Carbon regulation becomes the norm

Regulation of greenhouse gas emissions, including caps and prices, is spreading around the world.  Jurisdictions without something in place are becoming the exception.

Back in January I noted  that carbon pricing is in place or legislated in jurisdictions accounting for around a quarter of the world’s CO2 emissions from energy and industry (with typically around half of emissions in these jurisdictions being priced).    This is a huge increase from ten years ago, just before the introduction of the EUETS, when the corresponding figure was less than 1%.  But this trend is now set to go much further as limits on carbon emissions spread nationally in China and the USA.  A national emissions trading scheme in China is expected before the end of this decade, following on from the provincial trial schemes already established.  EPA regulation in the USA will introduce caps on emissions from the power sector to take effect from 2020.  These measures will mean that jurisdictions accounting for around two thirds of emissions will have some sort of cap or pricing in place.  (EPA regulation will be implemented at a state level.  It is likely to involve carbon pricing in some cases, although some states may pursue a more direct regulatory approach.  In either case emissions will be capped.)   This will have been achieved in only about a decade and a half.

With measures in place across the EU, the USA, and China, carbon caps and pricing will have become the norm rather than the exception.  This is likely to put pressure on other jurisdictions without such measures to act.  This trend is likely to be reinforced by agreement at the UNFCCC conference in Paris next year, even if, as expected, any agreement there stops short of legally binding emissions reduction targets.

Step chart

The blue line represents the proportion of global emissions of CO2 from energy and industry occurring in jurisdictions with carbon pricing (i.e. excluding other sources of CO2 such as deforestation, and other greenhouse gases).  The green line represents the proportion of emissions actually priced, which is typically about half the total, because some sectors are usually excluded from pricing.

Looking at broader climate legislation, including not just carbon pricing but also other measures aimed at emissions reduction or adaptation to climate change, a similar pattern of rapid growth is evident.  The number of pieces of climate legislation introduced in the last ten years was nearly four times that in the previous ten years.  And legislation has been introduced in both developed and developing countries.

Globe data on legislation

Source:  GLOBE study of climate change legislation (see notes)

There are, of course, many reasons to remain cautious about progress.  The outlook national scheme in China remains uncertain in the timing of its introduction (here assumed to be 2018), its stringency and its effectiveness.  US EPA regulation remains subject to legal challenge, although it seems likely to survive this.  Prices in the EUETS remain low.  And, even with all the present and prospective schemes in place, emissions reductions goals still fall short of what’s necessary to have a good chance of meeting the international commitment to limit the rise in average temperatures to 2 degrees (a goal which now looks extremely challenging in any case).

But there has been tremendous progress in a relatively short space of time.  Discussions on climate policy needs to recognise this context.  Any claims along the lines that either that “no-one’s doing anything” or “we’re the only ones doing something” are no longer valid.  There remains a pressing need to enhance this world-wide momentum, so that global emissions can peak and begin their ever more necessary downward track.  But acknowledging that should not obscure the remarkable progress that has been made.

Adam Whitmore –  30th September 2014


The spread of caps across the USA is shown as a single step on the above chart.  However the form of implementation of limits will in practice vary between states.  The caps under section 111(d) of the clean air act proposed by the EPA in the US are currently intensity based, with the conversion to mass based standards a matter of continuing discussion.  Implementation as the state level will vary, and it is likely that it will include pricing in some cases by not others.  Some states may join the Regional Greenhouse Gas Imitative (RGGI), others may include some kind of flexibility that in effect creates a carbon price in one state, or in a group, while others may pursue a more direct regulation of individual plants.

Japan and Mexico, which were deliberately excluded from the chart I posted in January as they have very low prices.  However they are included here.  The intention here is to give an indication of all the jurisdictions with action in some form.

The raw data for the chart showing the number of laws is here (see p.27):  (see p.27)

Tranmission links to enable wind power

The availability of long distance transmission systems can help smooth out variations in availability of wind power, but helps solar less.  Policy needs to enable long-distance links to allow continuing increases in wind generation.

One of the challenges of moving to an electricity system running with a large proportion of wind and solar is that electricity output from these sources is variable, because the wind does not always blow and the sun does not always shine at any particular location.  With small proportions of these sources on the system this is not much of a problem.  However as a system becomes more dependent on these variable renewables, production is potentially curtailed in periods of high output, while there is a shortage of power at other times.  I’ll refer in this post to wind and solar photovoltaic (PV) power, which is by far the predominant form of solar electricity.  The issues with concentrated solar thermal power are different because there is some potential for storage in the power plant.

Difficulties with balancing supply and demand typically begin to become more severe as each type of variable renewable power, for example wind, begins to supply more than (very roughly) around 15% of annual electricity.  In practice the extent of the challenge of integration depends on a range of factors.  One influence is the amount of inflexible generating capacity such as nuclear on the system.  Another influence is the correlation between variable supply and demand.  For example, there is some seasonal correlation between electricity demand and wind output in Europe (in winter demand is higher and it’s windier), and there is some daily correlation between electricity demand and solar PV output in Australia (the demand peak is in the daytime).  Nevertheless, in any case there will be challenges in reliably meeting demand while accommodating very large amounts of variable supply.

