COMPARATIVE ANALYSIS ON THE EFFICIENCY OF THE FIRM AND THEIRS DETERMINANTS OF EUROPEAN TRANSITION ECONOMICS

In this study we evaluate the efficiency of the firms with DEA technique and bootstrapping procedure in eleven transition economics among sixth industrial and service sectors in a cross-section 2005. At the second stage, we study the relationship between corrected efficiency scores and possible explanatory variables. On one hand, our result shows statistically significant differences in efficiency levels across countries and sectors. Likewise, significant levels of inefficiency are identified especially in service sectors (Hotels and Restaurants, Retail and Wholesale and Transport). On the others hand, the results of the analysis in second stage confirmed the existence of a significant relationship between the country that firms operate (country-effect) and firm ́s age (learning by doing). The latter relationship between age and efficiency has different behaviors according to the sector concerned, in the form of U shape in the Hotels and Restaurants, inverted U shape in Metal and Machinery and Wood and Furniture and linear in the Retail and Wholesale sector. Contrary perhaps to expectations, the results also allow us to conclude that there is no important statistically significant relationship between the efficiency and variables related to the strategies implemented by firms and the control of the property.


Introduction
In this paper we evaluate efficiency of the firms in transition economics. We take eleven countries as a subject of investigation. This choice is not only motivated by our comparative advantage in knowing and availability of database this countries better than any other, but also because it is one of the interesting and difficult changing countries. In Recently, Tvaronaviciene et al. (2009) analyze if institutional performance matters in Lithuania, Latvia and Estonia after the European Union accession. Authors take into account that different approaches to countries' development assessment might affect their comparison results. These authors explore how much Lithuanian, Latvian and Estonian ranking according to estimated development level differs due to variations of approaches applied, and how sensitive calculations are to institutional performance and current economic downturn.
Taking into account this empirical background, in this study, we use a two-stage approach. At the first stage we estimated sensitivity efficiency scores using bootstrapping Data Envelopment Analysis (DEA). We analyze distributions of both original and corrected efficiency scores (via kernel density estimation). At the second, relationship between corrected efficiency scores and possible explanatory variables. We found statistically significant differences in efficiency levels across countries and sectors. Likewise, significant levels of inefficiency are identified especially in service sectors (Hotels and Restaurants, Retail and Wholesale and Transport). Regarding the factors that explain the efficiency country-effect where the firm operates and the firm´s age have been significant. While the latter in firms belonging to four of the sixth sectors and with different behaviors in the form of curvilinear and linear shape.
The rest of the paper is organized as follows: section 1 shows characteristic of the countries, variable and data; section 2 presents the methodology; section 3 provides the main results and the last section offers the conclusions. S130 vakia, and Slovenia. Tables 1 and 2 show the main features of the analyzed countries they belong to the firms and the numbers of firms analyzed by both countries and sectors respectively.

S131
As complement figure 1 represent both countries by GDP per capita according to intensity of color (percentile #4, red color as high percentage of total value to percentile #1 light yellow as low percentage and representative percentage of firms by sector and country) As can be shown in Figure 1, percentile #4 is integrated by Czech and Slovenia countries. Percentile #3, Hungary, Slovakia and Estonia. Percentile #2, Latvia, Lithuania and Poland. Finally in the percentile #1, Romania, Moldova and Bulgaria. In relation to the representation of firms by both sectors and countries in the graphs on the right side of Figure 1

S132
To generate internationally comparable data, the questions in the Core questionnaire are asked in all countries and for all industries where the survey is implemented. In addition to this Core instrument, the Manufacturing Module and Services Module questions are asked to establishments in the manufacturing and services sectors, respectively.
The sampling methodology of the World Bank's Enterprise Survey generates sample sizes appropriate to achieving two main objectives: 1. A sample representative of the whole economy that substantiates assertions about the whole economy, not only about the manufacturing sector. The overall sample should include, in addition to selected manufacturing industries, services industries and other relevant sectors of the economy. 2. A large enough sample size for selected industries to conduct statistically robust analyses with levels of precision at a minimum 7.5% for 90% confidence intervals about 1 : (a) estimates of population proportions (percentages), at the industry level, and (b) estimates of the mean of log of sales at the industry level.
To analyze the efficiency levels the choice of output and input types follows Donthu and Yoo´s (1998) and Coelli et al. (2002) recommendations. The inputs are measured by three variables. The measure for labour is personnel cost (we can not use the number of employees variable as labour because there were a lot of missing values), being material and energy costs respectively the others inputs. Finally, output is measured by sales. All those variables are measured in US dollars according the PPP that OECD publishes. Let's note that using the number of employees (instead of the payroll) as a measure of size as other papers do, the sample sizes would have been reduced more than 40% due to missing values in the survey.
Concerning the explanatory reasons of the efficiency levels for second stage, although the requested information provides a vast array of possibilities, the actual information available imposes tight restrictions. After a careful analysis, we have chosen seven variables largely used in the literature: one of them make reference to the ownership structure (percentage of shares that the local owners hold), age of firm and five linked to the competitive characteristics (international quality certification, new product introduction, make joint venture, make licensing agreement and make outsourcing) with a binary character. Table A1 in appendix summarizes descriptive statistics of the selected variables.

