determination of eConomiC indiCators in the Context of Corporate sustainability performanCe

. This article is focused on determination of the most significant economic indicators influencing corporate sustainabil­ ity performance. Corporate sustainability performance is a multidimensional concept based on the original idea of sustainable development, replacing the traditional understanding of corporate performance only as capital appreciation for owners (share­ holders). Compared to the original concept of sustainable development which consists of environmental, social and economic performance, the so­called triple­bottom­line, it is broaden to the responsibilities and the impact of Corporate Governance on the corporate performance. The basic set of economic indicators has been constructed from a synthesis of resources developed by international organizations (Global Reporting Initiative, International Federation of Accountants) and research among manufacturing companies in the Czech Republic. The basic set of twenty­five key indicators is divided into seven groups: Costs, Investments, Economic Results, Asset & financial resources utilization, Suppliers reliability, Penalties and R&D expenses. Basic set of indicators was presented to 23 top­managers who quantified the potential effect of each indicator to the success and sustain­ ability of their companies. Through the methods of descriptive statistics knowledge of the particularities of each indicator was obtained. Correlation analysis and factor analysis were applied in order to eliminate information duplicity and dimensionality reduction. The result is a reduction in the number of economic indicators, so that the loss of information on the influence of the original indicators on the corporate sustainability is minimized. Corporate sustainability indicators are a tool for measuring and managing progress towards sustainability goals and environmental, social and economic impacts.


Introduction
Corporate economic performance indicators are going to remain one of the main interests of owners and inves tors. However, together with the information about cor porate governance, the environmental and social factors, it creates a complex picture of any company and it has its significance for other key shareholders, it brings transpa rency and sustainability into business (Kocmanova et al. 2011). The goal of this article is to define the most significant

Theoretical approach to corporate sustainability performance
The need for alternative performance measurement systems which would take the corporate influence on interest groups into account has increased in the time of crisis. According to Kruse and Lundbergh (2010) one of the reasons why corporations should consider their environmental and so cial performance as well is the fact that investors generally invest less money into corporations that do not follow this trend because they consider the level of risk higher. Based on the definition of sustainability performance published in the Report of the World Commission on Environment and Development (1987) established by the UN, corporate sustainability performance (CSP) is defined as corporate strategy which uses the best business techniques to fulfill and balance the needs of both the current and the future stakeholders. This presents a complex task of providing competitive product in a shortterm period and at the same protecting, maintaining and developing human and natural resources necessary for the future. CSP therefore measures the extent to which a corporation implements economic, environmental, social and corporate governance factors into its activities and to what extent it considers the impact of its activities on its surroundings. (Artiach et al. 2010;Labuschagne et al. 2005). CSP involves the triplebottom line concept which suggests balance of three aspects -en vironmental, social and economic -to reach sustainability in organizations (Elkington 1998).
The connection of economic and sustainability per formance is a subject of many theoretical and empirical studies. Kirchhoff (2000), Feddersen and Gilligan (2001), Fisman et al. (2008) state that economic benefits are reached by companies with high level of CSP by using brands and advertisements informing about the sustainability of their products. This way they support product differentiation. Turban and Greening (1997) show that high level of CSP allows companies to hire more innovative and motivated employees which is again reflected in the economic results. On the other hand there is the neoclassical approach stating that companies have only one social responsibility and that is increasing their profit (Milton Friedman). According to this approach CSP decreases economic performance be cause activities increasing CSP are costly (Friedman 1962;Alexander, Buchholz 1978;Becchetti et al. 2005). Investing into CSP means higher costs of improving conditions for the employees, donations, costs of supporting the community, introducing ecological processes and also opportunity costs of giving up socially irresponsible investments. From this point of view, investing into CSP goes against the interests of investors because reallocation of investors' resources of the particular company onto other stakeholders takes place (Aupperle et al. 1985;McGuire et al. 1988;Barnett 2007). Ullmann (1985) states there is no direct connection between CSP and economic performance. According to Ullmann the reason is the existence of many variables which influence both types of performance and that is why this connection should not exist.
It is important for companies to set measurable and rele vant goals of sustainable development and suitable metrics. Moreover, integrated reporting of financial and nonfinan cial information is needed (Hrebicek et al. 2011a(Hrebicek et al. , 2011b(Hrebicek et al. , 2011cDocekalova 2012). CSP indicators should reflect the reality of the company and its critical success factors as well as company values and culture. This is why their development should not be limited only to borrowing alre ady existing methods and norms. However, internationally acclaimed norms can be a certain lead when creating com pany's own suitable indicators.  Singh et al. (2009Singh et al. ( , 2012).

