410 likes | 537 Views
Sustainability Measurement and MCDM: The Case of the Wood-Based Industry in Europe. Roberto Voces, Luís Díaz-Balteiro, Jacinto González-Pachón, Carlos Romero. Research Group " Economics for a Sustainable Environment " Technical University of Madrid http://www.ecsen.es/.
E N D
Sustainability Measurement and MCDM: The Case of the Wood-Based Industry in Europe Roberto Voces, Luís Díaz-Balteiro, Jacinto González-Pachón, Carlos Romero Research Group "Economics for a Sustainable Environment" Technical University of Madrid http://www.ecsen.es/
Incoherent social preference within an environmental context Roberto Voces, Luís Díaz-Balteiro, Jacinto González-Pachón, Carlos Romero Research Group "Economics for a Sustainable Environment" Technical University of Madrid http://www.ecsen.es/
Scheme of this presentation 1 The context 2 The problem of incoherent preferences 3 Goal Programming formulations 4 Case study 5 Conclusions 3
The context (1/9) In an environmental context especially with a perspective of sustainability the points of view of several stakeholders must be taken account 5
The context (2/9) It is easier to obtain ordinal information about stakeholders preferences. Moreover this information is robust Ordinal and Global 6
The context (3/9) But what we actually need for management purposes are priority weights to be attached to each considered criterion (cardinal information) Cardinal and Global 7
The context (4/9) Due to cognitive constraints, cardinal information is usually derived from local information Cardinal and Local 8
The context (5/9) PREFERENCE INFORMATION CLASSIFICATION vs ORDINAL CARDINAL LOCAL GLOBAL vs 9
The context (6/9) PREFERENCE INFORMATION CLASSIFICATION ORDINAL CARDINAL LOCAL GLOBAL 10
The context (7/9) PREFERENCE INFORMATION CLASSIFICATION ORDINAL CARDINAL LOCAL GLOBAL 11
The context (8/9) GLOBAL m*11 .........m*1n RATIONAL DM m*21 ........ m*2n LOCAL ....................... m*n1 ....... m*nn Reciprocity mijx mji = 1 i, j mijxmjk = mik i, j, k Consistency 12
The context (9/9) • Hence in many cases we have two sources of information: • ORDINAL (i.e. rankings) very robust • CARDINAL (i.e. priority weights) very sensitive CONFLICT / INCOMPATIBILITY between the two sources of information A proposal of a new methodology for combining the two sources of information in a rational way 13
Section 2: The problem of incoherent preferences
The problem of incoherent preferences (1/7) DERIVED RANKING REVEALED RANKING m*11 .........m*1n m*21 ........ m*2n ....................... m*n1 ....... m*nn 15
The problem of incoherent preferences (2/7) INCOHERENT PREFERENCES Incompatible DERIVED RANKING REVEALED RANKING 16
The problem of incoherent preferences (3/7) SOLVING INCOHERENCES DERIVED WEIGHTS REVEALED RANKING Soft information Robust information 17
The problem of incoherent preferences (4/7) CONSENSUS WEIGHTS (variables) (Assignment values) (Ranking conditions) 18
The problem of incoherent preferences (5/7) Satisficing logic Compatibility search Goal Programming 19
The problem of incoherent preferences (6/7) Generaloptimisation problem based on a lp-distance p is the metric Feasible set (revealed ranking constraints) (normalization constraint) 20