Methodology
The research methodology consists of the following stages. First, a nonparametric methodology is used to measure the technical efficiency with DEA (see Charnes et al. 1978) to compare firms by sectors. Second, OLS regression will be used to determine the explanatory factors of corrected efficiency. The kernel distribution will provide a more specific analysis to know the shape of the industry efficiency distribution.

S133
To compute a production frontier and the technical efficiency indexes that it implies, can be done by econometric techniques or by mathematical programming. Each of these two approaches as its own advantages and disadvantages Authors as Lovell 1993 argue that "neither approach strictly dominates the other" meanwhile others such as Coelli, Rao and Battese 2002 suggest that "selection of the appropriate method should be made on a pragmatic basis". The last option has been chosen to carry on this analysis. Specifically, we use non-parametric DEA (Data Envelopment Analysis).
Some of the arguments in favour of this methodology lie in that it allows getting maximum information on the available data. In comparison with parametric approaches, DEA presents some virtues: It allows including multiple inputs and outputs with different unit measures, it explores the inefficiency origins quantifying over-use of inputs, and it does not require any initial hypothesis on the functional form of the production frontier 2 .
The current work uses an input-oriented model, since market in transition economics is better suited has a better intuition about the potential savings in inputs compared to the waste of resources, and because the different formats are subject to demand conditions, which leads them to adjust their inputs freely. According to Färe et al. (1994) we propose the following input-oriented DEA model to measure the technical efficiency: where: l = efficiency measure to be calculated for each DMU j, ujm = quantity of output m produced by DMU j, xjn = quantity of input n used by DMU j, and zj = intensity variable for DMU j.
Since the variable l is calculated for each DMU, the preceding formulation is estimated once for each DMU in the data set. Equations 1, 2 and 3 define a set of constraints for each output and input. A value of l = 1 means that a firm is considered efficient, while a value l < 1 means a firm is inefficient. For each DMU, we solve the linear program problem exposed.
The DEA method of estimating efficiency has received some opposition (see Henderson and Zelenyuk 2004). Traditional DEA assumes that no measurement error exists S134 and estimated production frontier is piecewise linear. Koop et al. (2000) mention that sensitivity of DEA outliers is no doubt one of the weaknesses of the DEA approach. As mentioned by (Daraio and Simar 2006), the DEA efficiency score is a performance indicator obtained by comparing each firm with the best performers of its objective group; i.e. the benchmarking is, in a certain sense, severe for each firm and sensitive to extremely performing firms to answer these comments, Wilson 1998, 2000) and others have introduced bootstrapping into DEA framework. Their methods, based on statically well-defined models, allow for consistent estimation of the production frontier, corresponding efficiency scores as well as standard errors and confidence intervals. The idea of the bootstrap is to simulate repeatedly the data generating process using the resampling method and applying the DEA to each simulated sample so that the resulting DEA efficiency indices mimic the sampling distribution of the original DEA efficiency indices. In this paper, we employ the bootstrap procedure suggested by Simar and Wilson 1998.

S137
The analysis of the average corrected efficiency of firms in each country in comparison with the activity, for sixth sector shows that; Slovenia and Latvia firms are above global frontier en table 5 sectors. Poland, Bulgaria and Slovakia in Table 4, Romania, Lithuania and Czech in Table 3, Hungary in Table 2 and finally Estonia and Moldova in Table 1. Figure 3 shows the levels of efficiency for countries above the global frontier in blue color or below in yellow. In the case of Hotels and Restaurants, Retail and Restaurants, Construction and Wood and Furniture over 50% of countries analyzed is above the global frontier. However, in the sectors Transport and Metals and Machinery less than 50% remain in that position.

