Energy Policy 60 (2013) 741752

Contents lists available at SciVerse ScienceDirect

Energy Policy
journal homepage: www.elsevier.com/locate/enpol

Reforming residential electricity tariff in China: Block tariffs

pricing approach
Chuanwang Sun a,b, Boqiang Lin c,d,n
a

China Center for Energy Economics Research, School of Economics, Xiamen University, Xiamen 361005, China
Institute for Studies in Energy Policy, Xiamen University, Xiamen 361005, China
c
Collaborative Innovation Center for Energy Economics and Energy Policy, Institute for Studies in Energy Policy, Xiamen University, Xiamen 361005, China
d
New Huadu Business School, Minjiang University, Fuzhou 350108, China
b

H I G H L I G H T S

We design a rising block tariff structure of residential electricity in China.

We set up a translog demand model to nd the non-linear effects on elasticities.
The higher income groups are less sensitive to price changes.
Block tariff structure generates more efcient allocation of cross-subsidies.
Block tariff structure supports the living standards of low income households.

art ic l e i nf o

a b s t r a c t

Article history:
Received 12 June 2011
Accepted 8 May 2013
Available online 14 June 2013

The Chinese households that make up approximately a quarter of world households are facing a
residential power tariff reform in which a rising block tariff structure will be implemented, and this tariff
mechanism is widely used around the world. The basic principle of the structure is to assign a higher
price for higher income consumers with low price elasticity of power demand. To capture the non-linear
effects of price and income on elasticities, we set up a translog demand model. The empirical ndings
indicate that the higher income consumers are less sensitive than those with lower income to price
changes. We further put forward three proposals of Chinese residential electricity tariffs. Compared to a
at tariff, the reasonable block tariff structure generates more efcient allocation of cross-subsidies,
better incentives for raising the efciency of electricity usage and reducing emissions from power
generation, which also supports the living standards of low income households.
& 2013 Elsevier Ltd. All rights reserved.

Keywords:
Block tariffs
Residential electricity
Price elasticity of power demand

1. Introduction
Different from the United States (Sueyoshi, 2010), China does
not have a perfect market for wholesale or retail power trade. The
Chinese state-owned power grid companies (State Power Grid
Company and China Southern Power Grid Company) are in charge
of the power transmission, distribution and retailing of the whole
country. The transactions are under the government-regulated
tariffs which are controlled by National Development and Reform
Commission (NDRC) rather than by the market.

n
Corresponding author: Collaborative Innovation Center for Energy Economics
and Energy Policy, Institute for Studies in Energy Policy, Xiamen University,
Xiamen, Fujian, 361005, China. Tel.: +86 5922186076; fax: +86 5922186075.
E-mail addresses: bqlin@xmu.edu.cn, bqlin2004@vip.sina.com (B. Lin).

0301-4215/$ - see front matter & 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.enpol.2013.05.023

In China, the electricity tariffs often play an important role as

government policy tools, for example, the substantial crosssubsidies1 in retail power tariffs. Government charges differential
retail tariffs for industrial and residential end-users. For the
industrial sector, the power tariff is higher than its long-term

1
In China, industrial and commercial power tariff is higher than residential
power tariff, from which residents obtain subsidies. That is the cross-subsidies we
discuss in the paper. On one hand, cross-subsidies make residents enjoy the lower
tariffs. On the other hand, it also makes sure the power companies protable.
Besides cross-subsidies, there are also different forms of subsidy in power sector in
China, for example subsidies for wind or solar power generation and directallowance for lowest income families. Since cross-subsidies are not able to
differentiate the targeted groups, it is lack of efciency. The high income people
who consume more power enjoy larger subsidies than the low income ones who
consume less power. Therefore, to design targeted power tariffs is the key to power
tariffs reform in China.

742

C. Sun, B. Lin / Energy Policy 60 (2013) 741752

marginal cost (LTMC). The residential end-user sector, however,

usually pays a power tariff that is lower than its LTMC. The stateowned power companies ensure their budget balance by taking
the gains from industrial sectors to compensate the loss from
residential sector. In other words, the residential end-users enjoy
subsidies from the industrial sector (Lin and Jiang, 2012). Lin et al.
(2009a) found that in 2007 the amount of retail power crosssubsidies was about US$32.5 billion.
The Chinese government adopts a at pricing mechanism in
residential sector. It indicates that paying the same tariff as the
lower income ones, the higher income households usually consume more electricity and enjoy more subsidies. Lin et al. (2009a)
found that 45% of the subsidies were given to the higher income
groups (accounting for 27% of population) while the lower income
persons (accounting for 22% of population) only received the 10.1%
subsides. Therefore, the at tariff for residential power consumption is not consistent with social equity, and leads to a lack of
allocation efciency in cross-subsidies.
Under the background of energy conservation and emission
control, China is promoting the reform of residential tariff for
making it more sensitive to price signals. It is now clear for the
government that a more active tariff reform is required to reect
the power supply cost better and to improve efciency2.
In November 2009, the government puts forward a plan for
reforming residential tariff with an intention to apply a rising
block tariff mechanism, in which the residential tariff will vary
according to the consumption levels. The reform plans a three-step
tariff system assigned to three consumption thresholds: basic
power consumption (BPC), normal power consumption (NPC),
and luxury power consumption (LPC). Under the plan for block
tariffs, the BPC will be charged at a subsidized price; the price will
be higher for any NPC exceeding the BPC limit; there will be a
further price increase at LPC thresholds.
The residential electricity block tariffs have been extensively
adopted in other countries, such as the United States and Japan3.
So the residential power tariff reform would update Chinese tariff
policy by meeting the global standards. The power tariffs are seen
as an important tool to achieve the objectives of social equity and
efciency. First, the rising block tariffs of residential electricity can
increase the economic efciency by charging a price that follows
the marginal cost pricing. Basically, residential electricity consumption tends to happen more around the peak time which adds
to the peak network capacity. Higher peak capacity leads to higher
marginal supply cost. The cost information cannot be reected in
the at tariff, while the rising block tariff can reect the increased
marginal cost. Second, the rising block tariff of residential electricity follows the inverse elasticity pricing, which means that
different prices are set for consumers with various price elasticities of demand. The consumers with a lower price elasticity of
demand have to purchase the electricity at a higher price than
those with a higher price elasticity under the rising block tariff
mechanism. The at tariff charges uniform price for each unit of
electricity consumption. In contrast, the rising block tariffs will be
able to identify the different responsiveness of demand for
electricity facing price changes. We can use Ramsey pricing to
analyze the above features of block tariffs.
Accordingly, estimating the price elasticity of electricity
demand for consumers of different incomes should be crucial for
the research on the rising block tariffs. We present a dataset for

2
The efciency here means a broad term containing economic, social, and
environmental objectives.
3
Source: http://www.tepco.co.jp/e-rates/individual/menu/home/home02-j.
html. Although the United States and Japan have power trading markets like PJM,
their tariffs of residential electricity are nearly in accordance with the forms of
increasing block tariffs.

the residential electricity consumption of China's 30 provinces

over the years 20002007, and employ a translog model of
residential power demand to estimate the price and income
elasticity.
The remainder of this study is organized as follows. Section 2
reviews the previous research. Section 3 describes China's fundamental conditions on residential electricity consumption, residential electricity tariff and income per capita. Section 4 analyzes the
inverse elasticity pricing model and residential utility curve.
Section 5 empirically studies the income and price elasticity of
residential electricity demand by a panel-data translog model.
Section 6 puts forward three proposals of the block tariffs for
residential electricity. Section 7 summarizes this study along with
future extensions.

