Environmental Policies that Maximise Social Welfare: the
Role of Intergenerational Inequality
USAEE Conference 2015
Frédéric Gonand
(University of Paris-Dauphine)
Pierre-André Jouvet
(University of Paris-Nanterre)
Pittsburgh, October 26th , 2015
1
Introduction (1/2)
• Carbon emissions can be curbed down through a public
intervention, e.g., a public decision increasing directly and
exogenously the fraction of renewables in the energy mix, or a
carbon tax influencing the optimal decisions of private agents.
• For a given target of reduction of carbon emissions, each policy
instrument triggers different aggregate effects on prices, GDP
growth and on intergenerational inequality.
• In this context, the social choice as concerns the optimal mix of
instruments that lessens carbon emissions is not necessarily trivial.
• We aim to determine the optimal social mix of instruments
lessening carbon emissions.
• Empirical GE-OLG model with energy sector and CO2 emissions,
parameterized on German data.
• Policy relevance.
2
Introduction (2/2)
• Dynamic GE setting with energy +environment: Böhringer and Rutherford
(1997), Böhringer and Löschel (2006), Otto, Löschel and Dellink (2007)… However,
literature often relies on static GE models that do not aim to account for
intergenerational redistributive effects.
• -> GE with OLG: Bovenberg and Heijdra (1998), Karp and Resai (2014). However,
literature often with theoretical approach + few generations, not mainly
designed to analyse interactions between GE-CO2 nor social choice.
• -> empirical, dynamic GE with OLG (Auerbach and Kotlikoff, 1987 / Carbone et
al., 2012 / Rausch, 2013). Our model close to the latter references BUT a) we
do not focus exclusively on carbon tax issues but also consider a rise in
renewables, b) modeling of carbon emissions, c) modeling of the
intertemporal social welfare, d) more than 60 cohorts on annual data.
• Aim: determine the optimal mix of instruments in an OLG-GE model
(see Van der Ploeg and Withagen (2014) in a Ramsey growth model but without OLGs
and without empirical parameterization).
3
The model
• Dynamic GE model with an energy module…
– Models the impact of numerous variables in the energy sector on growth,
savings, L supply, K per unit of efficient labor, aggregate substitution
between K and energy…
– Production function with K, L and Energy (nested CES function)
– Long-run macroeconomic equilibrium. 1 good.
• … an overlapping generations framework (OLG)…
– 60 cohorts defining optimally each year their level of consumption and
labour supply, in interaction with the conditions of the general equilibrium
– Dynamic equilibrium, intergenerational redistribution
• … and public finances : public spending (pensions; non ageing-related
public expenditures…); social contributions, income tax, carbon tax
• Modified version of Gonand & Jouvet (2015) (see July 2015 issue of
the JEEM) with renewables as policy variable, modeling of carbon emissions, and
of the intertemporal social choice (with variable aversion to social inequality and
variable discount rate applying to the welfare of future generations).
4
CES function with elasticity of substitution Capital / Energy
Capital demand
Labour
supply /
demand
Demographics
Overall
energy
efficiency
Energy
demand
(volume)
Demand of capital per
unit of efficient labour
Production function with a
CES nested structure
Real weighted end-use price of energy (past and future)
Real weighted end-use prices of…
Past consumption of oil
CES function
Profit maximization
Interest rate
Gross wage
Oil (fuel oil, diesel oil, RON 95)
Future energy
mix
Natural gas (household & industry)
around 60 overlapping
cohorts with increasing
life expectancy
Future demand for
energy (minus hydro,
wind, PV)
Electricity (households & industry)
Future demand for
electricity
Past consumption of biomass,
waste, biofuels, biogas
Oil, nat gas & coal prices
n
Future demand for non
electric energy
Past consumption of coal
Coal (steam & coking)
- Pension contrib
+ Pensions received
(variable age of ret.,
replacement &
contribution rates,
« décote »…)
Past consumption nat gas
Excise tax and VAT
Real end-use price of
renewables substit.
n-1
Carbon tax
Numerical
convergence to
equalize
demand of
capital per unit
of efficient
labour with
supply of
capital per unit
of efficient
labour
Past consumption of electricity
Future demand for
renewables substitutes
- Health contrib (agerelated)
End-use price excl. taxes
- Energy expenditures
Future demand for oil /
natural / coal
- Proportional tax
financing non-ageing exp
Transport, distrib / refining
costs
Future demand for oil
(Real) supply price
Future demand for
natural gas
+ Non ageing public
lump-sum exp.
