CenSES Annual Report 2015

CenSES annual report 2015

Education

2015 dissertation: Xiaomei Cheng, NHH

Xiaomei Cheng defended her PhD thesis in December 2015.

Her thesis is titled “Essays on Efficiency and Productivity in

Electricity Networks”.

The Norwegian electricity sector was reorganized and

deregulated in the beginning of the 1990s. The basic idea

behind the restructuring was to unbundle the services in

the value chain and expose some of them to competition.

The competitive part includes generation, wholesale and

retail/supply, while other functions, like transmission and

distribution, remained regulated. Since 1993, different

regulatory schemes have been used by the regulator NVE,

from rate of return regulation in the first years to the present

yardstick regulation. The efficiency incentives for the

network companies have gradually been strengthened, and

NVE has increasingly relied on benchmarking methodology

to evaluate the performance of the companies and to set

reasonable revenue caps.

The thesis contains five papers on topics related to the

performance, i.e., efficiency and productivity, in

Norwegian distribution companies using different

benchmarking approaches. The first paper compares Data

Envelopment Analysis (DEA) to the recently developed

StoNED (Stochastic Non-parametric Envelopment of Data)

method. The latter method can distinguish between

noise and inefficiency, but the paper illustrates that the

distributional assumptions often are inconsistent with

observed data, and that they can have a great effect on the

estimated efficiencies.

The second paper develops a measure of scale elasticities

based on StoNED. The analysis shows that the distribution

companies are predominantly too small, in line with

previous studies. Optimal company sizes decrease when the

analysis is adjusted for local cost drivers like weather and

topology, but this effect is small.

The third paper shows how measurement and

decomposition of Malmquist productivity indexes can

be implemented with StoNED. It also shows that the

distributional assumptions in StoNED will influence some,

but not all of, the productivity indices. The fourth paper

applies the developed method to data for 2004-2013 and

compares the results to those obtained with DEA and

SFA. All methods agree that the average productivity of

the distribution companies increased from 2004 to 2007,

and that it decreased thereafter. This is surprising, since

the efficiency incentives were strengthened with the

introduction of yardstick regulation in 2007. Although

the methods disagree somewhat on the cause for the

negative development, they all agree that there has been

technological regress for the entire period 2004-2013.

The last paper suggests an

alternative distributional

assumption about

inefficiency in order to

accommodate the negatively

skewed distributions that

are sometimes observed for

estimated residuals, and it

discusses the importance

of functional form in the

suggested approach via a

simulation study.

PhD student Xiaomei Cheng.