Proceedings of the 2017 International Conference “ECONOMIC SCIENCE FOR RURAL DEVELOPMENT” No 46 Jelgava, LLU ESAF, 27-28 April 2017, pp. 359-366 CLASSIFICATION OF SELECTED INDUSTRY COMPANIES LISTED ON THE STOCK EXCHANGE - THE IMPACT OF NORMALIZATION PROCEDURES Monika Zielinska-Sitkiewicz 1, Dr. oec. 1Warsaw University of Life Sciences Abstract. Competent assessment of the financial condition of stock companies becomes very important because trustworthy and timely information is expected by investors. Thus, it is necessary to use methods that streamline the analysis of the stock market. Taxonomic synthetic measures belong to such methods. One of the stages of the multivariate analysis is the data normalization. The choice of the normalization method is one of the most crucial steps for the researchers because it could influence the results of the analysis. In the present study, the author uses a synthetic taxonomic measure TMAI with selected normalization methods to compare the condition of 13 companies listed on the Warsaw Stock Exchange and create for them the rankings for the years 2013 and 2014. Comparing the results of the individual rankings, it can be stated that the selected normalization algorithm impacts the result of the obtained classification.

Key words: the data normalization, taxonomic measure, TMAI, food industry companies. JEL code: C10, G11 Introduction financial data of the selected companies of According to the Polish Information and this sector listed on the Warsaw Stock Foreign Investment Agency, the businesses of Exchange. the food sector generate over 13 % of the One of the groups of methods of the Polish GDP value and are one of the Multivariate Comparative Analysis is methods dominating industries in . Years after of linear ordering of objects. The first one who the Polish accession to the EU were a period proposed the synthetic measure of of increasing development of this branch of development for the comparison of the level economy with massive investments in of economic development of the selected modernization and expansion of food countries was Z. Hellwig (1968). The Hellwig production plants. What’s more, the development measure synthesises information development of commercial networks was also within the diagnostic variables and assigns the growth driver. Currently, the retail trade one aggregated measure to the analysed reached some growth barriers, and the phenomenon. W. Tarczynski and M. Luniewska condition of food companies largely depends (2006) have applied the analogical concept of on the fluctuations of the raw material prices the construction of the synthetic measure and the political situation on our eastern while building the TMAI measure for the border, which has a negative impact on the capital market, which determines the meat, dairy and fruit and vegetable industries. investment attractiveness of companies. Moreover, in 2014 a negative trend of limiting For all variables used in the algorithm of the market appeared, which resulted from the the synthetic index construction to be imposition of the embargo on food products mutually comparable in orders of magnitude from the EU by the Russian Federation. and devoid of denominations, their Taking into account: the Russian embargo, normalization is performed. However, the the increase in prices of agricultural raw change of the normalization formula of materials and the food price increases in diagnostic features may cause changes in the recent years and the changes in consumer layout of objects in the ranking, which cause preferences towards healthy , it is worth will involve neither the increase, nor the examining how the situation is shaped on the decrease of their evaluation. market of food production, analysing the