Electricity prices
Excise tax and VAT
Tax financing feed-in
tariffs
(Real) imports prices
End-use price excl. taxes
(Real) nal production prices
Future demand for coal
- Debt disimburst tax
Imports volume
Transport, distrib costs
(additional costs for
renewables)
National production volume
(Real) weighted
(wholesale) market price
Net annual income of each cohort
Intertemporal utility maximization
Consumption /savings and leisure /
working time of each cohort
Rates of marginality
Clean spark / dark
spread (-> main peaker
either coal of nat gas)
Reoptimisation in 2010 if new
informational set available
Aggregate capital supply per unit
of efficient unit
Optimal capital per unit of efficient labour
(after numerical convergence)
GDP (after numerical convergence)
Public debt reimbursement
Current and intertemporal welfare for each cohort
Intertemporal social welfare
Fuel costs, thermal efficiency,
carbon price (ETS EU),
emission factor, operational
costs, overnight investment,
cost of capital, lifetime,
utilisation rate. For wind and
PV: rise in productivity
(learning-by-doing). For
nuclear: productivity losses
(increased safety).
Feedin
tariffs
for
wind &
PV
Costs of production of electricity
out of coal, natural gas, oil, nuclear,
hydro, onshore wind, offwhore wind,
PV, biomass.
The policy scenarios
• Scenario A is the no-reform scenario. No energy policy aiming at
lowering CO2 emissions: i.e., no centrally implemented rise in
public targets for renewables in the energy mix, and no carbon tax.
• Scenario B adds to scenario A a centrally implemented rise in
the fraction of renewables in the energy mix (financed by a
specific feed-in tariff) from now on. Its achieves a reduction in carbon
emissions of 20% in 2050 as compared to 2009 (i.e., the year preceding the
public announcement of the reform in 2010).
• Scenarios C adds to scenario A a carbon tax created in 2015 and
increasing by 5% in real terms per year afterwards, achieving a
reduction in carbon emissions of 20% in 2050 as compared to 2009. The
income associated with the carbon tax is recycled through lower proportional
taxes on households’ gross income. This scenario does not encapsulate any
centralised policy in favour of renewables.
5
Results (1/6)
For a given target of CO2 emissions reduction, the effect on energy prices of a policy
increasing the fraction of renewables in the mix is lower than with a carbon tax, and its
effect on the structure of the mix is higher. However, the carbon tax, provided that it is
recycled to private agents, has a more favorable impact on economic growth.
7
Results (2/6)
8
Results (3/6): intergenerational redistributive effects
• An energy policy bolstering renewables weighs on private agents’
wellbeing but especially so for young and future generations.
More renewables = higher future energy prices and lower private agents’ income. Detrimental effect in the
short run relatively less pronounced for currently older generations (permanent income effect: the younger a
cohort today, the longer it will bear the cost of increasing energy prices).
• A carbon tax (fully recycled through lower taxes on income) displays
pro-youth intergenerational redistributive effects and is more
detrimental to currently relatively aged working cohorts and to
current retirees (-> more intergenerational redistributive effects).
Recycling a carbon tax through a lower proportional tax on income amounts, in absolute terms, to
distributing relatively more revenues to cohorts receiving higher income (i.e., currently aged and working
cohorts which are more productive than the younger ones). It equivalently amounts to redistributing less, in
absolute terms, to cohorts with relatively lower income (i.e., for young active cohorts and retired generations).
The net effect of the recycled carbon tax is thus positive for the current income of aged working cohorts at
any year in the model, but negative for the current income of young and retired cohorts. Consequently, the
influence on the permanent income of a recycled carbon tax is negative for the cohorts which are retired or
relatively aged but still active when the tax is implemented, and positive for the permanent income of future
generations.
9
Results (4/6): intergenerational effects
10
Results (5/6): social preferences
11
Results (6/6): intertemporal social choice
12
Conclusion
• We find that an utilitarist social planner prefers to achieve the target of
carbon reduction mostly by implementing a fully recycled carbon tax.
• However, we also show that this result does not hold for other social
preferences because of implied intergenerational redistributive effects.
For instance, a social planner that takes account of the welfare of future generations and is highly
averse to intergenerational inequality maximizes its welfare by implementing a relatively moderate
carbon tax and increasing in parallel the fraction of renewables in the electrical mix.
• These results have policy implications. While a recycled carbon tax
maximizes growth, it does not necessarily maximizes social welfare
because of its intergenerational redistributive implications. The optimal policy
depends on social preferences as concerns intergenerational inequality and the wellbeing of future
generations.
• Incidentally, our model also suggests that a mix of a carbon tax and of a centralized
policy favoring renewables is probably not enough to meet targets of carbon
emissions reduction of 70%/80% in 2050, as often advocated for in the public debate.
13
Thank you
14