« Back to Results
Marriott Philadelphia Downtown, Independence Ballroom II
Hosted By:
Association of Environmental and Resource Economists
emitting more CO2 will warm the earth, but by how much is uncertain despite large amounts of research
devoted to this area. How sensitive the climate is to CO2, the climate sensitivity, is a major determinant
of optimal policy stringency. Thus, uncertainty about the climate sensitivity and anticipated refinements
of our climate sensitivity beliefs will also be critical for contemporaneous climate policy.
Uncertainty and learning about the climate sensitivity have been studied by economists through
dynamic stochastic climate-economy models. Although the dynamics of our beliefs about the climate
sensitivity are surely important, economists have largely abstained from trying to accurately calibrate
this component of the climate-economy system due to computational burdens. This has resulted in stark
inconsistencies between climate-economy models suggesting relatively fast rates of learning, and the
reality that we have not honed down our beliefs for decades.
Here we calibrate the informational dynamics of a Bayesian climate policymaker. Using frontier
computational methods, we nest a general Bayesian learning framework into a dynamic stochastic
climate-economy model in a way that closely matches how climate scientists actually estimate the
climate sensitivity distribution using historical data. We generalize previous approaches by not making
distributional assumptions about climate sensitivity beliefs and the climate data generating process that
have resulted in artificially fast learning in climate-economy models. We show that our approach can
correctly recover a climate sensitivity distribution consistent with climate scientists’ current best
estimates, and it can accurately recover the dynamics of real world climate sensitivity beliefs.
Simulations of optimal policy and beliefs suggest that future learning will be several times slower than
suggested by previous work, and that future climate policy may not be as flexible to new observations of
CO2 and temperature as once believed.
growth, greenhouse gas emissions, and global warming. The present analytic climate economy
(ACE) competes quantitatively with numeric models used to derive the US federal social cost of
carbon. The analytic solution permits new insights into the evaluation of climate change, and it
overcomes numeric obstacles in incorporating uncertainties (Bellman’s curse of dimensionality).
The paper relates the optimal carbon tax directly to the characteristics of the carbon cycle and
the climate system.
Today’s policy advising remains in the hands of essentially deterministic models that explore
and average large samples of deterministic worlds. Yet, recent findings suggest that uncertainty
surrounding climate change could be the major driver of mitigation policy and welfare loss. In
highly stylized models, Pindyck (2013) and Weitzman (2009) argue that uncertainty outweighs
all other evaluation components and even makes the discount rate irrelevant. In contrast, ACE
shows that the integrated assessment’s sensitivity to discounting is even higher under
uncertainty than under certainty. ACE also show that nature’s stochasticity and epistemological
uncertainty imply opposing sensitivies, and that the Bayesian learning framework is the most
sensitive because updates change the long-run picture of the future.
Guided by the long run risk literature in asset pricing, ACE disentangles risk aversion from
consumption smoothing to calibrate the risk-free discount rate and risk premia separately.
Models lacking this feature are forced to either discount the future too highly, or to disrespect
the risk premia. I show that the relevant risk aversion for climate change evaluation is not Arrow
Pratt’s measure of risk aversion, but by how much Arrow Pratt risk aversion exceeds the desire
to smooth consumption over time (intrinsic aversion to risk). Higher moments of the uncertainty
distribution are evaluated with higher powers of such risk aversion.
show that new apartments on historical multi-century leases trade at a non-zero
discount relative to property owned in perpetuity. Descriptive regressions indicate that
new apartments with 825 to 986 years of tenure remaining are priced 4 to 6% below
new apartments under perpetual ownership contracts that are otherwise comparable.
We consider an empirical model in which asset value is decomposed into the utility of
housing services and a second factor that shifts with asset tenure and the discount rate
schedule. Exploiting the supply of new property with tenure ranging from multiple
decades to multiple centuries, we estimate the discount rate schedule, restricting it to
vary smoothly over time through alternative parametric forms. Across different
specifications and subsamples, we estimate discount rates that decline over time and,
to accommodate the observed price differences, fall to 0.5% p.a. by year 400-500. The
finding that households making sizable transactions do not entirely discount benefits
accruing many centuries from today is new to the empirical literature on discounting
and, with the appropriate risk adjustment, of relevance to evaluating climate-change
investments.
Climate Change: Connecting Theory with Empirics
Paper Session
Sunday, Jan. 7, 2018 8:00 AM - 10:00 AM
- Chair: Stephie Fried, Arizona State University
Calibrating Informational Dynamics: Learning About the Sensitivity of the Climate to Emissions
Abstract
Carbon dioxide (CO2) concentrations are projected to increase for the foreseeable future. We knowemitting more CO2 will warm the earth, but by how much is uncertain despite large amounts of research
devoted to this area. How sensitive the climate is to CO2, the climate sensitivity, is a major determinant
of optimal policy stringency. Thus, uncertainty about the climate sensitivity and anticipated refinements
of our climate sensitivity beliefs will also be critical for contemporaneous climate policy.
