Household Responses to Winter Heating Costs: The Remarkably Inelastic Demand for Space Heating - Dylan Brewer
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Household Responses to Winter Heating
Costs: The Remarkably Inelastic Demand
for Space Heating
Dylan Brewer∗
May 25, 2021
Abstract
I conduct a survey that presents research subjects with hypothetical costs to adjust
their thermostats. I estimate responses to the cost of heating and analyze the causes
for heterogeneity in household demand for energy services using the survey results as
a complete-information baseline. I find that even at the highest price level, half of the
participants exhibit zero response to price. On average, a 100 percent increase in the
cost of heating the home induces a 0.31 to 0.97 degree Fahrenheit (0.17 to 0.51◦ C)
reduction in the winter heating level, corresponding to a -0.005 to -0.014 elasticity.
Further, I find that participants’ experimental behavior with complete information can
explain observed real-world temperature settings, suggesting a limited role for infor-
mational barriers or salience issues in energy-service demand heterogeneity. Inelastic
demand suggests that energy efficiency policies may have high returns and that central-
ized demand-response policies may be required to address winter energy emergencies.
Further, individuals with higher temperature preferences are more price responsive,
suggesting that increasing block pricing policies for energy may reduce energy con-
sumption while minimizing the regressivity of energy pricing.
Keywords: Energy demand, thermostat, heating, heterogeneity, temperature
∗
School of Economics, Georgia Institute of Technology, brewer@gatech.edu, 221 Bobby Dodd Way, Room
224, Atlanta, GA 30332. Thank you to Soren Anderson, Joe Herriges, Joe Hamm, and Matt Oliver for useful
discussion. I am grateful for comments by seminar participants at the AERE virtual summer meetings,
the College of Charleston, and Michigan State University’s Environmental Science & Policy Program and
Department of Economics. The Environmental Science & Policy Program at MSU provided funding for the
experiment in the paper.
1
1 Introduction
Recent extreme winter-weather events in the United States have brought renewed at-
tention to infrastructure and policies related to winter heating. For example, a 2019 polar
vortex event brought extreme cold to the Midwest, leading utility companies in Michigan
and Minnesota to urgently request households to voluntarily reduce thermostat settings to
avoid widespread natural gas outages. In 2021, another polar vortex resulted in the failure
of energy infrastructure in Texas, leading to electricity outages for millions of Texans, with
total cost estimates as high as $295 billion (Perryman Group, 2021). In both cases, utilities
and policymakers had few options to address residential demand for heating when the system
was threatened.
This article investigates the use of pricing policies to reduce consumption of energy for
heating. Prices can be a powerful tool for conservation and energy reliability. By raising
energy prices when energy is scarce, consumers are given an incentive to curb consumption
and relieve strain on the system. This insight has lead to the development of prices that
vary by time of day and are highest during peak energy demand periods, which has received
significant attention in the energy policy literature (see e.g., Filippini (2011), Thorsnes et al.
(2012), Jang et al. (2016), Azarova et al. (2020), and Belton and Lunn (2020)).
Central to this narrative is the assumption that individuals respond to changing energy
prices and trade off the costs and benefits of energy consumption. Recent empirical work
challenges this assumption. One study of electricity-use data finds that 44 percent of studied
households did not respond to prices at all (Reiss and White, 2005). An analysis of natural
gas billing data finds that both low- and high-income households do not respond to prices
(Auffhammer and Rubin, 2018). An experiment in which Swedish renters were switched
from landlord-pay to tenant-pay electricity shows that while average electricity consumption
decreased by 24 percent, two-thirds of the reduction came from just 20 percent of the studied
households (Elinder et al., 2017). Following an energy-efficiency retrofit program in New
Zealand, 84 percent of households reported increasing their thermostats after the program
while 16 percent reported no change in behavior (Howden-Chapman et al., 2009). In a time-
of-use pricing experiment with Irish households, Prest (2020) finds that consumer awareness
2
of price changes was more important than price levels themselves for reducing electricity
demand.
Price responsiveness is low or zero for many energy users but high for a select group. Why
do some individuals fail to respond to prices while other individuals cut energy consumption
drastically when prices increase? There are two main potential explanations for low price
responsiveness: behavioral heterogeneity and preference heterogeneity. Either non-price-
responsive individuals are uninformed or face high costs to monitor energy prices, or these
individuals know prices and rationally choose not to respond because of high valuation of
energy services. The literature suggests a number of behavioral responses or informational
barriers for energy use. For example, Ito (2014) finds evidence that energy users respond
to average rather than marginal prices. Schleich et al. (2013) found that providing detailed
information on the amount of electricity consumed reduced average electricity consumption
by 4.5%, but that for half of households the informational treatment had no effect. Jessoe and
Rapson (2014) argue that consumers do not know prices or face a high cost of determining
energy prices. Allcott and Rogers (2014) find that social comparisons impact energy use
and observe behavior consistent with short attention spans. Finally, Allcott and Taubinsky
(2015) argue that consumers do not pay attention to energy prices when choosing light bulbs.
This paper tests whether individual heating-choice behavior is consistent with house-
holds having full information about the cost of energy. I conduct and analyze a nationally
representative survey in which participants make choices about how high to set their ther-
mostat during the winter when told the hypothetical cost of doing so. In this setting, energy
costs are easy to understand, cost-free to monitor, and salient. The survey environment is
clean of any potential confounding factors such as unobserved energy efficiency, thermostat
or meter placement, and attrition bias which makes it difficult to interpret results from the
field. The results from the survey serve as a fully informed benchmark to compare to real
temperature-setting behavior. If hypothetical temperature-setting behavior matches real
temperature-setting behavior, this provides evidence for heterogeneous preferences as the
primary driver for energy-use heterogeneity. If hypothetical temperature-setting behavior
differs from real temperature-setting behavior, this provides evidence for behavioral biases
or informational barriers as determinants of energy-use heterogeneity. The contribution of
3
a stated-choice survey is the ability to eliminate researcher uncertainty due to the potential
confounders of energy efficiency and behavioral responses.
From the survey responses, I find that under full-information conditions, 50 percent
of individuals reported they would not change the thermostat at any treatment cost. On
average, a 100 percent increase in the cost of heating the home induces a 0.31-0.97 (0.17 to
0.51◦ C) degree Fahrenheit reduction in the winter heating level, corresponding to an average
elasticity between -0.005 and -0.014. In addition, I find that reported actual household
temperature-setting behavior is consistent with realistic beliefs about the cost of heating.