To address this problem variable supply can be moved to different times, using storage, and to different places via transmission capacityDemand management can help, especially for a demand peaks, and some thermal back-up capacity is likely to be needed anyway in most cases.

The extent to which transmission links can help system balancing depends on the extent to which generation in different places is correlated.  If distant power sources tend to produce power at different times then transmission can greatly help smooth out variations in production, because when output is low in one location electricity can be moved there from another.  But if distant sources tend to produce power at the same time the advantages of long distance transmission will be reduced, because if output is low in one place there will be little electricity in the other place to move to fill the gap.  This is especially so if patterns of demand are also similar in the different places

Typically, correlations of output across regions for solar PV are much higher than for wind, especially as distances increase (see chart).  This is because solar output tends to depend strongly on time of day and season, which is quite similar even across an area as large as Western Europe, whereas wind varies with weather systems, which are less strongly related over distance of the order of 1000 miles or so.  Looking at the correlations between hourly output in different countries in Western Europe shows that the correlation between solar output in Germany and France is 88%, much greater than for wind at 44%.  Moving further away, the correlation between output in Germany and Spain is close to zero for wind, but still as high as 84% for solar.  The match of solar output is is actually greater than these figures imply due to the large number of times when solar output is zero for both countries.

Correlations of hourly output between different Eastern European countries are much greater for solar than for wind …

Correlations chart

These factors mean that increasing interconnection will tend to enable higher proportions of wind more than it will enable higher proportions of solar.  (This is assuming transmission is for load balancing.  Interconnection to transport power from sunnier to less sunny regions may well be valuable.  Also, this conclusion would change if intercontinental of transmission of large amounts of power were possible, but this remains a distant prospect.)

Incidentally, the correlation between solar and wind output is negative (when it’s windy it tends to be less sunny and vice versa), so having some of each on the system tends to increase the combined total that can be accommodated, compared with a system only having either one or the other.

A recent study of the USA by the National Oceanic and Atmospheric Administration (NOAA) confirmed very large benefits from High Voltage Direct Current (HVDC) long distance transmission for systems with large amounts of variable renewables.   The study used very detailed weather and load data, and data on existing power plants.  It concluded that with optimistic projections of wind and solar costs it is possible to reduce CO2 emissions by 82% with somewhat lower electricity costs, provided sufficient transmission capacity is in place.  The study emphasised the importance of the transmission network encompassing a large geographical area, such as the 48 contiguous US states, due to the large geographical scales over which weather is variable.

However, interconnectors are not always quick, cheap or easy to build.  They often link or cross different jurisdictions – US States, Chinese provinces or European countries – and will often link different types of electricity trading arrangement.  This can impose substantial barriers around permitting, and also around operation.  Policy can help the growth of wind by reducing these barriers and recognising the growing role of internconnection, although the precise policy measures necessary will be quite locally specific.  Enabling increased transmission is likely to be an important step in enabling the continuing growth of wind power in particular, and is likely to become increasingly urgent as growth continues.

Adam Whitmore – 16th September 2014


Thanks to Mathieu Ecoifier for providing the correlation coefficients between wind and solar in Europe.  Correlation coefficients are a rough and ready indicator of independence.  Actual effects on system operation will depend on many factors.

For analysis of the inverse correlation of wind and solar see for example Correlations Between Large-Scale Solar and Wind Power in a Future Scenario for Sweden, Joakim Widén, IEEE Transactions on Sustainable Energy, Vol. 2, No. 2, April 2011

The US study mentioned is Alexander MacDonald et. al., NOAA Earth System Research Laboratory, Low Cost and Low Carbon Energy Systems Feasible with Large Geographical Size (2014) – Presentation at Imperial College London 27th May 2014

Could rising aviation emissions be good for the environment?

international aviation sector is likely to require a substantial number of offsets to meet its goal of achieving carbon neutrality above a 2020 baseline.  If these offsets are forestry related there is the possibility of generating substantial biodiversity co-benefits.

Emissions from international aviation are around 2% of total emissions, and are expected to roughly quadruple by 2050, well above the expected growth rate of other sectors.  Faced with this prospect, and challenged by measures attempting to include international flights within the EUETS, the governing body for aviation the International Civil Aviation Organisation (ICAO) last year decided to look at using market based measures to cap net international aviation emissions at 2020 levels globally, with agreement to be reached by 2016.  The chart below illustrates the scale of the action needed to achieve this.  The blue line shows a scenario with high growth in emissions, which already includes efficiency gains from introducing new aircraft.  The dashed green line represents a lower emissions growth scenario.  The light blue area shows the potential contribution of new technologies and processes such as additional maintenance.  There is also a contribution from running the system more efficiently, with improved air traffic management and airport operation, shown by the brown area.  Such measures can in total probably reduce emissions growth by about 40%.  However this still leaves around 60% of emissions growth which is difficult to avoid by technology changes except in the long term.   (Reducing the growth in aviation services would also reduce emissions of course, but any set of policies that severely caps the number and length of journeys is likely to prove politically intractable.)


