Adapting to Climate Change: The Remarkable Decline in the U.S. Temperature-Mortality Relationship over the 20th Century

TLDR: This paper examined the temperature-mortality relationship over the course of the twentieth-century United States both for its own interest and to identify potentially useful adaptations for coming decades, finding that the mortality impact of days with mean temperature exceeding 80°F declined by 75 percent.

Abstract: This paper examines the temperature-mortality relationship over the course of the twentieth-century United States both for its own interest and to identify potentially useful adaptations for coming decades. There are three primary findings. First, the mortality impact of days with mean temperature exceeding 80°F declined by 75 percent. Almost the entire decline occurred after 1960. Second, the diffusion of residential air conditioning explains essentially the entire decline in hot day–related fatalities. Third, using Dubin and McFadden’s discrete-continuous model, the present value of US consumer surplus from the introduction of residential air conditioning is estimated to be $85–$185 billion (2012 dollars).

Figures

Table 7: Effect of Residential Air Conditioning on Heat-Related Mortality, By Age, Race, and Cause of Death, 1960-2004

Table 6: Interaction Effects Between Modifiers and Number of Temperature-Days Between 80-89°F and Above 90°F

Figure 1: Distribution of Daily Average Temperatures, 1900-2004

Table 3: Estimates of the Impact of High and Low Temperatures on Log Monthly Mortality Rate

Table 8: Robustness Analysis of the Effect of Residential Air Conditioning on the Temperature-Mortality Relationship, 1960-2004

Table 4: Estimates of the Impact of High and Low Temperatures on Log Monthly Mortality Rate, By US Climate Regions

Full text

UC Berkeley
UC Berkeley Previously Published Works
Title
Adapting to climate change: The remarkable decline in the US temperature-mortality
relationship over the Twentieth Century
Permalink
https://escholarship.org/uc/item/08c6t2hs
Authors
Barreca, A
Clay, K
Deschenes, O
et al.
Publication Date
2016-02-01
DOI
10.1086/684582
Peer reviewed
eScholarship.org Powered by the California Digital Library
University of California

1
Adapting to Climate Change: The Remarkable Decline in the U.S.
Temperature-Mortality Relationship over the 20
th
Century
*
Alan Barreca, Karen Clay, Olivier Deschenes, Michael Greenstone and
Joseph S. Shapiro
January 2015
Abstract
A critical part of adapting to the higher temperatures that climate change brings will be the deployment
of existing technologies to new sectors and regions. This paper examines the evolution of the
temperature-mortality relationship over the course of the entire 20
th
century in the United States both
for its own interest but also to identify potentially useful adaptations that may be useful in the coming
decades. There are three primary findings. First, the mortality impact of days with a mean temperature
exceeding 80° F has declined by about 70%. Almost the entire decline occurred after 1960. There are
about 14,000 fewer fatalities annually than if the pre-1960 impacts of high temperature on mortality still
prevailed. Second, the diffusion of residential air conditioning can explain essentially the entire decline
in hot day related fatalities. Third, using Dubin-McFadden’s discrete-continuous model, we estimate that
the present value of US consumer surplus from the introduction of residential air conditioning (AC) in
1960 ranges from $83 to $186 billion ($2012) with a 5% discount rate. The monetized value of the
mortality reductions on high temperature days due to AC accounts for a substantial fraction of these
welfare gains.
*
Alan Barreca (Tulane University, IZA, and NBER), Karen Clay (Carnegie Mellon University and NBER), Olivier
Deschenes (UCSB, IZA, and NBER), Michael Greenstone (Chicago and NBER), Joseph Shapiro (Yale and NBER).
Deschenes is the corresponding author and can be contacted at olivier@econ.ucsb.edu
. We thank Daron
Acemoglu, Robin Burgess, our discussant Peter Nilsson, conference participants at the “Climate and the Economy”
Conference and seminar participants at many universities for their comments. We also thank the editor and three
anonymous referees for their comments. Jonathan Petkun and Daniel Stuart provided outstanding research
assistance.