Previous work has studied the association between energy use levels and demographic
characteristics (e.g., Costa and Kahn, 2013a,b; Longhi, 2015), but this study is among the few
to examine the association between energy price elasticity and demographics. I analyze the
heterogeneous price responses and find that individuals with higher temperature preferences
are more price responsive. This suggests that increasing block pricing of energy or emissions
can reduce peak energy use and emissions while minimizing the regressive properties of energy
and emissions pricing programs. Other demographic characteristics of the respondents are
only weakly related to the measured elasticity, although urban respondents are more elastic
relative to rural respondents.
Finally, I propose a method to estimate participants’ mean perception of the true cost
of heating by regressing participants’ reported actual temperature settings on estimated
demand parameters for temperature. I find that demand for temperature under complete
information is a good predictor of reported actual temperature-setting behavior and that
the average participant perceives a non-zero cost of heating. This evidence suggests that
preference heterogeneity plays a large role in driving empirical observations of inelastic and
heterogeneous energy demand.
In the next section, I describe the empirical puzzle of energy demand heterogeneity. Sec-
tion three introduces the survey procedure and section four describes the survey data. In
section five, I estimate the elasticity of demand for winter heating, and I analyze hetero-
geneity of demand in section six. Section seven uses the estimated elasticities to infer the
participants perceived prices. Section eight discusses the policy implications and concludes.
4
2 The puzzle of energy demand heterogeneity
Empirical studies of energy demand consistently find a large degree of heterogeneity in
residential energy demand elasticity (e.g. Reiss and White, 2005; Howden-Chapman et al.,
2009; Elinder et al., 2017; Auffhammer and Rubin, 2018). It is not clear what causes this
heterogeneity. Total home energy use is a function of the outside temperature, the combined
efficiency of energy-using appliances and the efficiency of the building, and the intensity of
use of energy services. When a researcher observes a reduction in energy use as a response
to increased prices, it is unclear which mechanism causes that response: is the individual
purchasing more energy-efficient appliances or increasing the efficiency of the home (such as
though weatherization), or is the individual reducing the use of energy services by reducing
the thermostat, turning lights off, or cooking less? In particular, elasticities estimated using
monthly or yearly energy use data cannot determine the difference between a change in
energy efficiency and use of energy services. For example, the elasticities estimated in Reiss
and White (2005) cannot differentiate between an efficiency and intensity response.
Even if the researcher can determine whether the response is coming from energy ef-
ficiency or energy-use intensity, the behavioral mechanism is still unknown. To make a
fully-informed energy consumption decision, an individual must know the energy efficiency
of their home and appliances, the current price of energy, and how much their intended
behavior will change these conditions (Jessoe and Rapson, 2014). Studies often character-
ize inelastic demand for energy or energy efficiency as resulting from lack of information
(e.g., Schleich et al., 2013; Allcott and Rogers, 2014; Jessoe and Rapson, 2014; Allcott and
Taubinsky, 2015), but it is difficult to prove whether energy demand heterogeneity arises
because some individuals are not fully informed about changes in the price of energy or the
cost of energy services, or whether some individuals are informed but have inelastic demand
because of strong preferences for energy services.
Thus, observed patterns of energy demand heterogeneity may arise through either the
efficiency or energy-use intensity channels and may be explained by either behavioral inat-
tention or preference heterogeneity. Figure (1a) shows two hypothetical patterns in energy
use behavior: type 1 consumers appear responsive to price changes and type 2 consumers
5
appear unresponsive to price changes. Figure (1b) shows that demand for energy efficiency
conditional on fixed use of energy services could explain the patterns in the data. Individ-
uals may have inelastic demand for energy services, but their different demands for energy
efficiency could result in different energy consumption patterns. In contrast, figure (1c)
shows that the same data can be generated by heterogeneous behavioral perceptions of the
cost of energy services. If type 2 consumers perceive the cost of an energy service such as
indoor temperature setting to be close to zero, the result will be an inelastic price-energy
use relationship. Finally, figure (1d) shows how two different marginal benefits curves for
temperature setting may generate the patterns seen in the data. Different marginal benefits
curves reflect a preference-heterogeneity explanation.
This paper uses a stated-choice survey to explore the underlying causes of energy demand
heterogeneity. The benefit of using a survey to study energy-use behavior is that the causal
mechanisms for price responsiveness are clear and fully identified. In the survey, individuals
have all of the information required to make an informed choice of temperature use. In
addition, energy efficiency cannot be changed, so the observed behavior can be interpreted
as coming solely through the temperature-setting channel. By eliminating the potential con-
founders of energy efficiency and behavioral responses, I can test the preference heterogeneity
explanation.
3 Survey procedure
The survey participants comprise a nationally representative sample of US individuals
drawn from the Qualtrics Online Sample. I eliminated respondents if they failed Qualtrics
speeding checks, if they do not use heat at home in the winter, or if they provided poor-
quality responses (e.g., uninterpretable entries in free-response boxes). The final sample
includes 414 individuals.1 The survey took place in early March 2018, the end of winter for
most of the United States; thus, respondents completed the survey after making real heating
1
Qualtrics surveyed individuals until a quota of 600 completed the questionnaire without failing speeding
checks. 265 individuals failed Qualtrics speeding checks before reaching 600 quality responses. From here, I
eliminated 186 respondents (31%) who did not use heat, had missing responses, or poor-quality responses.
6
(a) (b)
(c) (d)
Figure 1: Hypothetical graphs showing how heterogeneous energy-use data observed in
panel (a) may be explained by energy efficiency investments in panel (b), or instead by
energy service use via heterogeneous behavioral cost perceptions in panel (c) or heterogeneous
preferences for energy services in panel (d).
7
decisions for several months.2
The survey begins by emphasizing the consequentiality of the data for use in science and
policy.3 Respondents affirm to “thoughtfully read and provide [their] best answers to the
questions in this survey,” or else they are removed from the survey pool. Participants answer
questions about their actual winter heating temperature settings before continuing to the
hypothetical choice survey. All temperatures in the survey were elicited in Fahrenheit, the
most common unit of temperature measurement in the United States. Next, I elicit each
individual’s temperature preference baseline by asking what temperature they would choose
if heating was costless:4
Imagine that you do not have to pay for heating your home during the winter. In
this situation, what temperature setting (degrees F) would you choose when you
are at home?