Materials and methods
As the above mentioned suggests, CSP is a multidimen sional concept and the research methods have to take this into account. Defining the economic indicators was realized in the following subsequent steps: first step me ant creating a set of economic indicators by analyzing the approach of international organizations Global Reporting Initiative and International Federation of Accountants. Then, the relevancy of these indicators was verified by a questionnaire survey. The aim of the third step was reducing the number of indicators which was realized by removing duplicity information by correlation analysis and also by factor analysis so that the loss of original in dicators information was minimal. The fourth step meant assigning weights to the key indicators because different indicators have different levels of importance in different companies. The impact on the total performance of each company also varies and assigning the weights approxi mates reality. The weights were set by point method as statistical testing of expert methods has shown that it does not cause any statistically relevant differences in the weights value results. As the last step aggregation methods were applied to combine key indicators into one aggregate indicator measuring economic performance and the benchmark was set.

Indicators reduction methods
To reduce dimensionality the correlation and factor ana lyses were applied. The purpose of the correlation analysis is disclosing multicollinearity of the key indicators and re moving the redundant key indicators from the model. High values of pair correlation coefficients, i.e. |r| > 0.8 suggest multicollinearity. To detect multicollinearity the variance inflation factor was also used (Variance Inflation Factor, VIF), which is easily detected from an inversion matrix of the correlation matrix. VIF are diagonal elements of such an inversion matrix (Clark 2004). The indicator with higher VIF value was removed from the model. Factor analysis is based on a simple idea to describe the behavior of a set of variables by using a smaller number of new variables -fac tors -and via theses come to conclusions about the mutual dependence of the original variables.
The factor model analysis is as follows: (1) where: is the original set of variables (variables are standardized, i.e. zero mean value and unit distribution), α i1, α i2, …, α im are factor loadings 1 , F 1, F 2, …, F m is m noncorrelated standardized factors, e i is specific (unique, error, residual) part of the variable x i . (OECD 2008;Skaloudova 2010).
There are various methods of carrying out the factor analysis, e.g. principal component analysis, principalaxis method, alpha method, image factoring etc. Before the com putation it is useful to decide whether the factor analysis is worth carrying out, i.e. the correlations of the variables are possible to explain by factors. Kaiser -Meyer -Olkin statistics (KMO statistics) is used for this as well as Barlett's test of sphericity. KMO is based on comparing the values of pair and partial correlation coefficients and reaches the values between 0 and 1. KMO statistics values are possible to interpret as follows: 0.90-1.00 using factor analysis is excellent, 0.80-0.89 very good, 0.70-0.79 medium level of usefulness, 0.60-0.69 average, 0.50-0.59 poor and 0.00-0.49 not acceptable. Bartlett's test of sphericity tests zero hypothesis that the correlation matrix of the variables is unitbased, i.e. correlation coefficients of the variables equ al zero and therefore the condition of mutual dependence of variables is not met which prevents applying the factor analysis (Skaloudova 2010).
An important decision has to be made when applying factor analysis and that is the number of factors. This step significantly influences the solution and interpretation of factor analysis results. To set the number of factors the so called Kaiser criterion is used. According to this rule only the factors that have eigenvalues greater than one are retained. The number of factors can also be defined from graphic pre sentation of eigenvalues of individual factors by using a scree plot. The borderline marking the suitable number of factors lies where the numerical drop between two factors is the most significant. The number of factors can also be set heuristically. Hendl (2009) states that factor analysis has three aims: 1. to analyze correlations of a number of variables by combining the variables so that the variables in one cluster strongly correlate and at the same time the variables in different clusters do not correlate; this means the set is characteristic for the particular fac tor variable, 2. to interpret the factors according to what variables are included in the particular cluster, 3. to summarize the variability of the variables by only a few factors. Inner reliability of suggested indicators was then checked by applying Cronbach's alpha. Inner reliability means that in dicators measuring the same phenomenon should have positi ve mutual correlations. For this purpose the Cronbach's alpha or reliability coefficient or consistency coefficient is applied. Cronbach's alpha reaches the values of 0 to 1. Cronbach's alpha is computed by the following formula: where: C is the average interitem covariance among the va riables, V is the average variance of all the variables. For standardized Cronbach's alpha the formula is: where: R is the average of all the correlation coefficients of the variables.