Source:  ICAO CAEP A38-WP/26, 2013

For this remaining emissions growth the only realistic option for capping net emissions at 2020 levels over the next few decades is likely to be the use of offsets.  Demand for offsets from aviation could reach some hundreds of millions of tonnes p.a. in the 2030s, and this demand would be reliable as well as large, given the steady growth in demand.  It could provide a much needed source of demand for international offsets, which is currently weak.  The cost of this to passengers is likely to be small.  Emissions from a transatlantic flight are very roughly around a tonne of CO2e per passenger, so this would add about $10 to the price of an economy class ticket assuming an offset price of $10/tonne, and less at current international offset prices, which are in the low single figures of dollars per tonne.

One source of offsets that looks particularly promising is reduced emissions from deforestation and degradation (REDD).  There has been a marked reduction in the rate of deforestation in Brazil (and some other jurisdictions) in the last decade, despite a slight increase last year.  The reductions in Brazil have been achieved through a variety of measures, including improved monitoring by remote sensing, new legal frameworks with better enforcement, more intensive agriculture and so forth.  But funding from governments, including Norway, Germany and the UK, has also played a useful role in reducing deforestation.  Future programmes will likely benefit from the additional funding that REDD offsets can provide. , although this funding will never be enough on its own.  And the scale of offsets available is potentially large.  For example, 500 million tonnes p.a. is equivalent to avoiding over 8,500sq. km of Amazon forest loss each year, compared with about 5,800 sq. km of forest currently lost in the Amazon region of Brazil last year (and an average annual loss of about 11,500 sq. km over the last ten years).

For a satisfactory scheme any offsets will of course need to be high quality, including meeting the usual tests of additionally, permanence and so forth, with adequate governance a prerequisite.  Buffers, exchange rates or risk premiums may be necessary to account for residual risks around permanence, leakage and other factors, or to realise an explicit goal of generating net benefits, with (for illustration) 1.5 tonnes of REDD offsets required for every tonne of aviation emissions.  This would somewhat increase the area protected for a given number of aviation emissions, assuming that REDD offsets are available at an appropriate price.

REDD programmes have the advantage that they help conserve biodiversity.  Indeed biodiversity benefits can be made an explicit criterion in programme design and selection.  This may, for example, include building on the current Climate, Community and Biodiversity (CCB) standard that is widely used in voluntary markets.  This would potentially allow an overall net gain for the environment if net carbon emissions were zero.  Reduced emissions from deforestation would match increased airline emissions, and biodiversity would additionally be preserved – hence the (deliberately provocative) title of this post.  Programmes can also provide opportunities for local communities, and the CCB standard is again relevant here.  Indeed appropriate community involvement in projects, ensuring local communities also benefit, is likely to be essential to any successful REDD programme.

Establishing that offsets issued now can be used after 2020 would provide valuable early demand for credits.  However given the early stages of development of proposals by ICAO it may be difficult to attract investors at present.

So far REDD has struggled to find adequate funding from carbon markets, despite discussion of allowing limited volumes of REDD credits under the California emissions scheme.  And significant challenges remain in any circumstances.  In particular, governance often remains difficult given the requirements for monitoring and permanence of REDD projects and programs.

Eventually some technical solution will be needed to enable aviation emissions to be reduced at source.  However in the meantime the chance to generate substantial additional benefits for biodiversity and other environmental goals by the judicious choice of forestry offsets to help meet aviation goals is an opportunity well worth further exploration.

Adam Whitmore – 10th July 2014

Thanks to Ruben Lubowski of Environmental Defense Fund for useful comments on this post.


The 1 tonne CO2e per transatlantic flight per economy class passenger figure is indicative, and depends on the multiplier applied to the CO2 emissions to represent other atmospheric effects of emissions at altitude.


The need for a trend break in the carbon intensity of energy use

Looking at past changes in each driver of total emissions emphasises the necessity of deploying very low carbon technologies. 

There has been some recent discussion about the Kaya identity, which breaks down emissions from energy use into four factors: population, income per capita, energy intensity of the economy (energy used to produce each unit of GDP), and carbon intensity of energy supply (carbon emissions per unit of energy), so:

Carbon emissions from energy use =

population x (GDP/capita) x (energy/GDP) x (carbon/energy)

This applies to energy, the largest source of man-made greenhouse gas emissions, but it’s not a useful tool for looking other sources of emissions such as deforestation, industrial process emissions, and other greenhouse gases.

Emissions from energy use have continued to grow to date because the growth in the first two factors in the identity has been greater than reduction in the last two.  The chart below shows the breakdown into the four factors of the change each decade in total global CO2 emissions  from fossil fuel combustion. Changes are measured in absolute terms – gigatonnes (Gt) of CO2 emissions per decade – rather than percentages.  The bar segments show the changes associated with each factor alone, holding the other factors constant.  Total changes, the net effect of the four factors, are marked by a triangle.  In each decade population has grown, as has income per capita.  This has been partially offset by a reduction in energy intensity.  The smallest effect to date has been the change in the carbon intensity of energy use.  There was a slight reduction in the emissions intensity of energy supply for three decades, reflecting in large part the growth of gas use, which has lower emissions per unit of energy than other fossil fuels, and, for part of the period, growth in electricity generation from nuclear.  However this trend reversed in the first decade of this century, mainly due to the growth of coal use in China.  This reversal, together with the larger effect of very strong per capita GDP growth and continuing population growth, resulted in an acceleration of the growth of emissions after the turn of the century.