This baseline temperature preference with no price can be thought of as a bliss point tem-
perature preference for heating. The respondents then see an example:
In this part of the survey, you will be asked to choose an indoor temperature
setting during the winter for when you are at home. Each question asks about a
scenario where heating is more or less expensive. The cost of heating your home
in each scenario is the monthly cost of increasing your thermostat setting by one
degree Fahrenheit while you are at home.
For example, if each degree Fahrenheit change costs $1 on your monthly heating
bill, the following changes to your thermostat setting would have these costs or
2
Prior to release of the survey, I conducted pre-testing with 200 student volunteers. Pre-test subjects
did not have difficulty understanding and responding to the questions, although several expressed that they
simply would not deviate from their preferred heating temperatures no matter the price.
3
Lewis et al. (2016) find that emphasizing consequentiality of the survey data is important for mitigating
hypothetical response bias.
4
The science and engineering literatures argue that temperature preference is determined by physiological
characteristics such as age (Taylor et al., 1995; Schellen et al., 2010), sex (Kingma and van Marken Lichten-
belt, 2015; Karjalainen, 2012, 2007; Fanger, 1970; Parsons, 2002; Cena and de Dear, 2001; Muzi et al., 1998;
Pellerin and Candas, 2003; Griefahn and Knemund, 2001; Nakano et al., 2002; Nagashima et al., 2002), diet
(Ringsdorrf Jr. and Cheraskin, 1982), and previous exposure (Young, 2010). There is some evidence that
temperature preferences of men and women differ by country (Beshir and Ramsey, 1981; Karjalainen, 2007;
Indraganti and Rao, 2010) and that individuals may be able to consciously alter the body’s internal response
to temperature (Kox et al., 2014).
8
savings:
Thermostat change Cost per degree Monthly heating bill change
Decrease thermostat by 4◦ F $1 Save $4
Decrease thermostat by 2◦ F $1 Save $2
Do not change thermostat $1 No change
Increase thermostat by 2◦ F $1 Spend $2
Increase thermostat by 4◦ F $1 Spend $4
I draw a low, medium, and high marginal cost from three independent uniform distributions
spanning $1 to $8 per month for a five degree Fahrenheit (2.8◦ C) change when they are
home.5 Respondents see a price and are asked to input their chosen temperature setting.
For example,
Choice #3: Imagine increasing your thermostat by one degree Fahrenheit will
increase your monthly heating bill by $1.60 (or changing your thermostat by five
degrees Fahrenheit will increase your heating bill by $8).
When a one degree change in temperature costs $1.60 per month, what tempera-
ture setting would you choose?
Remember that you said you would set your thermostat to 70 degrees Fahrenheit
if you weren’t paying for heating.
Respondents input their chosen temperature into a text-response box. After completing the
experiment, respondents supply their demographic information. Qualtrics compensates each
respondent a small sum after participating successfully. The full survey instrument is located
in the appendix.
5
The first price is a random draw from a U(1,2.67) distribution, the second price is a random draw from
a U(2.67,5.33) distribution, and the third price is a random draw from a U(5.33,8) distribution. These costs
of heating are based on estimates of the cost of heating holding housing attributes fixed using the Energy
Information Administration’s Residential Energy Consumption Survey as discussed in Brewer (2019). The
$1-to-$8 interval spans the lowest to highest reasonable costs per degree change in the average home. Real
costs of heating may be higher or lower than the hypothetical values, particularly in hot or cold areas
of the United States. Later, I test whether results vary based on geography and find no evidence that
price-responsiveness is related to latitude or longitude.
9
Table 1: Participant sample means and standard deviations.
Mean Std dev US
Income 79,925.13 (71251.68) 81,283a
Monthly heat bill 121.29 (145.81) 88.42b
Bliss point 70.77 (3.59) 70.49b
Actual temperature at home 69.70 (3.71) 70.01b
Household size 2.91 (2.85) 2.63a
Age 50.22 (16.60) 37.8a
Children 0.44 (1.11) 0.79c
Female 0.60 (.49) 0.51a
Non-white 0.29 (.45) 0.27a
Urban 0.76 (.43) 0.81a
High school 0.37 (.48) 0.27a
Some college 0.28 (.45) 0.21a
College 0.19 (.39) 0.27a
Graduate degree 0.10 (.30) 0.12a
Republican 0.28 (.45) 0.23d
Democrat 0.33 (.47) 0.29d
Respondents 414
Observations 1242
a: From the 2017 American Community Survey (ACS) five year profiles.
Age is median age. Urban-rural estimate from the 2015 ACS. b: From the
2015 Residential Energy Consumption Survey (RECS). Bliss point calcu-
lated from renters whose landlords pay for heat. c: From the 2018 Current
Population Survey. d: From the March 2018 Gallup Party Affiliation Poll.
4 Data
Table (1) displays summary statistics from the experimental sample after cleaning the
data.6 I recruited the initial sample to be nationally representative using sampling quotas
based on age, gender, race, education, political party, and fraction of rural respondents.
After cleaning, the sample is older, more female, and less educated than the national average.
Figure (2) displays kernel density plots of participants’ bliss point temperature preferences
and actual temperature settings. The distribution of bliss point temperatures appears to have
a higher mean than and similar variance to the distribution of actual temperature settings.
A Kolmogorov-Smirnov test of equivalence of distributions rejects the null hypothesis that
the distributions are the same (p-value = 0.001).
6
The data are jointly owned by the author and the Environmental Science and Policy Program at Michigan
State University. The data can be made available by request.
10Bliss point and actual heating temperatures
.15
.1
Density
.05
0
60 65 70 75 80
Degrees F
Bliss point Actual temperature
Figure 2: Kernel density of participants’ bliss point temperature preferences and real tem-
perature settings. 70 degrees Fahrenheit corresponds to 21.1◦ C. A Kolmogorov-Smirnov test
of equivalence of distributions rejects the null hypothesis that the distributions are equal
(p-value = 0.001).