2. Literature review
In this section, we include two parts. We rst summarize the
existing empirical researches on price elasticity of residential
electricity demand. In the second part, we summarize and compare the applications of inverse elasticity pricing approach in
public utility.
There are a lot of studies on demand functions of residential
electricity aiming at not only the developed countries, but also
some developing countries. Taylor (1975), Halvorsen (1975) and
Bohi (1981) conducted early research on the price elasticity of
residential electricity demand and nished the related work of
data classication. Bohi and Zimmerman (1984) found that the
short-term and long-term price elasticity of residential power
demand was 0.2 and 0.7, respectively. Narayana et al. (2007)
highlighted that the residential power demand is price elastic and
income inelastic in G7 countries. Garcia-Cerrutti (2000) pointed
out the minute impact of American household income on residential power demand based on the estimation of elasticity from
panel data. Silk and Joutz (1997) analyzed the impact of economic
variables, such as actual disposable income, power tariffs, temperature and real rate of interest, on residential power consumption. Filippini (1999) estimated a log-linear model employing
aggregated data referring to 40 cities in Switzerland between
1987 and 1990, and found that the price elasticity of residential
power demand was 0.3. The Australian National Institute of
Economic and Industry Research (2007) found that the price
elasticity of residential power demand was 0.25. Dilaver (2008)
investigated the interrelationship between the residential power
tariff, the household power consumption and the total household
expenditure in Turkey using the structural time series model.
Wasantha and Wilson (2009) found the value of income elasticity
of power demand was between 0.32 and 0.78 and the value of
price elasticity of demand was between 0.62 and 0.16 in Sri
Lankan. Bose and Shukla (1999) made a panel regression of the
parallel data of 19 states in India over 9 years to nd the estimated
result of the price elasticity of Indian residential electricity consumption. Dividing the samples by seasons, Filippini and Pachauri
(2004) took the income, power tariff and dummy variables that
reect household conditions as the independent variables into the
research. Holtedahl and Joutz (2004) studied the impact of household disposable income, population growth rate, residential power
tariff, urbanization rate and temperature condition on the power
demand of Taiwan residents.
However, there are few empirical analyses focusing on the
electricity demand of Chinese households who make up about a
quarter of all households in the world. We conduct a research on
estimating price and income elasticity of residential power
demand in China. Instead of adopting a loglinear model as most

C. Sun, B. Lin / Energy Policy 60 (2013) 741752

previous literatures do, we use a translog model to capture the

income effect on price elasticity.
With respect to cost recovery, social equity, energy efciency,
and emissions reduction, the reform to implement block tariffs is
expected to be a solution to deal with all concerns. In a perfectly
competitive market, marginal cost pricing is the best method to
maximize social welfare (Hotelling, 1938). Nevertheless, due to the
existence of natural monopolies in some public utilities, marginal
cost pricing will lead to prot losses and decrease the aggregated
welfare, requiring necessary government regulation. In terms of
economic efciency, Ramsey pricing as a second-best pricing
mechanism succeed in maximizing the total welfare subject to
the breakeven of public utilities regulated by government
(Boiteux, 1956; Egbert, 1991). Boiteux (1956) brought about the
RamseyBoiteux inverse elasticity pricing rule, which indicated
that the price markup was inverse to the price elasticity of
demand. Subsequently, the rule was further developed by
Baumol and Bradford (1970) and Mirrlees (1976). The inverse
elasticity pricing mechanism used to be commonly applied for
pricing the products of public utilities, such as power, railway, tap
water, etc. In the American generation market, the inverse elasticity pricing rule was implemented to have customers with
different price sensitivities face different prices (Berry, 2000,
2002). In the American public water service, the formulation of
the water block tariffs mechanism was based on marginal cost
pricing and inverse elasticity pricing rule (Kim, 1995). Matsukawa
et al. (1993) conducted empirical research of Ramsey pricing in
Japanese electricity utilities based on the estimation of electricity
cost and demand functions. Meanwhile, Sueyoshi (1999) examined
how much the current tariffs of Japanese electric power deviated
from the marginal cost pricing and Ramsey pricing. Faruqui and
George (2005) pointed out that the price gap between two price
levels in the proposal reecting the structure of the life-line power
tariff might be larger than the one only reecting the power
supply cost. Haney, et al. (2009) highlighted that reasonable tariff
structures would encourage efcient electricity use and reduce
CO2 emissions.
For setting electricity tariffs, the Ramsey pricing rule has been
replaced by some other power trading mechanisms in current
operations. Sueyoshi and Tadiparthi (2007), (2008), Sueyoshi and
Goto (2009), Goto and Sueyoshi (2009) and Sueyoshi (2010)
described how demand and supply sides determined an equilibrium point to determine the market price of electricity, considering a speculation of traders, a line limit, demand forecasting,
trader's collaboration, their learning processes and many other
factors that determined the price of electricity in the United States
and Japan. However, China's power trading arrangements can
hardly provide enough information to propose a marketization
reform currently. We conduct an inverse elasticity pricing
approach when alternative approaches that match the Chinese
situation are not readily available. Compared to the at pricing, we
put forward the rising block tariff proposals and evaluate the
advantages of this pricing mechanism.

743

expenditures of electricity per capita amounted to US$21.3,

accounting for 1.45% of disposable personal income and 1.64% in
consumption expenditure per capita6.
We analyze the residential electricity consumption, the residential electricity tariff, and the income per capita over the 30
provinces in 20077. According to their ranking of the gross
domestic product (GDP) per capita in 2007, the 30 areas are listed
on the horizontal axis of Fig. 1. It shows that provinces with higher
personal income on average consume more residential electricity
per capita. The bar represents the residential electricity consumption per capita in each area. There are relatively large disparities in
the income and power consumption per capita of these provinces.
In the top ten regions, except Shandong, each region's residential
electricity consumption per capita is more than 300 kW h, and six
of them even exceed 400 kW h. The residential electricity consumption per capita is less than 200 kW h in the nine regions with
the least GDP per capita. However, as the curve demonstrates, the
residential electricity expenditure maintains a stable proportion in
personal consumption expenditure on average, mostly between
1.5% and 2.0%. Therefore, the disparities in income exert impact on
residential electricity consumption per capita, while there is a
relatively stable interval for the proportion of residential electricity expenditure.
Using the data of China's 30 provinces over 20002007, we
further describe the relationship between the residential electricity consumption, residential electricity tariff and personal
income. The horizontal axis of Fig. 2 represents the residential
income per capita based on the constant price in 2000, and the
vertical axis represents the residential electricity consumption per
capita. The 240 sample points reect a highly positive correlation:
the higher the income level is, the more the electricity is
consumed.
Fig. 3 describes the relationship between the residential electricity consumption and the power tariff. The horizontal axis
expresses the average tariff of residential electricity, and the
vertical axis expresses the residential electricity consumption per
capita. Evidently, the distribution of 240 sample points appears
nearly random, which hardly represents an obvious correlation
between the power tariff and consumption. According to the
ranking of income, the 240 sample points are divided into three
groups. It exhibits a relatively explicit negative correlation in the
lower income group, which means the power consumption comes
down when the power tariff increases. In the middle income
group, the electricity tariff and consumption exhibit a quite weak
negative correlation. Furthermore in the higher income group, the
correlation is insignicant. To sum up, compared to the impact of
power tariffs on power consumption, the impact of income on
power consumption is likely to be much greater. Moreover the
impact of residential electricity tariff on power consumption is
more explicit in the lower income consumers, which is similar to
the result of Holtedahl and Joutz (2004) on Taiwan.