Uncertainty and learning about the climate sensitivity have been studied by economists through
dynamic stochastic climate-economy models. Although the dynamics of our beliefs about the climate
sensitivity are surely important, economists have largely abstained from trying to accurately calibrate
this component of the climate-economy system due to computational burdens. This has resulted in stark
inconsistencies between climate-economy models suggesting relatively fast rates of learning, and the
reality that we have not honed down our beliefs for decades.
Here we calibrate the informational dynamics of a Bayesian climate policymaker. Using frontier
computational methods, we nest a general Bayesian learning framework into a dynamic stochastic
climate-economy model in a way that closely matches how climate scientists actually estimate the
climate sensitivity distribution using historical data. We generalize previous approaches by not making
distributional assumptions about climate sensitivity beliefs and the climate data generating process that
have resulted in artificially fast learning in climate-economy models. We show that our approach can
correctly recover a climate sensitivity distribution consistent with climate scientists’ current best
estimates, and it can accurately recover the dynamics of real world climate sensitivity beliefs.
Simulations of optimal policy and beliefs suggest that future learning will be several times slower than
suggested by previous work, and that future climate policy may not be as flexible to new observations of
CO2 and temperature as once believed.
ACE - Analytic Climate Economy (with Temperature and Uncertainty)
Abstract
Integrated assessment of climate change analyzes the interactions of long-term economicgrowth, greenhouse gas emissions, and global warming. The present analytic climate economy
(ACE) competes quantitatively with numeric models used to derive the US federal social cost of
carbon. The analytic solution permits new insights into the evaluation of climate change, and it
overcomes numeric obstacles in incorporating uncertainties (Bellman’s curse of dimensionality).
The paper relates the optimal carbon tax directly to the characteristics of the carbon cycle and
the climate system.
Today’s policy advising remains in the hands of essentially deterministic models that explore
and average large samples of deterministic worlds. Yet, recent findings suggest that uncertainty
surrounding climate change could be the major driver of mitigation policy and welfare loss. In
highly stylized models, Pindyck (2013) and Weitzman (2009) argue that uncertainty outweighs
all other evaluation components and even makes the discount rate irrelevant. In contrast, ACE
shows that the integrated assessment’s sensitivity to discounting is even higher under
uncertainty than under certainty. ACE also show that nature’s stochasticity and epistemological
uncertainty imply opposing sensitivies, and that the Bayesian learning framework is the most
sensitive because updates change the long-run picture of the future.
Guided by the long run risk literature in asset pricing, ACE disentangles risk aversion from
consumption smoothing to calibrate the risk-free discount rate and risk premia separately.
Models lacking this feature are forced to either discount the future too highly, or to disrespect
the risk premia. I show that the relevant risk aversion for climate change evaluation is not Arrow
Pratt’s measure of risk aversion, but by how much Arrow Pratt risk aversion exceeds the desire
to smooth consumption over time (intrinsic aversion to risk). Higher moments of the uncertainty
distribution are evaluated with higher powers of such risk aversion.
How Do Households Discount Over Centuries? Evidence From Singapore’s Private Housing Market
Abstract
We examine Singapore's fairly homogeneous private-housing market andshow that new apartments on historical multi-century leases trade at a non-zero
discount relative to property owned in perpetuity. Descriptive regressions indicate that
new apartments with 825 to 986 years of tenure remaining are priced 4 to 6% below
new apartments under perpetual ownership contracts that are otherwise comparable.
We consider an empirical model in which asset value is decomposed into the utility of
housing services and a second factor that shifts with asset tenure and the discount rate
schedule. Exploiting the supply of new property with tenure ranging from multiple
decades to multiple centuries, we estimate the discount rate schedule, restricting it to
vary smoothly over time through alternative parametric forms. Across different
specifications and subsamples, we estimate discount rates that decline over time and,
to accommodate the observed price differences, fall to 0.5% p.a. by year 400-500. The
finding that households making sizable transactions do not entirely discount benefits
accruing many centuries from today is new to the empirical literature on discounting
and, with the appropriate risk adjustment, of relevance to evaluating climate-change
investments.
Discussant(s)
Laura Bakkensen
,
University of Arizona
Lint Barrage
,
Brown University
Stephie Fried
,
Arizona State University
Christian Gollier
,
Toulouse School of Economics
JEL Classifications
- Q5 - Environmental Economics