1 Corresponding author. Tel.: + 48 22 5937242; fax: + 48 22 59 372 22. E-mail address: [email protected]. 359 Proceedings of the 2017 International Conference “ECONOMIC SCIENCE FOR RURAL DEVELOPMENT” No 46 Jelgava, LLU ESAF, 27-28 April 2017, pp. 359-366 The main object of the presented article is The Table 1 shows description of eight to examine the sensitivity of the result of indicators recommended in the works by linear ordering on the applied procedure of Tarczynski and Luniewska (2004, 2006). variable normalization in the construction of In addition, the analysis used the the rankings of 13 food industry companies Operating Profit Margin Ratio ( OPMR ), listed on the Warsaw Stock Exchange. The calculated as ( Operating income/ Total study used the Taxonomic Measure of revenue)·100 . The increasing value of this Investment Attractiveness (TMAI), which indicator means the improvement of the allows using the comprehensive analysis of operational effectiveness of the unit’s companies based on the most important functioning, which in turn may prove the financial indicators, presenting it in the form existence of its significant development of a synthetic ranking. By calculating the potential. OPMR is, therefore, a stimulant. distance of each object from the model, it was Table 1 checked how the application of two different The selection of variables and their normalization methods of the same diagnostic impact on the general criterion variables affects the result of the The impact on the classification. The obtained results were Ratio Formula general compared to the model ranking proposed by criterion the expert using the correlation coefficient of Net Income/ Return on Shareholder stimulant Equity (ROE) the Spearman ranks. The study was Equity conducted for the years 2013 and 2014. Net Income/ Return on Average Total stimulant For the study, the following companies Assets (ROA) Assets were selected: Colian Holding SA, Duda SA, Inventory Net Sales/ Graal SA, Indykpol SA, Kruszwica SA, Turnover Average stimulant Ratio (ITR) Inventory Makarony Polskie SA, Mieszko SA, Mispol SA, (Average Liabilities Otmuchów SA, Pamapol SA, Pepees SA, Seko Liabilities/ Net stimulant Ratio (LR) SA, Wawel SA and Wilbo SA, which are listed Income)·365 Net Sales on the main market of the Warsaw Stock Asset Revenue/ Turnover Stimulant Average Total Exchange. The activities of the analysed Ratio (ATR) Assets companies are conducted mainly in Poland, Receivable Sales Revenue/ nominant (7 – and their profit and loss account is prepared in Turnover Average 10) Ratio (RTR) Receivables a spread-sheet. The companies representing Current Ratio Current Assets/ nominant (1.0 the alcohol industry were omitted. (CR) Current Liabilities – 1.2)

Synthetic TMAI measure – description Debt Ratio Total Liabilities/ nominant and results of the study (DR) Total Assets (57 %-67 %) Source: Based on papers by Tarczynski and Nine of the most important financial Luniewska, 2004, 2006 indicators were used for the calculation of the Building the expert ranking of companies of synthetic taxonomic measure TMAI in the the food sector, it has been stated that the research of food industry companies. They highest impact on the fundamental characterise the most important aspects of the assessment of the company’s condition comes company activity: profitability ( ROE, ROA, from the profitability indicators and then the OPMR ), liquidity ( CR ), efficiency ( ITR, LR, indicators relating to the working capital. It ATR, RTR ) and debt ( DR ). was posited that the relatively low information potential is in the indicators connecting the

1 Corresponding author. Tel.: + 48 22 5937242; fax: + 48 22 59 372 22. E-mail address: [email protected]. 360 Proceedings of the 2017 International Conference “ECONOMIC SCIENCE FOR RURAL DEVELOPMENT” No 46 Jelgava, LLU ESAF, 27-28 April 2017, pp. 359-366 long-term processes (e.g. income) in relation account both the scale of the variable to the balance sheet positions, which can measurement, and the characteristics of the result from one-time events, e.g. created on variable distribution, such as the arithmetic 31 December. In addition, some companies mean, standard deviation and the gap knowingly adopt the financial policy, which designated for the normalised values of makes the indicators go beyond some variables (Walesiak, 2004). In addition, it is contractual framework, which is not the worth looking at the results of the basic evidence of bad management. This case descriptive statistics and check the occurrence concerns, among others, the company Wawel of the outliers or extreme values. According to SA, which to a limited degree used foreign K. Kukula, L. Luty (2015), taking into account capitals to finance its activities and its Debt the final goal of the research, which is the rate (SZ) was the lowest in the entire group of ranking of objects, one should not use the the analysed companies. normalization methods that level the outliers Consequently, when creating the of the diagnostic features, because such classification of companies the ability of methods distort the real image of the spatial companies for the effective management of distribution of the studied complex the assets and to cover the current liabilities phenomenon impacting the sequence of the of the current assets, effectiveness of own objects under consideration. A completely equity and return on operating income were different view of the need for eliminating the taken into account. Tables 4 and 5 include the outliers, as the ones interfering with the layout of companies in the expert ranking for ranking, can be found in the article by Bak, the years 2013 and 2014. Szczecinska (2014). In the first stage of the study, the The Table 2 contains the results of the distributions of financial indicators were basic descriptive statistics calculated for all analysed, because when choosing the financial ratios for 2013. standardization formula one should take into Table 2 Descriptive statistics of variables in 2013

interquartile Standard coefficient Variable mean median min max Q1 Q3 range skewness kurtosis range deviation of variation

