MAKING CARBON TAXATION A GENERATIONAL WIN WIN
| Author | Jeffrey Sachs,Laurence Kotlikoff,Simon Scheidegger,Felix Kubler,Andrey Polbin |
| Date | 01 February 2021 |
| Published date | 01 February 2021 |
| DOI | http://doi.org/10.1111/iere.12483 |
INTERNATIONAL
ECONOMIC
REVIEW
February 2021
Vol. 62, No. 1
DOI: 10.1111/iere.12483
MAKING CARBON TAXATION A GENERATIONAL WIN WIN∗
By Laurence Kotlikoff, Felix Kubler, Andrey Polbin, Jeffrey Sachs,
and Simon Scheidegger1
Department of Economics, Boston University, U.S.A.; National Bureau of Economic Research,
U.S.A.; The Gaidar Institute, Russia; Department for Banking and Finance, University of
Zurich, Switzerland; The Russian Presidential Academy of National Economy and Public
Administration, Russia; Department of Economics, Columbia University, U.S.A.; Department
of Finance, University of Lausanne, Switzerland
Carbon taxation is mostly studied in social planner or infinitely lived-agent models, which obscure carbon
taxation’s potential to produce a generational win win. This article’s large-scale, dynamic 55-period, overlap-
ping generations model calculates the carbon tax policy delivering the highest uniform welfare gain to all cur-
rent and future generations. Our model features coal, oil, and gas, increasing extraction costs, clean energy,
technical and demographic change, and Nordhaus’ carbon/temperature/damage functions. Assuming high-end
carbon damages, the optimal carbon tax is $70, rising annually at 1.5%. This policy raises all generations’ wel-
fare by almost 5%. However, doing so requires major intergenerational redistribution.
1. introduction
Climate change presents grave risks to current and future generations.2Anthropogenic
global warming, associated with the human release of carbon into the atmosphere, is widely
viewed as the primary cause of climate change.3Since 1880, the planet’s average temperature
has risen by 1.8ºC, minimum levels of Arctic ice are declining by 12.8% per decade, 413 giga-
tonnes of ice sheets are melting annually, and the sea level may rise 8 feet by 2100. The 2018
∗Manuscript received January 2020; revised July2020.
1We thank seminar participants at the Bank of Italy, Oxford University, the Econometric Society 2019 Summer
Meetings in Seattle (the Seattle lecture), the Environmental Protection Agency, the University of Lausanne, the
Gaidar Institute, the University of Zurich, Stanford University,as well as the Toulouse School of Economics for their
extremely valuable comments and suggestions. Felix Kubler and Simon Scheidegger are generously supported by
grants from the Swiss Platform for Advanced Scientific Computing (PASC) under project ID “Computing equilibria
in heterogeneous agent macro models on contemporary HPC platforms”, the Swiss National Supercomputing Cen-
ter (CSCS) under project ID 995, and the Swiss National Science Foundation under project ID “Can Economic Pol-
icy Mitigate Climate-Change?”. Simon Scheidegger gratefully acknowledges support from the MIT Sloan School of
Management and the Cowles Foundation at YaleUniversity. Please address correspondence to: Felix Kubler, Depart-
ment for Banking and Finance, University of Zurich, Switzerland.
Email: fkubler@gmail.com.
2The perils include drought, extreme storms, floods,a rise in sea level, intense heat, wildfires, pollution, desertifica-
tion, the spread of disease, earthquakes, tsunamis,a rise in ocean acidity, the proliferation of insects, and mass extinc-
tions.
3According to NASA (see https://climate.nasa.gov), carbon dioxide in the atmosphere has increased by one-third
since 1950 and is now at its highest level in 650,000 years.
3
© (2020) by the Economics Department of the University of Pennsylvania and the Osaka University Institute of So-
cial and Economic Research Association
4kotlikoff et al.