Half (54 percent) of participants report that they set their actual thermostats equal to
their bliss point temperature preference. Figure (3) shows a plot of bliss point vs actual
temperature setting. 6.5 percent of participants report that the thermostat is higher than
the bliss point, perhaps because they did not understand the question or because they are
not in control of the thermostat.7 For experimental temperature settings, 50 percent of
participants continue to choose their bliss point temperature setting at the highest cost level
(including 73 percent of individuals who set their actual home temperature equal to their
bliss point).
7
Another possible explanation could be if participants choose temperatures hotter than they prefer to
satisfy another household member’s higher temperatures.
11Bliss point vs actual temperature
80
Bliss point (degrees F)
65 7060 75
60 65 70 75 80
Actual temperature (degrees F)
Participant 45 degree line
Figure 3: Scatterplot of bliss point temperature preferences and actual temperature settings
with a 45 degree line for reference. 54 percent of respondents set the thermostat equal to the
bliss point. 6.5 percent reported setting the thermostat greater than the bliss point. Random
noise has been added to the data to show clustering on common temperature choices such
as 70 degrees Fahrenheit (21.1◦ C).
125 Estimating demand for heating
The survey results provide points on each respondent’s temperature demand curve. The
most intuitive measure of an individual’s temperature response to a change in the price of an
additional degree is the semi-elasticity, or the degree change in the thermostat for a percent
change in price. I estimate the semi-elasticity using four methods and convert each to a
traditional elasticity for comparison to other studies.
For each choice c ∈ {1, 2, 3}, an individual i sees a price pricei,c to increase the thermostat
by one degree and chooses a temperature setting tempi,c . Thus, for a household i, I model
the choice of temperature setting as some demand function f (·) of the price:
tempi,c = f (pricei,c ) + εi , (1)
where εi is individual heterogeneity that is uncorrelated with pricei,c . The semi-elasticity is
∂temp
∂price
· pricei,c .
First, I pool the sample and estimate the mean semi-elasticity using ordinary-least-
squares and fixed-effects estimation. I estimate the following equation on the pooled tem-
perature choices:
tempi,c = α + βln(pricei,c ) + i .
Taking the derivative with respect to the price variable and solving for β reveals that β =
∂temp
∂price
· pricei,c . Thus with this functional form, the estimate β̂ serves as an estimate of the
average semi-elasticity.
Next, I estimate each individual’s unique semi-elasticity for temperature setting by cal-
culating the arc semi-elasticity directly using the midpoint formula. The arc semi-elasticity
between any two choices on the demand curve c and c − 1 is
∆c,c−1 tempi
Arc semi-elasticityc,c−1 = , (2)
%∆c,c−1 pricei
where ∆c,c−1 tempi = tempi,c −tempi,c−1 , the difference in temperature settings chosen by the
13
pricei,c −pricei,c−1
participant, and %∆c,c−1 pricei = pricei,c +pricei,c−1 , the percentage difference in researcher-
2
assigned energy price. This can be calculated directly for each pair of points on the demand
curve.8 One benefit of this approach is that the arc semi-elasticity uses information from
the bliss point choice (i.e., when price is zero), while the regression-based approaches cannot
because the log of zero is undefined. In addition, this approach provides a heterogeneous
and non-parametric measure of price responsiveness.
The final approach I use to measure semi-elasticity is an individual regression-adjustment
approach. For each participant i, I estimate the following equation separately with ordinary
least squares:
tempi,c = αi + βi ln(pricei,c ) + i . (3)
The estimate β̂i is an estimate of each individual’s mean semi-elasticity over individual i’s
experimental choices c. This method provides heterogeneous semi-elasticities but does not
incorporate information provided from the bliss-point choice.
Table (2) displays the average estimated semi-elasticities using all four methods. I boot-
strap the 95 percent confidence intervals of the averages using 1,000 replications and re-
sampling at the participant level. The estimates indicate that for a 100 percent increase
in the cost of heating, an individual reduces the thermostat setting by 0.31-0.97 degrees
Fahrenheit (0.17 to 0.51◦ C). This measurement corresponds to an elasticity between -0.005
and -0.014.9 This small average response is due to the large number of price-insensitive
participants and hides significant heterogeneity, which I analyze in the following section.
6 Heterogeneity analysis
Figure (4) displays a histogram of participants’ arc semi-elasticities, and figure (5) dis-
plays a histogram of participants’ regression-adjustment semi-elasticities. The distributions
8
See Allen and Lerner (1934) for a classic discussion on arc elasticities and semi-elasticities.
9
The elasticity is presented to compare to other papers in the literature. For temperature, the elasticity
is a poorly-defined concept because there is no natural zero consumption point; thus, using Celsius would
slightly alter the elasticity because the arbitrary zero point and scale changes.
14Table 2: Estimated semi-elasticities and elasticities. The interpretation of a semi-elasticity η1
is that for a 100 percent increase in price, the average participant will reduce the thermostat
by η1 degrees. 95 percent confidence intervals bootstrapped using 1000 replications with
sampling at the participant level.
Method Semi-elasticity Elasticity N
OLS -0.52 -0.0075 1,242
(-0.73,-0.31) (-0.0105,-0.0045)
FE -0.56 -0.0080 1,242
(-0.69,-0.43) (-0.0100,-0.0062)
Individual OLS -0.66 -0.0097 1,242
(-0.86,-0.46) (-0.0125,-0.0068)
Arc -0.69 -0.0100 1,242
(-0.97,-0.43) (-0.0142,-0.0063)
display a similar bunching of individuals completely unresponsive to prices with a significant
portion of more price-responsive individuals in the tail.
The distribution of elasticities is characteristic of those found in other energy settings.
Reiss and White (2005) estimate a similarly skewed distribution of annual elasticities for
electricity use with a mass of relatively price-insensitive households and a fat tail of more
elastic households. They also find that low-income households have more elastic demand
and conclude that households with space heating have a significantly more elastic demand
than other households. The experiment here shows that the skewed elasticity distribution
can be generated without the energy-efficiency responses included in a yearly elasticity.
I explore what drives heterogeneity in temperature response by regressing the arc semi-
elasticities on standardized vectors of the average price on the arc pricei,c,c−1 and participant
demographics. I use a Tobit maximum-likelihood estimation to account for the clustering
at zero in the dependent variable. Thus, denoting Z(·) as the function that transforms a
sample draw of a random variable into its z-score, I estimate the equation
Arc semi-elasticityi,c,c−1 = a + bZ(pricei,c,c−1 ) + dZ(demographicsi ) + ei,c,c−1 (4)
using maximum likelihood, treating all non-negative arc semi-elasticities as a corner solution.