ROE 0.026 0.047 -0.269 0.235 0.009 0.082 0.504 0.073 0.137 529.203 -1.133 1.508

ROA 0.029 0.025 -0.061 0.171 0.004 0.032 0.232 0.028 0.057 200.645 1.095 2.692

OPMR 0.205 0.182 0.101 0.381 0.163 0.219 0.280 0.056 0.082 40.186 0.961 0.446

RTR 6.636 5.773 3.373 13.502 4.130 8.727 10.129 4.597 3.272 49.299 0.995 0.008

ITR 10.860 11.504 2.806 29.106 5.211 12.424 26.299 7.213 6.794 62.557 1.563 3.823

LR 58.565 58.410 20.546 111.181 31.179 75.381 90.634 44.202 28.822 49.214 0.459 -0.394

ATR 1.382 1.251 0.761 2.628 0.998 1.474 1.866 0.476 0.614 44.401 1.127 -0.009

CR 1.335 1.058 0.860 2.575 0.964 1.601 1.715 0.637 0.512 38.368 1.400 1.570

DR 0.514 0.506 0.261 0.855 0.392 0.628 0.594 0.236 0.170 32.953 0.386 -0.145 Source: author’s calculations In 2013, extreme values were observed for and Asset Turnover Ratio (ATR), which is the indicators Return on Equity (ROE ), Return confirmed by the box-plot graphs no. 1 to 5. on Assets (ROA ), and the occurrence of the The remaining ratios assumed values that did outliers was stated for Operating Profit Margin not exceed the non-outlier range. The results Ratio (OPMR) , Inventory Turnover Ratio (ITR) of the basic descriptive statistics calculated for

1 Corresponding author. Tel.: + 48 22 5937242; fax: + 48 22 59 372 22. E-mail address: [email protected]. 361 Proceedings of the 2017 International Conference “ECONOMIC SCIENCE FOR RURAL DEVELOPMENT” No 46 Jelgava, LLU ESAF, 27-28 April 2017, pp. 359-366 applied financial indicators for 2014 are presented in Table 3.

Box-plot ROE 2013 Box-plot ROA 2013 Box-plot OPMR 2013 0,3 0,20 0,40

0,18

0,16 0,35 0,2 0,14

0,12 0,30 0,1 0,10

0,08 0,25

0,0 0,06

0,04 0,20 Median = 0.047 0,02 Median = 0.025 Median = 0.182 -0,1 25%-75% = 25%-75% = 25%-75% 0,00 0,15 (0.0086, 0.0823) (0.0039, 0.0316) = (0.1635, 0.2188) Non-outliers range = -0,02 -0,2 Non-outliers range = Non-outliers range (0.0065, 0.1568) -0,04 (0.0031, 0.0526) 0,10 = (0.1006, 0.2866) Outliers -0,06 Outliers Outliers Extreme value -0,3 -0,08 Extreme values 0,05 Extreme values Source: author’s calculations Source: author’s calculations Source: author’s calculations Fig. 1. Box-plot ROE 2013 Fig. 2. Box-plot ROA 2013 Fig. 3. Box-plot OPMR 2013

Box-plot ITR 2013 Box-plot ATR 2013 32 2,8 30 2,6 28 26 2,4 24 2,2 22 20 2,0

18 1,8 16 14 1,6

12 Median = 11.504 1,4 Median = 1.251 10 25%-75% 25%-75% 1,2 8 = (5.2112, 12.4243) = (0.9976, 1.474) 6 Non-outliers range 1,0 Non-outliers range 4 = (2.8061, 14.9116) = (0.7614, 1.474) 0,8 2 Outliers Outliers 0 Extreme values 0,6 Extreme value Source: author’s calculations Source: author’s calculations Fig. 4. Box-plot ITR 2013 Fig. 5. Box-plot ATR 2013 Table 3 Descriptive statistics of variables in 2014

interquartile Standard coefficient Variable mean median min max Q1 Q3 range skewness kurtosis range deviation of variation

