Barcelona 12 15 May 2003 DATA ENVELOPMENT ANALYSIS

Barcelona 12 -15 May 2003 DATA ENVELOPMENT ANALYSIS IN THE BENCHMARKING OF ELECTRICITY DISTRIBUTION COMPANIES M. Sc. Jukka Lassila M. Sc. Satu Viljainen M. Sc. Samuli Honkapuro Prof. Jarmo Partanen M. Sc. Jukka Lassila FI Session 5 – Block 2 1

Barcelona 12 -15 May 2003 Overview Introduction Evaluation of the present DEA-model Developments of the present DEA-model Interruption costs Conclusions M. Sc. Jukka Lassila FI Session 5 – Block 2 2

Barcelona 12 -15 May 2003 Finland – Electricity distribution companies • • The number of electricity distribution companies: ~ 100 Average length of the network: 3 700 km (123… 49 000 km) Average number of customers: 31 000 (766… 314 000) 3 years experience of efficiency benchmarking (1999, 2000, 2001) M. Sc. Jukka Lassila FI Session 5 – Block 2 3

Barcelona 12 -15 May 2003 The factors of the efficiency benchmarking by DEA-model EFFICIENCY SCORE (0… 1) Operational costs Power quality (interruption time) Distributed energy M. Sc. Jukka Lassila FI Session 5 – Block 2 Number of customers Length of the network 4

Barcelona 12 -15 May 2003 The efficiency scores of the Finnish distribution companies The average is 0. 830 M. Sc. Jukka Lassila FI Session 5 – Block 2 5

Barcelona 12 -15 May 2003 The effects of the efficiency benchmarking (1/2) 1) Directing effects companies tend to pay attention to factors that are used in the DEAmodel 2) Efficiency score affect directly to the reasonable return on capital M. Sc. Jukka Lassila FI Session 5 – Block 2 6

Barcelona 12 -15 May 2003 The effects of the efficiency benchmarking (2/2) Example: Operational costs of a company are 200 M€/a. A) Efficiency score is 1. 0 Impact on allowed return = (1. 0 - 0. 9) * 200 M€ = 20 M€/a B) Efficiency score is 0. 72 Impact on allowed return = (0. 72 - 0. 9) * 200 M€ = -36 M€/a M. Sc. Jukka Lassila FI Session 5 – Block 2 7

Barcelona 12 -15 May 2003 Problems of efficiency benchmarking with DEA-model • The directing effects of benchmarking are not equal for all the companies - There are large numbers of companies for which the efficiency scores do not depend on power quality - Power quality affects the efficiency scores randomly • The changes in the directing effects differ from one year to another • The present efficiency benchmarking method has to be developed M. Sc. Jukka Lassila FI Session 5 – Block 2 8

Barcelona 12 -15 May 2003 Problems of efficiency benchmarking with DEA-model Price of outage [€/customer, h] The number of companies that have insignificant factors in the DEA-model M. Sc. Jukka Lassila FI Session 5 – Block 2 9

Barcelona 12 -15 May 2003 Developing the DEA-model (1/2) M. Sc. Jukka Lassila FI Session 5 – Block 2 10

Barcelona 12 -15 May 2003 Developing the DEA-model (2/2) • Principle changes in the model - power quality can be measured as a interruption costs - power quality is not a separate factor in the model - interruption costs are added to operational costs Power quality becomes meaningful and almost equally important factor for each company M. Sc. Jukka Lassila FI Session 5 – Block 2 11

Barcelona 12 -15 May 2003 Number of companies having insignificant factors in efficiency benchmarking M. Sc. Jukka Lassila FI Session 5 – Block 2 12

Barcelona 12 -15 May 2003 Price of outages in developed DEA-model • For most companies price of outages is between 4… 6 €/customer, h • Corresponding prices of outages in the present DEA-model are 0… 500 €/customer, h M. Sc. Jukka Lassila FI Session 5 – Block 2 13

Barcelona 12 -15 May 2003 Conclusions • The directing effects of benchmarking have to be predictable and equal for each company • This presentation introduced a solution to a problem concerning equality - basic idea was change the way in which power quality is handled in the DEA-model • Future research activities include improving the predictability and taking investment into account in the efficiency benchmarking M. Sc. Jukka Lassila FI Session 5 – Block 2 14