4. The inverse elasticity pricing and residential utility curve

3. Residential electricity consumption in China
In 2009, the households consumed 457.1 billion kilowatt-hours
(kW h) electricity, accounting for 12.5% of the country's total
power consumption4. China's residential electricity consumption
per capita was 303.8 kW h, and the average retail tariff of residential electricity was US$70.0 per thousand kW h5 in 2009. The
4
5

Data source: Statistics express of national electric industry 2009.

Data source: Regulation report of power tariff implementation 2009.

4.1. Inverse elasticity pricing and non-linear pricing

Non-linear pricing like two-part pricing8 or multistep pricing
describes that there is a non-linear relationship between the
6

Data source: China Statistical Yearbook 2010.

Data source: China Energy Statistical Yearbook 2008, Regulation report of
power tariff implementation 2008.
8
The two-part pricing is a pricing mechanism in which the xed cost and
marginal cost are respectively charged on the condition of payment balance.
For instance, two-part pricing is used in the industrial electricity by capacity
7

744

C. Sun, B. Lin / Energy Policy 60 (2013) 741752

Fig. 1. Chinese provinces residential electricity consumption in 2007.

Fig. 2. The residential electricity consumption and the income per capita.

consumption and the aggregated expenditure. Conversely, the at

pricing indicates that the total expenditure is proportional to
consumption (see Fig. 4). The block tariff mechanism is a kind of
non-linear pricing, which assigns different power tariffs for several
levels according to the amount of the power consumption.
According to the inverse elasticity pricing rule, the price
premium to marginal cost is proportional to the reciprocal of the
price elasticity of demand (Resende, 1997). The equation is given as,
P m MC m

1


1 m
Pm

where Pm represents the average price of commodity m, and MCm is

the marginal cost of commodity m. m represents the price elasticity.
is the Lagrange multiplier, and =1 is called Ramsey multiplier and
it is a constant lower than 1.
In accordance with non-linear pricing theory (Wilson, 1996),
due to the heterogeneity of consumers, their consumption varies
even when they are assigned the same price. Therefore, the price
elasticity m is not only associated with the function concerning
the price but also related to the type of the consumer (Brown and
(footnote continued)
power tariff and quantity power tariff, and the mobile communication price is
divided into a monthly rental fee and call charge.

Sibley, 1986). Then the problem is given by:

P m MC m

1


1 m P m ;
P m

where, denotes the type of the consumer. With the different

purchasing power and consumption preference, the different
consumers demand functions are different. In other words, the
demand Qm is a function related to . Simultaneously, the relation
between Pm and is given by the inverse demand function P m Q m .
With respect to the obvious impact of income on electricity demand,
the article classies the heterogeneous consumers by their income.
Due to the different electricity demands, the different incomes will
also lead to different price elasticities m P m ; . This is crucial for
the formulation of the block tariffs research (Kim, 1995).
Through adding up the various consumers surplus and producers surplus in accordance with the consumers heterogeneity, the
article obtains the inverse elasticity pricing equation for different
consumers (Wilson, 1996). The equation is given as Kim (1995):
P 1 MC 1
P 2 MC 2
P MC
 1
 2

P1
P2
P

where, 1; ; N denotes the consumers type; P represents the

average price aiming at different type of consumers; MC denotes
the marginal cost; and denotes the price elasticity of demand.

C. Sun, B. Lin / Energy Policy 60 (2013) 741752

745

Fig. 3. The residential electricity consumption and the electricity tariff.

R represents the Ramsey multiplier. We substitute P into

Eq. (4), and we obtain the equation (Qi et al., 2010):



MC  j j 1j j k R
FC
6

j jR
j j

Total expenditure

Two part pricing

Non-linear pricing
Flat pricing

where k Q =P j j

Non-linear pricing
Two part pricing

4.2. The residential utility curve

Block tariffs

Block tariffs

Flat pricing

Consumption
Fig. 4. The at pricing and the non-linear pricing.

Due to the cross-subsidies, the residential power tariff is lower

than its LTMC in China. We apply the inverse elasticity pricing rule
to maximize the social welfare on condition of removing the crosssubsidies and ensuring the suppliers payment balance.
P MC  Q FC 0

Eq. (4) reects the payment balance of power suppliers, where

Q denotes the consumption of and FC denotes the xed cost.
Based on Eq. (2), the price for is given by,
P

MC  j j
j jR

We take the q1 and q2 as the dividing points to divide the

residential power consumption into three levels, basic power
consumption (BPC), normal power consumption (NPC), and luxury
power consumption (LPC). Three power tariffs, p1p2 and p3 are
set for these three blocks, respectively. In other words, when the
consumption is less than or equal to q1, the power tariff is p1;
when the consumption is in the interval from q1 to q2, the tariff of
the consumption above q1 is p2;and when the consumption is
more than q2, the tariff of the consumption above q2 increases
to p3. Eq. (7) indicates the relationship between residential power
consumption q and the residential expenditure E.
8
q  p1
q q1
>
<
q1  p1 qq1  p2
q1 o q q2
E
7
>
: q  p q q  p qq  p
q2 o q
1
1
2
1
2
2
3
According to Eq. (4), the average price corresponding to
each type of consumer should be higher than the LTMC. But
affected by various factors, such as the equipment utilization

746

C. Sun, B. Lin / Energy Policy 60 (2013) 741752

Fig. 5. The impact of residential electricity block tariffs on residential utility.

ratio9 and the line loss of power transmission and distribution, the
LTMC of residential power is always much higher than the current
power tariff. Lin et al. (2009a) estimated that the LTMC of
residential power is US$0.154 per kW h in 2007. Therefore, if the
tariff of BPC is higher than this standard, it will make the residential
electricity unaffordable for the poorest. With respect to the developed countries experience, the tariff of rst-step power consumption is lower than its LTMC in the rising block tariff mechanism.
Drawing on Taylor's (1975) description of indifference utility
curve, Fig. 5 reects the different residents utility situations in the
at tariffs and the proposed rising block tariffs. The horizontal axis
represents the residential power consumption, and the vertical
axis represents the consumption of other goods. As shown in
Fig. 5, q1 and q2 divide the residential electricity consumption into
three parts: BPC, NPC and LPC. The line AB displays the current
budget constraint line before implementing the rising block tariffs
(with the at tariffs, the budget constraint line is a straight line).
The slope of line AB expresses the ratio of at tariff to average
price of other goods (p0/p). There are two kinds of consumers.
With relatively lower income and less power consumption, the
type I consumers indifference curve MN is tangent to AB at L.
With relatively higher income and more power consumption, the
type II consumers indifference curve SR is tangent to AB at T10.
We assume two block tariff structures: AFGH and ACDE. The slope
of AF is larger than that of AB, so the rst-step power tariff in the
9
The residential power consumption is characteristic of obvious gap between
on-peak and off-peak. The peak load units only turn on for providing the peak
electricity supply. Take the peak load unit in pumped-storage power station as an
example, its utilization rate is 1100 hours annually, around one third of the unit in a
common hydroelectric power station.
10
In fact, because the income of type I and type II consumers are different,
their budget constraint lines won't always be the same. But our analysis does not
aim to compare the utility between the type I and type II consumers. Hence, using
the line AB to represent their budget constraints lines could simplify the gure, but
not make any impact on the result.