ROE 0.076 0.064 0.006 0.220 0.034 0.085 0.214 0.051 0.059 77.492 1.501 2.411

ROA 0.045 0.033 0.002 0.166 0.020 0.040 0.165 0.020 0.048 106.651 1.943 3.381

OPMR 0.207 0.196 0.110 0.409 0.154 0.223 0.299 0.069 0.086 41.483 1.345 1.782

RTR 7.126 5.509 3.134 17.421 4.556 8.360 14.288 3.804 4.115 57.743 1.723 2.791

ITR 11.131 10.966 2.144 27.592 5.737 14.135 25.447 8.398 6.703 60.220 1.187 2.494

LR 51.422 54.859 19.908 84.615 31.063 67.470 64.706 36.407 20.685 40.227 -0.216 -1.038

ATR 1.444 1.186 0.680 2.602 1.017 1.947 1.922 0.930 0.667 46.174 0.933 -0.615

CR 1.806 1.391 1.012 4.000 1.057 2.414 2.988 1.357 0.995 55.118 1.290 0.643

DR 0.474 0.469 0.166 0.854 0.350 0.580 0.688 0.230 0.197 41.637 0.241 0.034 Source: author’s calculations In 2014, the occurrence of extreme values Margin Ratio (OPMR), Inventory Turnover was stated only by the indicator Return on Ratio (ITR) and Receivable Turnover Ratio Assets (ROA ) (Fig. 6). Furthermore, the (RTR ), which is documented by the box-plot outliers were observed for the indicators: type of graphs from no. 7 to 10. Return on Equity (ROE ), Operating Profit

1 Corresponding author. Tel.: + 48 22 5937242; fax: + 48 22 59 372 22. E-mail address: [email protected]. 362 Proceedings of the 2017 International Conference “ECONOMIC SCIENCE FOR RURAL DEVELOPMENT” No 46 Jelgava, LLU ESAF, 27-28 April 2017, pp. 359-366

Box-plot ROE 2014 Box-plot ROA 2014 Box-plot OPMR 2014 0,24 0,18 0,45

0,22 0,16 0,40 0,20 0,14 0,18 0,35 0,16 0,12 0,30 0,14 0,10 0,12 0,08 0,25 0,10 0,06 0,08 Median = 0.064 Median = 0.033 0,20 Median = 0.196 0,06 25%-75% = 0,04 25%-75% 25%-75% 0,15 0,04 (0.0336, 0.0852) = (0.0202, 0.0401) = (0.1539, 0.2226) Non-outliers range= 0,02 Non-outliers range Non-outliers range 0,02 (0.0057, 0.1566) = (0.0016, 0.0435) 0,10 = (0.1099, 0.3242) 0,00 0,00 Outliers Outliers Outliers -0,02 Extreme values -0,02 Extreme values 0,05 Extreme values Source: author’s calculations Source: author’s calculations Source: author’s calculation Fig. 6. Box-plot ROE 2014 Fig. 7. Box-plot ROA 2014 Fig. 8. Box-plot OPMR 2014

Box-plot LR 2014 Box-plot ITR 2014 18 30 28 16 26 24 14 22 20 12 18 16 10 14 12 8 Median = 5.509 Median = 10.966 25%-75% 10 25%-75% 6 = (4.5557, 8.3596) 8 = (5.7366, 14.1347) Non-outliers range 6 Non-outliers range 4 = (3.1339, 12.2533) 4 = (2.1443, 14.7625) Outliers 2 Outliers 2 Extreme values 0 Extreme values Source: author’s calculations Fig. 9. Box-plot RTR 2014 Source: author’s calculations Fig. 10. Box-plot ITR 2014 In the second step, due to the specificity of Two normalization formulas were selected variables, Current Ratio (CR ), Debt Ratio (DR ) for the further analysis in order to compare and Receivable Turnover Ratio (RTR ) were the ranking results 1: individually transformed from nominants into • standardization stimulants. Description of the financial indicators and transformations of variables zij = (xij − x j /) s j (1) being nominants into stimulants used in the • Weber standardization study was included in the author’s work (Chrzanowska, Zielinska-Sitkiewicz, 2014). zij = (xij − Me j ) ,1 4826 MAD j (2) In the next stage, when choosing the normalization formulas, the specificity of the The classic standardization and Weber variables (financial ratios) was what guided standardization cause the unification of the us, and well as the following demands: values of all variables in terms of variation • bringing the order of variable magnitudes measured with the standard deviation or the to the state of comparability; absolute median deviation . This means the • possibility of normalization of the elimination of variation as the basis for characteristics adopting the positive and differentiation of objects (Walesiak, 2014). It negative values or only the negative ones; is recommended to use the Weber • possibility of normalization of the standardization in order to eliminate the

characteristics adopting the zero value 1 (Kukula, 2000). x j s, j - arithmetic mean, standard deviation for j variable, Me , MAD – median and absolute median deviation

