The B.E. Journal of Economic
Analysis & Policy
Symposium
Volume 10, Issue 2 2010 Article 19
DISTRIBUTIONAL ASPECTS OFENERGYAND CLIMATE POLICY
On the Economics of Climate Policy
GaryS. Becker∗
Kevin M. Murphy†
Robert H. Topel‡
∗
University of Chicago,gbecker@uchicago.edu
†
University of Chicago,kevin.murphy@chicagobooth.edu
‡
University of Chicago,robert.topel@chicagobooth.edu
RecommendedCitation
GaryS. Becker, Kevin M. Murphy,and Robert H. Topel (2010) “Onthe Economics of Climate
Policy,” The B.E. Journal of Economic Analysis & Policy: Vol. 10: Iss. 2 (Symposium), Article
19.
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
On the Economics of Climate Policy∗
Gary S. Becker, Kevin M. Murphy, and Robert H. Topel
Abstract
We analyze the central features of economic policies to mitigate climate change. The basic
structure of Pigouvian “carbon pricing” is shown to follow from a standard Hotelling problem
for the intertemporal pricing of an exhaustible resource. We extend this analysis to consider the
strength and timing of research incentives, the costs of implementation delay and the impact of anticipated
future technologies on current carbon prices. We study a variety of issues related to the
valuation of climate investments, including uncertainty as to the future timing and distribution of
climate impacts and the appropriate social rate of discount for valuing policies. Under reasonable
circumstances the insurance properties of climate investments may warrant unusually low discount
rates. We use the same framework to argue that policy makers in developing countries will discount
the expected returns from climate investments more heavily, because such investments have
weaker insurance value in the developing world.
∗
Prepared for a conference on “Distributional Aspects of Energy and Climate Policy” jointly sponsored
by the University of Chicago, Resources for the Future, and the University of Illinois. We
are grateful for comments from our discussant, Michael Greenstone, conference participants, Don
Fullerton, and an anonymous referee. We also acknowledge support from George J. Stigler Center
for the Study of the Economy and the State and from the University of Chicago Energy Initiative.
Gary S. Becker, University Professor, Booth School of Business and Department of Economics,
University of Chicago. Kevin M. Murphy, George J. Stigler Distinguished Service Professor,
Booth School of Business and Department of Economics, University of Chicago. Robert H. Topel,
Isidore and Gladys J. Brown Distinguished Service Professor, Booth School of Business, University
of Chicago.
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
1. Introduction
Energy is essential to the maintenance and spread of economic welfare. At a
point in time, individuals in richer countries such as the US and Canada use much
more energy than individuals in poorer countries. Over time, long run growth in
living standards is strongly associated with rising energy use, especially in
developing countries.1
There is little to indicate that these patterns might change,
so that future growth and the escape of developing countries from current levels
of poverty hinge on the existence and use of abundant energy supplies.
Yet rising worldwide demands for energy run up against new evidence of
the social costs of energy use. The broad consensus of scientific research is that
the continued dependence of economic activity on carbon-based fuels and their
associated emission of greenhouse gases (GHG) create risks of substantial future
changes in earth’s climate, along with associated harm to the welfare of future
generations. This “new” knowledge of anthropogenic climate change has
motivated national and international efforts to regulate the use of carbon-based
energy sources, and to promote the development and use of “clean” energy
alternatives. For example, the Obama administration recently “committed” the
US to achieve an 80 percent reduction in carbon emissions by 2050, even while
enabling legislation to begin the regulation of such emissions languishes in
Congress. In Europe, an incipient “cap-and-trade” market for emissions permits
is in place, while the state of California is developing unilateral action along the
same lines. Broadly-based international efforts have met with little success, as
evidenced by the failure of the Kyoto (2000) and Copenhagen (2009) negotiations
to achieve implementable frameworks for reducing GHG emissions.
These initiatives must confront several daunting challenges to the
successful design and implementation of a useful energy-climate policy. First,
current generations—who are the ones that get to decide—must be convinced that
the future costs of climate change are worthy of current concern. This hasn’t been
achieved. Second, current generations must agree to forego use of abundant
1
Our point is that economic development (almost) universally expands energy use. Energy use
per unit of income—sometimes called the “energy intensity” of income—is generally declining,
both worldwide and within countries. This reflects both technical advances in energy use and
changes in the composition of GDP within countries. But GDP growth rates over long periods
almost always exceed the rate of decline in energy intensity of GDP, so that overall energy
demand rises with income, especially in developing countries. For example, between 1980 and
2007 GDP growth in China has averaged 10 percent per year, while energy intensity has declined
at “only” 5.3 percent per year. The corresponding figures for India are 6.1 percent GDP growth
and 2 percent decline in energy intensity. In the U.S. GDP growth has averaged 2.94 percent since
1980, and energy intensity of GDP has declined at 2.03 percent. For a complete tabulation of
energy intensity see World Bank data at:
http://data.worldbank.org/indicator/EG.GDP.PUSE.KO.PP.KD
1
Becker et al.: On the Economics of Climate Policy
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
carbon-based energy sources in order to mitigate uncertain harm to generations of
the distant future. We know of no good examples of such sacrifice. Third, even
if current generations are convinced that mutual sacrifice would be a good thing,
effective policies require the largely voluntary yet global cooperation of nations in
a setting where non-cooperation offers substantial rewards. Again, we are
unaware of the success, or even the formation, of similar policies.
These problems are created by the fact that climate is a “global public
good.” In terms of climate impact earth’s atmosphere doesn’t much care where
GHG emissions come from—a ton of carbon emitted in India has the same impact
as one from Canada, so the atmosphere is over-used by all in a classic example of
the tragedy of the commons. Meaningful efforts to successfully correct this
externality must then hinge on collective and harmonized action by nations worldwide.
Yet efforts at cooperation are hampered by the same free-rider incentives
that created the problem in the first place—the benefits of carbon-based energy
use are current and highly focused, while the social costs are greatly delayed,
difficult to (currently) discern or measure, and highly dispersed. In addition, as
we argue below, the social costs of climate change and the benefits of mitigation
policies are not uniform, which leads to divergent valuations of social investments
in “climate capital.” This is especially true when, as here, the distributions of
returns on such social investments are country-specific and highly uncertain.
Then policies such as widely-discussed carbon taxes or cap-and-trade schemes
offer much different risk-reward tradeoffs to developing countries, such as China
or India, than to developed countries like the US. These divergent valuations help