Source: IPCC (see notes [1])

Globally, population will continue to grow in the coming decades, as will income per head, with the world economy likely to roughly triple in size by 2050 [2].  If historic trends continue then falls in energy intensity energy will balance some of this growth of GDP, resulting in total energy use growing by about 70% over the period to 2050.  There is doubtless more that can be done to reduce energy intensity, but even doubling this rate from its past level, which would be an extraordinary achievement, would only approximately stabilise energy use, and thus emissions if there is no change in carbon intensity.

This leads a huge amount needing to be done by reductions in carbon intensity to achieve emissions reduction targets.  Emissions need to more than halve globally by 2050 if the agreed target of limiting temperature rise to two degrees centigrade is to be reached, and emissions will need to continue reducing thereafter.  Global average carbon intensity needs to fall by a corresponding amount, even if huge gains in energy efficiency are sufficient to keep total energy use constant, which seems unlikely.  Less ambitious emissions reductions targets still in practice will require large reductions in carbon intensity,  especially if progress on reducing energy intensity is slower.  Such reductions are made all the more challenging by the huge inertia in the energy system due to the scale and long lifetimes of existing infrastructure, with high emission existing energy sources often lasting decades.  This in turn implies that increasingly most new infrastructure will need to be very low carbon if the required changes to carbon intensity are to be made in only a few decades

However carbon intensity of energy use is currently almost unchanged from its level a quarter of a century ago.  Of the four factors driving emissions this has had by far the least effect – the net effect of the red bars in the above chart is much smaller than the other three factors.  If emissions are to be reduced, this factor needs to become as large as, or larger than, the other three factors.  A transformation of the energy system towards much lower carbon intensity really is necessary.  In this respect the falling cost of renewables, especially solar, and progress in delivering batteries with improved technology and lower costs represent some of the most encouraging developments in securing a low carbon economy that we have yet seen.

Those who point out the urgency of the transformation to low carbon energy systems are right.  Those who point out the difficulty of achieving the huge required trend break in carbon intensity are also right.  The combination of urgency and scale of required reductions in carbon intensity, along with the need for faster improvements in the efficiency of energy use, is, at its simplest, what makes the problem of reducing emissions from the energy sector so challenging.

Adam Whitmore – 27th June 2014

Notes and references

[1] The chart is from IPCC, Fifth Assessment Report Working Group 3, and can be found here:

Income is converted into common units using purchasing power parity (PPP) exchange rates.

[2] The extrapolation of the formula to 2050 is indicatively as follows.   World GDP is expected grow at roughly 3% p.a., roughly tripling by 2050.  This comprises a 30% growth in population to just over 9 billion over the period, with GDP per capita growing at around 2.3% p.a. to more than double by 2050.  Data is from OECD modelling see

With no change in energy intensity, energy use would also triple.  However, energy intensity has historically reduced at about 1.5% p.a. (based on World Bank data – other sources give broadly similar figures).  This would reduce growth in energy demand to around 70%.  Doubling the rate of energy intensity reduction to 3.0%, a huge increase, would only succeed in reducing the growth rate of emissions to around zero, thus stabilising emissions at their current level.  Any reductions in emissions would still need to come from changes in carbon intensity.

Types of carbon pricing (part 3 of 3)

This is the last of three posts summarising the differing features of carbon pricing instruments – emissions trading (cap-and-trade), carbon taxes, and hybrids – and commenting on some of the implications for existing carbon pricing schemes.  The three posts together can be found as a pdf file here.

This post looks at the benefits of hybrid schemes that include elements of both quantity and price management, and what this implies for the design of carbon pricing schemes around the world.  In practice most carbon pricing schemes are to some extent already hybrids, but here a preferred form of hybrid is suggested.

It has long been recognised there is no need to restrict policy to a pure price (carbon tax) or pure quantity (cap-and-trade) instrument, and that a hybrid of the two can have benefits[1].  In particular an emissions cap combined with a price floor can limit total emissions, and so reduce the risk of crossing cumulative thresholds where damage becomes very high, while ensuring that the price never falls below estimates of the current marginal cost of damage.  It is thus likely to form a better approximation to the cost of damages than either a tax or a cap alone, more closely conforming to the basic principle that carbon prices should approximate to the costs of damages[2].  It also allows the cap to specify a (global) social choice about the acceptable level of risk, while continuing to price emissions below that threshold so as to stimulate investment to abate emissions which, even if the worst of the risks are avoided, impose costly damage.  And it gives a clear signal about the scale of the challenge and the transformational change necessary to meet it.

A price ceiling may also make sense if it prevents unduly costly abatement that reflects only the particular circumstances of a regional ETS over a limited time horizon.  However a cap is there to prevent steeply rising damages above a particular threshold, so it is necessary that the integrity of an overall global target is maintained.  This can be supported by setting aside allowances from within the cap in a cost containment reserve, as is already done in California.