1 Corresponding author. Tel.: + 48 22 5937242; fax: + 48 22 59 372 22. E-mail address: [email protected]. 363 Proceedings of the 2017 International Conference “ECONOMIC SCIENCE FOR RURAL DEVELOPMENT” No 46 Jelgava, LLU ESAF, 27-28 April 2017, pp. 359-366 distorting effect of the outlier or extreme Table 5 observations. Results of the TMAI measure and the ranking of the studied In the last step of the analysis, the companies for 2014 Taxonomic Measure of Investment Attractiveness was calculated for two studied periods. The calculations did not include any weighs for the applied financial indicators (Luniewska, Tarczynski, 2006). The ranking Normalization Normalization formulas: standardization Weber standardization results for 2013 and 2014 with the expert Name TMAI nr TMAI nr EXPERT ranking ranking EXPERT classification are presented in Tables 4 and 5: COLIAN HOLDING 0.1142 8 0.0502 9 7 DUDA 0.0942 9 0.1928 4 9 Table 4 Results of the TMAI measure and the GRAAL 0.1619 5 0.1844 6 3 ranking of the studied IDYKPOL 0.1146 7 0.1851 5 5 companies for 2013 KRUSZWICA 0.1337 6 0.2644 2 10

MAKARONPL 0.1895 3 0.1531 8 4

MIESZKO 1 - - - - -

OTMUCHÓW 0.1814 4 0.1665 7 8

PAMAPOL 0.0195 11 0.0000 12 11

Normalization Normalization formulas: standardization Weber standardization PEPEES 0.0675 10 0.0173 11 6 Name TMAI nr TMAI nr EXPERT ranking ranking EXPERT SEKO 0.2213 2 0.2259 3 1 COLIAN HOLDING 0.2488 6 0.3376 6 3 WAWEL 0.4389 1 0.4236 1 2 DUDA 0.2961 2 0.3897 3 9 WILBO 0.0000 12 0.0184 10 12 GRAAL 0.2520 5 0.3333 7 8 Source: author’s calculations IDYKPOL 0.2435 7 0.2899 10 11 Comparing the results of the individual KRUSZWICA 0.1991 10 0.3717 4 4 rankings, it can be stated that the selected MAKARONPL 0.2794 3 0.3322 8 6 normalization algorithm impacts the result of MIESZKO 0.2101 9 0.3395 5 5 the obtained classification. In order to perform OTMUCHÓW 0.2271 8 0.3082 9 10 a detailed analysis of changes in the position PAMAPOL 0.0000 13 0.0000 13 13 of the surveyed companies, depending on the PEPEES 0.1017 11 0.1846 11 7 adopted formula of transformation of the SEKO 0.3523 1 0.4118 2 2 variables, the Spearman rank correlation WAWEL 0.2688 4 0.5103 1 1 coefficients were calculated between the WILBO 0.0918 12 0.0223 12 12 Source: author’s calculations assessments of companies designated with the TMAI measure and the expert method. Tables 6 and 7 contain the obtained values.

1 At the request of the company, on December 31, 2014, the board of the Stock Exchange in Warsaw made the resolution to exclude the Mieszko shares from the trading. Therefore, there are no financial results of the company for 2014.