explain the current lack of progress in international negotiations over climate
policy, such as Copenhagen (2009), and what we believe are the limited prospects
for cooperation going forward.
From (very) high altitude, the economics of anthropogenic climate change
is a standard problem of externality—current users of carbon-based fuels do not
bear the environmental costs of energy consumption, so they use too much of the
stuff, and too little of “clean” alternatives. The problem occurs because some
resource—here the atmosphere—is unpriced and so overused. The idealized
textbook market intervention is to price the overused resource, equating the
private and social costs of its use. This can be accomplished via a Pigouvian tax
(or its equivalent) equal to the marginal external cost of using a unit of carbonbased
fuels. The resulting ideal “carbon-price” would exactly balance the benefits
of additional carbon emissions, which occur now, against their costs, which are
spread over the near and distant future.
This basic solution to the externality problem is familiar and
straightforward. It is central to virtually all serious national and international
2
The B.E. Journal of Economic Analysis & Policy, Vol. 10 [2010], Iss. 2 (Symposium), Art. 19
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
policy proposals to deal with climate change, including the Waxman-Markey2
and
Lieberman-Warner3
bills in the US House and Senate, the design of cap-and-trade
policies in the EU, and the tentative framework discussed in the recent
Copenhagen negotiations. But its conceptual simplicity is superficial—the actual
design and implementation of such policies faces daunting challenges and
unresolved questions. Our analysis seeks to contribute to a number of unresolved
issues in the design and effects of policies to mitigate climate change. These
include:
1. Valuing future costs: The social costs of current GHG emissions are
uncertain and spread over the distant future. How should current policy
value the costs of climate damage, which will fall mainly on future
generations? Should the future benefits of investments in “climate
capital” be discounted at market rates or, as some have argued, at much
lower rates? How will these valuations differ across major countries, the
large majority of which must cooperate to achieve efficient policy
outcomes?
2. Uncertainty: How does the great uncertainty regarding the extent and
costs of future environmental harm affect current strategies, social
investments, and valuations?
3. Catastrophic climate change: Among the uncertainties is the possibility of
catastrophic outcomes that could greatly reduce future living standards or
endanger future populations. How should current policy value and
mitigate these possibilities?
4. Market responses to climate policies: At its barest level, “carbon pricing”
is a market-based solution that relies on market responses to efficiently
designed incentives. How will markets respond to policy-generated
incentives? Will market responses enhance or constrain the effects of
policies?
5. Innovation incentives, policy design and the costs of delay: How does an
efficient policy affect research incentives and the pace technical progress
in alternative energy sources and in mitigation? If technical breakthroughs
are likely to be the ultimate solution to the energy “problem,” are
incentives to innovate harmed by delays in implementing an optimal
policy? How should the prospect of future innovation affect current
policy?
The paper is organized as follows. Section 2 develops the basic features
of optimal carbon pricing, which we relate to a standard Hotelling problem for the
2
http://www.govtrack.us/congress/bill.xpd?bill=h111-2454
3
http://www.govtrack.us/congress/bill.xpd?bill=s110-2191
3
Becker et al.: On the Economics of Climate Policy
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
intertemporal pricing of a depleting resource. We extend the analysis to consider
the strength and timing of research and development incentives, the costs of delay
in implementing an optimal policy, and the impact of anticipated future
technologies, and their form, on current carbon prices. In Section 3 we analyze a
variety of issues related to the valuation of climate policies, including uncertainty
as to the future timing and distribution of climate impacts. We pay particular
attention to the appropriate social rate of discount for valuing policies, showing
that under certain reasonable circumstances the insurance properties of climate
investments may warrant unusually low discount rates. We use the same
framework to argue that policy makers in developing countries will discount the
expected returns from climate investments more heavily, because such
investments have weaker insurance value in the developing world. Section 4
concludes.
2. Features of Efficient Climate Change Policies
The scientific foundations for anthropogenic climate change indicate that current
emissions of GHGs create environmental and other costs that are (1) greatly
delayed, (2) very long-lasting, and (3) highly uncertain.4
This is because the flow
of CO2 to the atmosphere has a long lasting impact on the stock of atmospheric
CO2, as reabsorption is very slow. In turn, the growth in global temperature lags
the atmospheric stock of CO2 because, for example, melting of ice caps reduces
earth’s albedo (reflectivity) and oceans warm slowly. Many costs are likely to lag
a rise in temperature—for example, rising sea levels would be driven by melting
of ice caps, which would follow a prolonged warming period. Finally,
uncertainty as to environmental feedbacks and other impacts includes the
prospects for “catastrophes” of various forms. Because of these features, policies
that would mitigate these effects must balance costs and benefits over hundreds of
years, and subject to large and costly contingencies, which make the problem of
policy design a good deal more daunting than the usual project evaluation.
2.1 Carbon Pricing as an Exhaustible Resource Problem
To illustrate central elements of dynamic carbon pricing and its connection to key
assumptions about preferences, growth and technology, consider a simple
certainty framework in which the target cap on atmospheric concentration of
GHGs at some endogenous future date T (say in T=200 years) is the goal of
environmental policy. Denoting the concentration of GHGs at date t by Qt, this
4
See Archer (2007) and (2009) for useful summaries of the state of climate science research on
global warming.
4
The B.E. Journal of Economic Analysis & Policy, Vol. 10 [2010], Iss. 2 (Symposium), Art. 19
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
terminal condition is TQ Q . Stated in this way, the optimal policy solves a
Hotelling problem for allocating the use of an exhaustible resource over time—
where here the exhaustible resource is the capacity of the atmosphere to “safely”
hold a given concentration of GHGs.5
Let the private social (consumer plus producer) surplus from current
emissions, q, be ( , , )t t t tV q b y where tb is the unit cost of carbon-free energy
sources at date t and ty is income. Finally, let the technology for mitigating
emissions be represented by the cost 1
( )t t tC s 
, where ts is the amount of period-t
emissions avoided through mitigation activities and t indexes the evolving
efficiency of mitigation—higher values of t reduce the costs of emissions
mitigation. For example, ts might be the amount of period t emissions that are
eliminated by sequestration or other technologies, and  makes the process more
efficient. With these definitions, the policy problem is to maximize the present
discounted value of social surplus.
(1)
1
,
0
( ( , , ) ( ))
. .
rt
t t t t t t t
q s
t
t t t t T
MaxW V q b y C s e dt
s t Q q s aQ and Q Q
 