Standardization allows the marginal effects of the regression to be easily compared. The
15Distribution of arc semi−elasticities
50
45.41
30 40
Percent
23.43
20
16.43
10
8.454
6.28
0
η>0 η=0 −1marginal effects from this estimation are interpreted as the change in arc semi-elasticity for
a one-standard-deviation change in the predictor variable while holding the other predictor
variables constant.10
I explain heterogeneity as a function of bliss-point temperature preference, average monthly
heating bill, income, age, household size, number of children living at home, gender, race,
urban/rural status, education, and political party. Figure (6) plots the estimated marginal
effects with the 95 percent confidence intervals bootstrapped using 1,000 replications with
repeated sampling at the participant level. Most strikingly, individuals with a one-standard-
deviation-higher bliss point temperature have on average a -0.42 higher arc semi-elasticty
(i.e., are more elastic), all else equal. Higher-income and higher-education respondents
have less elastic demand, all else equal, although the confidence intervals for the education
marginal effects include zero. Participants living in urban areas are more responsive to price
changes. Older participants have less elastic demand, with a one-standard-deviation increase
in age corresponding with a 0.23 lower arc semi-elasticity, all else equal. The marginal ef-
fects estimates of average heating bill, participant gender, race, number of children, latitude,
longitude, and household size have confidence intervals that contain zero.
Political party is not a strong determinant of elasticity, with Republicans, Democrats, and
Independents having statistically indistinguishable elasticity measures when controlling for
other covariates. In two papers, Costa and Kahn estimate heterogeneous energy use patterns
by political ideology. First, Costa and Kahn (2013a) show that total household electricity use
is lower for politically progressive households. Second, Costa and Kahn (2013b) show that
politically progressive homeowners are more responsive to non-price nudges. The survey
in this paper measures a different dimension of energy use, but nonetheless the lack of
heterogeneity by political group is surprising. It is possible that in the literature, total
energy use and ownership of energy-efficient appliances are correlated with local progressive
energy-efficiency policies and thus reflect these policies rather than individual behavior. In
this estimation, I include many controls that are correlated with ideology and whose influence
may be spuriously attributed to ideology (e.g., urban or rural).
10
The marginal effect I estimate is the “unconditional” average partial effect ∂E(Arc semi-elasticity|Z(x))
∂Z(xj ) where
x is a matrix of predictor variables and xj is a single predictor variable using the results provided in
Wooldridge (2010).
17Heterogeneous demand for heat
−.048
Mean price −.48
Bliss point −.047
Mean heat bill .29
Income −.2
HH size .19
Age .048
Num children .047
Female .11
Non−white −.29
Urban .1
Grad degree .078
College .28
Some college .3
High school −.095
Republican −.11
Democrat −.19
North (lat) −.064
East (long)
More −.8 −.6 −.4 −.2 0 .2 .4 .6 .8 Less
elastic elastic
Marginal effect
Figure 6: The marginal effects from a Tobit estimation of the estimated arc semi-elasticities
on average price and participant demographics. The marginal effects from this estimation
can be interpreted as the change in arc semi-elasticity for a one-standard-deviation change in
the predictor variable holding the other predictor variables constant. 95 percent confidence
intervals are bootstrapped using 1000 replications with sampling at the participant level.
18In this survey, age plays a large role in determining elasticity whereas sex does not. The
science and engineering literatures focus on measuring differences in temperature preference
and sensitivity, but differences in behavior are often ignored. For example, a group of people
may, on average, be able to detect a difference in temperature in a laboratory more readily,
but this does not translate necessarily to differences in thermostat-setting behavior. Indeed, I
find here that men and women do not respond to prices differently after other characteristics
have been controlled for despite numerous findings that women prefer higher temperatures
than men.11 Prior studies’ findings may reflect how temperature decisions are made in
settings that affect multiple individuals with heterogeneous temperature preferences or other
barriers to adjusting the thermostat.12
7 Estimating the perceived cost of heating
I now use the hypothetical survey choices with respondents’ reported actual temperature
settings to estimate the average perceived actual cost of heating. To estimate perceived
actual cost of heating, I impose some structure on temperature demand. Suppose that an
individual i’s demand for temperature tempi is a linear function of the cost of heating pricei :
tempi = blissi + γi pricei (5)
where γi ≤ 0. Equation (5) states that an individual reduces the temperature setting from
their bliss point preference blissi as the cost of maintaining that temperature increases.13 I
observe each participant’s temperature setting and price information for four survey instances
j (including the choice of bliss point temperature when the price is zero), and the reported
11
See Karjalainen (2012) for a review of this literature.
12
For example, Kingma and van Marken Lichtenbelt (2015) discuss temperature demand in shared office
buildings, and Karjalainen (2007) finds that women are less likely to change the thermostat settings than
men are.
13
I derive the linear demand function in equation (5) from a model (similar to Brewer (2019)) where
individuals consume a numeraire good xi and temperature setting tempi with cost per degree Fahrenheit
pricei . Each individual has income yi and bliss point temperature preference blissi . Let the utility function
take the form ui (xi , tempi , blissi ) = xi − 2γ1 i (tempi − blissi )2 . The first order conditions for maximization
of this utility function subject to the budget constraint yi ≥ xi + pricei tempi returns the linear demand
function in equation (5).
19¨ i . The perceived true price of heating
actual (non-experimental) temperature setting temp
¨ i is unobserved.14 I estimate the average perceived price in a two-stage procedure, by
price
first estimating the demand parameters γi from equation (5) using the experimental data
and then using actual temperature choices and estimates γ̂i from the first stage to estimate
the mean perceived temperature.
