Agricultura 5(2007) Copyright 2007 by University of Maribor The decision support system for supplementary activities on organic farms Karmen Pažek1 and črtomir rOzmAn1 University of Maribor, Faculty of Agriculture, Vrbanska 30, 2000 Maribor, Slovenia In this paper we present the decision support system for ranking food processing projects on organic farms. The system is based on simulation modeling and multi criteria decision analysis. The deterministic simulation system KARSIM EKO .0 (DSM) consists out of deterministic production simulation models that enable different types of costs calculations for organic production and on farm food processing in the framework of supplementary activities. Simulation models results were further evaluated using a qualitative multi-attribute modeling methodology, supported by the software tool DEX - i and quantitative analytical hierarchical process - AHP, supported by Expert Choice 2000TM software. The analysis showed that by using current model input parameter the combination of apple cider, apple vinegar and plum brandy (business alternative ) results with the best multicriteria evaluation (EC = 0,413 and DEX - i evaluation = very good). Key words: simulation modeling, KARSIM EKO 1.0, Expert Choice (AHP), DEX - i, organic farming, supplementary activities INTRODUCTION Since the beginning of the 990s, organic farming has rapidly developed in almost all European countries. More than 5.4 million hectares were managed organically by almost 43 000 farms in the 25 countries of the European Union. This constituted 3.3 % of the agricultural area and 32 % of the farms in the EU (FiBL 2005). Organic farming is becoming more and more popular also among Slovene farmers. By the end of 2005 the number of farms practicing exclusively organic farming or taking up organic farming system amounted to more than 1 600 (or 1.85 % of all farms in Slovenia). The recent analysis showed that the average size of an Slovene organic farm is 3.4 ha. The data suggests that 4.7 % arable land in Slovenia is controlled by organic inspection body (MKGP 2006). In such a manner, the consumer preference for organic fruit food products on Slovene market is also high (Adamič 2000). In the practice the experience has shown that the organic farm products must be processed in to be sold more effciently. The fnancial cost beneft analysis (CBA) is thus a necessity before a particular project on investment into food processing is undertaken. Some authors conducted studies where CBA methodology was used for decision making about ranking the food processing alternatives. Pažek (2003) analyzed 27 different food processing alternatives on Slovene organic farms using fnancial CBA. The only criterion in this study for the selection of the best investment alternative was the net present value (NPV). Recent works on decision support system in agriculture used a variety of methodological approaches. Rozman et al. (2002) combine simulation modeling and mathematical programming for selection of best apple orchard system. Turk and Rozman (2002) consider the capacity of simulation Corresponding author: karmen.pazek@uni-mb.si modeling and CBA for planning and decision making in fruit production. All before mentioned studies used fnancial criterion as prevailing criterion in the decision process. Thus, cost-beneft analysis informs the decision making process, but it does not by itself make decisions. However, one should not automatically pursue the most fnancial effcient investment project, without weighing effciency against the other important criteria that affect complete project desirability. Therefore in real world situation the decision on investment is rarely made on the basis of only fnancial criterion. In the case of food processing there are many other criteria that must be taken into consideration besides fnancial indicators such as NPV (Tiwari et al. 999). Rogers and Bruen (1998) suggest the use of multi-criteria analysis to take into account different yet relevant criteria - even if these cannot be related to monetary outcomes - to be its main advantage. The most commonly applied approaches of MCDA are the analytical hierarchical process - AHP and the utility theory (Tiwari et al. 1999). Both before mentioned MCDA methods evaluate alternatives in empirical form. In the opposite, the approach based on the expert system for multi-attribute decision-making (DEX), uses qualitative variables and “if then” decision rules (Bohanec et al. 1995, Bohanec et al. 2000). In agriculture DEX - i was applied for the evaluation of some impacts by using genetically modifed crops, with special emphasis on soil biology and ecology. The