 
   

where r is the rate of interest used to discount future environmental costs (the rate
of return on investments in environmental capital) and a is the rate at which
atmospheric GHGs are reabsorbed. We shall have much more to say about r later,
but for now we simply take it as given without pondering how large or small it
could or should be.6
Letting ( )tV  denote the derivative of ( )tV  with respect to tq , the basic
solution to (1) has a familiar structure:
5
There need not be a fixed target level of GHG for this analysis to apply. As long as the effects of
climate change are simply a function of the stock of GHG at some future date (200 years in our
example), the optimal program will need to solve this same Hotelling problem given the optimal
level of GHG at the terminal date.
6
The appropriate rate of return on social investments in climate capital will depend on insurance
properties of the investment’s return. Projects that pay off by mitigating climate-related
catastrophes may have discount rates that are well below the market rates of return on other risky
assets, and even below the risk free rate. We take up these issues in Section 3, below.
5
Becker et al.: On the Economics of Climate Policy
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
(2a)
( )
0
1 1 ( )
0
( , , )
( )
r a t
t t t t
r a t
t t t t
V q b y P e
C s P e

  
 
 
In (2a) both the marginal value of using and the marginal cost of eliminating a
unit of emissions are equated to the period-t “carbon price” ( )
0
r a t
tP e P
 , which
represents the scarcity value of a “unit” of the otherwise unpriced absorptive
capacity of the atmosphere. This price is the outcome of an ideal Pigouvian tax or
cap-and-trade system, so that tP equates the marginal benefit of q to current users
and the (present value of) incremental costs imposed on future generations.7
2.2 Carbon Pricing, Timing and the Returns to Innovation
The fact that the socially optimal carbon price rises at the rate of interest (plus
absorption) is a well-known property of this and other exhaustible resource
problems—Nordhaus (2007) refers to the rate of growth r+a as the “net carbon
interest rate.8
It is a condition for intertemporal efficiency in the use and
mitigation of emissions, equating the value of benefits from creating incremental
emissions (due to energy use) to the present value of costs. Less appreciated is
how this property of an optimal policy impacts the social value of innovations,
incentives to innovate and the cost of waiting to implement the policy.
To fix ideas with a not-entirely-fanciful example, think of a current
investment in technology that could eliminate one unit of carbon emissions at
some arbitrary future date, t—a one-period “carbon eating tree.”9
The present
value of this unit reduction in future emissions is ( )
0 0
rt r a t rt at
tPe P e e P e  
  ,
which implies that the time profile of values is independent of the interest rate, r.
If a=0, the present discounted value of the innovation is independent of how far
in the future it pays off, t, because the value of the gain rises at the interest rate.
And if a>0 the present value actually rises with t because the time-t value of the
innovation rises faster than the interest rate. In effect, the value of the innovation
is undiscounted.
The result is even stronger if innovation is scalable. Think of an
innovation that would reduce the incremental cost of mitigation at some future t,
7
Nordhaus (2007a,b) and (2008) are good summaries of the state of economic modeling applied to
global warming and climate policy. Many of the same analytical tools appear in Stern (2006).
8
The original statement is in Hotelling (1931). Treating Q as an exhaustible resource, the
condition is that “owners” of the resource must be indifferent between selling a unit today and
holding it for future use. Here tP is the per-period shadow value of relaxing the constraint on
GHG concentrations, Q .
9
E.g. Dyson (2008).
6
The B.E. Journal of Economic Analysis & Policy, Vol. 10 [2010], Iss. 2 (Symposium), Art. 19
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
raising t by ln td  >0. The present value of this cost reduction (per unit change
in ln  ) is
1 1
0
( )rt
t t t t
rt
t t
at
t
e C s s
e Ps
P e s
   




Here the unit value of the innovation is independent of r, and also of t if a=0, but
the reduction in incremental cost is scalable because it applies to all units, ts . But
ts rises over time because tP is increasing, so the present value of the innovation
also rises with t—the gain has larger present value the farther in the future it
occurs. Applied to all periods, a scalable technical advance in mitigation that
applies from the present day forward is worth
(3) ln 0
0
at
t
t
W P s e dt

 
Even with a=0, the value of the gain applies to the quantity of mitigation in all
future periods with equal weights. If a>0 then future periods get more weight
than the present. And of course the result applies to other types of innovation,
such as an advance that would improve the consumption efficiency (surplus) per
unit of emissions, like a change in fuel efficiency of cars. All such gains are
valued at the rising Pigouvian price tP , which “undoes” the effect of discounting
in present value calculations.
These conclusions may appear anomalous, and it is easy to extend the
policy problem (1) to include factors that would cause the optimal atmospheric
price Pt to rise more slowly, so that delay is more costly. For example, if we
amend the optimal policy to include possible environmental damages ( )tD Q from
rising atmospheric concentrations along the trajectory to T, then the current
optimal carbon price is continuously updated, incorporating the impact of current
emissions on future damages:
1ln
( )t
t t
d P
r a P D Q
dt

  
where ( )tD Q > 0 is the current period marginal damage from emissions,
equivalent to the effective current period “rental price” of atmosphere. This
7
Becker et al.: On the Economics of Climate Policy
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
reduces the growth rate of tP relative to the standard Hotelling solution, but
without negating the broader point, which is that innovations yield greatest value
when the carbon price is high, and in the optimal policy that price is rising over
time.
The central lesson about valuing progress in (3) is simple, but it has
important implications for interpreting both the form of policy responses to
climate change as well as the urgency with which those policies are implemented
and the costs of delay. The slow progress of international negotiations in
gatherings such as Kyoto and Copenhagen is widely lamented, as is similarly
slow progress in crafting and adopting enabling legislation in the US and other
countries. Do these delays adversely impact incentives to find “solutions” in the
form of technologies that would reduce the carbon impact of energy
consumption? Is a sense of urgency warranted?
If we assume that slow progress toward implementing a policy is just
that—slow progress that will eventually result in widely applied carbon pricing
that reflects social costs—then equation (3) indicates that the costs of delay are
small. To illustrate, use (3) for the present discounted value of a cost-saving
innovation that requires substantial up-front R&D effort. Realization of these
incentives requires two things: an initial incentive, 0P , that signals the current
scarcity value of emissions, and a commitment to a time path for that value in the
future. Given these, delaying the start of this payoff for d years would not much
affect its present value, even if the initial d years of a payoff stream were
foregone. And if the whole program is simply pushed back the payoff would
likely rise because the initial price dP would increase by more than simple
interest (because interim unpriced emissions tighten the ultimate constraint),
which also raises s. The result is that the current value of innovation incentives
and the social gain from innovations are not much harmed and may even be
increased by delay.
This analysis assumes that an optimal carbon pricing program is
eventually implemented—that negotiations and legislation result in something
useful in terms of price signals and commitment to policy. The message is not
that delay is costless—the tighter constraint and necessarily higher carbon price
caused by delay demonstrate the costs—but rather that the returns to climaterelated
innovations are not much reduced by delay in implementing well-designed
incentives. And if the ultimate efficiency gain is likely to derive from currently
unforeseen innovations driven by carbon pricing, rather than simply by business
as usual along a rising price path, it is likely that delays of a few years don’t much
impact that outcome. Put differently, it is far more important to get the form of
policy right—including believable commitments to the level and time path of
future carbon prices—than to get a policy done quickly.
8
The B.E. Journal of Economic Analysis & Policy, Vol. 10 [2010], Iss. 2 (Symposium), Art. 19
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
2.3 Factors Affecting the Impact of Policy
Conditions (2a) for the rate of change price of the carbon price embed the
properties of evolving demand and supply for carbon-based fuels, via the social
surplus ( )tV z, and the
evolution of technology. It’s worth being explicit about these, because
expectations of how demand and technology will evolve in the future are essential
ingredients of current policy and the optimal level and timing of mitigation
activities. Using the usual notation for time rates of change (e.g.
dq
q
dt
 )
displacement of (2a) gives the rates of change of emissions (q) and mitigation (s):
(2b)
[ ]
[ ] [1 ]
t t t t tt
t t t t
q r a b y
s r a
  
  
  
 
    
   
Here,  is elasticity of emissions with respect to the cost of its substitute, b,  is
the income elasticity of demand for emissions generating activities,  is the
elasticity of mitigation supply (the inverse of the elasticity of marginal cost) and
 is the price elasticity of current emissions, q, which embeds both supply and
demand responses to changes in P.10
Equations (2b) have several important implications. First, absent technical
progress in reducing emissions ( 0t