Using the experimental data, the first-stage ordinary least squares estimation of
tempi,j = δi + γi pricei,j + ξi,j (6)
for each participant provides an unbiased estimate γ̂i if the error term ξi,j is independent of
pricei,j .15 In the second stage, I regress actual temperature settings on a constant term µ,
the individual’s bliss point blissi , the first-stage estimate γ̂i , and an interaction of γ̂i with
an indicator variable equal to one if the individual does not pay for heating:
temp ¨ i = 0)γ̂i + ζi ,
¨ i = µ + λblissi + φ1 γ̂i + φ2 1(price (7)
where ζi is an error term. Thus, φ̂1 is an estimate of the mean perceived cost per degree
Fahrenheit that induced the actual temperature setting if the individual paid for heating. If
the perceived cost of heating is low or zero, this is suggestive evidence that individuals have
incorrectly low beliefs about the cost of heating. For individuals who do not pay for heating,
the estimate of the mean perceived price is φ̂1 + φ̂2 .16 Furthermore, the estimate λ̂ should
be equal to one if individuals choose their bliss point when the price is zero.
Table (3) displays estimates from the two-stage procedure. I estimate an average per-
ceived monthly cost per degree Fahrenheit of $0.65 or equivalently a monthly cost of $3.25
per five degrees Fahrenheit which is roughly in the middle of the experimental values used.
In addition, the estimate of the coefficient on bliss point λ̂ = 0.92, which is consistent with
14 ¨ i
In this section, I use the “double-dot” notation to indicate data that are non-experimental; hence, temp
¨
is the individual’s actual temperature setting at home and pricei is the true cost per degree (only observed
for those who do not pay for heat).
15
The term δi is an individual-specific constant that can be interpreted as the estimated temperature
setting when price is zero—the bliss point.
16
If individuals who do not pay for heating were randomly selected, the perceived price should be uncor-
related with the chosen temperature and thus φ̂1 + φ̂2 = 0.
20Table 3: Estimates of equation (7), an OLS regression of real temperature setting on a
constant term, the individual’s bliss point, the first-stage estimate of temperature demand
term γ̂i , and an interaction of γ̂i with an indicator variable equal to one if the individual
does not pay for heating. The coefficient on γ̂i is an estimate of the mean perceived price
for individuals who pay for heating. 95 percent confidence intervals are bootstrapped using
1000 replications with sampling at the participant level.
¨ i
y = temp (1)
blissi 0.92
(0.85,0.98)
γ̂i 0.65
(0.44,0.87)
¨ i = 0)γ̂i
(price -0.35
(-0.73,0.08)
Constant 5.18
(1.08,9.70)
2
R 0.74
N 414
the hypothesized value of 1. The procedure shows that the estimates of demand from the
experiment predict real temperature settings well, with an R-squared value of 0.74.
The results from the two-stage procedure reflect reasonable and large perceptions of
the cost per degree Fahrenheit. In addition, half of participants report that they set their
actual thermostats equal to their bliss point temperature preference when at home. Of these
individuals, 70 percent were similarly unresponsive to the cost of heating in the experiment.
Taken together, this suggests a limited role for behavioral misperceptions of energy costs,
though it does not rule them out.
8 Implications and conclusions
The survey reproduces energy-use heterogeneity distributions comparable to those seen
in actual energy-use data. Half of participants report that they set their actual thermostats
equal to their bliss point temperature preference when at home. Of these individuals, 70
percent were similarly unresponsive to the cost of heating in the experiment. This is evi-
dence that for these 70 percent of individuals, there is some perceived negative preference
21for deviating from their temperature bliss point in excess of the savings that they could
have made in the experiment. These participants’ behavior is consistent with a rational zero
response to the cost of heating at the relevant price level. Under perfect-information con-
ditions, energy-use behavior displays significant heterogeneity and unresponsiveness. Thus,
programs or policies designed to remove informational barriers (for example, programs that
offer in-home-displays for time-varying energy prices as in Jessoe and Rapson (2014) and
Prest (2020)) may have less effect on winter heating relative to other energy-using behav-
iors.
The paper finds that individuals, on average, set their thermostats consistent with having
complete cost information. It is not likely that individuals know the exact cost-per-degree
change on the thermostat, but over time most people have adjusted their behavior based
on feedback from energy bills. Second, more than half of all individuals are completely
unresponsive to prices. People simply do not like to be cold. Energy service demand is
highly inelastic, a 100 percent increase in the cost of heating reduces thermostat settings
by 0.31 to 0.97 degrees Fahrenheit (0.17 to 0.51◦ C), corresponding to a -0.005 to -0.014
elasticity. The cost of heating is low enough to take heating for granted, but it is likely that
even if the cost of heating was to dramatically increase (perhaps due to a pollution fee),
behavior would respond very little. Inelastic demand for energy services does not mean that
prices are ineffective during normal winter weather; instead, it means that the benefits from
energy services are high. As long as the inelasticity does not arise from an artificial barrier
such as false information about the energy cost savings, individuals will make the proper
tradeoff between costs and benefits from energy use when facing prices that reflect the full
external costs of energy use.
Inelastic demand implies that the gains from policies targeting home energy efficiency
are likely high. If energy-use behavior is fixed for many individuals, energy efficiency savings
are large and will not be cannibalized by a rebound effect. It is not clear whether house-
holds optimally adopt energy-efficiency upgrades (i.e., whether an “energy-efficiency gap”
exists), but a recent review of the literature did not find much evidence that individuals
systematically fail to adopt energy-efficiency upgrades (Gillingham and Palmer, 2014).
Alternatively, demand-response policies in which the utility secures centralized control of
22energy-using appliances may be preferred to reduce consumption because they rely on house-
hold participation rather than energy-use elasticity to achieve reductions. Recent work has
shown that opt-in rates for winter demand response are highly responsive to offering partic-
ipation incentives (Srivastava et al., 2020). Furthermore, similar demand-response programs
have high compliance and acceptability rates (Sarran et al., 2021) relative to voluntary re-
quests to reduce energy consumption (Gyamfi and Krumdieck, 2011). These policies also
offer the utility the ability to respond to emergency conditions such as an unexpected cold-
wave or supply-side disruption immediately—a significant advantage relative to time-varying
pricing policies.
Another implication of these findings is that increasing block pricing can be used to re-
duce energy use without large incidence for a bulk of users with inelastic energy demand.17
The largest determinant of elasticity in the experiment was having a high bliss point tem-
perature preference, implying that individuals with larger energy-service demand are more
price responsive. By increasing the price of energy for higher-demand users who are most
price-responsive, a regulator or regulated energy provider can reduce load (and corresponding
emissions) without increasing payments from inelastic users. For example, a carbon tax with
a zero-price carbon allowance may reduce the regressivity of the policy without sacrificing
efficiency gains.