Also, the advantages of a tax may be better realised by a hybrid scheme than by a pure tax for simple pragmatic reasons.  Emissions trading schemes are already in place in many jurisdictions and it is likely to be easier to move to a hybrid scheme that at least gains some of the benefits of a tax than to abolish an ETS and replace it completely with a tax.  Indeed many ETSs already have elements of price management.

Price floors and ceilings can be stepped, creating a full price schedule, where additional emissions are considered to create different costs.  However estimates of the appropriate price levels are subject to large uncertainties.  And the stock of GHGs is the atmosphere varies relatively little with additional emissions over short time periods even in large jurisdictions.  This suggests that a stepped floor may imply a precision in the specification of desirable trade-offs which is not really justified by the precision of the available data.

There may be further advantages from introducing hybrids in enhancing stability and thus credible commitment with price containment mechanisms.  They can reduce the pressure to review a scheme because either prices or emissions levels are proving outside expected ranges.  On the other hand price containment may make linking of different emissions trading schemes more complex.

Implications for existing schemes

So what does a policy reflecting best design practice look like?  Larger jurisdictions should look to put in place an emissions trading scheme with a long term declining cap, consistent with long term goals for limiting the global stock of greenhouse gases.   This should be complemented by a lower limit on price in the form of an auction reserve price increasing over time.  For smaller jurisdictions a tax, or at least fairly tight bounds on the price under an ETS, may make more sense in the short term, probably with the eventual goal of moving towards an ETS linked (directly or indirectly) with other schemes.  Making additional allowances available at high prices (a ceiling) is likely to be beneficial if these can be drawn from a pool of allowances previously set aside.  However the quantity of additional allowances available at a higher price should be limited to protect the integrity of the overall cap, at least in larger schemes.

This rough outline gives a framework for a broad, high level assessment of current schemes.  The scheme that approximates most closely to best practice design at the moment is California.  Quebec offers an example of moving to an ETS linked to a larger scheme, in this case California, and also includes the good design features of the California scheme.  RGGI has broadly appropriate in structure, but the levels of both price floors and ceilings look much too low to adequately price the damages due to additional emissions.  Similarly the price ceiling in the Alberta scheme looks too low at present, and moves towards reform are welcome.

The EU ETS remains the world’s largest carbon pricing scheme, with a cap declining according to a long term goal, and as such is very valuable.  It would benefit greatly from a price floor, although this is unlikely to prove politically feasible.

British Columbia’s carbon tax seems appropriate to its circumstances and is at something like an appropriate level, although it could seek long term to integrate into an emissions trading scheme covering Canada and (preferably) the rest of North America.  In the meantime Mexico may be justified in choosing a tax due to its administrative simplicity, but the starting level looks much too low.  South Africa may also be right to favour a tax, at least at present, given its greater simplicity.  New Zealand would be justified in considering a carbon tax in place of its ETS, and certainly current low prices under the ETS look inappropriate.

China appears to be moving in the right direction with its preference for emissions caps including recognition of the need for price containment, though its particular approaches to price containment will take some time to become established.

So, much is being done, but there is potential for reform to improve the functioning of carbon pricing schemes.  Continuing expansion of limits on emissions in China is of enormous importance.  And the California scheme offers a model of good design which many could follow.

Adam Whitmore – 13th May 2014

[1] The original analysis in this area is Roberts M.J. and Spence M., 1976 Effluent charges and licenses under uncertainty.  Journal of Public Economics 5 193-208.

[2] Working Paper No. 48  See Climate Strategies Briefing Paper ( Grubb, M. (2012). Strengthening the EU ETS. Creating a stable platform for EU energy sector investment. Climate Strategies Full Report ( for a discussion of this issue in the specific context of the EUETS.

Types of carbon pricing (part 2 of 3)

This post is the second of three summarising the differing features of carbon pricing instruments – emissions trading (cap-and-trade), carbon taxes, and hybrids – and commenting on some of the implications for existing carbon pricing schemes.  The three together can be found as a pdf file here.

Carbon taxes define a set price for all emissions from a jurisdiction and a particular time.  For jurisdictions accounting for a small proportion of emissions and looking at limited time horizons – for example British Columbia over the next 5 years – variations in emissions will have little effect on the stock of GHGs in the atmosphere, so there is little likelihood that any incremental emissions will lead to a dangerous threshold being reached.  Consequently the cost per tonne of damage is quite constant over different levels of emissions (the marginal damage function is very flat).  In contrast, an excessively high price under a cap may prove economically damaging.  This implies a constant price set by a tax may price damage appropriately.

Indeed given the dependency of damage on the stock of emissions such arguments apply to quite large jurisdictions over quite long timescales.  The limited variation in damages per tonne as emission vary over quite large ranges compared with annual emissions from any one jurisdiction is among the main reasons that many favour taxes rather than quantity limits [1].

Price stability

A carbon tax also addresses many of the drawbacks of an ETS by providing price stability.  This may stimulate investment more efficiently than a volatile price, because it can be built into companies’ financial models with greater confidence.  It also provides governments with greater revenue stability.  It is likely to make revenue neutrality for governments,(i.e. no change in total tax burden) easier to manage.  Revenue neutrality is often stated as an objective of carbon pricing, and appears to be an important factor in continuing political support for the carbon tax in British Columbia.