 ) and with negligible reabsorption (a=0),
the growth rate of mitigation is proportional to the rate of interest. The factor of
proportionality is the elasticity of mitigation supply, t , so optimal mitigation
grows more rapidly when supply is more elastic or when the rate of interest is
high. But with t

>0 the growth rate of mitigation is augmented by anticipated
technical progress (the rate of decline in costs) in emissions reduction. Given the
dependence of emissions mitigation on technology and research, and expectations
that costs of emissions reductions actually will fall over time, this means that
“waiting” to achieve emissions reductions is a central element of dynamically
efficient policy. In a broader context, however, the magnitude of  is
endogenous to current policy, because it is an outcome of current and future R&D
10
In a competitive market  will be given by the harmonic mean of supply and demand
elasticities,
1 1 1
( )S D    
  .
. , as well as changes in the availability of substitutes,

9
Becker et al.: On the Economics of Climate Policy
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
efforts, and our previous discussion indicates that incentives to innovate are
powerful, provided that innovators can collect on the value of their innovations.
Similarly, the benefits of deferral are larger when the elasticity of
mitigation supply is large, and especially when large values of t are likely to
evolve from future technical advances. Large values of t mean that marginal
costs of mitigation at a point in time do not rise sharply with s—mitigation
activities are easily “scalable”—so there is not much cost to sharply ramping up
mitigation in later periods when emissions reductions will be most valuable. But
when t is small the marginal cost of mitigation increases rapidly with s—there is
a large cost penalty if mitigation efforts are concentrated in fewer periods. Then
it is worthwhile to do things in smaller pieces by spreading mitigation activities
over time. Then the optimal policy is to ramp up mitigation efforts sooner rather
than later.
This interpretation of the elasticity t is the “certainty” equivalent of a
broader point about scalable technologies—they can be deployed as needed on
large scale without much cost penalty. As we show below, development of highly
scalable (high  ) technologies is especially valuable if we extend the analysis to
incorporate uncertainty and the possibility that future environmental effects of
GHGs may turn out to be much more costly than currently anticipated, or that
low-probability but high-damage outcomes may occur. Then scalable
technologies to reduce emissions have high option value, precisely because they
can be deployed on a large scale when mitigation is most critical. Notice also that
technical advances that enhance scalability ( ) or enhance the efficiency of
mitigation ( t

) are complementary—the social benefits from higher t

are
proportional to  , and conversely.
Similar implications apply to the “value” side of (2b). Since the price of
emitting carbon rises at the “net” interest rate r+a, this price rise induces
conservation in proportion to the price elasticity  . Note that  embeds both
production and consumption responses to carbon pricing—for example, the fact
that carbon-based fuels are abundant and in fairly inelastic supply on the world
market suggests that  is likely to be small.11
Together with demand growth
( 0ty

 ) on the world market, the implication is that substantial conservation
relative to business-as-usual is unlikely. Then policy success is critically
dependent on technical advances that would promote mitigation by enhancing 
11
That is, the burden of P is likely to fall on suppliers of carbon emitting energy sources, who
would supply roughly the same quantities at substantially lower after-tax prices. Then imposition
of emissions pricing may not much impact fuel use or emissions through conservation.
10
The B.E. Journal of Economic Analysis & Policy, Vol. 10 [2010], Iss. 2 (Symposium), Art. 19
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
and  , or substitution toward non-carbon-based energy alternatives with
declining costs ( t tb

< 0). In other words, the point of optimal emissions pricing
is not so much to induce conservation on the demand side—which is likely to
have small effects—as it is to guide the research incentives that will result in
greater supply of clean energy alternatives in the long run.
2.4 Evolving Expectations and Changing Incentives
For a given rate of interest the rate of growth of the optimal carbon price is
determined. The other key to incentives is 0P —the initial or current carbon
price—which determines the level of the entire future price path. This is affected
by the entire array of technology and substitution effects, and the way they are
anticipated to evolve, as in (2b). Factors that reduce the anticipated growth of q
or raise the growth of s will reduce 0P and delay the ultimate date T when net
additions to the stock Q optimally cease. For example, the expected emergence of
technologies ( ) that make s more scalable allow for lower net emissions (q-s)
along a flatter path, making T longer, and so on.
The prices Pt that support optimal net emissions in problem (1) could be
generated by an ideal set of emissions taxes or by cap-and-trade determination of
an emissions price. Though we don’t wish to join a full debate over the relative
merits of carbon taxes versus cap-and-trade schemes—see Nordhaus (2007c) for a
good discussion—our framework does highlight some key issues that have not
been emphasized in previous literature. While we have framed the optimal policy
in a certainty-equivalent framework, the fact that the optimal initial price, P0,
depends on expectations of future market responses and technologies means that
an efficiently updated policy should adjust the current price level, P0, as
information evolves. For example, an innovation that reduces expected future
mitigation costs will reduce P0, exactly as a new “find” that increases the future
availability of an exhaustible resource (such as oil) will reduce its current price,
even if the newly discovered units are not currently recoverable. In an ideal cap
and trade framework in which total acceptable emissionsQ (as opposed to yearby-year
emissions) are fixed and unchanging, and tradable over time, the
collection of information and the formation of expectations about such future
innovations and technologies is decentralized to market participants—a clear
advantage in terms of incentives. But the possibility of governments
manipulating the variable they control, Q , invites rent-seeking, which is the
foundation of many economists’ critique of cap-and-trade schemes and their
consequent preference for tax-based incentives.
11
Becker et al.: On the Economics of Climate Policy
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
Yet tax-based schemes also have powerful disadvantages. Tax-based
carbon pricing sacrifices the substantial advantage offered by market-based
expectations—when carbon prices are set by governments the formation of
“expectations” that determine both the level and growth of the optimal emissions
tax is necessarily centralized in government, which is responsible for setting and
updating the entire price (tax) path Pt. There is little reason to believe that
governments would do well in this regard, and the opportunities and incentives
for inefficient choices and rent-seeking appear to us just as powerful with taxes as
with cap-and-trade.
2.5 Discounting Future Climate Costs and Returns
Ignoring reabsorption, a, for any given future marginal damage from incremental
emissions, say TP = $500 per ton of CO2 emissions in T=100 years, the strength
of initial incentives, 0P , is determined by the rate of interest; 0
rT
TP e P
 . Higher
r means low 0P and a gradual ramping up of incentives and responses. Low r
means that conservation and mitigation efforts are more front-loaded. If we base
r on historical market rates of return on physical and human capital, then a value
in the neighborhood of r=.06 is reasonable. This yields 0P =$1.24 if 100 $500P  .
In contrast, the UK government’s 2006 Stern Review of the Economics of Climate
Change argued that policies should reflect much lower interest, r = .015, based on
the Review’s notion that that it is ethically improper to heavily discount the costs
that current emissions impose on future generations. Then the current tax or price
is .015 100
0P $500 e 
  = $112, which is almost 100 times larger than with r=.06.
As pointed out by Nordhaus (2007b) and Weitzman (2007), among others, this
philosophical choice of a (very) low discount rate for investments in climate
capital accounts for virtually all of the differences between the Stern Review’s
draconian recommendations for current action and the more gradualist policies
advocated by other economists. Much then hinges on the choice of a social rate
of discount for climate capital, r, which we take up below.12
3. Valuing Future Climate Damages
The fact that current economic activity and policy affect uncertain climate
outcomes and costs over vast time periods may be the most daunting challenge of
climate policy. Possible future outcomes—including the possibility of
12
Moreover, since the analysis in the Stern Review assumes that the cost of GHG concentrations
are proportional to GDP and that GDP grows substantially over time, the implied net discount rate
is even smaller (and is essentially zero). See Nordhaus (2007b) for a clear discussion of this issue.
12
The B.E. Journal of Economic Analysis & Policy, Vol. 10 [2010], Iss. 2 (Symposium), Art. 19
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
environmental catastrophes that could harm large populations or greatly reduce
productivity—must be both envisioned and valued, and then balanced against the
current cost of mitigating such harms.
We address three issues related to the current valuation of uncertain future
damage. First, given the possibility of various types of catastrophe, what does
economic analysis say about the costs that should be incurred today in order to
avoid them? At standard discount rates, events that have even substantial impacts
on productivity and population-wide living standards in the distant future have
only small present value. We show that these values substantially increase,
however, when climate-related damages are unequally distributed, when future
lives are at risk, and when we allow for uncertainty as to when the damaging
events might occur (holding constant the expected time to occurrence). Even at
“market” rates of discount, not-implausible values for the magnitudes of future
catastrophes imply substantial current willingness to pay to avoid them.
We then extend the analysis to the valuation of climate investments with
uncertain future returns. We show that appropriate social discount rates for
investments in climate capital may be well below market returns on other forms
of capital, reflecting the insurance value of climate investments. We also find that
distribution matters. The global public good nature of harmonized climate
policies is challenged by heterogeneity of valuations—projects that have high
insurance value to developed countries because they reduce future risks are likely
to be much less valuable to developing countries, for whom the possibility of
rapid economic growth is likely more important.
3.1 The Costs of Future Catastrophic Outcomes
Future catastrophic outcomes may include substantial damages to productive
capacity, sustained reductions in economic growth, threats to living standards or
lives of particular populations, or permanent environmental harm that reduces
welfare for any given level of economic activity. To frame these possibilities, we
begin with the standard infinite-horizon model of intergenerational utility that
underlies most work in economic growth and climate policy.13
Write the current
value of generational welfare over the indefinite future as:
(4) 0
0
( ) t
t
t
U u c e dt

