بررسی پویایی های بهره وری در بخش مهمان نوازی ایتالیایی . مطالعه موردی منطقه ای

This paper introduces a methodology to describe and compare the economic relative performance of the hospitality sector of the Italian regions during the period 2000–2004. Dynamics of the hospitality sector of each region is represented by the evolution of its economic efficiency. The investigation involves the following steps: a static Data Envelopment Analysis (DEA) to estimate the pure economic efficiency; two different notions of distances between time series and hierarchical clustering techniques are used to classify the economies in the sample. By using a correlation-based distance, three main clusters are detected, while two clusters are identified when the average distance is used. The trend patterns, identified by employing the correlation distance, can be interpreted in terms of exogenous factors that influence the economic efficiency of the group of regions, causing shocks picked up by the high volatility as well as structural breaks. By employing the average distance, one infers information on the cluster that have had similar efficiency values over the period under analysis. This efficiency can be also interpreted in terms of a particular type of hospitality management as well as the firm structure. Following the analysis, some policy and management implications are presented.

The hospitality sector plays an important role in the Italian economy as a revenue generator. Federalberghi (2010) emphases that the Italian hospitality sector, expressed in terms of number of hotel rooms, ranks fourth after the United States, Japan and China. Besides, amongst the European countries, Italy is a leader in terms of hotel dimension and quality (number of stars). This motivates the interests of the this paper to examine the economic efficiency of the Italian hospitality sector. This question is particularly important in the light of an increasing awareness of sustainability issues that challenge the need for a further expansion of tourism infrastructure that may exploit finite and no-renewable natural resources (e.g. Bruni, Guerriero, & Patitucci, 2011). As a matter of fact, within the time span between 2000 and 2004, supply capacity has grown by 7.9%, reaching two million beds-place in 2004 (ISTAT, 2011). Since the seminal work on Data Envelopment Analysis (DEA) by Charnes, Cooper, and Rhodes (1978), empirical research papers have focused on efficiency in the manufacturing sector, health services, educational institutions, the services sector and private organizations such as banks. The analysis of efficiency in the tourism and hospitality sector has been growing during the last two decades (see Barros, 2005a and Barros, 2005b; Fuentes, 2011 or Pulina, Detotto & Paba, 2010) for a literature account). In this paper the dynamic evolution of the efficiency of the hospitality sector in the Italian regions is explored. The dynamic of the efficiency is explored in two steps. In the first step, following the work by Banker, Charnes, and Cooper (1984) a Data Envelopment Analysis (DEA) is applied to all the regions in the temporal window 2000–2004. Assuming a variable return to scale frontier of efficiency, the pure technical efficiency (PTE) is obtained for each region and period. This information allows one to quantify the efficiency level of the regions with respect to its own performance over time, as well as the performance of the relatively most efficient regions and periods. In the second step, the regions are clustered according to the temporal evolution of their efficiency. Two measures of distance between the time series of the hospitality sector in each Italian region are employed: the correlation and the supremum distance. These two measures are complementary to understand the dynamic evolution of the relative efficiency of the regions. Dynamics of two regions are close with respect to the correlation distance (Gower, 1966) if they have similar trend behavior across the time period. The supremum distance, on the other hand, groups regions in corridors along the whole period of study. If the supremum distance among the efficiency of a group of regions is equal to 10, it means that across the different periods, no one of the regions were separated more than 10 points of efficiency. Whereas the correlation distance gives information about the trends of the efficiency, the supremum distance informs on how different the dynamics of regional efficiency was during the period of study. Then, both distances give complementary information about the dynamics of the regions. On the one hand, if a group of regions have small correlation distance among them, this can be interpreted as economies having similar responses to external shocks affecting their efficiency. On the other hand, if a group of regions are “close” with respect to the supremum distance, this means that they have followed almost the same trajectory during the period under study, although they could have had different trends. Even though there is an increasing concern with efficiency in the literature of tourism and hospitality (see Barros, 2005a and Barros, 2005b; Pulina et al., 2010; Fuentes, 2011 for a literature account), so far a few studies have expored the dynamic evolution of efficiency. Tsaur and Sheng-Hshiung (2001) study the efficiency of the 53 international tourist hotels in Taiwan from 1996 to 1998 and the time effect is introduced computing the average of the inputs and outputs during the three years. Hwang and Chang (2003) compute the efficiency change in year 1994–1998 for 45 Taiwan hotels using the Malmquist productivity decomposition. The authors use this temporal information to organize the 45 hotels into 6 clusters according to the efficiency change during the period 1994-1998 and the final relative efficiency in 1998. Thus, they identify in the two extremes; hotels with high competitiveness and a fast pace of progress as hotels in the “right track” and hotels with low competitiveness and worse pace of progress as firms with managerial deficiencies. Barros (2005b) explores the evolution of the efficiency of a hotel chain through two alternatives: on the one hand he uses a Malmquist productivity index to decompose the total productivity change in technical efficiency change and technological change and, on the other hand, the author analyzes the changes of the total productivity measures across the time with a Tobit model. Assaf and Agbola (2011) study the efficiency of a sample of 31 Australian hotels during the period 2004–2007. They employed the 31 × 4 = 124 observations in one DEA analysis, comparing the efficiency of the same hotels across the temporal window of 4 years. The authors use a truncated regression for showing that large hotels located in Australian cities are the conditions for being more efficient. This finding is consistent with the study by Barros (2006) and suggests that big hotels located in cities tend to be more efficient than those small in remote areas. Barros used information from Portuguese hotels between 1998 and 2002 to estimate a translog frontier model. Stochastic frontier analysis has provided an instrument for exploration of the dynamic, through the data panel study of different specification of cost and production functions. Perez-Rodriguez and Acosta-Gonzalez (2007) explored the cost efficiency and economic scales of the lodging industry on the island of Gran Canaria during the period 1991–2002 using a stochastic cost frontier model. The authors show statistically that efficiencies vary in time and that the mean cost inefficiency decreased over time. The paper is organized as follows. Section 2 introduces the DEA methodology and applies it to the hospitality sector of the 21 Italian regions. Starting from these results, Section 3 analyses the dynamics of the economic efficiency for these economies by introducing two different metric distances and hierarchical clustering techniques. The final section includes concluding remarks, policy and management implications of the results and future research.