decision support system incorporates economic and ecological impacts of cropping systems (Bohanec et al. 2004). The same author presented a decision-support tool SMAC Advisor for the assessment of coexistence between genetically modifed (GM) and conventional maize. The assessment is based on a DEX - i multi-attribute model (Bohanec et al. 2006). The MCDA for solution of some farm management decision problems is also described by other researcher (Hererro et al. 1999, Mazzeto and Bonera 2003), while the combination of 5 THE DECISION SUPPORT SYSTEM FOR SUPPLEMENTARY ACTIVITIES ON ORGANIC FARMS simulation models and the multi-criteria method is discussed by Pažek et al. (2006), Rozman et al. (2006) and Klajić et al. (2000). This paper presents application of simulation model KARSIM EKO .0 for cost analysis and investment planning on organic farms in combination with multi criteria decision models. The simulated alternatives are additionaly evaluated with multi attribute decision tools: Expert Choice decision support system (AHP) and DEX-i expert system. METHODOLOGY One of the goals of the research is to develop a computer-based decision support system for the assessment of fnancial, technological, ecological and market impact with special emphasis on supplementary activities on organic farms. Using the technologic-economic simulation modeling one can obtain information about the system itself and its responses to different model input parameters (Csaki 1985, Rozman 2004). The relationships between system elements (in this case input material, human labor) are expressed with a series of technological equations that are used for calculation (estimation) of input usage and outputs produced. For f-nancial and technological analysis of the food processing on Slovene organic farms the computer simulation model KARSIM EKO 1.0 was developed (Pažek 2005). There are two basic sub-model: the sub-model of specifc farm products and the sub-model of food processing into different fnal organic products. The developed model enables calculation of the most important economic parameters (break even price, coeffcient of proftability, fnancial result,…) and fnancial indicators for the alternative evaluation (investment costs, Net Present Value, internal rate of return). The KARSIM EKO .0 output data represent some of the input parameter of analyzed farm business alternative in multi-criteria decision analysis. The decision simulation model (DSM) structure is shown in Figure . The food processing investment alternatives are further evaluated with multi attribute decision model. A multi-attribute model is a hierarchical structure that represents the decomposition of the decision problem into sub-problems, which are smaller, less complex and possibly easier to solve than the complete model (Figure 2). The variants are decomposed in specifc parameters (criterion, attribute, objective) and evaluated separately for each single parameter. The fnal variant evaluation is provided with the combine proceeding. The provided value presents the portfolio for selection of most suitable variant (solution) (Bohanec and Rajkovič 995). For evaluation of simulated alternatives two methodological tools are applied: DEX - i expert System and AHP based Expert Choice (EC). DEX is an expert system shell for multi-attribute decision making that combines the “traditional” multi-attribute decision making with some elements of Expert Systems and Machine Learning. The distinguishing characteristic of DEX is its capability to deal with qualitative models. Instead of numerical variables, which typically constitute traditional quantitative models, DEX uses qualitative BUSINESS ALTERNATIVES CROP, FRUIT AND ANIMAL PRODUCTION BASIC FARM DATA FARM DATA BASE PRODUCTION TECHNICAL PARAMETERS PRODUCTION MODELS PRODUCTION COSTS CALCULATIONS(ENTERPRISE BUDGETS) PROCESSING TECHNICAL PARAMETERS PROCESSING ENTERPRISE BUDGETS COSTS CALCULATIONS FOR ALTERNATIVES REQUIRED INVESTMENTS FINANCIAL CBA (NPV CALCULATION) Fig. 1: The structure of deterministic simulation model (DSM) for cost calculations and planning on organic farms KARSIM EKO 1.0 16 THE DECISION SUPPORT SYSTEM FOR SUPPLEMENTARY ACTIVITIES ON ORGANIC FARMS Alternative evaluation Utility function 1 r Aggregate attribute Financial X, r Partial utility function F (X„ X,) I ^ r Net Present Value X[ Investment costs X2 F (x«» x»> xio> xu> xu) Procedure X3 Technological X, F 12000 Extremely high NPV (€) 0-5000 Low 5001-10000 Average 10001-20000 High > 20000 Extra high Human labour (hours) 0-50 Low 51-100 Average 101-200 High > 200 Very high Equipment requirements (equipment) 0-2 Simple 3-4 Average 5-6 Demanding > 6 Very