	−0.3321	0.0443
Percentage of large accommodation properties	−0.2875	0.0131
GDP	−0.0022	0.000
Serviced apartment sector
Constant	1.3665	0.0116
International attractiveness	−0.2642	0.0172
Percentage of large accommodation properties	−0.3321	0.0567
GDP	−0.0026	0.0010

Note: GDP = gross domestic product.

As highlighted above, accommodation properties of a larger size usually have strong economies of scale and can generate additional savings. Finally, the results indicate that accommodation properties that are located in states and territories that have better economic conditions are more efficient. Initially, this result was expected as a strong economy helps attract more investments into the area. It also helps improve the infrastructure required to attract more international visitors to the area. The above findings provide an insight into the rationale for WA experiencing the highest efficiency increase in both the hotel sector and the guest house and motel sector and also achieving the second highest increase in the serviced apartment sector. Over the last few years, the state recorded the highest economic growth among all Australian states and territories (IBISWorld, 2009). From this finding, it is also not surprising that the ACT seems to perform strongly, despite the fact that it has a low percentage of international tourists and also a low percentage of large accommodation properties. The economy of this territory was also one of the best performing economies among all Australian states in the last few years. Most accommodation properties in this territory are located in Canberra, which is the capital city of Australia. Given that Canberra is also the main center for government administration and defense, this helps attract many meeting and convention customers to hotels in the area.

Conclusion

This study provided, for the first time, a rich analysis of the efficiency of the Australian accommodation industry across various states and territories. We used the distance function approach, which allows for the inclusion of multiple outputs. We showed that the hotel sector is the most efficient accommodation sector in Australia, followed by the guest house and motel sector and the serviced apartment sector. We also showed that NSW, VIC, QLD, and the ACT are the most efficient across the three accommodation sectors included in the study. On the other hand, SA, TAS, and the NT are the least efficient. The study also discussed the degree of improvement in efficiency for the various states and territories. Furthermore, the study analyzed the sources of efficiency variations between the various states and territories. We showed that those states or territories that are more internationally attractive, have a higher percentage of large accommodation properties, and possess better economic conditions are better performers.

These findings could be of particular importance to tourism policy makers in the various states and territories. There was always a call for a study that benchmarks the Australian accommodation industry; however, none of the previous studies in the literature has addressed this issue. Therefore, we expect the results of this study to be used as a starting point for further investigation into the performance of the industry. The low efficient states and territories can benefit from the results by investigating the sources of efficiency advantage in the more efficient states and territories. The low performing states might also consider investigating how some states and territories have achieved a strong improvement in efficiency on each of the sectors analyzed. Investors in the industry might also use the results as guidance for future investments. For example, investors might want to consider opening new properties in states that are internationally attractive to international tourists, as these states seem to enjoy higher performance.

Tourism associations in each state, particularly the low performing ones, might also consider increasing the percentage of large accommodation providers to help improve the overall performance of the industry. They might also need to promote their tourism industry better to international tourists in order to attract stronger sources of revenues. It seems also that some sectors like the serviced apartment sector need more attention than other accommodation sectors. Currently, this sector in Australia still lacks consistency and clarity in the ratings, a lack of understanding by the wider investment community, and a lack of market representation. Future research will extend the analysis to include other accommodation sectors in the analysis of technical efficiency in the Australian accommodation industry.

Footnotes

A.George Assaf, PhD (assaf@ht.umass.edu), is an assistant professor in the Isenberg School of Management, University of Massachusetts-Amherst, Amherst, Massachusetts. Frank W. Agbola (Frank.Agbola@newcastle.edu.au) is a senior lecturer in the Newcastle Business School, University of Newcastle, Callaghan, New South Wales, Australia.

Notes

References

Alvarez

Crespi

(2000). Determinants of technical efficiency in small firms. Small Business Economics, 20, 233-244.

Anderson

I. R.

Fish

Michello

(1999). Measuring efficiency in the hotel industry: A stochastic frontier approach. International Journal of Hospitality Management, 18, 45-57.

Anderson

I. R.

Lewis

Parker

(1999). Another look at the efficiency of corporate travel management departments. Journal of Travel Research, 37, 267-272.

Anderson

I. R.

Randy

Fok

(1999). Hotel industry efficiency: An advanced linear programming examination. American Business Review, 18, 40-48.

Assaf

Barros

C. P.

Josiassen

(2010). Hotel efficiency: A bootstrapped metafrontier approach. International Journal of Hospitality Management, 30, 819-826.

Assaf

Knežević

L C

. (2011). Privatization, market competition, international attractiveness management tenure and hotel performance: Evidence from Slovenia. International Journal of Hospitality Management, 30, 391-397.

Assaf

A. G.

Agbola

F. W.

(2011a). Modelling the performance of Australian hotels: A DEA double bootstrap approach. Tourism Economics, 17(1), 73-89.

Assaf

A. G.

Agbola

F. W.

(2011b). Estimating and bootstrapping Malmquist indices: A case of total productivity in the Australian hotel industry. Tourism Analysis, 16, 295-304.

Australian Bureau of Statistics. (2010). Tourist accommodation, Australia (Catalogue # 8635.0). Sydney, Australia: Author.

10.