The problem of incoherent preferences (7/7) Generaloptimisation problembased on al-distance (Chebychev) Feasible set (revealed ranking constraints) (normalization constraint) 21
GP formulations (1/4) CONSENSUS WEIGHTS (variables) …….. (Derived Assignment) (Revealed Ranking) GOALS CONSTRAINTS 23
GP formulations (2/4) Weighted Goal Programming formulation Feasible set (derived assignment goals) (revealed ranking constraints) (normalization constraint) 24
GP formulations (3/4) Min-Max Goal Programming formulation Feasible set (derived assignment goals) (discrepancy) (revealed ranking constraints) (normalization constraint) 25
GP formulations (4/4) Extended Goal Programming formulation [0,1] control parameter Feasible set (derived assignment goals) (discrepancy) (revealed ranking constraints) (normalization constraint) 26
Case study (1/9) The general purpose of the study was to analyze the sustainability of the wood industry in Europe. Data for 14 indicators in 17 European countries were obtained. In order to obtain weights for each indicator, a questionnaire was sent to 104 experts from 22 countries. Moreover, the questionnaire asked about a whole ranking of indicators (revealed ranking) Individual priority weights were obtained by using Saaty AHP. From these (normalized) weights a ranking of indicators can be obtained (derived ranking) 28
Case study (2/9) INDICATORS • Economic indicators • Environmental indicators • Social indicators • Innovation (sustainability) indicators 29
Case study (3/9) INDICATORS I1 Gross value added (percentage of the total manufacturing) I2 Energetic efficiency (Purchases of energetic products/Production) I3 Dependency of industrial rounwood (Imports/Apparent consumption) I4 Unitary average wage (weighted by per capita income) I5 Gross value added per employee (Gross value added/Number of employees) I6 Intensity in labor force (percentage of labor costs respect to production) I7 Investment (Total investment/ Gross value added) I8 Acquisition of built-in technology (gross investment in machinery and equipment/Number of firms) 30
Case study (4/9) INDICATORS I9 Innovative enterprises (percentage of the total number of enterprise) I10 Effects of the innovation (turnover of the innovative firms like percentage of the total turnover I11 Patents (patent applications to the EPO during the year 2003) I12 External competitiveness (Balassa index) I13 Total waste (total waste generated /gross value added) I14 Environmental protection 31
Case study (5/9) 0.157 1st REVEALED RANKING 0.133 2nd I12 I12 I10 I11 I11 I10 I14 I14 I13 I13 I4 I5 I5 I8 I9 I3 I9 I1 I6 I6 I2 I7 I1 I3 I7 I2 I4 I8 3rd 0.133 4th 0.133 0.133 5th 0.054 6th 7th 0.045 8th 0.044 9th 0.038 10th 0.036 11th 0.035 12th 0.022 13th 0.020 DERIVED RANKING 14th 0.017 32