Determinants of corrected efficiency
Having analyzed the dynamic evolution of corrected efficiency, the next step is to study what their determining factors are. The dependent variable , θ i t is the corrected efficiency. In this sense we avoid one of the problems in the estimation of the second stage, since our dependent variable it is not truncated. As can be seen in Table 4, there were significant statistical differences in relation to the country where firms do business (country-effect). Also the firm's age in four sectors (Hotels and Restaurants, Metals and machinery, Wood and Furniture sectors appears as a determinant factor of efficiency, but with different behavior as will be see later. However, the set of dummy variables that characterize the strategic behavior of firms in terms of organizational practices have little or no relevance. For firms belonging to Construction sector new product introduction is positively relationship with efficiency level. In the case of firms belong-

Fig. 3. Levels of efficiency firms by both sectors and country according global frontier
Journal of Business Economics and Management, 2013, 14(Supplement 1): S128-S144 S138 Coef.

S139
ing to sector Hotels and Restaurants perform joint venture is related to higher levels of efficiency. Make license agreements is also positively relationship with efficiency in Retail and Wholesale sector. These two strategies may lead to transfer of technology and managerial expertise.
Despite this low significance of the variables related to the strategies implemented by firms could be related to several factors, the low percentage of companies which occasionally puts out these actions and the presence of country effects and age as dominant. The percentage of shares in local power is not significant in any sector. Therefore greater control of the property is not associated with higher levels of efficiency.
Deepening on the latter effect, two opposite results may be considered. This is, the absence of statistical significant that age has on efficiency in the sectors of Construction and Transportation 6 . A possible explanation for this might relate to the older firms may have more experience in the market (learning by doing), but may be less flexible and dynamic than younger, which could affect the firm's ability to adjust to the market. However, the introduction of new products has a positive effect on efficiency in the Construction sector, while firms that have licensing agreements are more efficient in the Transport sector. In relation to the sectors where the firm´s age is associated with efficiency levels, Figure 4 shows the different trajectories.
In the case of sector Hotels & Restaurants relationship is U-shaped where the average age is in the negative slope (red vertical line) this relationship between efficiency and age may indicate the need for a significant level of expertise in market knowledge to become more efficient. In other words, established firms maintain a significant market power. In the other three sectors the relationship between efficiency and age is positive, inverted U-shaped for sectors Metals & Machinery and Wood and Furniture where the average is in the positive slope and linear trajectory for the sector Retail and Wholesale.
In this sense, generally speaking, greater seniority affords the firm greater know-how, which can lead to a greater capacity for developing its activities in a more efficient way. Firm experience, might be related to technical efficiency since firm production might involve "learning by doing".

Conclusions
In this study we evaluate de efficiency firms with DEA technique and bootstrapping procedure in eleven transition economics among sixth industrial and service sectors in a cross-section 2005. At the second stage, we study the relationship between corrected efficiency scores and possible explanatory variables. Benchmark analysis in first stage made it possible to detect resource management carried out by managers in a sector with turbulent environment. We found statistically significant differences in efficiency S140 levels across countries and sectors in benchmark analysis. Likewise, significant levels of inefficiency are identified especially in service sectors (Hotels and Restaurants, Retail and Wholesale and Transport).
In relation to the determinants of efficiency, and according to Caves and Barton (1990) and Caves (1992), several studies have developed a strategy to identify the determinants of efficiency 1) Factors outside the company, such as the degree of competitiveness in the market where the company operates. 2) Characteristics of the company as the company size, type of organization, greater or lesser intensity of the investments made and the benefits of localization. 3) Deviations business dynamics of the situation of long-run equilibrium. These deviations may be the consequences of changes in demand facing the firm, or the consequence of certain production strategies such as the degree of technological innovation. 4) Public or private property. The degree of intervention in the management of the business manager can affect the degree of efficiency in the use of productive factor. The results of the analysis in second stage confirmed the existence

S141
of a significant relationship between the country that firms operate (country-effect) and firm´s age. Contrary perhaps to expectations, the results also allow us to conclude that there is no important statistically significant relationship between the efficiency and variables related to the strategies implemented by firms and the control of the property, perhaps due to the dominance influence of the country-effect and the experience (learning by doing) The results of this paper should be considered with some caution as a consequence of the cross-section analysis. Given the consequences that the efficiency has for a firm a main recommendation will be the need to be very conscious of the huge possibilities that do exist for a large part of the firms in transition economies, and the significant amount of resources that could be saved 7 . A second recommendation could be to continue to investigate to the low significance of strategy firms. In this line, panel data is perhaps the more appropriate way to obtain stylized facts but perhaps the difficulty of preparing such information is an important limitation. Leopoldo LABORDA CASTILLO. He works as STC at the World Bank and as Associated Professor at the University of Alcala (Spain). He received his PhD from University of Alcala. His research focuses on productivity and efficiency analysis in public and private sector.