1 Corresponding author. Tel.: + 48 22 5937242; fax: + 48 22 59 372 22. E-mail address: [email protected]. 364 Proceedings of the 2017 International Conference “ECONOMIC SCIENCE FOR RURAL DEVELOPMENT” No 46 Jelgava, LLU ESAF, 27-28 April 2017, pp. 359-366 Table 6 classification using the classic standardization Values of the Spearman rank correlated with the expert system. Therefore, correlation coefficients between the one should be advised to consider the ranking results for 2013 adjustment of the normalization methodology of the financial ratios in the case of outliers, or Normalization formulas the extreme values, if the comparative studies

standardi zation Weber standardi zation expert ranking of the companies are conducted periodically.

standardization 1 0.817 0.687 Conclusions Weber standardization 1 0.878 1) The conducted study shows that the use of expert ranking 1 different normalization formulas of Source: author’s calculations variables can cause the change of the Table 7 results of the company classification, which Values of the Spearman rank correlation coefficients between results neither from the data structure the ranking results for 2014 change, nor the effectiveness modification of their operations. Conducting the analysis based on the financial indicators, one Normalization formulas should characterise the distributions of the standardi zation Weber standardi zation expert ranking studied variables calculating the descriptive standardization 1 0.842 0.896 statistics, which enables for the group of Weber standardization 1 0.731 studied companies to identify the outlier expert ranking 1 and extreme values. If they are stated, one Source: author’s calculations should check whether the change of the For 2013 and 2014, quite a high normalization procedure from the classic convergence of both rankings was obtained, standardization (used in literature quite during the construction of which the classic widely with the TMAI calculation) into the standardization and the Weber standardization Weber standardization does not lead to a was used. Moreover, in 2013 the classification more accurate classification. obtained by using the Weber standardization 2) Subjectivity in the construction of formula indicated the share of companies taxonomic synthetic metres affects the closer to the expert ordering, almost equally system of companies in the rankings. The recognizing the best and the least innovatively correct classification can be ensured by the attractive companies. It is worth mentioning integration of expert opinions, supported that in 2013, compared to 2014, much more by a detailed financial analysis in the outlier and extreme values were noted in the stages of selecting the group of financial distributions of financial ratios. In 2014, the indicators. The selection of the compliance of TMAI rankings with the expert normalization method should be preceded ordering was reversed. There were less by the examination of the distributions of companies with the values strongly deviating variables and it is possible to consider the from the others in the group of the studied potential use of weights. companies than in 2013. The obtained

1 Corresponding author. Tel.: + 48 22 5937242; fax: + 48 22 59 372 22. E-mail address: [email protected]. 365 Proceedings of the 2017 International Conference “ECONOMIC SCIENCE FOR RURAL DEVELOPMENT” No 46 Jelgava, LLU ESAF, 27-28 April 2017, pp. 359-366 Bibliography 1. Bak., I., Szczecinska, B. (2014). Analysis of Tourist Attractiveness of Polish Voivodship Capitals in Poland, Wiadomosci Statystyczne, (12), pp. 80–95. 2. Chrzanowska, M., Zielinska-Sitkiewicz, M. (2014). An Application of Selected Synthetic Measures for Classification of Major Development Companies Listed on the Warsaw Stock Exchange, Folia Oeconomica nr 1(298), Lodz, pp. 99-113 3. Hellwig, Z. (1968). Zastosowanie metody taksonomicznej do typologicznego podzialu krajow ze wzgledu na poziom rozwoju oraz zasoby i strukture wykwalifikowanych kadr. Przeglad Statystyczny (4), pp. 307- 327. 4. Kukula, K. (2000). Metoda unitaryzacji zerowanej, PWN, Warszawa. 5. Kukula, K., Luty, L. (2015). The Proposal for the Procedure Supporting Selection of a Linear Ordering Method, Przeglad Statystyczny (2) pp. 219 - 231. 6. Tarczynski, W., Luniewska ,M. (2004). Dywersyfikacja ryzyka na polskim rynku kapitalowym, Placet, Warszawa. 7. Tarczynski, W., Luniewska, M. (2006). Metody wielowymiarowej analizy porownawczej na rynku kapitalowym, PWN, Warszawa. 8. Walesiak, M.(2004). Decision Problems in a Cluster Analysis Procedure, Ekonometria (13), pp. 55-59. 9. Walesiak, M. (2014). Data Normalization in Multivariate Data Analysis. An Overview and Properties, Przeglad Statystyczny (4), pp. 363–372

1 Corresponding author. Tel.: + 48 22 5937242; fax: + 48 22 59 372 22. E-mail address: [email protected]. 366