 
13
Pindyck and Wang (2009) provide a dynamic general equilibrium approach to valuing
catastrophic outcomes, including parameterized distributions for both the arrival rate and
distribution of harm from catastrophes. Weitzman (2009) allows the distribution of future harm
from climate change to have “fat tails”, which can greatly impact the current value of avoidance.
13
Becker et al.: On the Economics of Climate Policy
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
In (4)  is the rate of time preference, or in an intergenerational context the rate
at which earlier generations discount the well being of later ones, and tc
represents the per-capita flow of goods and services (consumption) available to
generations alive at future date t, which may include valuations of environmental
factors. We continue to abstract for the moment from issues of uncertainty.
One form of calamity that can be represented in (4) is a permanent
reduction in future living standards that is known to commence at some future
date T, say in 100 years. So assume that future productivity is reduced by a
constant percentage, resulting in a permanent change in future consumption of
lnd c from T onward. For example ln .01d c   represents a permanent 1percent
reduction in per-capita income and consumption. Assume a constant
elasticity of the marginal utility of consumption,  , and steady state economic
growth of g. Then we can apply the Ramsey Equation linking the equilibrium
interest rate to time preference and economic growth, r g   . The current
value of this harm as a fraction of current (time zero) national income is:
(5)
( )
0
0 0
1
ln
( )
r g T
dU e
d c
c u c r g
 

 
where 1
( ) lnr g d c
 is the damage valued at date T and ( )r g T
e 
discounts the
date-T value to the present, allowing for economic growth. How large is (5)?
Assume r=.06 and g=.02—fairly standard values in a growth framework—and
let lnd c =-.01 (a one percent permanent reduction in future incomes). Then with
T = 100 years the right side of (5) is equal to -.0046, or about half of one percent
of current income. For the US with a national income of about $13 trillion, this
implies a present discounted value of future harm of about $59 billion. Viewed as
a long term project to avoid such damage, the expenditure flow at 6 percent
interest is about $3.6 billion. Cutting the horizon to T=50 years substantially
impacts the estimates. Then a permanent 1 percent reduction in future income is
worth about 3.4 percent of current income ($440 billion), or a flow expenditure of
$26.4 billion per year. By comparison, with these same parameters a current
permanent reduction in consumption of one percent would be worth roughly
$3.25 trillion or a flow of expenditure equal to roughly $195 billion per year.
Adding uncertainty about when such climate-related damages might occur
substantially raises the present value of avoiding them.14
To demonstrate this in a
simple way, hold constant the expected time until damage occurs at T=100 years,
but assume that the damage is equally likely to commence at any future date.
14
Karp (2009) makes a related point.
14
The B.E. Journal of Economic Analysis & Policy, Vol. 10 [2010], Iss. 2 (Symposium), Art. 19
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
This implies an arrival rate (hazard) for the damaging event of 1
h T 
 . The
present value of future expected damages as a fraction of current income is then:
(6) 0
0 0
1 1 1
ln
( ) 1 ( )
dU
d c
c u c r g T r g

   
Using the same values as above, a permanent 1 percent reduction in living
standards with expected time to occurrence of T=100 years has present value
equal to 5 percent of current income, which is roughly 11 times greater than when
the damage was known to commence in 100 years. This is worth about $650
billion to the US in 2010, equivalent to a flow expenditure of $39 billion per year
at 6 percent interest. At T=50 (h=.02) the cost is 8.3 percent of current national
income, or a flow of $65 billion. Our point is that uncertainty over the time at
which climate change will have an adverse effect can greatly increase its current
valuation.
Formulas (5) and (6) express the current value of marginal losses in future
per capita consumption—everyone consumes one percent less than otherwise.
This is consistent with most of the existing analysis of valuing climate costs,
where those costs are framed in terms of reductions in future GDP, or costs as a
fraction of GDP, as if the burden of climate impacts is equally spread among the
future population. But much of the concern about climate-related damages has to
do with the distribution of harm, where some groups are harmed much more than
others. Concave u(c) means that reductions in c have rising marginal cost to those
who experience them, so a given reduction in aggregate income is more costly
when it is highly concentrated. For example, with  =2 a catastrophe that reduces
incomes by half among 2 percent of the population is twice as costly as an acrossthe-board
reduction in living standards of one percent, even though both events
reduce overall per-capita income by the same amount (one percent). Taken a step
further, a climate-related catastrophe that reduces future national incomes by one
percent by killing off one percent of the population, while leaving others
unharmed, may be very costly. Such catastrophes are not “marginal,” reducing
everyone’s income proportionally. Instead they wipe out consumer surplus—or
in the extreme case the value of life—for a swath of the population.
A framework for valuing such catastrophes is provided by the economic
literature on the value of a statistical life (VSL), which measures people’s
willingness to pay for a reduction in the probability of death that would save one
“statistical life.” For example, if in a population of 10,000 persons each would be
willing to pay $600 per year to reduce the per-year probability of accidental death
by 1 in 10,000, then VSL = $6 million, which is about the value used by the US
Environmental Protection Agency for cost-benefit analyses of regulations or
projects that would reduce mortality risks. Murphy and Topel (2006) use this
15
Becker et al.: On the Economics of Climate Policy
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
value to calibrate the value of a life-year ( ) ( ) / ( )v c u c u c , which is the
“consumer surplus” achieved by being alive and consuming amount c, where c
includes leisure and other factors that people value. They find that the value of a
life year is about six times current income, so if we think of ( ) ( )v c c c the data
suggest that ( )c 6 at current income levels. Then the above calculations
would increase by at least a factor of 6 for life-threatening events that cause an
equally-calibrated reduction in future “income.”
The analyses in Murphy and Topel (2006) and Hall and Jones (2007) also
indicate that ( )c rises with income, so the value of life is income elastic. Then
if future generations are richer than us, the value of lives saved from mitigating
future catastrophes will be proportionally greater than today. For example, an
income elasticity of  =1.2 and long run economic growth at 2 percent yields
( )c 9.0 in 100 years. The result is that a randomly occurring event that causes
a “concentrated” change in future costs because of climate-related mortality has
much higher current value:
(7) 0 0
0 0
( )1 1
ln
( ) 1 ( )
dU c
d c
c u c r g T r g