AFGH proposal is higher than p0 (based on the assumption that the

average price level of other goods is constant). On the contrary, the
slope of AC is smaller than that of AB, so the rst-step power tariff
in the ACDE proposal is lower than p0. The slope of CD is equal to
that of FG. And both of them are larger than that of AB so that both
of the second-step power tariff in the two proposals are the same
and higher than p0. The slope of DE is equal to that of GH, so the
third-step power tariffs are the same. The only difference of both
proposals is the rst-step power tariff. But we will nd that just
this little difference creates the obvious improvement of the utility
with the ACDE proposal.
No matter which type they consume, their budget constraint
line AB is superior to AFGH, which indicates that without the block
tariffs the indifference curve will always be tangent to the budget
constraint line at a point with a higher utility level. So the proposal
AB is better for improving the consumer's welfare than AFGH.
Then, if we modify the proposal AFGH into proposal ACDE by
reducing the rst-step tariff, the utility will be improved and even
superior to the proposal AB. As for the type I consumers, their
electricity consumption is less than q1. While the slope of AC is
lower than that of AB, the lower power tariff of AC will promote
their quality of life because, that type I consumers with low
income make more use of electricity. In other words, the budget
constraint line AC will be tangent to the higher indifference curve
M1N1, so that the utility level of type I consumers improves. As for
the type II consumers, compared with the curve AB, the rst-step
AC and a part of second-step CD are above AB. So the indifference
curve SR climbs to S1R1 to be tangent to curve CD at T1. It means
that the reduction of power consumption from T to T1 can promote
the utility of type II consumers. To sum up, Fig. 5 suggests that the
ACDE proposal with relatively low rst-step power tariff is much
better. As Taylor (1975) said, based on consumer's indifference
utility curve, the reasonable block tariffs of electricity can improve
the utility level of most people to promote social welfare.

5. Empirical analysis on elasticity of residential electricity

demand
5.1. Demand model and data
To evaluate the price elasticity and income elasticity of residential power demand, the traditional empirical approach estimates a standard log-linear demand model using aggregate level
data (Narayana et al., 2007; Silk and Joutz, 1997). Although the
standard log-linear model is easy to implement in order to obtain
the elasticity, the constant coefcients that just present the
average demand elasticity among the full sample and hardly
provide any information on personal heterogeneity. From Fig. 3,
the higher income consumers are less sensitive than those with
lower income under the at tariff, which implies that the price
elasticity, instead of being constant, is likely to vary with the
residential income levels.
Croissant (2000) and Liu (2012) modeling the demand function
use a translog form, which is more exible to capture the effects of
price and income on elasticities. The translog demand model is
given as:
lnC it i 1 lnI it 2 2 lnP it 2 1=23 lnI it  lnP it T it

where, Cit is the residential power demand per capita in region i on

year t; Iit represents the personal income; Pit denotes the residential electricity tariff; T is the time variable; i represents intercept
and regional effects. The price elasticity P;it and income elasticity
I;it of each residential power demand can be derived as:
P;it 22 lnP it 1=23 lnI it

C. Sun, B. Lin / Energy Policy 60 (2013) 741752

Table 1
Results of panel data unit root test.

Variables Method
lnC it

lnI it

lnP it

LLC
Fisher
ADF
LLC
Fisher
ADF
LLC
Fisher
ADF

P-value of original
series

P-value of rst order differential

series

0.3258
0.9999

0.0000
0.0000

1.0000
1.0000

0.0000
0.0004

0.1568
0.0781

0.0000
0.0008

Note: LLC (Levin, Lin and Chut) is a homogeneous unit root test. FisherADF is a
heterogeneous unit root test.

I;it 21 lnI it 1=23 lnP it

10

The data is based on the residential power consumption, the

residential electricity price and the income per capita during the
period 20002007 of China's 30 provinces (except Tibet and
Taiwan)11. Similarly, Bose and Shukla (1999), Filippini (1999),
Garcia-Cerrutti (2000) and Narayana et al. (2007) studied the
elasticity of residential power demand using the provincial or state
panel data.
In order to avoid spurious regression, stationarity tests are
conducted before the regression. The result (Table 1) shows that all
the variables are I(1) under the 1% signicance level. Furthermore,
the Pedroni Test (Pedroni, 1999; Pedroni, 2004) and Kao Test (Kao,
1999) are used to test the panel co-integration relation. The
result (Table 2) manifests that the statistic is signicant either by
Pedroni Test or by Kao Test (on the 1% signicance level), so the
co-integration relation is proved.

5.2. Empirical results and robustness check

The translog demand model (Model 1) using Fixed Effects
estimation is the baseline model. The empirical results (Table 3)
indicate that all the coefcients are signicant on the 95% condence interval. The interaction term (3) is positive, and the
quadratic terms (1 and 2) are negative. It shows that the income
elasticity of residential power demand I;it is negatively related to
the residential income and positively to the residential power tariff.
As for the higher income consumers, their residential electricity
demand is less responsive to the change of income. Decrease of
residential power tariff creates an increase in purchasing power.
Hence, either the reduction of power tariff or improvement of
income exerts a similar impact on income elasticity. The price
elasticity of residential power demand P;it is negatively related to
the residential power tariff and positively to residential income. Due
to the substitution effect of price on demand, the rising price leads
to the drop of demand, and the price elasticity of demand is usually
negative12. That is, the negative correlation between the price
elasticity and the residential power tariff reects that the higher
residential power tariff brings about the higher absolute value of
11

The residential power tariff slightly varies with time in each province.
Relatively, it is more effective to reect the disparity of power tariff by the regional
panel data. The power consumption per capita is from China Energy Statistics
Yearbook for related years. The residential power tariff data are adopted from China
State Electricity Regulation Commission (SERC), which are adjusted to the constant
2000 price by consumption price index of residential water and electricity of each
province. The income per capita data are from National Statistics Yearbook for
related years, based on the constant 2000 price by the provincial GDP deator.
12
The lower the price elasticity is, the higher its absolute value will be.