demanding Process (steps in processing technique) 0-5 Simple 6-10 Average 11-15 Demanding > 15 Very demanding 7 THE DECISION SUPPORT SYSTEM FOR SUPPLEMENTARY ACTIVITIES ON ORGANIC FARMS sources (i.e., past selling experiences and climatic characteristics). The following qualitative scales were used for non-numerical sub-attributes (Table 2). After each attribute has been assigned with its scales (qualitative value), the utility functions (knowledge base) are defned. The utility function is conducted for each level in the hierarchy (partial utility function for aggregate attributes and overall utility function for the whole model except for the lowest level in the hierarchy). The decision rules are presented in complex form where the asterisk “*” means any value and >= equal or better (Table 3). Finally, attribute values for each alternative are put into Table 2. Qualitative scales for non-numerical attributes Market objective poor; average; good; excellent Spring frost frequency high; average; low Hailstone frequency high; average; low the DEX-i evaluation table and the analysis is ultimately conducted. Another applied MCDA method is analytical hierarchical process (AHP) with application of decision support software Expert Choice 2000TM. The AHP is a decision-aided method which de-composes a complex multi-factor problem into a hierarchy. The AHP method determines the priorities of each alternative with the assigned weight for each alternative by analyzing the judgmental matrices using the ad- Table 3. The decision rules for organic farm planning problem Human labor intensity * * high high low low low low Project evaluation negative * * * unacceptable <=bad very high unacceptable bad unacceptable <=bad very high * bad unacceptable <=bad <=average unacceptable bad unacceptable good * acceptable excellent very good >=good high acceptable excellent very good excellent low acceptable good very good excellent low acceptable >=good excellent vanced mathematical theory of eigenvalues and eigenvectors. It interprets the eigenvector associated with the largest eigenvalue as the priorities that indicate the importance of each alternative in accomplishing the objective. AHP combines both subjective and objective judgments in an integrated framework based on ratio scales from simple pair-wise comparisons. Saaty (1980) developed the following steps for applying the AHP: 1. Defne the problem and determine its goal. 2. Structure the hierarchy from the top (the objectives from a decision-maker’s viewpoint; i = 1, …, m ob jectives) through the intermediate levels (criteria on which subsequent levels depend) to the lowest level which usually contains the list of alternatives. 3. Construct a set of pair-wise comparison matrices (size n x n) for each of the lower levels with one matrix for each element in the level immediately above by using the relative scale measurement (for each objective i, compare the j = 1, …, n alternatives and determine their weights aij with respect to objective i). 4. There are n x (n - ) judgments required to develop the set of matrices in step 3. Reciprocals are automatically assigned in each pair-wise comparison. 5. Hierarchical synthesis is now used to weight the eigenvectors by the weights of the criteria and the sum is taken over all weighted eigenvector entries corresponding to those in the next lower level of the hierarchy (the fnal alternative weights (priorities) Wj with respect to all the objectives by W = a, w + a w2 + j lj  2j … + a w ). 6. Having made all the pair-wise comparisons, the consistency is determined by using the eigenvalue, ? max, to calculate the consistency index, CI as follows: CI = (? - n) / (n - 1), where n is the matrix size. max Judgment consistency can be checked by taking the consistency ratio (CR) of CI. The CR is acceptable, if it does not exceed 0.0. The alternatives are then ordered by the Wj, with the most preferred alternative having the largest Wj. The EC software allows us to enter the data for each alternative into the so-called Data Grid, where individual objectives can be entered directly. In this case the intensities or possible qualitative values of decision attributes at the lowest level in the hierarchy are compared in the pair wise comparison matrix (and not the alternatives). This feature enables the usage of the same qualitative scales for qualitative attributes as in the DEX-i model. The data (attribute values) is the entered for each alternative. The use of the Data Grid combines the power of the hierarchy and the pair-wise comparison process with the ability to evaluate hundreds or even thousands of alternatives that would be diffcult to compare