We
propose
strategies
to
estimate
and
make
inference
on
key
features
of
heterogeneous
effects
in
randomized
experiments.These
include
best
linear
predictors
the
using
machine
learning
proxies,
average
sorted
by
impact
groups,
characteristics
most
least
impacted
units.The
approach
is
valid
high
dimensional
settings,
where
are
proxied
(but
not
necessarily
consistently
estimated)
predictive
causal
methods.We
post-process
these
proxies
into
estimates
features.Our
generic,
it
can
be
used
conjunction
with
penalized
methods,
neural
networks,
random
forests,
boosted
trees,
ensemble
both
causal.Estimation
based
repeated
data
splitting
avoid
overfitting
achieve
validity.We
use
quantile
aggregation
results
across
many
potential
splits,
particular
taking
medians
p-values
other
quantiles
confidence
intervals.We
show
that
lowers
estimation
risks
over
a
single
split
procedure,
establish
its
principal
inferential
properties.Finally,
our
analysis
reveals
ways
build
provably
better
through
learning:
we
objective
functions
develop
construct
effects,
obtain
initial
step.We
illustrate
tools
learners
field
experiment
evaluates
combination
nudges
stimulate
demand
for
immunization
India.
American Economic Review,
Journal Year:
2019,
Volume and Issue:
109(12), P. 4178 - 4219
Published: Nov. 26, 2019
We
estimate
the
causal
effects
of
acute
fine
particulate
matter
exposure
on
mortality,
health
care
use,
and
medical
costs
among
US
elderly
using
Medicare
data.
instrument
for
air
pollution
changes
in
local
wind
direction
develop
a
new
approach
that
uses
machine
learning
to
life-years
lost
due
exposure.
Finally,
we
characterize
treatment
effect
heterogeneity
both
life
expectancy
generic
inference.
Both
approaches
find
mortality
are
concentrated
about
25
percent
population.
Proceedings of the National Academy of Sciences,
Journal Year:
2021,
Volume and Issue:
118(2)
Published: Jan. 11, 2021
Recent
dramatic
and
deadly
increases
in
global
wildfire
activity
have
increased
attention
on
the
causes
of
wildfires,
their
consequences,
how
risk
from
might
be
mitigated.
Here
we
bring
together
data
changing
societal
burden
United
States.
We
estimate
that
nearly
50
million
homes
are
currently
wildland–urban
interface
States,
a
number
increasing
by
1
houses
every
3
y.
To
illustrate
changes
affect
air
pollution
related
health
outcomes,
these
linkages
guide
future
science
policy,
develop
statistical
model
relates
satellite-based
fire
smoke
to
information
monitoring
stations.
Using
model,
wildfires
accounted
for
up
25%
PM
2.5
(particulate
matter
with
diameter
<2.5
μm)
recent
years
across
half
some
Western
regions,
spatial
patterns
ambient
exposure
do
not
follow
traditional
socioeconomic
gradients.
combine
stylized
scenarios
show
fuel
management
interventions
could
large
benefits
impacts
climate-change–induced
approach
projected
overall
temperature-related
mortality
climate
change—but
both
estimates
remain
uncertain.
use
results
highlight
important
areas
research
draw
lessons
policy.
The Economic Journal,
Journal Year:
2021,
Volume and Issue:
131(640), P. 3241 - 3273
Published: April 27, 2021
Abstract
We
provide
nationwide
causal
estimates
of
air
pollution's
effect
on
short-run
productivity
for
China's
manufacturing
sector
from
1998
to
2007.
Using
thermal
inversions
as
an
instrument,
a
1
µg/m3
decrease
in
PM2.5
increases
by
0.82%
(elasticity
−0.44).
Increased
hiring
attenuates
the
elasticity
−0.17.
Differential
effects
trade
shock
coastal
versus
inner
regions
imply
pollution
output
1.43.
Simulating
dynamic
general-equilibrium
model
yields
−0.28
with
respect
PM2.5.
An
exogenous
1%
gross
domestic
product
0.039%.
Nature Communications,
Journal Year:
2018,
Volume and Issue:
9(1)
Published: Nov. 16, 2018
Local
air
quality
co-benefits
can
provide
complementary
support
for
ambitious
climate
action
and
enable
progress
on
related
Sustainable
Development
Goals.
Here
we
show
that
the
transformation
of
energy
system
implied
by
emission
reduction
pledges
brought
forward
in
context
Paris
Agreement
change
(Nationally
Determined
Contributions
or
NDCs)
substantially
reduces
local
pollution
across
globe.
The
NDCs
could
avoid
between
71
99
thousand
premature
deaths
annually
2030
compared
to
a
reference
case,
depending
stringency
direct
controls.
A
more
2
°C-compatible
pathway
raises
number
avoided
from
178-346
2030,
up
0.7-1.5
million
year
2050.
Air
morbidity,
mortality,
agriculture
globally
offset
costs
policy.
An
integrated
policy
perspective
is
needed
maximise
benefits
health.
OECD Economics Department working papers,
Journal Year:
2019,
Volume and Issue:
unknown
Published: Dec. 12, 2019
This
study
provides
the
first
evidence
that
air
pollution
causes
economy-wide
reductions
in
market
economic
activity
based
on
data
for
Europe.
The
analysis
combines
satellite-based
measures
of
with
statistics
regional
at
NUTS-3
level
throughout
European
Union
over
period
2000-15.
An
instrumental
variables
approach
thermal
inversions
is
used
to
identify
causal
impact
activity.
estimates
show
a
1μg/m3
increase
PM2.5
concentration
(or
10%
sample
mean)
0.8%
reduction
real
GDP
same
year.
Ninety-five
per
cent
this
due
output
worker,
which
can
occur
through
greater
absenteeism
work
or
reduced
labour
productivity.
Therefore,
results
suggest
public
policies
reduce
may
contribute
positively
growth.
Indeed,
large
benefits
from
uncovered
compare
relatively
small
abatement
costs.
Thus,
more
stringent
quality
regulations
could
be
warranted
solely
grounds,
even
ignoring
terms
avoided
mortality.