Sample definition
The research is focused on companies of the group 27.1 Manufacture of electric motors, generators, transformers and electricity distribution and control apparatus accor ding to CZNACE and also on companies with more than 250 employees. This group of economic activities is divided into two subgroups 27.11 Manufacture of electric motors, generators and transformers a 27.12 Manufacture of electri city distribution and control apparatus. The basic set con sists of 32 companies. 65.2% of the addressed companies are majorityowned by international subjects. Before the questionnaire survey itself it was known from the basic set that majority of the companies had the form of Ltd. 82.6% of questionnaires were filled in by Ltd companies and 17.4% by stock companies, see Table 1. There was one cooperative in the basic set as well, but this company did not take part in the survey. One of the definition criterion of the basic set was number of employees. The survey is only focused on companies with more than 250 employees. The variable of number of employees has been recoded into five intervals and Table 1 shows the value frequency of each interval. Majority of companies fall between 250-750 employees. Two companies of the interval above 2251 employees stated these numbers of employees: 3200 and 7500.

Defining key indicators of corporate economic performance
The basic set of economic performance indicators was de fined on the basis of synthesis of knowledge gained in the preresearch stage (results published in Kocmanova and Docekalova (2012)) and approaches of Global Reporting Initiative and International Federation of Accountants. Resources consumption is described by indicator EN1 -Costs. Indicator EN2 -Investments is focused on the investmenteffectiveness. Indicator EN3 -Economic results measures how successfully the resources were transformed and valorized. Effectiveness of using pro perty and financial resources is described by EN4. Cooperation with suppliers is an important factor for all companies of manufacturing industry and that is why indicator EN5 -Supplier reliability has been included in the basic set of indicators. Indicator EN6 -Penalties describes the financial impact of irresponsible behavior of a company. Also indicator EN7 -R&D expenses has been included in the economic indicators. See Appendix Table A1 for the list of indicators. These indicators were then presented through questionnaire survey to topma nagers of corporations described in Table 1.