747

elasticity13. Similarly, the positive correlation between the price

elasticity and the residential income shows that the residential
power demand of those with lower income is more sensitive.
To explore the possibility of variation in price and income
elasticity and check the robustness of our results, we estimate
further demand models. First, to control for regional effects, we
adopt the urbanization level (Uit) and the climate conditions
TEM it as explanatory variables in Model 2. Second, we use
personal expenditure Expit instead of personal income to measure the income effects in Model 3. Thirdly, to avoid the specied
error, we estimate time effect by time dummy variables in
Model 4. Fourthly, we apply a standard log-linear form to evaluate
the average demand elasticity in Model 5.
The results are robust to these changes. The coefcients of the
interaction term are positive, signicant, and robust across Model
14, which indicates that the interaction effects of income is an
important determinant of price elasticity of residential power
demand. The residential power demand of high income person
is hardly sensitive to the changes of price. The coefcients of price
quadratic term is negative, which implies that the rising block
tariff for high income residents can help raise the awareness of
their energy conservation. The estimates of the average temperature reveal that a 1% increase in temperature is related to a 0.11
0.22% increase in residential electricity demand. The positive
estimates of the urbanization level indicate that individual electricity demand varies with the urbanization even when facing
constant income and price. The positive time effects imply a
upward trend of power demand. The results of Model 5 show that
the average income elasticity of demand is about 0.47 and the
average price elasticity of demand is about 0.12, which are close
to the results of existing studies (Wasantha and Wilson, 2009;
Bose and Shukla, 1999; Lin, 2003). Since the coefcients in Model
2 are more signicant than other translog models, we use them to
evaluate the price and income elasticity of demand.
Figs. 6 and 7 plot elasticity values for the 240 panel data points
provided by 8 years (2000 to 2007) of record for the 30 provinces. It
is found in Fig. 6 that the price elasticity drops or its absolute value
increases with rising power tariff and decreasing income. Like the
results of Model 5, most of the scatter points are distributed around
P;it 0:12. Moreover, in a segment of samples representing high
income and low power tariff, the values of P;it are above zero, which
is similar to demonstration of Fig. 3. Since the demand of the higher
income groups is less responsive to the current power tariff change,
their electricity consumption might be in accordance with inefcient
outcomes. If the power tariff for the higher income group appropriately rises, the absolute value of price elasticity will increase. The
energy-saving awareness of higher income consumers will be raised,
and the use of power will be more efcient. Reiss and White (2005)
put forward the same viewpoint on the increasing block tariffs of
electricity of California State in the United States. Fig. 7 illustrates that
the income elasticity I;it drops with rising income and falling power
tariff. Meanwhile, most of the samples are distributed around
I;it 0:47, which is similar to the average income elasticity estimated by Model 514.

6. The proposals of residential electricity block tariffs

Based on the reference of Lin et al. (2009a), the residents will
be divided into ve classes by income. According to the China
Population Statistics Yearbook 2008, we obtain the average
13
At a relatively higher tariff level, the demand will become more responsive
to the change of power tariff.
14
A scenario analysis of the elasticity turning conditions is shown in
Appendix 1.

748

C. Sun, B. Lin / Energy Policy 60 (2013) 741752

Table 2
Results of panel data co-integration test.
Pedroni Test:
Alternative hypothesis mode

Statistic

Statistical value

P-value

Results of homogeneous alternative test (common AR coefs)

Panel PP-statistic
Panel ADF-statistic
Group PP-statistic
Group ADF-statistic

2.695938
5.312438
8.308613
8.909563

0.0001
0.0000
0.0000
0.0000

Results of heterogeneous alternative test (individual AR coefs)

Kao Test:

t-Statistic
4.016517

ADF statistic

P-value
0.0000

Table 3
Summary estimators of Model 15.
Model 1
lnI it 2
lnP it 2
1=2lnIit lnP it

nn

0.0447
(0.0211)
0.1272nn
(0.0580)
0.3294nn
(0.1438)

Model 2

Model 3

nnn

0.0558
(0.0265)
0.2189nnn
(0.0698)
0.5261nnn
(0.1717)

Model 4

0.1099nnn
(0.0231)

0.0542
(0.0191)
0.1457nnn
(0.0520)
0.3793nnn
(0.1300)

0.0514nnn
(0.0118)
0.3207nnn
(0.0713)

ln EXP it 2
1=2ln EXP it lnP it
lnI it
lnP it
lnU it
ln TEM it
0.0942nnn
(0.0146)

Model 5

nnn

0.5357nnn
(0.1041)
0.1256nnn
(0.0389)
0.0314nnn
(0.0069)

0.0219
(0.0620)
0.1684nn
(0.0792)
0.1004nnn
(0.0080)

0.0404
(0.1024)
0.2244nnn
(0.1323)

0.4736nnn
(0.0371)
0.1182nn
(0.0502)
0.5363nnn
(0.0720)
0.1117nnn
(0.0266)
0.0333nnn
(0.0034)
Fig. 6. The price elasticity of residential power demand.

Note: Standard errors are reported in parentheses and coefcients with

n
are statistically signicant at the 1%, 5%, and 10% levels, respectively.

nnn nn

and

population of each family in income groups15. Then, the monthly

average residential electricity consumptions of households of
different incomes are estimated based on the condition of their
household electric appliances in Table 4.
We put forward three proposals according to the different
income groups (Table 5). Based on the analysis on the residential
utility curve (Fig. 5), the main discrepancy among these three
proposals lies in the rst-step tariff for the low income consumers.
The rst-step of Proposal 1 carries the most extensive coverage,
and the population exclusively sharing the favorable rst-step
power tariff (p1) will account for 73%. That of Proposal 3 makes
pointed references to the lowest-income groups, the BPC of which
just exclusively covers 22% of the population.
The residential electricity block tariff reform in China is going to
assign three-step tariffs, p1, p2 and p3 for the corresponding threestep power consumption, BPC, NPC and LPC. Considering the expenditure on electricity as Eq. (7), the article gures out the block tariffs
scheme (p1, p2 and p3) of each step subjected to ensuring the power
suppliers payment balance using Eq. (4), based on the data from the
annual report of China State Electricity Regulation Commission
(SERC). The blocks designing in detail is shown in Appendix B.
As seen in Table 6, the block tariffs and the corresponding
consumption of these three proposals are quite different,

15

Based on our assumption, there are 371 million households in China in 2007.
The difference between this value and the data from China Population Statistics
Yearbook 2008 is very slight (about 0.5%).

Fig. 7. The income elasticity of residential power demand.

especially the p1 and BPC16. In Proposal 1, the rst-step considers

the groups with annual income below US$1500 per capita. The
BPC ranges from 0 to 90 kW h per month, and the corresponding
power tariff is US$0.131 per kW h. In Proposal 2, the BPC up to a
limit of 60 kW h per month, and the corresponding power tariff is
16
The proposed tariffs shown in Table 6 would eliminate fully the 2007 level
cross-subsidies of around US$32.5 billionwhich is related to the difference
between the LTMC of US$0.154 per kW h and the actual average retail tariff.
Therefore, the tariff levels of these proposals are much higher than the actual tariff.

C. Sun, B. Lin / Energy Policy 60 (2013) 741752

749

Table 4
Population proportion and monthly power consumption in different income groups.

Income per capita annual (US $)

Population proportion (%)
Household monthly power consumption (kW h)
Persons per family

Lowest income

Lower income

Middle income

Higher income

Highest income

o 450
22
35
4.5

450750
26.5
60
4

7501500
24.5
90
3.5

15003000
18
150
3

43000
9
180
2.5

Table 5
Three proposals of the block tariffs based on different income groups.
Group

Proposal 1

Proposal 2

Lowest income
Lower income First-step (73%)n
Middle income
Second-step
Higher income
(18%)
Highest
Third-step (9%)
income

Table 8
The impact on residential power expense.
Proposal Step

Proposal 3
First-step (22%)

First-step (48.5%)

Second-step
(69%)

Second-step
(42.5%)
Third-step (9%)

Third-step (9%)

Note: The percentage in parentheses denotes the population ratio for each
income group.