Case study (6/9) Weighted Goal Programming formulation Min (n1+p1)+(n2+p2)+(n3+p3)+(n4+p4)+(n5+p5)+(n6+p6)+(n7+p7)+(n8+p8)+(n9+p9)+ (n10+p10)+(n11+p11)+(n12+p12)+(n13+p13)+(n14+p14) Feasible set 3rd 1st 2nd w9-w1 0.01 w9-w2 0.01 w9-w3 0.01 w9-w4 0.01 w9-w5 0.01 w9-w6 0.01 w9-w7 0.01 w9-w8 0.01 w9-w10 0.01 w9-w11 0.01 w9-w12 0.01 w9-w13 0.01 w9-w14 0.01 …… w5-w1 0.01 w5-w2 0.01 w5-w3 0.01 w5-w4 0.01 w5-w6 0.01 w5-w7 0.01 w5-w8 0.01 w5-w10 0.01 w5-w11 0.01 w5-w13 0.01 w5-w14 0.01 w12-w1 0.01 w12-w2 0.01 w12-w3 0.01 w12-w4 0.01 w12-w5 0.01 w12-w6 0.01 w12-w7 0.01 w12-w8 0.01 w12-w10 0.01 w12-w11 0.01 w12-w13 0.01 w12-w14 0.01 w1+n1-p1=0.036 w2+n2-p2=0.020 w3+n3-p3=0.054 w4+n4-p4=0.133 w5+n5-p5=0.133 w6+n6-p6=0.045 w7+n7-p7=0.038 w8+n8-p8=0.018 w9+n9-p9=0.133 w10+n10-p10=0.044 w11+n11-p11=0.157 w12+n12-p12=0.133 w13+n13-p13=0.022 w14+n14-p14=0.035 …… --------------- (assignment goals) (revealed ranking constraints) …… …… --------------- --------------- …… 33
Case study (7/9) 0.143 0.157 1st REVEALED RANKING 0.133 0.133 2nd I12 I10 I11 I14 I13 I5 I8 I9 I1 I6 I2 I3 I7 I4 3rd 0.123 0.133 4th 0.105 0.133 0.100 0.133 5th 0.090 0.054 6th 7th 0.070 0.045 8th 0.065 0.044 9th 0.050 0.038 10th 0.045 0.036 11th 0.030 0.035 12th 0.025 0.022 REVISED DERIVED WEIGHTS (WGP) 13th 0.015 0.020 14th 0.005 0.017 34
Case study (8/9) Min-Max Goal Programming formulation Min D Feasible set 1st w9-w1 0.01 w9-w2 0.01 w9-w3 0.01 w9-w4 0.01 w9-w5 0.01 w9-w6 0.01 w9-w7 0.01 w9-w8 0.01 w9-w10 0.01 w9-w11 0.01 w9-w12 0.01 w9-w13 0.01 w9-w14 0.01 …… n1+p1≤ D n2+p2 ≤ D n3+p3 ≤ D n4+p4 ≤ D n5+p5 ≤ D n6+p6 ≤ D n7+p7 ≤ D n8+p8 ≤ D n9+p9 ≤ D n10+p10 ≤ D n11+p11 ≤ D n12+p12 ≤ D n13+p13 ≤ D n14+p14 ≤ D w1+n1-p1=0.036 w2+n2-p2=0.020 w3+n3-p3=0.054 w4+n4-p4=0.133 w5+n5-p5=0.133 w6+n6-p6=0.045 w7+n7-p7=0.038 w8+n8-p8=0.018 w9+n9-p9=0.133 w10+n10-p10=0.044 w11+n11-p11=0.157 w12+n12-p12=0.133 w13+n13-p13=0.022 w14+n14-p14=0.035 …… (assignment goals) (discrepancy) (revealed ranking constraints) …… …… …… 35
Case study (9/9) 0.143 0.136 0.157 1st REVEALED RANKING 0.133 0.126 0.133 2nd I12 I10 I11 I14 I13 I5 I8 I9 I1 I6 I2 I3 I7 I4 3rd 0.123 0.116 0.133 4th 0.105 0.105 0.133 0.100 0.100 0.133 5th 0.090 0.090 0.054 6th 7th 0.070 0.076 0.045 8th 0.065 0.066 0.044 9th 0.050 0.056 0.038 10th 0.045 0.046 0.036 11th 0.030 0.036 0.035 12th 0.025 0.026 0.022 REVISED DERIVED WEIGHTS (WGP / Min-Max) 13th 0.015 0.016 0.020 14th 0.005 0.006 0.017 36
Conclusions (1/) • In an environmental context with a sustainability perspective it is • necessary to take into account: • Several criteriaof different nature • Several social groups or stakeholders with different preferences with • respect the above criteria Aggregation of individual preferences into a collective one is a crucial Issue in this field
Conclusions (2/) • Due to the characteristics of the situational context, the stakeholders • usually provide their preferences in two ways: • Ordinal (rankings) Robust information but incomplete • Cardinal (and local; i.e. pc matrices) Sensitive but complete In many cases, there is a clash (incompatibility) between the two sources of information Both source of information are empirically worthwhile
Conclusions (3/) A critical issue is to combine or to take advantage from both empirical sources We have proposed a method bases upon Goal Programming to undertake the above task • The proposed method presents some advantages: • It is non-interactive • The computational burden implies to solve LP of moderate size • All the compromise consensus obtained have a clear preferential • interpretation
THANK YOU FOR YOUR ATTENTION! COMMENTS ARE WELCOME jgpachon07 http://es.linkedin.com/in/jgpachon