2
I. Introduction
The accumulation of greenhouse gases (GHGs) in the atmosphere threatens to alter the climate
dramatically and in a relatively short period of geological time. While much attention has been devoted
to reducing GHG emissions, comparatively little has been devoted to understanding how societies will
adapt to climate change. Adaptation, according to the Intergovernmental Panel on Climate Change
(IPCC), is defined as "adjustment in natural or human systems in response to actual or expected climatic
stimuli or their effects, which moderates harm or exploits beneficial opportunities" (IPCC 2007). These
adjustments can take the form of alterations in the uses of existing technologies and/or the invention of
new technologies. The poor state of knowledge about adaptation opportunities and adaptation’s costs
proves a challenge both for developing reliable estimates of the costs of climate change and for
identifying solutions to the risks that climate change poses.
The health and broader welfare consequences of increases in temperatures are an area of
special concern. For example, the identification of adaptation opportunities that can reduce the human
health costs of climate change is recognized as global research priority of the 21st century (WHO 2008;
NIEHS 2010). This need is especially great in developing countries where high temperatures can cause
dramatic changes in life expectancy (Burgess et al. 2014). High temperatures, beyond their health
consequences, can have a range of other negative consequences, including causing workers to be less
productive, making it difficult for children to study, and generally leading to less pleasant lives (Hsiang
2010, Sudarshan et al. 2014).
This paper provides the first large-scale empirical evidence on long-run adaptation opportunities
through changes in the use of currently existing technologies. The empirical analysis is divided into three
parts. The first part documents a remarkable decline in the mortality effect of temperature extremes:
the impact of days with a mean temperature exceeding 80° F has declined by about 70% over the course
of the 20th century in the United States, with almost the entire decline occurring after 1960. The result
is that there are about 14,000 fewer fatalities annually than if the pre-1960 impacts of mortality still
prevailed. At the same time, the mortality effect of cold temperatures declined by a substantially
smaller amount. In effect, U.S. residents adapted in ways that leave them largely protected from
extreme heat.
The second part of the analysis aims to uncover the adaptations that muted the relationship
between mortality and high temperatures. We focus attention on the spread of three health-related
innovations in the 20th century United States: residential electricity, access to health care, and

3
residential air conditioning (AC). There are good reasons to believe that these innovations mitigated the
health consequences of hot temperatures (in addition to providing other services). Electrification
enabled a wide variety of innovations including fans, refrigeration, and later air conditioning. Increased
access to health care allowed both preventative treatment and emergency intervention (e.g.,
intravenous administration of fluids in response to dehydration, see Almond, Chay and Greenstone
(2006)). Air conditioning made it possible to reduce the stress on their thermoregulatory systems during
periods of extreme heat.
The empirical results point to air conditioning as a central determinant of the reduction of the
mortality risk associated with high temperatures during the 20
th
century. Specifically, the diffusion of
residential AC after 1960 is related to a statistically significant and economically meaningful reduction in
the temperature-mortality relationship at high temperatures. Indeed, the adoption of residential air
conditioning explains essentially the entire decline in the relationship between mortality and days with
an average temperature exceeding 80 °F. In contrast, we find that electrification (represented by
residential electrification) and access to health care (represented by doctors per capita) are not
statistically related to changes in the temperature mortality relationship.
The mortality analysis is conducted with the most comprehensive set of data files ever compiled
on mortality and its determinants over the course of the 20th century in the United States or any other
country. The mortality data come from newly digitized state-by-month mortality counts from the United
States Vital Statistics records, which is merged with newly collected data at the state level on the
fraction of households with electricity and air conditioning and on the number of doctors per capita.
These data are matched to daily temperature data, aggregated at the state-month level, for the 1900-
2004 period.
These data are used to fit specifications that aim to produce credible estimates of the
relationship between mortality rates and high temperatures, as well as the adaptations that modify that
relationship. Specifically, the baseline specification includes state-by-month (e.g., Illinois-by-July) fixed
effects and year-by-month (e.g., 1927-by-March) fixed effects, so the estimates are identified from the
presumably random deviations from long-run state-by-month temperature distributions that remain
after non-parametric adjustment for national deviations in that year-by-month's temperature
distribution. The baseline specification also includes a quadratic time trend that varies at the state-by-
month level and in the preferred specification state-level per capita income that is allowed to have a
differential effect across months. Further, the models control for current and past exposure to
temperature, so the estimates are robust to short-term mortality displacement or “harvesting”.