Table 6
The rising block tariff structure of the three proposals.
3
Proposal Level Interval of consumption per month
(kW h)

Block tariffs (US

$/kW h)

BPC
NPC
LPC

090
90150
More than 150

0.131 (p1)
0.247 (p2)
0.402 (p3)

BPC
NPC
LPC

060
60150
More than 150

0.108 (p1)
0.252 (p2)
0.327 (p3)

BPC
NPC
LPC

035
35150
More than 150

0.052(p1)
0.244 (p2)
0.298 (p3)

Table 7
Results of sensitivity analysis of changes in LTMC (US$/kW h).
Change rate of
LTMC (%)

+20
+10
0
10
+20

Proposal 1

Proposal 2

Proposal 3

BPC

NPC

LPC

BPC

NPC

LPC

BPC

NPC

LPC

0.146
0.138
0.131
0.125
0.117

0.270
0.255
0.247
0.239
0.230

0.427
0.414
0.402
0.386
0.377

0.122
0.114
0.108
0.102
0.993

0.275
0.263
0.252
0.246
0.238

0.354
0.341
0.327
0.320
0.312

0.058
0.055
0.052
0.050
0.057

0.266
0.254
0.244
0.236
0.228

0.313
0.305
0.298
0.290
0.281

US$0.108 per kW h. In Proposal 3, the rst-step only considers the

lowest-income groups (accounting for 22%) whose income per
capita is below US$450 annual. The interval of BPC is from 0 to
35 kW h per month, and p1 is US$0.052 per kW h, which is lower
than the former at tariff. From the analysis result of the
residential utility curve, the Proposal 3 exhibits the highest likelihood of helping the poorest17.

17
Whether the residential tariff structure is in the light of these three
proposals or in a at rate of US$0.154 per kW h (instead of the actual average of
US$0.070 per kW h), the traditional cross-subsidies between residential and
industrial sectors are fully eliminated, i.e. the cross-subsidies for each tariff band
sum to zero. However, under different tariffs structures, there are various crosssubsidies for each tariff block. Under the at rate, the cross-subsidies for each tariff
block are zero. But under these three block tariffs proposals, the subsidies among

First-step
Secondstep
Thirdstep
First-step
Secondstep
Thirdstep
First-step
Secondstep
Thirdstep

Flat tariffs

Block tariffs

Power expense
(US $)

Ratio
(%)

Power expense
(US $)

Ratio
(%)

4.54
10.51

1.71
1.61

6.63
24.78

2.50
3.80

12.60

1.57

40.75

5.08

3.46
8.24

1.67
1.56

4.46
18.30

2.15
3.46

12.60

1.57

39.96

4.98

2.45
6.88

1.63
1.51

1.90
14.52

1.26
3.19

12.60

1.57

39.27

4.89

Furthermore, we carry out a sensitivity test by simulating the

tariff structures with different LTMC to check the robustness of the
results above. The sensitivity test is necessary for two reasons.
First, in China the ratio of coal-red power units to the total
installed capacity is approximately 80%, and the recent uctuation
of steam-coal price has a marked impact on the residential
electricity LTMC. Second, residential electricity LTMC is an important variable in the inverse elasticity pricing approach.
In Table 7, results show that when the LTMC varies by 10%, the
block tariffs of three proposals would only change by 36%, and
their variations are positively correlated. Furthermore, the tariff
structures remain unchanged. We also get similar results for LTMC
with a 20% change. The sensitivity test indicates that the variation
in LTMC only slightly inuences the quantitative results and our
proposals are robust.
After implementing the block tariffs, distinct power tariffs are
assigned to households with different incomes. The household
expenditure in electricity will change18. We estimate the ratios of
the electricity expenditure to total consumption expenditure for
the at and block tariffs, respectively. Table 8 shows that in the at
tariffs, the ratios for residents with different income levels are very
close. Furthermore, the ratio of higher income residents is even
slightly lower than that of the lower income residents. However,
when proposed rising block tariffs are set, consumers with higher

(footnote continued)
each block still exist. Generally speaking, there are negative subsidies for highincome people, since their average price is higher than US$0.154 per kW h. The
low-income group with a block price below US$0.154 per kW h would obtain
positive subsidies.
18
Because of the substantial cross-subsidies, the current residential power
tariffs are below the breakeven. The block tariffs analysis are based on the
breakeven prices, assuming that the cross-subsidies has been removed. This is
the reason why power expenses for low income groups under proposed tariffs
should have higher power expenses than that under the current at power tariff.

750

C. Sun, B. Lin / Energy Policy 60 (2013) 741752

Table 9
The impact on saving energy and mitigating emissionsa.
Proposal 1 Proposal 2 Proposal 3
Electricity conservation (billion kW h)
43.71
Reduction of CO2 emission (million tons) 40.73

33.86
31.55

38.69
36.05

a
Since the tariff levels under the three proposed rising block structures are
higher than the actual rate, some residents become more sensitive and try to
reduce their own power consumption. Assuming that the energy structure of
power generation does not change, the carbon dioxide emissions embodies in per
kW h power will be constant. Therefore, a small level of power consumption
indicates a reduction of emissions.

income should pay a higher proportion of power expenditure in

their total consumption expenditure. It is noteworthy that in
Proposal 3, the power expenditure proportion of the lowestincome group is lower compared to that with at tariffs. In this
proposal, power consumption of lowest-income groups can be
raised without more expenditure so that their utility level can be
directly improved. Lin and Yang (2009b) demonstrated that the
increasing power consumption of lower income group has a
clearly positive impact on advancing the sustainable development
of society.
Witte and Marques (2010) pointed out in absence of clear and
structural incentives the average efciency of the utilities falls in
comparison with utilities which are encouraged by incentives. In
the at pricing, the residential electricity demand of the higher
income group is less responsive to the changes of electricity tariff.
But in the proposed block tariffs, there is an obvious improvement
that the price elasticity will reach 0.19, which implies that the
higher income groups are motivated to respond to the increased
electricity tariff. The experience of the developed countries shows
that the consumer will be affected as the power expenses
accounted for over 5% (Lin et al., 2009a). In Table 8, the proportion
of power expense to the total expenditure in highest income group
approaches about 5% indicating that the highest income consumers have an energy-saving impetus to conscientiously reduce the
unnecessary electricity consumption. Therefore, the mechanism of
the block tariffs helps to reallocate electricity resources among the
different income groups and improve energy efciency.
Narayana et al. (2007) suggested that pricing policies in developed countries were able to improve the power efciency and curb
carbon emissions. The implementation of rising block tariffs has a
certain effect on emissions reduction. The amount of residential
power consumption reached 362.27 billion kW h in 2007. If Proposal
1 is carried out, almost 43.71 billion kW h will be cut down as well as
40.73 million tons of carbon dioxide in Table 9. As for Proposal 2,
almost 33.86 billion kW h will be cut down as well as 31.55 million
tons of carbon dioxide. If we implement Proposal 3, almost 38.69
billion kW h will be cut down as well as 36.05 million tons of carbon
dioxide. To sum up, the block tariffs are able to curb carbon emissions
10.712.1% of residential electricity sector (Lin and Sun, 2010).
Although the amount of carbon emission reduction is not signicant
in comparison with the huge increment of China, the block tariffs still
have a incentive effect on carbon saving.

