Energy demand in China has grown at an
alarming rate over the past fifteen years.
In 1995, the Chinese economy consumed
33.25 quadrillion Btus of energy, and by
2011, that number had more than tripled to
109.62 quadrillion Btus (EIA, 2013). China
surpassed the U.S. in terms of greenhouse
gas emissions in 2006 and in terms of total
energy consumed in 2009. Though forecasts
vary about how quickly China’s energy demand
will grow in the future, even at half
of the recent rate, China’s energy use would
double in 18 years.
Existing forecasting models appear to
have underestimated China’s recent growth
in energy demand. For example, in its 2000
International Energy Outlook, the Energy
Information Administration of the US Department
of Energy predicted that China
would consume 55 quadrillion Btus of energy
in 2005, just 5 years later. China’s
actual consumption was nearly 25 percent
higher, at 67.92 quadrillion Btus. Similarly,
some academics have noted that one factor
contributing to the spike in oil prices in
2008 was higher than expected demand for
oil from China (Hamilton, 2013).
It is important to understand factors that
drive energy demand in China in order to
improve forecasts and to evaluate policies
that might alter the path of energy consumption.
Given the level of central planning
in the country, a first-order question is
how well neoclassical models of householdor
firm-level energy consumption apply to
China.
In this paper, we focus on residential en-
* Auffhammer: University of California Berkeley and
NBER, 207 Giannini Hall, Berkeley, CA 94730-3310,
auffhammer@berkeley.edu. Wolfram: University of California
Berkeley and NBER, Haas School of Business,
Berkeley, CA 947201900, wolfram@haas.berkeley.edu.
Acknowledgements: We thank Joshua Blonz for valuable
research assistance. All remaining errors are ours.
ergy consumption and investigate how income
growth, particularly among households
close to the bottom of the income distribution,
affects adoption of energy-using
consumer durables. Consumers in urban
China have recently acquired energy-using
assets at an alarming rate. For example,
there were 8 air conditioning units for every
100 households in 1995, and by 2009,
there were 106 units for every 100 households
(Auffhammer, 2014). Similarly, vehicle
ownership in urban China has risen
at almost 40 percent per year between 2000
and 2010, helping fuel China’s rapid growth
in oil consumption (China Statistical Yearbook,
2001 and 2011).
A previous literature has documented an
S-shaped relationship between household
income or expenditure level and ownership
of appliances, cars and other energy-using
assets (e.g. McNeil and Letschert, 2010).
The S-shape is consistent with decisionmaking
in which households at very low
levels of income do not allocate additional
income to acquire energy-using assets, but
past a certain threshold, households become
much more likely to use income gains
to acquire refrigerators, cars or electric water
heaters (Gertler, Shelef, Wolfram and
Fuchs, 2013).
This paper explores behavior consistent
the S-shape in China. We combine
province-level data on rural appliance penetration
levels between 1999 and 2010 with
province-level data on income distributions
over the same time period. We use across
and within province variation to investigate
the relationship between poverty rates,
controlling for the average income, and
residential energy consumption. In spite
of tremendous overall reductions, China’s
poverty alleviation has been uneven, so we
have reasonable variation in income distributions
across provinces and over time.