in pair wise comparison matrix. Alternatives priorities are established relatively to each covering objective by using ratio scaled rating intensities (scales). A real-world case The real organic part time farm in north east Slovenia was considered to apply KARSIM EKO .0 and MCDA methods using the presented methodological approach. In order to perform analysis 3 business alternatives on sample part time organic farm were selected: - the alternative 1 (80% of apples and 20% of plum on 2 ha grassland orchard) presented apple and plum processing into apple cider, apple vinegar and plum brandy, - the alternative 2 presented apple and plum processing into apple cider and plum brandy and - the alternative 3 presented the combination of dry fruit (apple) and plum brandy. RESULTS AND DISCUSSION In the frst phase for every analyzed business alternative the costs are calculated using the KARSIM EKO .0. In the second phase the CBA was conducted for 15 years at 14% discount rate (Table 4). The CBA results show the fnancial feasibility of business alternatives; assuming that expected prices and yields would be achieved and that products would be successfully marketed. 18 THE DECISION SUPPORT SYSTEM FOR SUPPLEMENTARY ACTIVITIES ON ORGANIC FARMS The simulation results were further evaluated with multi-attribute decision model (DEX - i and AHP). Since the main results from a simulation model are numerical (investment fnancial indicators and labor input) the qualitative values must be assigned to each quantitative parameter in order to enable further analysis in DEX - i decision model. This is conducted with classifcation algorithm based on classifcation intervals. Table 4. The results of fnancial CBA analysis of business alternatives on a sample organic farm Alternative 1 Alternative 2 Alternative 3 Investment costs (€) 7 512 1 761 11 492 NPV (€) 4 934 2 920 101 Labor (hours) 1 391 996 1 179 Ranking of alternatives 1 2 3 The DEX - i and AHP evaluation of alternatives (table 5 and 6) result in same ranking of alternatives (alternative 1, alternative 3, alternative 2). These ranking differs from CBA ranking (alternative , alternative 2, alternative 3). The ranking of alternatives based on DEX - i and AHP is in this case the same. However, the AHP produces more detailed ranking which can be useful when the alternatives are close to each other and would probably produce the same qualitative DEX - i evaluation. Therefore, the applied AHP methodol- Table 5. DEX – i project evaluation of business alternatives with importance weights of aggregate attributes Attribute Alternative 1 Alternative2 Alternative 3 Investment project appraisal very good acceptable good Financial indicators (*W=44,7 %) excellent good excellent NPV very high high very high Investment costs low low low Labor intensity (*W=6,4 %) very high very high very high Technological requirements (*W=11,9 %) unacceptable unacceptable unacceptable Equipment requirements very demanding demanding very demanding Processing process average demanding demanding Market objective (*W=24,4%) excellent average good Risk objective (*W=12,6%) low low low Spring frost probability average average average Hailstone dangerous low low low Ranking 1 3 2 Table 6. The Expert Choice AHP business alternatives evalua tion for the sample organic farm ?Wac 0.413 0.240 0.347 1 3 2 Weight (Wa) 0.438 0.065 0.122 0.245 0.130 ab Alternative 1 0.396 0.230 0.333 0.573 0.333 Alternative 2 0.169 0.540 0.333 0.191 0.333 Alternative 3 0.435 0.347 0.333 0.236 0.333 Wa - weight; ab - alternative priority, ?Wac - alternative priority with respect to individual objective (objectives with no sub-levels are assed directly from pair wise comparison matrices) ogy should bring unequivocal clarity to the decision which business alternative should be favored and implemented on a sample part time organic farm (Table 6). The AHP does not exclude any alternatives it only ranks them according to the defned hierarchy and relative importance of decision criteria. On the contrary, using DEX - i expert system it can be defned which combination of attribute values is not acceptable for the decision maker. Thus, the DEX - i assessment can be used for exclusion of “unacceptable” alternatives, but the shortcoming of DEX - i is its inability (in contrast to AHP) to separate between alternatives with the same qualitative evaluation. CONCLUSIONS The integrated simulation model KARSIM EKO .0 combined with multi-criteria decision analysis present the suitable methodological tool for decision support system on organic farms. The system takes into consideration different independent criteria and enables ranking of farm business