Barros

C. P.

(2005). Evaluating the efficiency of a small hotel chain with a Malmquist productivity index. International Journal of Tourism Research, 7, 173-184.

11.

Barros

C. P.

(2006). Analyzing the rate of technical change in the Portuguese hotel industry. Tourism Economics, 12, 325-346.

12.

Barros

C. P.

Alves

F. P.

(2004). Productivity in the tourism industry. International Advances in Economic Research, 10, 215-225.

13.

Barros

C. P.

Dieke

(2008). Technical efficiency of African hotels. International Journal of Hospitality Management, 27, 438-447.

14.

Barros

C. P.

Mascarenhas

M. J.

(2005). Technical and allocative efficiency in a chain of small hotels. International Journal of Hospitality Management, 24, 415-436.

15.

Barros

C. P.

Santos

C. A.

(2006). The measurement of efficiency in Portuguese hotels using data envelopment analysis. Journal of Hospitality & Tourism Research, 30, 378-400.

16.

Bell

A. R.

Morey

R. C.

(1995). Increasing the efficiency of corporate travel management thought macro-benchmarking. Journal of Travel Research, 33(3), 11-20.

17.

Bernard

Jenson

B. J.

(1999). Exceptional exporter performance: Cause, effect, or both? Journal of International Economics, 47(1), 1-25.

18.

Brown

J. R.

Ragsdale

C. T.

(2002). The competitive market efficiency of hotel brands: An application of data envelopment analysis. Journal of Hospitality & Tourism Research, 26, 260-332.

19.

Chen

C. F.

(2007). Applying the stochastic frontier approach to measure hotel managerial efficiency in Taiwan. Tourism Management, 28, 696-702.

20.

Coelli

Rao

O’Donnell

C. J.

Battese

(2005). An Introduction to efficiency and productivity analysis. New York, NY: Springer.

21.

Fare

Primont

(1995). Multi-output production and duality: Theory and applications. Boston, MA: Kluwer Academic.

22.

Ghemawat

Khanna

(1998). The nature of diversified business groups: A research design and two case studies. Journal of Industrial Economics, 1, 35-61.

23.

Hwang

S. N.

Chang

T. Y.

(2003). Using data envelopment analysis to measure hotel managerial efficiency change in Taiwan. Tourism Management, 24, 357-369.

24.

IBISWorld. (2009). Hotel accommodation in Australia (Report No. H5711). Sydney, Australia: Author.

25.

International Labour Organization. (1989). Productivity and training in the hotel, catering and tourism sector (Vol. 70). Geneva, Switzerland: Author.

26.

Jovanovic

(1982). Selection and evolution of the industry. Econometrica, 50, 649-670.

27.

Khanna

Palepu

(2000). Is group affiliation profitable in emerging markets? An analysis of diversified Indian business groups. Journal of Finance, 55, 867-891.

28.

Lovell

C. A. K.

Richardson

Travers

Wood

L. L.

(1994). Resources and functionings: A new view of inequality in Australia. In Eichhorn

(Ed.), Models and measurement of welfare and inequality (pp. 787-807). Berlin, Germany: Springer.

29.

Morey

Dittman

(1995). Evaluating a hotel GM’s performance. Cornell Hotel & Restaurant Administration Quarterly, 36(5), 30-35.

30.

O’Donnell

Coelli

(2005). A Bayesian approach to imposing curvature on distance functions. Journal of Econometrics, 126, 492-523.

31.

Pérez-Rodríguez

J. V.

González

E. A.

(2007). Cost efficiency of the lodging industry in the tourist destination of Gran Canaria (Spain). Tourism Management, 28, 993-1005.

32.

Reynolds

(2003). Hospitality-productivity assessment using data envelopment analysis. Cornell Hotel and Restaurant Administration Quarterly, 44, 130-137.

33.

Reynolds

Thomson

(2007). Multiunit restaurant productivity assessment using three-phase data envelopment analysis. International Journal of Hospitality Management, 26(1), 20-32.

34.

Rosenbaum

S. M.

Spears

D. L.

(2006). An exploration of spending behaviors among Japanese tourists. Journal of Travel Research, 44, 467-473.

35.

Shang

J. K.

Hung

W. T.

C. F.

Wang

F. C.

(2008). Ecommerce and hotel performance: Three-stage DEA Analysis. Service Industries Journal, 28, 529-540.

36.

Sigala

Airey

Jones

Lockwood

(2004). ICT paradox lost? A stepwise DEA methodology to evaluate technology investments in tourism settings. Journal of Travel Research, 43, 180-192.

37.

Spiegelhalter

Best

N. G.

Carlin

van der Linde

(2002). Bayesian measures of model complexity and fit (with discussion). Journal of the Royal Statistical Society, Series B: Statistical Methodology, 64, 583-639.

38.

Thrane

Farstad

(2011). Domestic tourism expenditures: The non-linear effects of length of stay and travel party size. Tourism Management, 32(1), 46-52.

39.

Wagner

(2005). Exports and productivity: A survey of the evidence from firm level data (Discussion Paper No. 319). Hamburg, Germany: Hamburg Institute of International Economics.

40.

Wang

Hung

W. T.

Shang

J. K.

(2006). Measuring the cost efficiency of international tourist hotels in Taiwan. Tourism Economics, 12, 65-85.