7. Conclusion and future extensions

The Chinese government controls the residential power tariff
for 25% of world households. But this government-controlled at
power tariff is under criticism in terms of efciency and equity. By
estimating the price elasticity of residents for power consumption,
we propose three block tariff structures for the reform of residential electricity tariffs in China. We adopt a translog demand model
to capture the impacts of income on price elasticity. We nd that

the average price elasticity is around 0.12 and the sensitivity of

high-income residents to price variation is lower than that of lowincome residents. In addition, we nd that the results are robust.
We provide three proposals for three different income structures of residents. Our analysis indicates that increasing the tariffs
of higher income groups could stimulate their awareness of energy
conservation. Compared with the at tariffs, rising block tariff can
increase the efciency of subsidies and provide the opportunity for
CO2 emission control. Having compared the three proposals, we
nd that it is crucial to determine the rst-step power tariff, and
setting a life-line tariff for the neediest could guarantee an
affordable minimal level of electricity consumption and lift their
living standards.
For countries like China, policy makers should take social stability,
affordability, fairness, energy efciency as well as cost recovery into
consideration (Jiang and Lin, 2012). Thus, the rising power tariff
reform as we have proposed is necessary for realizing this goal.
However, our research is only an initial study on the rising
block tariff reform of China. Many further issues are still left to be
analyzed.
First, in addition to the poorest consumers, the higher income
consumers also share the rst-step subsidized tariff, which leads
to certain welfare leakages. Nevertheless, the loss can be decreased
by applying appropriate policies targeting at the intended consumers. Depending on the regulation requirement, there could be
many tariff designs. For instance, in view of efciency and social
equity, the tariff schemes could be much more detailed. In another
example, assuming the acceptance of residents of the reformed
tariffs, the constraint of budget balance could be slightly relieved
by reducing instead of totally eliminating the cross-subsidies.
Another problem with the block tariff structure is penalization
of the residential customers in a family with many members,
which will affect social equity. The practical solution currently is to
allow the household to install more than one electricity meters or
enjoy the extra allowance from government.
Furthermore, the power tariff reforms in China cannot be
achieved too fast, and the progressive reform should be appropriate
for China's development. Given the various conditions of residents
income and current power tariffs among regions, the residential
electricity block tariffs should be in line with local conditions rather
than being imposed uniformly in the whole country.

Acknowledgements
The paper is supported by National Social Science Foundation
of China (Grant No. 12&ZD059), Ministry of Education Foundation
of China (Grant Nos. 10GBJ013, 13YJC790123 and 12JJD790027),
National Natural Science Foundation of China (Grant Nos.
71203186, 71073131 and 71203187), Fundamental Research Funds
for the Central Universities (Nos. 2010221051 and 201122G008),
Social Science School of Xiamen (Grant No. [2013]29).

Appendix A. Scenario analysis

As indicated in Eqs. (9) and (10), Figs. 6 and 7 plot elasticity
values for the 240 panel data points provided by 8 years (2000 to
2007) of record for the 30 provinces. The scenario analysis below
is supported by further investigations into price and income
elasticities which fall to the sensitive or unusual values.
Price elasticity varies with the power price and personal
income. If one's personal income is around US$1000, his price
elasticity of power demand would change from 0.1 to 0 when the
electricity price drops from US$0.080 to US$0.064 per kW h in
Fig. A1. When the electricity price drops below US$0.064 per

C. Sun, B. Lin / Energy Policy 60 (2013) 741752

751

are programmed under GAMS (The General Algebraic Modeling

System) version 22.1.
First, it is important to dene the sets of residents of different
income levels in these three proposals. The ve income-level residents
(Lin et al. 2009a) shown in Table 4 are remerged into three groups
(rst-step, second-step and third-step) for each proposal. For simplicity, we will not clarify each proposal subscript in the following
derivations. Table 5 shows the proportion (i) of each group (i). The
average income (Ii), the original household power consumption level
q0i and the average number of people per family (ni) in each group
could be estimated. Then we obtain the original total amount of
residential power consumption (Q0).
Q 0 Q 0i

11

Q 0i q0i 
Fig. A1. Scenario analysis of price elasticity.

i  N
ni

12

here, Q 0i denotes the original total amount and group amount of

residential power consumption, and N represents the total population.
Second, in order to fully eliminate the cross-subsidies, the block
tariffs should ensure the suppliers payment balance, as shown in
the equation below.
P i MC i  Q i FC 0

13

here, Pi and MCi denote the average power price and marginal cost
of each group. Qi denotes new group amount of residential power
consumption. FC is the xed cost. The equation implies that total
revenue equals total costs.
Thirdly, the inverse elasticity pricing method is written as
follows.
P i MC i 
 i j R
Pi
Pi

Fig. A2. Scenario analysis of income elasticity.

kW h, his price elasticity will turn positive. This unusual feature

implies this person is not willing to reduce his power consumption
in response to a price change. However, for another one whose
income is around US$3000, the price turning point of positive
price elasticity will rise to US$0.121 per kW h. It indicates that the
personal power consumption behavior differs by income levels.
Thus, evaluating the elasticity turning conditions for different
income groups should be important when we design the block
tariffs.
Like price elasticity, income elasticity also varies with the
power price and personal income as shown in Fig. A2. It is found
that when the electricity price is about US$0.100 per kW h, the
income elasticity is 0.3, 0.4 and 0.5 for the resident whose income
is around US$3500, $1500 and $600, respectively. With an unusually low income elasticity value (below 0.5), the residential
power consumption will be insensitive to income change. Therefore, it is important to pay attention to this insensitive group and
try to make them more sensitive when we design the block tariffs.

MC i  ji j
ji jR

This section presents the derivation of the tariff values and

consumption levels based on the inverse elasticity pricing method
and the conclusions of Taylor (1975) utility curve. The equations

15

here, i represents the price elasticity for each group, and R is the
Ramsey multiplier.
Following Kopsakangas-Savolainen (2004) and Qi et al. (2010),
we introduce a scaling term ki.
ki Q i  P i ji j

16

Fourthly, according to the results of Model 2, we obtain the

relationship among price elasticity, income and power tariff.
i 0:4378  lnP i 0:2631  lnI i

17

The above equation is important when we design the block

tariffs, since the original insensitive and higher-income group
should be more sensitive to the price signals. In particular, under
new tariffs structure they will not have an unusual price elasticity
as before.
Fifthly, we consider the household power expenditure of each
group (Ei).
E1 Q 1  p1
E2 q1 
E3 q1 

Appendix B. Blocks designing

14


2  N
N
 p1 Q 2 q1  2
 p2
n2
n2

18
19


3  N
N
N
 p3
 p1 q2 q1  3
 p2 Q 3 q2  2
n3
n3
n2

20
here, Pi denotes the each block tariff corresponding to three-step
power consumption, BPC, NPC and LPC. q1 and q2 are the thresholds levels of consumption which represent the second and third
steps of the tariff blocks.

752

C. Sun, B. Lin / Energy Policy 60 (2013) 741752

For each group, Pi is determined by Ei and Qi.