Furthermore, some of the other advantages of emissions trading may also prove less compelling for smaller jurisdictions.  They are more likely to be technology takers, playing a limited role in stimulating new technology, which will frequently be deployed globally, implying any strategic signal for technology development created by a cap is less relevant.  And appropriate measures for shielding of emissions intensive trade exposed industries against carbon leakage remain entirely possible under a tax.

Administrative simplicity

A tax may also prove administratively simpler than an ETS, because an ETS requires allowances to be tracked whereas a tax simply requires emissions to be monitored.  This is intrinsically simpler than tracking allowances in any case, and may be made more so by the existence of existing systems for taxing energy use.  For large economies the administrative costs of an ETS are likely to be a small proportion of the total scheme costs, but this may not be the case in smaller economies.  Simplicity may also be an appealing feature for jurisdictions with less developed administrative capacity, which may struggle to implement an ETS.

Fit with complementary measures

A carbon tax may also fit better with complementary measures, such as those to encourage deployment of renewables.  Unanticipated increases in renewables deployment can reduce the carbon price under an ETS in a way that is not possible with a tax, and indeed this is one of the factors that has contributed to lower prices under the EUETS [2].  (A related argument that the unilateral UK carbon tax, known as carbon price support, does nothing to reduce emissions because the cap is set at the EU level is less clear-cut, for example because of current surpluses under the EUETS, the potentially endogenous nature of future caps, and the risk of lock in from investment). [3]

Drawbacks to carbon taxes

However there are also drawbacks to taxes.  Setting the price of emissions at the level of damages is sound in principle.  However there is an order of magnitude uncertainty about what that cost of damage is.  Even if the damage is fairly constant (the slope of the curve is almost flat) there is still a risk of (greatly) over-pricing or under-pricing the damage, although If taxes are primarily intended to reach a certain target level of emissions by adjusting them over time this may be less of a concern.

And for larger jurisdictions the advantages of emissions trading remain – there is, by design, no limit on emissions under a carbon tax, so there is a risk of crossing thresholds of atmospheric concentration with consequences of very high damage costs.  In principle this risk may be mitigated by the possibility of increasing taxes rapidly as the threshold is approached.  However it may not be possible for governments to signal such an increase, or to implement it, especially as there would need to be an increase across all major jurisdictions to avoid crossing a global threshold of atmospheric concentration.  Furthermore such an increase may not be anticipated by investors in infrastructure, leading to difficulties in making large, rapid reductions in emissions even in the case of very high taxes.

Similarities between taxes and quantity limits

Under both a tax and an ETS learning is possible.  If the tax is not producing sufficient abatement then it can be increased, if a cap it producing low prices it can be tightened.  A tax may have some advantages in this respect as it can be adjusted annually, but something like the five year rolling cap introduced in Australia appears to offer opportunities for an ETS to show similar flexibility, so there does not appear to be a clear cut advantage for either type of instrument.

And in both cases it may be politically difficult to set the carbon price at an adequate level.  Taxes are rarely popular, although the British Columbia carbon tax seems to have done better than most.  And there will always be concerns about setting a cap too tight, risking higher prices and distorting growth.  This will be exacerbated by the interest of both governments and companies in being optimistic about economic growth and industrial production.

Comparison of properties of price and quantity instruments

The circumstances which favour emissions trading and taxes are summarised in the table below.

Factors where higher values favour caps (and lower values favour taxes) Because …
Share of global emissions covered Increased proportion of atmospheric GHG stock covered
Time periods for which policy is committed (including future targets) Increased proportion of atmospheric GHG stock covered
Length of life of investments Increased emissions lock-in, so larger contribution to GHG stock
Importance of strategic signal for technology and infrastructure development A cap can give clearer signals on longer term abatement
Variation in abatement costs over time The flexibility on timing offered by an ETS may help firms abate at lower cost
Administrative capacity Jurisdictions with higher administrative capacity will find the additional administrative burden of an ETS less onerous
Factors where higher values favour taxes (and lower values favour caps) Because …
Rate of decay of atmospheric GHG stock A higher decay rate of the gas in the atmosphere diminishes the effects of uncertain emissions on outcomes.
Discount rate A higher discount rate diminishes the importance of future damage from with uncertain emissions.
Frequency of policy review Adjustments to taxes can reduce expected deviations in emissions trajectories.
Importance of stable price signal for current investment A tax gives a constant price signal (though subject to amendment)


Some of the drawbacks of both an absolute cap and a pure tax come from the rigidity of either the quantity or price that is set.  In almost all markets supply and demand both vary with price.  However under a pure ETS the supply of allowances remains constant irrespective of the price (zero elasticity of supply), whereas under a carbon tax the variation of emissions is unlimited at the same price (infinite elasticity of supply).   Very few markets function this way, and a carbon market need not.  It is perfectly possible to set a price schedule which varies with the price of allowances.  Schemes which include both price and quantity limits are referred to as hybrid schemes, and these are reviewed in the next section.

Adam Whitmore – 5th May 2014



[1] For a detailed analysis of this issue see Newell and Pizer, Regulating Stock Externalities Under Uncertainty Resources for the Future, May 2000

[2] The recession appears to have been the main factor leading to lower prices under the EUETS.  However renewables deployment also appears to have played a a role.