U.S. Government’s National Climate Assessment projected potentially massive costs to the
economy, the ecosystem, health, infrastructure, and the environment.4Recent estimates put
2100 climate damages at one-quarter of global GDP.5
Inspired and instructed by Bovenberg and Heijdra (1998), Bovenberg and Heijdra
(2002), Heijdra et al. (2006), Leach (2009), and Sachs (2014), this article develops a large-scale
overlapping generations (OLG) climate-change model to realistically assess the potential
generational win wins available from carbon taxation. The OLG model appears better suited
for studying carbon taxation than the standard frameworks—the social planner (SP) and
infinitely lived agent (ILA) models. In the OLG framework, generations are selfish. They
care solely about their own well-being when alive and care nothing for the welfare of their
immediate, let alone far distant, progeny. Hence, their imposition of negative externalities on
future generations comes naturally. The OLG framework also highlights a key point that has
been obscured in all but a few analyses of carbon taxation. Carbon taxation, coupled with
the appropriate intergenerational redistribution, can make all current and future generations
better off.
Our OLG model has 55 overlapping generations. A single consumption good (corn) is pro-
duced with capital (unconsumed corn), labor, and energy. The energy is clean or dirty. Clean
energy is produced using capital, labor, and fixed natural resources (e.g., windy areas), which
is proxied by land. Labor and land are in fixed supply. Corn and clean energy experience tech-
nical change (total factor productivity (TFP) growth), which can happen at permanently dif-
ferent rates. As in Golosov et al. (2014), we explicitly model dirty energies, in our case coal,
oil, and natural gas.6Each has finite reserves, and each is subject to increasing extraction cost.
Apart from our energy supply side and OLG structure, our model matches Nordhaus (2017).
In particular, it adopts Nordhaus’ modeling of carbon emissions, temperature change, and
temperature-induced economic damage. It also adopts Nordhaus’ projections of the global
population with appropriate assumptions about its distribution by birth cohort.
Calculating Pareto improvements is a standard procedure for determining optimal policy
responses to negative externalities. Anthropogenic climate change is arguably the planet’s
worst negative externality. We use Auerbach and Kotlikoff’s (1987) lump-sum redistribution
authority (LSRA) to derive the largest uniform (across all current and future generations)
welfare increasing (UWI) carbon tax, where welfare changes are measured as compensating
consumption differentials. We also present results for two alternative means of distributing
efficiency gains from controlling CO2emissions.7The first allocates all efficiency gains uni-
formly to current generations (the born). The second allocates efficiency gains uniformly to
future generations (the unborn). Depending on the size of the damages, optimal carbon policy
can depend on how efficiency gains are shared.
Our policy cannot readily be supported as a solution to an SP’s problem since achieving
a uniform welfare gain across all or some subset of generations (with, in the latter case, no
welfare losses to any generation) depends on the no-policy path of generation-specific utility.
However, the status quo path of utility is not an argument of standard social welfare func-
tions. This is clear in the SP welfare functions posited in typical integrated assessment models
(IAMs), such as Nordhaus’ (2017). Those functions place no weight, positive or negative,
on the status quo path of welfare. Consequently, there is nothing in IAMs that guarantees
a Pareto improvement. On the contrary, IAMs’ SPs would use carbon policy to redistribute
resources across generations, even in the absence of carbon damage where such policy is able
to better achieve the SP’s desired intergenerational distribution of welfare.
4https://nca2018.globalchange.gov.
5https://www.ngfs.net/en/communique-de-presse/ngfs-publishes-first- set-climate-scenarios- forward-looking-
climate-risks-assessment- alongside-user. In one of our models below, which features an extremely large,but, unfortu-
nately, highly plausible damage function, we produce climate damages of this order of magnitude.
6Golosov et al. (2014) combine oil and natural gas.
7Our three focal Pareto improvements help characterize the infinite set of such improvements. Which Pareto im-
provement and its associated carbon tax cum redistribution policy is chosen by society are possibly viewed as made by
an SP,although how such a planner would gain dominion over all current and future generations is unclear.
generational win-win carbon taxation 5
Other models, for example, Golosov et al. (2014), examine optimal taxation in ILA mod-
els. If there is a single (representative) agent, welfare results are identical to those obtained
in the SP framework. However, one needs to find the correct taxes to implement the first-best
allocation. The ILA model relies, implicitly, on intergenerational altruism (see Barro, 1974).
