Service Quality Regulation in Electricity Distribution

1
Service Quality Regulation in Electricity
Distribution

• Necmiddin BAGDADIOGLU
• Orçun SENYÜCEL

2
Objectives

• Incorporate service quality measure into
  electricity regulation. New in literature
  Growitsch et al (2008), Coelli et al (2008-Draft)
  
• Determine technical efficiency of Turkish
  electricity distribution utilities
• Focus on exogeneous determinants of inefficiency
• Analyze effects of electricity losses and illegal
  usage on TE.

3
Turkish Electricity Reform

• Electricity Sector Reform and Privatization
  Strategy Paper (2004) TEDAS ? 2012
• Transitory period 20 utilites through mergers of
  79 distribution utilities.
• ESRPSP mergers determined by operational
  problems, technical financial features.
• Turkey accession country. EU Energy Acquis
• EMRA has not announced regulatory framework

4
Briefly SFA v DEA

• Average Cost (all noise)
  Syrjanen, M., P. Bogetoft, P. Agrell (2006)

5
Briefly SFA v DEA

• Deterministic frontier (all ineff ? u)
  Syrjanen, M., P. Bogetoft, P. Agrell
  (2006)

6
Briefly SFA v DEA

• Stochastic frontier (both noise v and
  ineff u) Syrjanen, M., P. Bogetoft, P.
  Agrell (2006)

7
Briefly SFA

• Two component error terms, first captures
  statistical noise
• Second captures effects of TE.
• Half normal, exponential, truncated dist.

8
Distance Functions

• DF Distance of the prod to PPB
• Two different types input output DF
• Input DF How much input vector can be contracted
  (output constant)
• Output Vice versa.

9
Distance Functions
Kumbhakar Lovell (2003)
10
Distance Functions

• Deviations from 1 is technical inefficiency
• h(.) represents deviation ?exp (-u)
• exp (-u) one of the component error terms.

11
Distance Functions

• Adding random error term, imposing homogeneity
  rest.
• We preffered translog input DF.

12
Methodology

• Following Coelli, (M outputs K inputs)

13
Methodology

• Following Coelli and Battese,
• Two environmental variables

14
Models

• Model I Input TOTEXLIEU (TOTEXL)
• Model II Input Interruption Time (ITC)
• Output Energy supplied (ENG) and number of
  customers (CUST)
• Environmental factors
• Village Cust Density (VCD)
• Geographic Conditions (GEO)

15
Model I
16
Model II
17
Descriptive Statistics
Variable Obs Mean Std. Dev. Min Max
TOTEXL (million TL) 120 160 159 30 891
Energy Supplied (GWh) 120 4744 3596 628 17987
Number of customers [phone] 812 323 3719
ITC 120 2.60 2.14 0.07 11.88
VCD 120 0.216 0.106 0 0.419
18
Model I
Variables Coefficient t-ratio
Constant 0.5223 0.7
ln ENG (y1) -0.3429 -2.8
ln CUST (y2) -0.7294 -4.8
y12 -0.1342 -0.3
y22 -1.0085 -1.5
y1y2 0.3645 0.7
VCD 3.8507 2.0
GEO 6.4172 1.7
0.6373 2.8
0.9023 23.5
LLF -36.0925
Note , and denotes significance at the
1, 5 and 10 levels.
19
Model I
RTS0.93
20
Model II
Variables Coefficient t-ratio
Constant 0.6918 13.2
ln ENG (y1) -0.3988 -3.1
ln CUST (y2) -0.5408 -3.7
y12 -0.2412 -0.4
y22 -1.6727 -2.3
y1y2 0.7813 1.3
ln (ITC/TOTEXL) 0.2745 6.4
ln(ITC/TOTEXL)2 0.1682 2.7
ln (ITC/TOTEXL)y1 -0.1434 -1.0
ln (ITC/TOTEXL)y2 0.3120 2.9
VCD 5.8604 4.4
GEO 1.5384 6.3
0.2293 3.3
0.9129 19.8
LLF 12.7273
Note , and denotes significance at the
1, 5 and 10 levels.
21
Model II
RTS1.06
22
Average efficiency scores
QoS has significant effect TE decreased by
16.5 LLR test also states QoS important
Quality excluded model Quality included model
Small utilities 0.60 0.43
Medium utilities 0.91 0.77
Large utilities 0.91 0.84
Total 0.79 0.65
23
Average efficiency scores
Utility Quality excluded Quality included Quality included (Cost of Losses Excluded) Losses and Illegal Usage ()
Dicle Edas 0.11 0.10 0.50 62.04
Vangolu Edas 0.16 0.12 0.47 61.95
Aras Edas 0.33 0.19 0.48 31.71
Coruh Edas 0.87 0.40 0.75 13.70
Firat Edas 0.64 0.47 0.52 14.67
Camlibel Edas 0.89 0.67 0.81 10.52
Toroslar Edas 0.90 0.79 0.95 15.38
Meram Edas 0.90 0.71 0.89 8.58
Baskent Edas 0.91 0.77 0.95 11.26
Akdeniz Edas 0.89 0.78 0.92 11.21
Gediz Edas 0.93 0.89 0.98 7.62
Uludag Edas 0.93 0.91 0.93 9.09
Trakya Edas 0.89 0.72 0.89 10.52
AYEDAS 0.93 0.93 1.00 10.80
Sakarya Edas 0.92 0.84 0.94 12.33
Osmangazi Edas 0.93 0.87 0.94 7.42
Bogazici Edas 0.90 0.87 1.00 17.32
Menderes Edas 0.91 0.78 0.87 9.66
Goksu Edas 0.92 0.80 0.83 10.78
Yesilirmak Edas 0.89 0.48 0.81 12.28
24
Conclusion

• QoS impact on TE. GEO VCD are crucial
  environmental variables.
• Excl. losses and illegal electricity usage
  overestimates TE.
• Privatization Eight utilities are established
  far from the optimal size and have low average
  efficiency scores (0.43). TPA may merge other six
  utilities.