In this paper, a methodology has been introduced to explore the dynamical behavior of the economic efficiency of the hospitality sector in the Italian regions, during the period 2000-2004. This tool allows one to construct clusters according to two measures of distance between the trajectory efficiency of the regions: the correlation and the average distance. The correlation distance clusters together regions where the time series of the measure of efficiency are correlated. The average distance clusters together regions having a similar level of efficiency during the whole period. All regions, except three outliers have been clustered in three groups according to the correlation distance. Then, the evolution of the average efficiency has been taken into consideration, in order to identify some common features which may have determined regions belonging to the same cluster to respond to shock in a similar way. Considering the evolution of the overnight of stays during the period of interest, as an indicator for external shocks, it has been noticed that regions belonging to Cluster C (i.e. Apulia, Trento, Veneto and Bozen) show an opposite pattern with respect to the evolution of such an economic indicator. It may be possible that the terroristic attack in 2001 have moved tourism flows away from central regions with large cities such as Florence and Rome, thus causing an opposite movement of the efficiency in the peripheral regions. On the one hand, when considering the average distance, regions are grouped into two clusters, with the majority of regions belonging to the first cluster. When looking at the evolution of the average efficiency over the period under consideration, the path of the efficiency of all the regions, but those belonging to Cluster 2, is rather smooth. On the other hand, Sicily and Aosta Valley, which belong to the second cluster, are characterized by a large decrease of the efficiency in the period taken into consideration. This may be due to the high seasonality of these regions, which may cause inefficiencies in the utilization of resources. Finally, all regions are segmented according to the fact that their belonging to the clusters identified with the correlation distance rather than the average distance. Regions which show a similar trend and dynamics share common features in terms of entrepreneurial structure (i.e. number of stars, beds, rooms) as well as for the seasonality. Structural factors rather than the mere geographical location of hotels indeed affects the evolution of hotel efficiency over time. The results obtained in this paper suggests future research into two lines. On the one hand, the exploration of the data in the Italian hospitality sector has indicated the possibility to explore new relationships between investments, labor, revenues and value added among the regions. On the other hand, new distances can be employed to extract different information from this empirical data and another data sources. In particular, if longer time series were available, the evolution of the clusters could be further investigated. (see Brida et al., 2010 and Brida et al., 2011). Limitation of the present research includes the relatively short time span of the data set and the use of aggregate data that may not be entirely represent tourism demand and supply.