 

   
With a constant hazard rate and an expected arrival time of T = 100 years, with
1.2  and 0( )c =6, a “catastrophic” event that reduces per-capita output by
killing off dlnc = -.01 of the population has present value equal to 36 percent of
current income. Letting T = 1000—a catastrophe that could occur every thousand
years, on average—the current value is about 4.5 percent of income. And of
course the value is highly sensitive to the choice of r: a reduction in the discount
rate from .06 to .04 raises the current value of avoiding such a catastrophe from
4.5 percent to 22 percent of current income.
These results indicate that uncertainty over the future timing, magnitude
and distribution of losses from climate change can greatly impact our assessments
of current cost, even if future costs are discounted at conventional rates of return
of, say, 6 percent.
3.2 Discounting the Returns on Climate Capital
One of the most controversial aspects of debates over climate change policy is the
appropriate social discount rate to be applied to future damages. At the extreme
among economists, the Stern Review’s advocacy for a very low interest rate of
r=.015 accounts for almost all of its severe recommendations. Behind the Stern
Review recommendations is the notion that the welfare of future generations
16
The B.E. Journal of Economic Analysis & Policy, Vol. 10 [2010], Iss. 2 (Symposium), Art. 19
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
should be weighted equally with current ones, it being ethically repugnant (in
Stern’s view) to discount their welfare. Then the only source of a positive
discount rate on real cash flows is the growth of consumption over time, because
future generations are richer than we and utility is concave— >0 in our earlier
notation. And in the Stern recommendations even that is given little weight in
reaching the desired result. Similarly, non-economists such as Archer (2009)
have argued that economic analysis is itself ill-equipped to deal with
intertemporal valuations spanning a generation or more. Like Stern, Archer
argues for effectively zero discounting because current action is a moral
imperative.
The slowly ramping policy profiles offered by Nordhaus (2008) and others
are based on higher discount rates that reflect historical long run returns on other
types of capital. In contrast, while critiquing the analytical foundations of the
Stern rates, Weitzman (2009) offers a “Dismal Theorem” based on the possibility
of extreme catastrophes that drive consumption near zero and the marginal utility
of consumption beyond the moon. Policies that can avoid such outcomes can
have unbounded value under particular assumptions about the distribution of
climate effects on c—they should have “fat tails”—and the rate at which marginal
utility rises as consumption falls. The more general and useful point is that
uncertainties about the distributions of climate damage and the payoffs from
mitigation investments may greatly affect valuations. Climate policies that
effectively insure against large downside risks (when the marginal utility of
consumption would be large) needn’t have large expected returns, so the typical
market benchmarks for r might be inappropriate for valuing investments in
climate capital. We return to this point shortly.
From an economic and empirical perspective the choice of a discount rate
is not about the philosophical choice of the correct ethical weight to be applied to
the welfare of our and other peoples’ great-grandchildren, nor is it about the way
we “should” discount marginal dollars of their income because they will be richer.
As in all analyses that must balance costs and benefits, the issue is opportunity
cost. The fact that costs and returns are so uncertain and widely spaced in time
adds practical difficulties but not conceptual ones.
Consider a current project costing $1 million that would reduce the impact
of climate change 100 years from now. Assume that, absent the project, the
resulting climate change would impose a real cost of $20 million on future
generations. The logic of Stern (2006) and Archer (2009) suggests that we
should implement the project if, and only if, we value giving $1 to the current
generation less than we value giving $20 to the future generation. That is, the
question of whether the mitigation project is worthwhile allegedly depends on the
relative values we place on the consumption of current and future generations.
17
Becker et al.: On the Economics of Climate Policy
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
This is not correct—our choice does not depend on our relative preference for
current versus future generations.
Assume we wish to give the future generation the $20 million benefit they
would derive were we to implement the project today. If undertaken, the rate of
return on the current project is 3 percent, which is the solution for r in the
equation $1x 100r
e = $20. But if the market rate of interest is 6 percent, $1million
invested at the market rate of return would yield $403 million in 100 years,
compared to the $20 million benefit generated by the climate mitigation project.
This means that future generations would gain (a lot) if the current generation
were to forego the climate project and invest in other assets that yield higher
returns. Alternatively, it would take only $49,000 invested at 6 percent to provide
the future generation with the $20 million needed to compensate them for the
harm from climate change. Our point is that it is the market rate of return—not
our attitudes toward future generations or our moral view of discounting—that
determines the appropriate discount rate. To evaluate climate mitigation policy
with a lower rate of return unnecessarily harms either current or future
generations, or both. Future generations would not thank us for investing in a
low-return project.
It is appropriate to discount the costs and benefits of climate change
policies at a “market” rate of return because the market rate measures the
opportunity cost of such investments—returns available from investing the same
amount in physical or human capital—so long as such opportunities exist. But
what “market rate” should we use? At the low end one might benchmark by the
risk free rate as represented by the returns on government bonds. An alternative
would be the much higher historical returns on risky investments such as physical
capital or equities. Offered the opportunity to invest for the benefit of our greatgrandchildren
in 2110—who by any reasonable expectation will be much richer
than us15
—would we opt for Treasury bills and an annual return of perhaps 3
percent when the historical equity premium consistently provides long run returns
in the neighborhood of 6 to 8 percent? Most would choose equities.
Yet the fact that most of us would choose equities reflects an implicit but
appropriate (in this context) belief that those assets correctly gauge the
opportunity cost for long-term financial investments, including allowance for risk.
The weakness in this argument is that it is not obvious that the risks and returns
on climate investments align with those on other physical assets or equities. If the
returns on climate investments are uncorrelated with returns on the market
portfolio, or if by eliminating calamitous harm to overall productivity and living
standards climate investments pay off exactly when other productive assets do
15
At 1.5 percent annual growth, per capita income in 2110 will be about 4.5 times the current
level. At 2 percent the multiple is 7.4. Growth rates in developing countries such as China or
India are expected to be much higher.
18
The B.E. Journal of Economic Analysis & Policy, Vol. 10 [2010], Iss. 2 (Symposium), Art. 19
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
not, then the appropriate rate of return and discount rate for climate projects
should be lower than for other assets, perhaps even lower than the risk-free rate.
Further, though the climate impacts of investments in climate capital are global,
the risk properties of those effects, and hence their value, may differ greatly
across the countries whose participation in global agreements is essential.
3.3 Expected Social Returns on Climate Capital
The risk properties of the returns on climate investments derive from at least four
stochastic drivers: (1) global economic growth, because greater growth likely
means greater emissions; (2) the impact of emissions on climate; (3) the impact of
climate on environment, productivity and welfare; and (4) the effectiveness of
current investments in mitigating future harm.
To illustrate the determinants of an appropriate discount rate for climate
investments, consider a standard asset pricing framework for valuing a current
(time 0) project that offers uncertain returns at some future date, F.16
Assume that
the current generation can invest in  units of a climate project, with current cost
( )K  and marginal cost ( ) ( )k K  . The investment offers uncertain future
returns of x per unit, where x may be interpreted as the project’s future impact on
GHG concentrations, or other measures that would mitigate climate impacts.
With this setup, the social planner’s intertemporal problem is
(8)  0 0( ( ), ) ( ), ( )F FMax U u y K E u y x x