E
Pi i
Qi

21

References
Australian National Institute of Economic and Industry Research (2007). The Own
Price Elasticity of Demand for Electricity in NEM Regions. Australian National
Electricity Market Management Company.
Baumol, W.J., Bradford, D., 1970. Optimal departures from marginal cost pricing.
American Economic Review 60, 265283.
Berry, S.K., 2000. Stranded costs, access charges, and Ramsey pricing in the US
electric utility industry. The Quarterly Review of Economics and Finance 40,
503517.
Berry, S.K., 2002. Generation search costs and Ramsey pricing in a partially
deregulated electric utility industry. Journal of Economics and Business 54,
331343.
Bohi, D.R., 1981. Analyzing Demand Behavior: A Study of Energy Elasticities. Johns
Hopkins Press.
Bohi, D.R., Zimmerman, M., 1984. An update on econometric studies of energy
demand. Annual Review of Energy 9, 105154.
Boiteux, M., 1956. Sur la gestion des Monopoles Publics astreints a lquilibre
budgtaire. Econometrica 24, 2240.
Bose, R.K., Shukla, M., 1999. Elasticities of electricity demand in India. Energy Policy
27, 137146.
Brown, S.J., Sibley, D.S., 1986. The Theory of Public Utility Pricing. Cambridge
University Press.
Croissant, Y., 2000. Fonction de demande et surplus: une estimation sur des
donnes de panel pour les transports urbains franais. conomie & Prvision
4, 5366.
Dilaver Z. (2008). Residential Electricity Demand of Turkey: A Structural Time
Series Analysis. Surrey Energy Economics Centre (SEEC), Department of
Economics, University of Surrey, Working Paper.
Egbert, D., 1991. The optimality of BoiteuxRamsey pricing. Econometrica 59,
99121.
Faruqui, A., George, S.S., 2005. Quantifying customer response to dynamic pricing.
The Electricity Journal 4, 5363.
Filippini, M., 1999. Swiss residential demand for electricity. Applied Economics
Letters 6, 533538.
Filippini, M., Pachauri, S., 2004. Elasticities of electricity demand in urban Indian
households. Energy Policy 32, 429436.
Garcia-Cerrutti, L.M., 2000. Estimating elasticities of residential energy demand
from panel county data using dynamic random variables models with heteroskedastic and correlated error terms. Resource and Energy Economics 22,
355366.
Goto, M., Sueyoshi, T., 2009. Economies of scale and productivity growth in
Japanese electricity distribution. Energy Policy 8, 31303138.
Halvorsen, R., 1975. Residential demand for electric energy. The Review of
Economics and Statistics 57, 1218.
Haney, B., Jamasb, T., Pollitt, M.G., 2009. Smart Metering and Electricity Demand:
Technology, Economics and International Experience. University of Cambridge
Electricity Policy Research Group Working Paper, EPRG0903, Cambridge Working Paper in Economics 0905, January.
Hotelling, H., 1938. The general welfare in relation to problems of taxation and of
railway and utility rates. Econometrica 6, 242269.
Holtedahl, P., Joutz, F.L., 2004. Residential electricity demand in Taiwan. Energy
Economics 26, 201224.
Jiang, Z.J., Lin, B.Q., 2012. China's energy demand and its characteristics in the
industrialization and urbanization process. Energy Policy 49, 608615.
Kao, C., 1999. Spurious regression and residual-based tests for cointegration in
panel data. Journal of Econometrics 90, 144.

Kim, H.Y., 1995. Marginal cost and second-best pricing for water services. Review of
Industrial Organization 10, 323338.
Kopsakangas-Savolainen, M., 2004. The welfare effects of different pricing schemes
for electricity distribution in Finland. Energy Policy 32, 14291435.
Lin, B.Q., 2003. Structural changes, efciency improvement and electricity demand
forecasting. Economic Research Journal 5, 5765.
Lin, B.Q., Jiang, Z.J., 2012. Designation and inuence of household increasing block
electricity tariffs in China. Energy Policy 42, 164173.
Lin, B.Q., Jiang, Z.J., Lin, J., 2009a. Targeting electricity tariff subsidy contributes to
enhance energy fairness and efciency. Finance Research 11, 118.
Lin, B.Q., Sun, C.W., 2010. Evaluating carbon dioxide emissions in international
trade of China. Energy Policy 38, 613621.
Lin, B.Q., Yang, F., 2009b. The effect of power industry on the sustainable
development of Chinese economy. World Economy 7, 313.
Liu W.W. (2012). U.S. Gasoline Demand and Policies: An Application of Semiparametric Modeling and Estimation. Dissertation. State University of New York
at Binghamton.
Matsukawa, I., Madono, S., Nakashima, T., 1993. An empirical analysis of Ramsey
Pricing in Japanese electric utilities. Journal of the Japanese and International
Economies 3, 256276.
Mirrlees, J.A., 1976. Optimal tax theory: a synthesis. Journal of Public Economics 4,
327358.
Narayana, P.K., Smythb, R., Prasad, A., 2007. Electricity consumption in G7
countries: a panel cointegration analysis of residential demand elasticities.
Energy Policy 35, 44854494.
Pedroni, P., 1999. Critical values for cointegration tests in heterogeneous panels
with multiple regressors. Oxford Bulletin of Economics and Statistics 61,
653670.
Pedroni, P., 2004. Panel cointegration, asymptotic and nite sample properties of
pooled time series tests, with an application to the PPP hypothesis. Econometric Theory 20, 597625.
Qi, F., Zhang, L.Z., Wei, B., Que, G.H., 2010. Study on electricity tariffs based on
Ramsey pricing theory. Power Demand Side Management 2, 2427.
Reiss, P.C., White, M.W., 2005. Household electricity demand, revisited. Review of
Economic Studies 72, 853883.
Resende, M., 1997. Ramsey pricing and regulator's social welfare weightsan
empirical application. Review of Industrial Organization 12, 413416.
Silk, J.I., Joutz, F.L., 1997. Short and long-run elasticities in US residential electricity
demand: a co-integration approach. Energy Economics 19, 493513.
Sueyoshi, T., 1999. Tariff structure of Japanese electric power companies: an
empirical analysis using DEA. European Journal of Operational Research 118,
350374.
Sueyoshi, T., 2010. An agent-based approach with collaboration among agents:
estimation of wholesale electricity price on PJM and articial data generated by
a mean reverting model. Energy Economics 5, 10251033.
Sueyoshi, T., Goto, M., 2009. Can environmental investment and expenditure
enhance nancial performance of US electric utility rms under the clean air
act amendment of 1990? Energy Policy 11, 48194826.
Sueyoshi, T., Tadiparthi, G.R., 2007. An agent-based approach to handle business
complexity in US wholesale power trading. IEEE Transactions on Power
Systems 3, 532543.
Sueyoshi, T., Tadiparthi, G.R., 2008. An agent-based decision support system for
wholesale electricity market. Decision Support Systems 2, 425446.
Taylor, L., 1975. The demand for electricity: a survey. The Bell Journal of Economics
6, 74110.
Wasantha, P.P.A., Wilson, C., 2009. Estimating short and long-term residential demand
for electricity: new evidence from Sri Lanka. Energy Economics 32, S34S40.
Wilson, R.B., 1996. Nonlinear pricing and mechanism design. Handbook of
Computational Economics 1, 253293.
Witte, K.D., Marques, R.C., 2010. Designing performance incentives, an international
benchmark study in the water sector. Central European Journal of Operations
Research 18, 189220.