However, such altruism begs the question of why appropriate climate policy is not already in
place.8The choice of OLG versus ILA frameworks would be of little importance for carbon
policy if both frameworks produced identical or very similar policy prescriptions. We com-
pare an otherwise identical economy that is populated by a single ILA to our OLG model and
show that, depending on the rate of time preference and the magnitude of climate damages,
the two frameworks can produce dramatically different optimal carbon policies with the ILA
policy potentially harming some generations to help others.
Our UWI solution may usefully be compared with the recent proposal by the Climate
Leadership Council (CLC)9to tax carbon annually and annually rebate all revenues in equal
amount to all Americans. As described below, our model includes such annual revenue re-
bates, albeit to owners of dirty energy reserves. However, the rebates are insufficient to
guarantee a Pareto improvement, let alone a UWI. On the contrary, for older generations
who are too old to benefit from carbon mitigation, a carbon tax cum lump-sum redistribution
of revenue produces a deadweight loss per Diamond et al. (1974).
1.1. Overview of Findings. Under our baseline, business-as-usual (BAU) calibration, cli-
mate damages are initially 0.17% of output, peaking 231 years after the policy is initiated at
7.73% of output.10 Dirty energy represents 97% of total energy in year 0, 86% in year 50,
46% in year 100, and 0% after year 130, when the value of additional dirty energy extraction
exceeds its cost. Relative to their initial stocks, BAU extraction reduces, by year 130, coal, oil,
and gas reserves by 70%, 75%, and 85%, respectively. Hence, BAU entails burning most of
the planet’s fossil fuels. Our BAU simulation also predicts that gas production will rise over
the next 20 years and then steadily declines. As for coal production, it rises over the next 45
years and then sharply declines. Finally, oil production falls for 20 years, rises for the following
40 years, and then gradually declines through year 130, which is the last year that dirty energy
is produced. These surprising dynamics reflect the different extraction costs for gas, coal, and
oil, as well as the precise dynamics of the price of energy.
Given the significant uncertainty (see Lontzek et al., 2015) surrounding climate damage, we
also consider larger damage functions where the quadratic coefficient in the damage function
by Nordhaus (2017) is multiplied by either 3 or 6. Below, we refer, admittedly loosely, to the
three damage functions as 1×,3×, and 6×damages.11 As Lontzek et al. (2015) indicate, sensi-
tivity analysis is hardly a perfect substitute for formally modeling climate-damage uncertainty
in an OLG setting, which is our major near-term research goal.12 In our 6×BAU simulation,
damages are initially 1% of output, peaking at 18.8% in year 220. The BAU course of dirty
energy production and the extent of reserve exhaustion are quite similar for all three (1×,3×,
and 6×) damage functions.
With our baseline, the Nordhaus (2017) damage function, the optimal UWI carbon tax
starts at $23, rises 2% per year and raises the welfare of all current and future generations
8Distinct infinitely lived dynasties would try to free-ride on one another both within and across regions. However,
as shown, for example, by Bernheim and Bagwell (1988), intermarriage between altruistic dynasties produces altru-
istic linkages across dynasties, which eliminate the free-rider problem. One can also question the ILA framework on
empirical grounds. See, in this regard, Altonji et al. (1997), Altonji et al. (1992), Hayashi et al. (1996), Abel and Kot-
likoff (1994), and Gokhale et al. (1996).
9https://clcouncil.org/our-plan.
10 Throughout this article, we take the year 2015 as period 0.
11 The specification for damage as share of GDP in Nordhaus (2017) is Dt=1−1
1+π1TA
t+π2(TA
t)2with parameters
values π1=0, π2=0.00236, where the term TA
treferences the Celsius change since 1900 in global mean surface tem-
perature. The reference,below, to an “mx Damage Function” means we set π2to 0.00236 times m.
12 With uncertainty,the optimal UWI carbon tax will be higher to insure future generations against extreme carbon
damage. Indeed, given the downside tail risk, a positive carbon tax would likely be warranted as self-insurance even
where expected future carbon damages negative.
To continue reading
Request your trial