          
The representative individual in (8) derives utility from income (consumption) y
and the state of the environment  , both of which can be affected by current
climate investments. The factor  <1 reflects pure time preference between the
present (t=0) and future (t=F), and E is the expectations operator reflecting
uncertainty over the joint distributions of y,  and x. We interpret this social
valuation problem as country-specific, so that the distributions of outcomes may
be quite different for, say, China than for the US.
The choice of investment in the climate project solves
(9)
 
 
( )
( ) cov ,f
F F
r
F F F
k E m X
e E X m X



 
16
See Cochrane (2005) for a clear presentation of asset pricing and discounting issues.
19
Becker et al.: On the Economics of Climate Policy
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
where fr is the risk-free rate of return, Fm is the marginal rate of substitution
between future and current consumption, and FX is the future generation’s value
of the income and environmental payoffs on the investment:
(10) ( )F F F FX y x   
where /F yu u   is future willingness to pay for environmental improvements.
Divide (10) by ( )k  to obtain marginal returns per dollar invested
( / ( )F FR X k  ) and solve for the required return on the environmental asset,
which yields the familiar CAPM form for required expected returns on the
investment, Er :
(11) ,
,
cov( , )
( )
(1 )( )
E f F F
f m R M f
M m R M f
r r m R
r r r
r r r


 
  
   
In (11), Mr is the market rate of return on equities and var( )M fr r m  is the
equity premium. The term , cov( , ) / var( )m R m R m  is the environmental
project’s “beta.” We have expressed  in terms of the covariance of R with m
instead of the more traditional covariance with growth in income because of the
presence of environment in welfare, ( )u .
According to the third line in (11), the required expected return on the
environmental asset will be smaller than the market rate so long as its market
“beta” ( ,m R ) is smaller than 1.0. This has the usual risk-return interpretation—
if the environmental asset offers greater payoff than the market when m is high,
then it reduces risk and should have a lower than market expected return. While
this may seem likely, so E Mr r is plausible, the first line of (11) offers a more
aggressive point about risk and return for investments in climate projects—the
expected return on an environmental project may fall below the risk-free rate if
cov( , )m R >0. Because m falls with income, positive covariance of m and R is not
relevant for most financial assets. But climate projects are alleged to have the
potential of averting disasters, so they may pay off precisely in states of the world
where willingness to pay, m, is greatest. For example, if climate change may
greatly reduce future productivity and living standards, or cause widespread harm
and death in some states of nature, then projects that avert such outcomes (see
equation (10)) may be highly valued even if the payoff is rare—they have low
.
20
The B.E. Journal of Economic Analysis & Policy, Vol. 10 [2010], Iss. 2 (Symposium), Art. 19
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
expected return but high market value because they pay off when mitigation of
damage is most valuable.
3.4 Scalable Technologies and the Effectiveness of Current Investments
Our earlier discussion emphasized the importance of “scalable” technologies in
mitigating extreme climate outcomes. Scalable technologies are particularly
likely to warrant low rates of discount because they can greatly reduce the risk of
extreme climate outcomes. This point can be demonstrated in the current context
by putting a bit more structure on the form of future environmental harm and the
technology of mitigation.
Let future income in the absence of mitigation be F F F Fy y Q  , where
FQ is the future stock of global atmospheric GHGs above some base and F > 0 is
the extent of future economic damage per unit of FQ .17
Both FQ and F are
currently unknown— FQ is determined by global economic growth and carbonbased
energy use, while the distribution of F represents current uncertainty
about the future cost of GHG concentrations. Their interaction means that
extreme values of F can cause future environmental “catastrophes” when FQ is
large. In anticipation of such damage, we assume that the current generation can
invest in units of climate capital,  , which can be combined with future variable
inputs FZ to mitigate ( , )F FG Z units of environmental harm once FQ and F
are known. For example, FG may represent units of FQ removed from the
atmosphere, or emissions that are avoided by deploying clean energy
technologies. Assume that mitigation has constant returns, so
( , ) ( )F F F FG Z g z  , where /z Z  . Let F be the fixed future cost of
deploying the technology. If deployed, future income net of mitigation is:
(12)  ( )F F F F F F Fy y Q g z z       
Let ( )Fg z be iso-elastic; 1
( )F F F Fg z A z

 , where FA is the unknown future
productivity of currently chosen environmental capital. Then the optimal date-F
choice of z yields
(13) 1 1
F F F F F F Fy y Q A  
   
      
17
If future harm is convex in FQ then average harm per unit, ( )F FQ , will be increasing in FQ .
We ignore this so not to complicate the analysis.
21
Becker et al.: On the Economics of Climate Policy
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
where 1
(1 )   
  is the elasticity of supply of mitigation, which indexes the
scalability of the investment project—a perfectly scalable (constant marginal cost)
project has 1  . The marginal return on the environmental asset is then
 
1
( )
F F
F
A
R
k


 