alternatives. The real value of presented decision model is in its capability to conduct different kinds of “what if” analysis in farm food processing projects. The use of both MCDA approaches can bring additional information into the decision-making framework (for instance the “unacceptable” alternatives can be excluded with the use of the DEX - i model, while the precise ranking of remaining alternatives is based strictly on the AHP Expert Choice model). In presented paper both MCDA methods favored alternative , which got the highest DEX - i and AHP (EC) evaluation. The application of the proposed decision support system (combination of KARSIM EKO 1.0, AHP and DEX - i model) would increase the accuracy of information needed for developing farm business plans and that in addition it would help preventing many inappropriate decisions being made on organic farms. 9 THE DECISION SUPPORT SYSTEM FOR SUPPLEMENTARY ACTIVITIES ON ORGANIC FARMS REFERENCES 1. Adamič N. Analysis of the possibilities of marketing or- ganic products in the north-eastern part of Slovenija (Graduation Thesis). Maribor: Faculty of Agriculture, University of Maribor, 2000. 2. Bohanec M, Messéan A, Angevin F, Žnidaršič M. SMAC Advisor: A decision-support tool on coexistence of genetically-modifed and conventional maize. Proc. Information Society IS 2006;9-12. 3. Bohanec M, Džeroski S, Žnidaršič M, Messéan A, Scatasta S, Wesseler J. Multi-attribute modeling of economic and ecological impacts of cropping systems. Informatica 2004;28:387-92. 4. Bohanec M, Zupan B. A function - decomposition me- thod for development of hierarchical multi - attribute decision models. Decision Support Systems 2004;36: 25-33. 5. Bohanec M, Zupan B, Rajkovič V. Applications of quali- tative multi - attribute decision models in health care. I. J. Med. Inf. 2000;58-59: 191-05. 6. Bohanec M, Rajkovič B. “Multi Attribute Decision Mo- deling: Industrial Application of Dex.” Informatica 1999;23:448-91. 7. Bohanec M, Rajkovič B. Semolič B, Pogačnik A. “Knowledge - based portfolio analysis for project evaluation.” Information & Management 1995;28:293-02. 8. Bohanec M, Rajkovič V. Večparametrski odločitveni modeli. Organizacija 1995;7:427-38. 9. Csaki C. Simulation and system analysis in agriculture. Elsevier, Amsterdam, 1985. 10. FiBL - Forschungsinstitut für biologischen Landbau (http://www.organic-europe.net/europe_eu/statistics. asp, 08.12.2006). 11. Herrero M, Fawcett RH, Dent J.B. “Bio-economic eva- luation of dairy farm management scenarios using integrated simulation and multiple-objectives models.” Agricultural Systems 1999;69:169-88. 12. Klajić M, Bernik I, Škraba, 2000. A. Simulation Ap- proach to Decision assessment in Enterprises. Simulation 2000;75/4:99-20. 13. Mazzeto F, Bonera R. MEACROS: a tool for multi-cri- teria evaluation of alternative cropping systems. Eur. J. Agr. 2003;18:379-87. 14. MKGP ANEK - Akcijski načrt razvoja ekološkega kme- tijstva v Sloveniji do leta 205. Ljubljana, Slovenia, 2006. 15. Pažek K, Rozman Č, Borec A, Turk J, Majkovič D, Bavec M, Bavec F. The Multi criteria models for decision support on organic farms. Biol. Agr. Hort. 2006;24/1:73-89. 16. Pažek K. KARSIM EKO 1.0, Internal Database; Univer- sity of Maribor, Faculty of Agriculture, 2005. 17. Pažek K. The fnancial analysis (CBA) of supplemen- tary activities on organic farms (Master Thesis). Maribor: Faculty of Agriculture, University of Maribor, 2003. 18. Rogers M, Bruen M. Choosing realistic values of indif- ference, preference and veto thresholds for use with environmental criteria within ELECTRE. Eur. J. Oper Res. 1998;107:542-51. 19. Rozman Č, Pažek K, Bavec M, Bavec F, Turk J, Majkovič D. The Multi-criteria analysis of spelt food processing alternatives on small organic farms. J. Sust. Agr. 2005;28/2:159-79. 20. Rozman Č. Application of simulation models and posi- tive mathematical programming for the economic analysis of fruit production (dissertation). Maribor: Faculty of Agriculture, University of Maribor, 2004. 21. Rozman Č, Tojnko S, Turk J, Par V, Pavlovič M. Die Anwendung eines Computersimulationsmodells zur Optimierung der Erweiterung einer Apfelplantage unt-er den Bedingungen der Republik Slowenien. Berichte über Landwirtschaft 2002;80/4:632-42. 22. Saaty T.L. The analytic hierarchy process. McGraw Hill, New York, 1980. 23. Tiwari DN, Loof R, Paudyal GN. Environmental-eco- nomic decision making in lowland irrigated agriculture using multi-criteria analysis techniques. Agric. Syst. 1999;60:99-112. 24. Turk J, Rozman Č. A feasibility study of fruit brandy production. Agricultura 2002;8 (1/1):28-33. Received: January 23, 2007; Accepted in fnal form: April 17, 2007 20