[3] Separately it is also been argued that there are circumstances in which a tax may prevent capture of rents by oligopolistic fuel producers better than an ETS.  However it is not clear that the conditions in which such considerations prevail apply in practice given the current structures of gas and coal markets in particular.  See Goulder reference quoted in my previous post for a discussion of this issue.

Types of Carbon Pricing (Part 1 of 3)

This post is the first of three summarising the differing features of carbon pricing instruments – emissions trading (cap-and-trade), carbon taxes, and hybrids – and commenting on some of the implications for existing carbon pricing schemes.  The three together can be found as a pdf file under the carbon pricing pages of this site.

These posts focus on the differences between types of carbon pricing.  However there are many commonalities, and any type of well-designed carbon pricing is usually preferable to none.   Political circumstances will often play a major role when choosing the best approach in practice – a scheme which cannot be introduced for political reasons cannot be regarded as optimal in any practical sense – and a pragmatic approach to carbon pricing is likely to be the most productive.  Furthermore carbon pricing has quite a short history compared with many forms of regulation (the EUETS, the first large scale carbon pricing scheme, started less than 10 years ago).  For now there should not be undue concern about a wide diversity of approaches to carbon pricing, because this variety enables more to be learned about how different designs work in practice.

The role of uncertainty

The standard theory on the choice between a cap and trade scheme and carbon taxes(1) frames the problem as maximising total net benefits of pricing under uncertainty.  If the behaviour of the market were known in advance setting prices or quantities would yield the same result – one could set quantity knowing the price that would result, or vice versa.  However when market responses are uncertain, as they always are in practice, the two instruments have quite different properties.

When the abatement costs differ greatly depending on the amount of abatement required (high slope of the marginal abatement cost curve) a tax will tend to be preferred.  This is because setting a cap a little too high or a little too low could result in either excessively high prices for little benefit or missed opportunities if the tax is set too low.  In contrast when the damage costs are rapidly increasing as emissions rise (there is high curvature of the damage function) a cap will tend to be preferred.  This is because under a tax the price might be set too high, proving economically costly for little benefit, or too low, leading to very high damages as threshold levels of pollution are reached.

The problems caused by uncertain outcomes have been highlighted by the emergence of a large surplus of allowances under the EUETS, in which a fixed cap has led to unexpectedly low prices.

A long term cumulative cap to prevent dangerous thresholds being crossed

The prospect of rapidly increasing damages implies that globally and over the long term an emissions cap may have significant advantages for limiting greenhouse gas emissions, especially as avoiding severe damage requires deep cuts in emissions compared with business as usual.  The costs of the damages from climate change cannot be known in advance with any certainty, but seem likely to increase very rapidly (and highly non-linearly) as the concentration of greenhouse gases in the atmosphere increases and large irreversible damages, such as the melting of ice caps, are locked in.   This requires a limit on cumulative emissions (a global cumulative carbon budget) to prevent such large damages being realised, including those from natural thresholds being crossed.  There is inevitable scientific uncertainty about exactly where each threshold is, so attitudes to risk will also be important in setting the cap.

Addressing such rapidly increasing damages ideally requires a global annual cap reducing over time to be set, such that the cumulative total cap (area under the curve) corresponds to a limit on the cumulative emissions above which dangerous thresholds may begin to be crossed.  (The situation is complicated by the need to take account of sinks and other forcings such as aerosols, and further by the dependence of damage on the path of the stock over time.)   This type of approach informs the analysis of the limits of the cumulative amount of fossil fuel that can be burnt (see the recent Fifth Assessment Report from the IPCC (ref 2)).

The effect of decisions taken now on the stock of a pollutant over time is particularly relevant for climate change.  Much energy infrastructure has a very long life, so a decision now will influence emissions for decades.  Furthermore a larger proportion of CO2 emitted stays in the atmosphere for centuries, so emissions determined by current investment decisions will affect the stock in the atmosphere over all relevant timescales.

There is no binding global agreement to establish such a cumulative cap (nor does there appear likely to be).  However an increasing number of emissions trading schemes in major economies with stringent long term emissions goals are being established, and may provide over time the best approximation to the ideal of a global emissions limit that is likely to be available.  For example the EU has cap decreasing at 1.74% p.a. with a goal of 80-95% reduction by 2050, California also has a clear 2050 goal of 80% reduction from 1990 levels, and China is increasingly expressing emissions reduction ambitions and implementing them in the form of emissions caps, regionally at present but with ambitions to move to national limits.  In contrast a series of carbon taxes would give much less certainty of staying below any threshold.


Emissions trading may also allow more flexibility in how and when emissions are reduced.  Banking provisions and multi-year compliance periods, which feature in most scheme designs, can allow firms to make choices about when to abate and how much, giving them flexibility in reducing costs in ways which are difficult to replicate under a tax.