 .18
According to (13), the extent of mitigation increases with the
scale of the investment in climate capital,  , the state of future productivity, FA ,
and with the damage from future GHG concentrations, F . For given values of
FA , and F , mitigation is greater when the project is more “scalable”—that is,
when  is large.
Some investments such as reducing current GHG emissions are not
scalable in that they cannot be cheaply adjusted once the values of FA and F are
realized;  is small and marginal cost rises sharply with mitigation efforts.
Others—such as development of clean energy technologies or investments in the
capacity to remove carbon from the atmosphere or sequester emissions—can be
deployed in large scale based on the future demand for climate mitigation. A
perfectly scalable technology,  =1, would provide a great deal of insurance by
effectively truncating the distribution of harm in what would otherwise be the
most damaging states of nature, when F is large, yielding F F Fy y   because
excess concentrations of GHG are eliminated. Even technologies that are not
perfectly scalable can have substantial value. Larger values of 1
(1 )   
 
mean that the payoff from implementing the technology is more sensitive to the
realized marginal value of environmental improvements, F , which enhances the
positive covariance between m and R . This further reduces the implied rate of
discount for such projects because highly scalable technologies provide additional
insurance—they are deployable as needed by varying FZ . The magnitude of this
advantage depends on the uncertainty about F (and FA ). Greater uncertainty
raises the (current) value of ex-post scalability.
This is our earlier point about the value of scalable technologies—research
and development investments in mitigation technologies that can be deployed in
large scale in the event that damages are large can offer important insurance
against looming catastrophe. Such projects should not be heavily discounted.
18
With fixed cost, F , RF=0 for low values of A and  because the technology will not be
deployed. At the other extreme, marginal returns are zero when A and  are very large, or when
 is large, because all excess emissions are mitigated. It is also plausible that scalable (high )
technologies are more costly to develop, so current costs are ( , )K   .
22
The B.E. Journal of Economic Analysis & Policy, Vol. 10 [2010], Iss. 2 (Symposium), Art. 19
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
More broadly, much of the discussion surrounding climate investments stresses
that they are meant to avert future catastrophes, which we interpret to mean that
they yield dividends when willingness to pay, m, is highest. Our analysis
provides a positive economic case for discounting them lightly. This is, we think,
the less extreme and more relevant implication of Weitzman’s “Dismal Theorem”
analysis.
3.5 Heterogeneous Valuations
Much has been made of the fact that effective climate policies require harmonized
international efforts, because earth’s atmosphere is a global public good. Freeriding
and the tragedy of the commons aside, our analysis suggests an additional
impediment to cooperation and harmonization, based on heterogeneous valuations
of climate investments between developed and developing countries.
Consider China (CH) and the United States (US) as extremes of the
relevant development scale. In terms of our notation above, any reasonable
growth scenario implies CH US
m m ; that is, China will continue to grow faster
than the US, so the ratio of future to current marginal utility of consumption is
lower in China. And states of nature where China grows fastest correspond to the
smallest values for CH
m —these are the good states of nature from China’s
perspective, because they are rich. The danger of harmful climate outcomes
increases with global GHG emissions, which increase with economic growth. So
it is reasonable to assume that future GHG concentrations will be greatest if China
(and India, and others) grows rapidly, possibly approaching the living standards
and energy consumption now observed in the US. This means that “good” states
of nature from China’s growth perspective are, climate-wise, most damaging to
the US, especially if US living standards are harmed. Climate projects valued
highly by the US because of strong insurance properties (cov( , )US
m R is high)
may have little current value to China because cov( , )CH
m R is weak or even
negative. In effect, a world with greater climate damage is one in which China
gets rich, and they are more willing to bear the future cost that the US would like
to avoid.
This discussion can be framed in terms of current efforts to establish a
harmonized price for carbon emissions. With rapid economic growth, China’s
real future willingness to pay for a unit reduction in GHG concentrations may be
equal to that of the US. But China discounts this return more heavily, because it
only occurs when China prospers. Expressed as a preference for a current tax on
GHG emissions, policy makers in China and the rest of the developing world will
rationally prefer a lower (or no) tax that is less of a hindrance to attaining
prosperity, while the US and other developed countries prefer a higher tax that
23
Becker et al.: On the Economics of Climate Policy
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
insures the prosperity they already have. Even ignoring other challenges to
international accord, the likelihood of an effective global policy in such a world
appears to us slim.
4. Conclusions
The basic designs of economic policies to mitigate anthropogenic climate change
and its effects are not novel. They are rooted in well-understood methods for
dealing with externalities, which remedy market failure by pricing an otherwise
over-used resource—here the capacity of the atmosphere to safely absorb GHG
emissions. Implementing such policies is more challenging, for two basic
reasons. First, the external costs of GHG emissions are global rather than local,
so useful policies that would price or regulate current emissions require
harmonized action worldwide. Second, the harms that policies seek to value and
internalize are both highly uncertain and spread over future generations—all of
the benefits of using carbon-based energy occur today, while the possible social
costs are far removed.
Our analysis has sought to extend previous work on both the form and
substance of climate policies, particularly in the area of valuing uncertain future
costs of GHG emissions and the benefits of policies that would mitigate those
costs. An important finding is that “gradualist” policies advocated by most
economists—setting a low initial emissions price that would rise at a “market”
rate of interest—are based on the implicit assumption that returns on climate
investments have a similar payoff structure to other forms of investment in
physical or human capital. This assumption ignores the possible insurance value
of social investments in climate capital, which may pay off precisely when other
forms of capital do not.
References
Archer, David, Global Warming: Understanding the Forecast, Malden, MA:
Blackwell Publishing, Ltd., 2007
Archer, David, The Long Thaw, Princeton, N.J.: Princeton University Press, 2009
Cochrane, John, Asset Pricing, Princeton, N.J.: Princeton University Press, 2005
Dyson, Freeman, “The Question of Global Warming”, The New York Review of
Books, June 12, 2008
24
The B.E. Journal of Economic Analysis & Policy, Vol. 10 [2010], Iss. 2 (Symposium), Art. 19
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM
Hall, Robert E. and Charles I. Jones, “The Value of Life and the Rise in Health
Spending, Quarterly Journal of Economics, 122, 2007, pp 39-72
Hotelling, Harold, “The Economics of Exhaustible Resources”, Journal of
Political Economy, April, 1931, pp 137-175
Karp, Larry, “Sacrifice, Discounting and Climate Policy: Five Questions”,
Working Paper, U.C. Berkeley, July 2009
Murphy, Kevin M. and Robert H. Topel, “The Value of Health and Longevity”,
Journal of Political Economy, October, 2006, pp 871-904
Nordhaus, William, “To Tax or Not to Tax: Alternative Approaches to Slowing
Global Warming”, Review of Environmental Economics and Policy, 1 (1),
Winter, 2007, pp 26-44. (2007a)
Nordhaus, William, “A Review of the Stern Review on the Economics of Climate
Change”, Journal of Economic Literature 45 (3), 2007, pp 686-702.
(2007b)
Nordhaus, William, The Challenge of Global Warming: Economic Models and
Environmental Policy, Working Paper, Yale University, 2007. (2007c)
Nordhaus, William, A Question of Balance: Weighing the Options on Global
Warming Policies, New Haven, CT: Yale University Press, 2008
Pindyck, Robert S. and Neng Wang, “The Economic and Policy Consequences of
Catastrophes”, Working Paper, Massachusetts Institute of Technology,
September, 2009
Stern, Nicolas, The Stern Review on the Economics of Climate Change, H.M.
Treasury, London, 2006.
Weitzman, Martin, “The Stern Review of the Economics of Climate Change”,
Journal of Economic Literature 45 (3), 2007
Weitzman, Martin, “On Modeling and Interpreting the Economics of Catastrophic
Climate Change,” Review of Economics and Statistics, February, 2009, pp
1-19
25
Becker et al.: On the Economics of Climate Policy
Brought to you by | University of Chicago
Authenticated | 128.135.12.127
Download Date | 4/25/14 8:16 PM