Number of indicators reduction
The questionnaire survey was based on evaluating the signi ficance of the indicators, i.e. how much the factors of these indicators support the corporate sustainability performan ce. Rating method has been used. Experts expressed their opinions on the basis of a predetermined scale <0; 10>.
The first step of data processing was a quality check car ried out in order to find out whether there were erroneous, missing or distant values in the data. The following statistic measures were computed in order to get the basic know ledge of key indicators: -Measures of central tendency (arithmetic mean and median), -Measures of variability (range R, standard deviation s and variation coefficient Vx), -Measures of shape (skewness skew and kurtosis kurt). These characteristics of key economic indicators are in Appendix Table A2. Evaluating the significance of Economic result shows the lowest variability (Vx = 83.9% and s = 0.8) of all the indicators which were presented in the survey. The respondents agreed on the key significance of this indicator and its impact on success and performance of a corporation ( x = 9.4 and x  = 10.0). The highest variation coefficient of the economic performance indicators is detected in ROCE and this indicator is evaluated as the least significant ( x = 2.0 and x  = 1.0). ROI has symmetrical distribution. Investments are the most asymmetrical indicator (skew = -1.6), this indi cator also has the highest coefficient of kurtosis (kurt = 3.4).
The basic set of key economic indicators included twen ty five indicators. Key indicators that correlate the most (r > 0.79) are listed in Table 2. Besides that Economic result and Revenue correlate highly positively (r = 0.789) as well as Economic result with Total costs (r = 0.774). After evalua ting the values of pair correlation coefficients and VIF values it was decided that Total costs, Revenue, ROE, Receivables turnover ratio and Stock turnover would not enter into furt her stages of constructing the aggregated indicator. The next step of reducing the number of indicators was factor analysis. To extract factors the principal com ponent analysis was selected. This method organizes the noncorrelated factors (components) according to their variance so that the first factor has the highest variance and the last one the lowest one. The principal component analysis exists individually as well; factor analysis can be considered as its extension.
Number After removing these economic indicators KMO statistics increased from 0.502 to 0.699 and at this value it is meaningful to apply factor analysis. This is confirmed by Bartlett's test, because based on its result the zero hypothesis that variables do not depend on each other can be rejected, see Table 3. Table 4 which shows the communalities makes it obvious that factors explain the variability of Personal costs the best (87.0%) and only by 49.6% in case of Turnover. Ten extracted components explain the total variance of the original variables and the first three with eigenvalue higher than 1 explain 71.77% of total variance, see Table 5.
Scree plot shows all the extracted components with their eigenvalues. The graph clearly shows the turn between the first and the second component. It would therefore be suffici ent to consider only one component. According to the Kaiser criterion three components should be selected -this appro ach is also more convenient considering the total explained variance, see Appendix Figure A1.
In Table 6 showing the factor matrix the factor weights lower than 0.3 are neglected. The key indicators selection is based on their component correlation and only the ones with the factor weight higher than 0.7 which can be considered sufficiently high, are included in the aggregate indicator.
Two key indicators Cash Flow and ROA (Return on assets) are included in the aggregate indicator and are combined into one component. Reliability of this solution was verified by Cronbach's alpha (α), which approximates the recommended limit of 0.7, see Table 7.

Key indicators aggregation
Weights (v i ) are computed by point method according to this formula: where: b i is an average number of points assigned by the respondents to i KPI. For each aggregate indicator it is necessary to compute normalized weights so that: The computed weights suggest that Cash Flow (v = 70.8%) is much more significant than ROA (v = 29.2%).
There are three methods of aggregating indicators. Additive method of aggregation is a linear method based on the summary of weighted and normalized subindica tors. An important condition when using the linear additive aggregation is preference independence of individual sub indicators. (OECD 2008) The problem of indicators com pensation can be solved by the multicriteria aggregation method which does not allow subindicators compensation at all (Munda, Nardo 2005, 2009. A compromise between fully compensational and noncompensational approach to the aggregate indicator construction is offered by the geometry aggregation method which defines the aggregate indicator as a product of individual subindicators raised to a higher power by the particular weight value. To create the aggregate indicator measuring economic performance the geometrical aggregation method was selected. The key indicators have to be transformed to the same units -%. Cash flow is related to added value.
Aggregate indicator measuring economic performance is described as follows: where: x 1 -Cash flow/added value; x 2 -EBIT/assets.

Conclusions
The aim of this article was to define economic indicators which influence corporate sustainability performance the most. To meet this aim an expert evaluation with a subsequent statistic evaluation was used. By applying statistic methods (correlation and factor analyses) it was found out that the original set of 25 key indicators can be substituted by only two -Cash flow a ROA. These indicators are therefore the top key ones for corporate sustainability from the point of view of the topmana gement. It is important to emphasize that it cannot be said about any of the indicator sets that it is optimal as the development and application of indicators should be a dynamic process which supports decision making and company management more than the goal itself. Searching for the balanced set of indicators is a complex process. Provided it is allowed for the indicator set deve lopment process to take long time, it can reduce its dy namics and reliability. Once a small, good and balanced set of simple indicators is created, the real effort should be made creating the evaluation process, providing the indicators are a base of a constructive dialogue among the organizational levels and mainly coming up with a way to improve the values of these indicators.