Given the string of costly energy emergencies in the United States sparked by extreme
winter weather, more research on heating behavior is needed. Rather than relying on volun-
tary requests for reductions, utilities and policymakers should develop evidence-based strate-
gies to reduce winter peak demand as well as respond to extreme winter events or supply-side
disruptions. While prices may be useful for managing normal day-to-day heating demand,
they are unlikely to provide the relief necessary to resolve an acute shortage.
17
Increasing block pricing charges a higher marginal cost per unit of energy for consumption of units of
energy over a threshold. It essentially provides users with an allowance of cheap energy each billing period
before having to spend more on additional energy consumption.
23References
Allcott, H. and T. Rogers (2014, October). The short-run and long-run effects of behav-
ioral interventions: Experimental evidence from energy conservation. American Economic
Review 104 (10), 3003–37.
Allcott, H. and D. Taubinsky (2015, August). Evaluating behaviorally motivated policy:
Experimental evidence from the lightbulb market. American Economic Review 105 (8),
2501–38.
Allen, R. G. D. and A. P. Lerner (1934). The concept of arc elasticity of demand. The
Review of Economic Studies 1 (3), 226–230.
Auffhammer, M. and E. Rubin (2018, January). Natural gas price elasticities and optimal
cost recovery under consumer heterogeneity: Evidence from 300 million natural gas bills.
Technical report, Energy Institute at Haas.
Azarova, V., J. J. Cohen, A. Kollmann, and J. Reichl (2020). Reducing household electricity
consumption during evening peak demand times: Evidence from a field experiment. Energy
Policy 144, 111657.
Belton, C. A. and P. D. Lunn (2020). Smart choices? an experimental study of smart meters
and time-of-use tariffs in Ireland. Energy Policy 140, 111243.
Beshir, M. and J. Ramsey (1981). Comparison between male and female subjective estimates
of thermal effects and sensations. Applied Ergonomics 12 (1), 29 – 33.
Brewer, D. (2019). Equilibrium sorting and moral hazard in residential energy contracts.
Working paper https://www.dylanbrewer.com/wp-content/uploads/2018/10/Brewer -
JMP.pdf.
Cena, K. and R. de Dear (2001). Thermal comfort and behavioural strategies in office
buildings located in a hot-arid climate. Journal of Thermal Biology 26 (4), 409 – 414.
International Thermal Physiology Symposium.
24Costa, D. L. and M. E. Kahn (2013a). Do liberal home owners consume less electricity? a
test of the voluntary restraint hypothesis. Economics Letters 119 (2), 210 – 212.
Costa, D. L. and M. E. Kahn (2013b). Energy conservation “nudges” and environmentalist
ideology: Evidence from a randomized residential electricity field experiment. Journal of
the European Economic Association 11 (3), 680–702.
Elinder, M., S. Escobar, and I. Petre (2017, March). Consequences of a price incentive
on free riding and electric energy consumption. Proceedings of the National Academy of
Sciences 114 (4), 3091–3096.
Fanger, P. (1970). Thermal comfort: Analysis and applications in environmental engineering.
Danish Technical Press.
Filippini, M. (2011). Short- and long-run time-of-use price elasticities in Swiss residential
electricity demand. Energy Policy 39 (10), 5811–5817. Sustainability of biofuels.
Gallup (2018, March). Party affiliation. https://news.gallup.com/poll/15370/party-
affiliation.aspx, accessed 2020-05-07.
Gillingham, K. and K. Palmer (2014, January). Bridging the energy efficiency gap: policy
insights from economic theory and empirical analysis. Review of Environmental Economics
and Policy 8 (1), 18–38.
Griefahn, B. and C. Knemund (2001). The effects of gender, age, and fatigue on susceptibility
to draft discomfort. Journal of Thermal Biology 26 (4), 395 – 400. International Thermal
Physiology Symposium.
Gyamfi, S. and S. Krumdieck (2011). Price, environment and security: Exploring multi-
modal motivation in voluntary residential peak demand response. Energy Policy 39 (5),
2993–3004.
Howden-Chapman, P., H. Viggers, R. Chapman, D. ODea, S. Free, and K. OSullivan (2009).
Warm homes: Drivers of the demand for heating in the residential sector in New Zealand.
Energy Policy 37 (9), 3387 – 3399. New Zealand Energy Strategy.
25Indraganti, M. and K. D. Rao (2010). Effect of age, gender, economic group and tenure
on thermal comfort: A field study in residential buildings in hot and dry climate with
seasonal variations. Energy and Buildings 42 (3), 273 – 281.
Ito, K. (2014, February). Do consumers respond to marginal or average price? evidence from
nonlinear electricity pricing. American Economic Review 104 (2), 537–63.
Jang, D., J. Eom, M. Jae Park, and J. Jeung Rho (2016). Variability of electricity load
patterns and its effect on demand response: A critical peak pricing experiment on korean
commercial and industrial customers. Energy Policy 88, 11–26.
Jessoe, K. and D. Rapson (2014, April). Knowledge is (less) power: Experimental evidence
from residential energy use. American Economic Review 104 (4), 1417–38.
Karjalainen, S. (2007). Gender differences in thermal comfort and use of thermostats in
everyday thermal environments. Building and Environment 42 (4), 1594 – 1603.
Karjalainen, S. (2012). Thermal comfort and gender: a literature review. Indoor Air 22 (2),
96–109.
Kingma, B. and W. van Marken Lichtenbelt (2015, August). Energy consumption in build-
ings and female thermal demand. Nature Climate Change 5, 10541056.
Kox, M., L. T. van Eijk, J. Zwaag, J. van den Wildenberg, F. C. G. J. Sweep, J. G. van der
Hoeven, and P. Pickkers (2014). Voluntary activation of the sympathetic nervous system
and attenuation of the innate immune response in humans. Proceedings of the National
Academy of Sciences 111 (20), 7379–7384.
Lewis, K. E., C. Grebitus, and R. M. Nayga (2016, Dec). U.s. consumers preferences for
imported and genetically modified sugar: Examining policy consequentiality in a choice
experiment. Journal of Behavioral and Experimental Economics 65, 18.
Longhi, S. (2015). Residential energy expenditures and the relevance of changes in household
circumstances. Energy Economics 49, 440 – 450.