4
Although quasi-experimental variation in AC adoption is unavailable, three sets of additional
results lend credibility to the findings about the importance of residential AC. First, residential AC
penetration rates do not affect the mortality consequences of days with temperatures below 80 °F,
suggesting that the adoption of AC is not coincident to factors that determine the overall mortality rate.
Second, the protective effect of residential AC against high temperature exposure is substantially larger
for populations that are more vulnerable (i.e., individuals age 65 or older and blacks, relative to whites).
Third, residential AC significantly lessened mortality rates due to causes of death that are physiologically
and epidemiologically related to high temperature exposure (e.g., cardiovascular and respiratory
diseases). In contrast, residential AC is not associated with causes of death where there is little evidence
of a physiological or epidemiological relationship with high temperature exposure (e.g., motor vehicle
accidents or infectious diseases).
The third part of the analysis develops a measure of the full consumer surplus associated with
residential AC, based on the application of Dubin-McFadden’s (1984) discrete-continuous model. This
analysis is conducted with household-level Census data on AC penetration rates and electricity
consumption, as well as data on electricity prices. We find that AC adoption increases average
household electricity consumption by about 1,100 kwh or 11.6%. We estimate that the gain in consumer
surplus associated with the adoption of residential AC ranged from $5 to $10 billion (2012$) annually at
the 1980 AC penetration rate, depending on the assumptions about the shape of the long run electricity
supply curve. This translates into an increase in consumer surplus per U.S. household in 1980 of $120 to
$240. The present value of US consumer surplus from the introduction of residential AC in 1960, which is
the first year in which we measure the AC penetration rate, ranges from $83 to $186 billion ($2012) with
a 5% discount rate.
The paper contributes to several literatures. First, a nascent literature that aims to uncover
adaptation opportunities that are available in response to climate change with existing technologies
(e.g., Auffhammer and Schlenker (2014), Klein et al. (2014), Hsiang and Narita (2012)). Second, there is a
voluminous literature that explains the tremendous increases in life expectancy over the course of the
20
th
century that has to date not recognized the systematic role of air conditioning (e.g., Cutler et al.
2006). Third, an important literature has examined the welfare consequences of technical progress in
household production, especially in appliances (e.g., Bailey (2006), Coen-Pirani et al. (2010), Greenwood
et al. (2005)).
The paper proceeds as follows. Section II presents the conceptual framework where we review
the physiological relationship that links temperature and health, and the mechanisms that link the

Frequently asked questions

Q1. What have the authors contributed in "Adapting to climate change: the remarkable decline in the u.s. temperature-mortality relationship over the 20 century" ?

This paper examines the evolution of the temperature-mortality relationship over the course of the entire 20 century in the United States both for its own interest but also to identify potentially useful adaptations that may be useful in the coming decades. Third, using Dubin-McFadden ’ s discrete-continuous model, the authors estimate that the present value of US consumer surplus from the introduction of residential air conditioning ( AC ) in 1960 ranges from $ 83 to $ 186 billion ( $ 2012 ) with a 5 % discount rate. Deschenes is the corresponding author and can be contacted at olivier @ econ. The authors thank Daron Acemoglu, Robin Burgess, their discussant Peter Nilsson, conference participants at the “ Climate and the Economy ” Conference and seminar participants at many universities for their comments. 

Q2. Why is air conditioning the prominent technology used to reduce the risks of heat stress?

Due to the strong connection between ambient temperature and heat-related illness, air conditioning is probably the most prominent technology used to reduce the risks of heat stress. 

Q3. How many percent of the dwellings had electricity?

Eight-five percent of urban and rural non-farm dwellings had electricity, while only 10 percent of farm dwellings had electricity. 

Q4. What was the first thing that was done to bring water into many households?

electricity access made the pumping of water feasible on a wide scale, bringing running water into many households for the first time. 

Q5. Why do the authors focus on doctors per capita?

The authors focus on doctors per capita, largely because they provided, and continue to provide, the majority of patient care outside and inside hospitals. 

Q6. In what year did the last year of the electricity program exist?

In 1956, the last year national summaries are available, 99 percent of all dwellings and 96 percent of farm dwellings had electricity. 

Q7. What is the relationship between temperature and mortality?

Exposure to cold temperatures causes cardiovascular stress due to changes in blood pressure, vasoconstriction, and an increase in blood viscosity (which can lead to clots), as well as increased levels of red blood cell counts, plasma cholesterol, and plasma fibrinogen (Huynen et al. 2001). 

Q8. Who had the contact with the patients?

Health stock was largely attributable to primary care doctors, who had the most contact with the patients and could treat some heat-related complications, like heat exhaustion and heat stroke. 

Q9. What is the important aspect of the study?

It could also raise overall health capital, which would help populations tolerate the additional stress from exposure to temperature extremes. 

Q10. What is the definition of a temperature-mortality relationship?

exposure to temperatures outside these ranges or exposure to temperature extremes for prolonged periods of time endangers human health and can result in mortality. 

Q11. What is the role of the heat-related mortality?

any of the innovations could reduce heat-related mortality indirectly by raising health capital throughout the year and, thus, mitigating mortality risk from a heat-related health shock (or any health shock for that matter). 

Q12. How do the authors estimate the welfare effects of AC?

the authors estimate the welfare consequences of the reductions in hot day mortality by multiplying the number of avoided fatalities by the value of a statistical life. 

Q13. What is the reason for the population shift to the South?

it has been claimed that the availability of residential AC is a major reason for the population shift to the South over the last several decades (Gordon (2000), Holmes (1998)).