Strategic signals

Emissions caps can have the further advantage of giving a stronger strategic signal that emissions will have to decrease to much lower levels in the long term.  The changes required to achieve this often fundamental and transformational rather than marginal.  The signal provided by a long term quantity limit may prove effective in stimulating investment in technology development, physical infrastructure, grid operating regimes and other longer term elements of a low carbon economy.   These will require many other policy interventions, and will not be achieved by carbon pricing alone.  However a cap can be useful in making the case for these measures by defining the scale of the challenge (although it can have the weakness of not incentivising measures that go beyond the cap, which is a point I’ll return to in looking at hybrid instruments).

In contrast a tax, even if effective in signalling marginal changes, may not signal more fundamental change to the same extent, although a defined escalator on a tax may go some limited way towards this.  For example, very high fuel taxes have played a role in incentivising improved fuel efficiency in vehicles, but fleet efficiency standards (in effect a declining cap on emissions intensity) have also played an important role, and the expectation of the need to move to very low levels of emissions seems to have been important in stimulating the world motor vehicles industry to put vast resources into developing electric vehicles.  Similarly the expectation of very substantial decarbonisation of the power sector created by quantity targets appears to be driving necessary discussions and early action on grid design, trading arrangement reform, and system operation.

Addressing competitiveness concerns

There may also be some advantage from greater administrative ease in addressing concerns about competitiveness of emissions intensive trade exposed industry through the allocation of free allowances.  In principle the same outcomes can be achieved with a tax by setting thresholds above which the tax is payable, as, for example, under the proposed South African carbon tax.  However providing such shielding under an ETS may be politically, legally or administratively simpler under an ETS.  For example, it may be politically difficult to be seen to “give tax breaks to big polluters”.

Offsets and linking

Other proposed advantages of emissions trading are less compelling relative to a carbon tax.  For example, offsets can be included under a tax, as is proposed in South Africa and Mexico, as well as under an ETS.  There may be potential to link emissions trading schemes.  However at present trading schemes remain diverse with wide dispersion of prices and limited prospects for direct linkage.  And under a tax governments can easily look to the levels of taxes elsewhere and take that into account in setting their own tax rates, with some potential for linking taxes by means of credits if this is desired.

Quantity limits as an expression of non-monetary values

Among the most compelling reason for choosing caps is that the consequences of climate change imply choices about issues that are not captured by an economic cost benefit analysis looking at maximising net benefits.  Choices are ultimately about the effects some people now impose on others, the legacy current generations leave for the future and how can this be balanced against the needs of the present.  This necessarily requires the debate to address how acceptable we find the risk of melting ice caps or the loss of the Amazon forest.  While economic analysis may inform some of these choices it cannot make them, because in the end they are not only about money.  Under this framework an emissions trading scheme is an instrument to achieve a goal that is necessarily specified outside the framework of net monetary benefits.  This is represented much more directly by limits on cumulative emission than by a carbon, even though uncertainties about the effects of a particular atmospheric concentration remain.

Drawbacks to quantity limits

However an ETS also has drawbacks.  Prices can be very volatile, because abatement is typically a small proportion of emissions making the price the result of a small difference between two numbers (the cap and business as usual emissions), one of which is rigidly fixed and the other of which is highly uncertain(3).  Such volatility is likely to persist, even with provisions to bank allowances, which are intended to smooth out price fluctuations, and with other provisions such as overlapping or rolling compliance periods.  For example, banking is a feature of the EUETS, and prices have still been volatile, although banking has helped sustain Phase 3 prices above zero.

Highly volatile prices are undesirable because they increase the risk of investments in abatement, and hence their costs, leading to decreased economic efficiency.  Volatile prices may also bias the form of abatement towards shorter term expenditure, such as fuel switching, rather than longer term investment.  They also make government finances more difficult to plan where auctions are used, and make a revenue neutral carbon pricing scheme, often an objective of policy, more difficult to sustain.

More fundamentally, an emissions trading scheme may fail to price emissions correctly in some circumstances because it fails to give any incentives to reduce emissions further below the cap.  If emissions are below the cap, allowances are not scarce, and the price drops to close to zero.  However emissions below the cap impose a cost and so should be priced(4).  This is evident under the EUETS at the moment where further abatement would clearly have value not signalled by the current price.  This problem of under-pricing damage is especially severe given the limited time horizons and incomplete commitment that are part of emissions trading schemes in practice.

Such drawbacks may carry particular weight where the advantages of an ETS seem less compelling.  My next post will look at what the alternative of carbon taxes might deliver, and the circumstances in which this might be a more appropriate policy choice.

Adam Whitmore – 28th April 2014


1 This basis of the choice between price and quantity instruments was first laid out clearly in by Martin Weitzman  in one of the most widely cited papers in the environmental economics literature (Weitzman, M.L. 1974 Prices vs. quantities, Review of Economic Studies 41 (4) 477 -491).     A good recent survey of the merits of different approaches is in Carbon Taxes vs. Cap and Trade: A critical Review, Lawrence H. Goulder Andrew Schein  Working Paper 19338

3.  See Grubb, M. (2009). Reinforcing carbon markets under uncertainty: the role of reserve price auctions and other options for a discussion of this point.

4  This is measured by the Social Cost of Carbon.  There are large uncertainties in estimating what this is, and it is difficult to account for non-market impacts, but it nevertheless provides a useful indication of the cost of damage, at least as a lower bound – see page below in this section for a discussion of the social cost of carbon.