26Muzi, G., G. Abbritti, M. P. Accattoli, and M. dell’Omo (1998, Aug). Prevalence of irrita-
tive symptoms in a nonproblem air-conditioned office building. International Archives of
Occupational and Environmental Health 71 (6), 372–378.
Nagashima, K., T. Yoda, T. Yagishita, A. Taniguchi, T. Hosono, and K. Kanosue (2002).
Thermal regulation and comfort during a mild-cold exposure in young Japanese women
complaining of unusual coldness. Journal of Applied Physiology 92 (3), 1029–1035.
Nakano, J., S. Tanabe, and K. Kimura (2002). Differences in perception of indoor environ-
ment between Japanese and non-Japanese workers. Energy and Buildings 34 (6), 615 –
621. Special Issue on Thermal Comfort Standards.
Parsons, K. C. (2002). The effects of gender, acclimation state, the opportunity to adjust
clothing and physical disability on requirements for thermal comfort. Energy and Build-
ings 34 (6), 593 – 599. Special Issue on Thermal Comfort Standards.
Pellerin, N. and V. Candas (2003). Combined effects of temperature and noise on human
discomfort. Physiology and Behavior 78 (1), 99 – 106.
Perryman Group (2021). Preliminary estimates of economic
costs of the february 2021 Texas winter storm. Technical re-
port. https://www.perrymangroup.com/media/uploads/brief/
perryman-preliminary-estimates-of-economic-costs-of-the-february-2021-texas-winter-st
pdf accessed 05-24-2021.
Prest, B. C. (2020). Peaking interest: How awareness drives the effectiveness of time-
of-use electricity pricing. Journal of the Association of Environmental and Resource
Economists 7 (1), 103–143.
Reiss, P. C. and M. W. White (2005). Household electricity demand, revisited. The Review
of Economic Studies 72 (3), 853–883.
Ringsdorrf Jr., W. and E. Cheraskin (1982). Vitamin c and tolerance of heat and cold:
Human evidence. Orthomolecular Psychiatry 11, 128–131.
27Sarran, L., H. B. Gunay, W. O’Brien, C. A. Hviid, and C. Rode (2021). A data-driven study
of thermostat overrides during demand response events. Energy Policy 153, 112290.
Schellen, L., W. D. V. M. Lichtenbelt, M. G. L. C. Loomans, J. Toftum, and M. H. D.
Wit (2010, July”). Differences between young adults and elderly in thermal comfort,
productivity, and thermal physiology in response to a moderate temperature drift and a
steadystate condition. Indoor Air 20 (4), 273–283.
Schleich, J., M. Klobasa, S. Glz, and M. Brunner (2013). Effects of feedback on residential
electricity demandfindings from a field trial in Austria. Energy Policy 61, 1097 – 1106.
Srivastava, A., S. Van Passel, R. Kessels, P. Valkering, and E. Laes (2020). Reducing win-
ter peaks in electricity consumption: A choice experiment to structure demand response
programs. Energy Policy 137, 111183.
Taylor, N. A., N. K. Allsopp, and D. G. Parkes (1995). Preferred room temperature of
young vs aged males: The influence of thermal sensation, thermal comfort, and affect.
The Journals of Gerontology: Series A 50A(4), M216–M221.
Thorsnes, P., J. Williams, and R. Lawson (2012). Consumer responses to time varying prices
for electricity. Energy Policy 49, 552–561. Special Section: Fuel Poverty Comes of Age:
Commemorating 21 Years of Research and Policy.
United States Census Bureau (2011-2017). American community survey.
https://www.census.gov/programs-surveys/acs/, accessed 2020-05-07.
United States Census Bureau (2018). Current population survey.
https://www.census.gov/programs-surveys/cps.html, accessed 2020-05-07.
United States Energy Information Administration (2015). Residential energy consumption
survey. https://www.eia.gov/consumption/residential/, accessed 2020-05-07.
Wooldridge, J. (2010, September). Econometric analysis of cross section and panel data (2
ed.). Cambridge, Mass: MIT Press.
28Young, A. J. (2010). Homeostatic responses to prolonged cold exposure: Human cold ac-
climatization (Supplement 14: Handbook of Physiology ed.)., pp. 419–438. John Wiley
and Sons, Inc.
29Qualtrics Survey Software https://msu.co1.qualtrics.com/Q/EditSection/Blocks/Ajax/GetSurveyPrin...
Default Question Block
THANK YOU FOR YOUR INTEREST IN OUR RESEARCH STUDY!
What the study is about: You are being asked to take part in a research survey
investigating a variety of issues relevant to the environment. Please read this
page carefully and feel free to ask any questions you may have before starting
the survey. By completing this survey, you voluntarily agree to participate in this
research study.
What we will ask you to do: If you decide to continue, you will answer a series
of questions about your perceptions, your experiences, and demographics.
Your answers will be anonymous. This survey is anonymous--we will not collect
1 of 17 4/12/2019, 4:11 PMQualtrics Survey Software https://msu.co1.qualtrics.com/Q/EditSection/Blocks/Ajax/GetSurveyPrin...
your name, phone number, or other specific identifiers. Data from this study will
be securely stored. Only the researchers will have access to the data. There is
no anticipated risk to you.
Taking part is voluntary: Taking part in this study is completely voluntary. If you
decide to take part, you are free to withdraw at any time. You will be
compensated through the survey panel that you belong to after completing this
survey. Please note that if you fail attention or speeding checks, you will not
receive your incentive. Questions about your incentive should be directed
through your panel membership where you have an account as a panelist.
If you have questions: This study is conducted by the students of a course at
Michigan State University and supervised by Dr. Joseph Hamm. If you have
concerns or questions about this study, such as scientific issues, how to do any
part of it, or to report an injury, you may contact Dr. Hamm at jhamm@msu.edu
or by postal mail at:
Michigan State University
557 Baker Hall, 655 Auditorium Road
East Lansing, MI 48824
If you have any questions or concerns about your role and rights as a research
participant, would like to obtain information or offer input, or would like to
register a complaint about this study, you may contact--anonymously if you
wish--Michigan State University’s Human Research Protection Program at
517-355-2180, Fax 517-432-4503, or email irb@msu.edu or postal mail at:
4000 Collins Rd. Ste. 136
Lansing, MI 48910
Please note the project number (i055631) in any correspondence about the
survey.
2 of 17 4/12/2019, 4:11 PMYou can also read
