Can emissions trading produce adequate carbon prices?

Prices under emissions trading schemes have been low to date.  Sometimes this may be because systems are new, but the EUETS is long established and needs to demonstrate that it can now produce adequate prices. 

Prices under emissions trading systems around the world have so far remained low.  The chart below shows carbon pricing systems arranged in order in increasing price, with prices on the vertical axis shown against the cumulative volume covered on the horizontal axis.  Carbon taxes are shown in purple, emissions trading systems in green.  It is striking that all of the higher prices are from carbon taxes, rather than emissions trading systems.

Prices under Emissions Trading Systems and Carbon taxes in 2016


Source:  World Banks State and Trends of carbon pricing report[1].  Prices are from mid-2016.

Prices in the largest emissions trading system, the EUETS have been around $5-6/tonne, and prices in the Chinese pilot schemes have been similar and in some cases even lower, although with little trading.  The price under the California and Quebec scheme (soon to be joined by Ontario) is somewhat higher.  However, this is supported by a floor set in advance and implemented by an auction reserve price.  If this price floor were not present a surplus of allowances would very likely have led to lower prices.  The Korea scheme has had very low trading volumes, so does not provide the same sort of market signal found under more liquid schemes.

In contrast, a wide range of carbon taxes are already at higher levels and in some cases are due to increase further.  The French carbon tax, which covers sectors of the economy falling outside the EUETS, is planned to reach €56/tCO2 (US$62/tCO2) in 2020 and €100/tCO2 (US$111/tCO2) in 2030[2].  In Canada a national lower limit on carbon prices for provinces with an explicit price-based system (not shown on the chart) is due to reach $50 per tonne in 2022[3]. The UK carbon price floor, which covers power sector emissions, was due to rise to substantially above current levels, but is currently being kept constant by the Government, mainly because the price under the EUETS is so low.

Increases such as those due in France and Canada will bring some carbon taxes more in line with the cost of damages, and thus to economically efficient prices.  The cost of damages is conservatively estimated at around $50/tonne[4], rising over time (see here for a discussion of the social cost of carbon and associated issues).  The increases will also bring prices more into line with the range widely considered to be necessary to stimulate adequate low carbon investment[5].

Low prices under emissions trading systems have been attributed to a range of factors, including slower than expected economic growth and falling costs of renewables[6].  However these factors do not explain the consistent pattern of low prices across a variety of systems over different times[7].

While it is difficult to derive firm evidence on why this pattern should be present, two factors seem plausible.  The first is systematic bias in estimates – industry and governments will expect more growth that actually occurs, costs will be overestimated, and these tendencies will be reflected in early price modelling, which can often overstate likely prices.

But the second, more powerful, tendency appears, based on anecdotal evidence, to be that there is an asymmetry of political risk.  The political costs of unexpectedly low prices are usually perceived as much less than those of unexpectedly high prices, and so there will always be tendency toward caution, which prevents tight caps, and so leads to prices being too low.

This tendency is difficult to counteract, and has several implications for future policy.

First, it further emphasises the value of price floors within emissions trading systems.  Traditional environmental economics emphasises the importance of uncertainty around an expected level of abatement costs or damages.  If decision makers are not in fact targeting expected average levels, but choosing projections of allowance demand above central expectations then the probability of very low prices is increased, and the case for the benefits of a price floor is stronger.

Second, it implies that it is even less appropriate than would anyway be the case to expect the carbon price alone to drive the transition to a low carbon economy.  Measures so support low carbon investment, which would in any case be desirable, are all the more important if the carbon price is weak (see here for a fuller discussion of the value of a range of policy measures).   While additional measures do risk further weakening the carbon price, they should also enable reduced emissions and tighter caps in future.

Third, it requires governments to learn over time.  Some low prices may reflect the early stage of development of systems, starting slowly with the intention of generating higher prices over time.  However this does require higher prices to eventually be realised.

The EUETS has by some distance the longest-established system, having begun eleven years ago and with legislation now underway for the cap to 2030, by which time the system will be 25 years old.  The EU should be showing how schemes can be tightened over time to generate higher prices.  However it now looks as though the Phase 4 cap will be undemanding compared with expectations (see previous posts).  The recent vote by the European Parliament’s ENVI committee failed to adopt measure that are adequate to redressing the supply demand balance, with tweaks to the market stability reserve unlikely to be enough.  This undermines the credibility of cap-and-trade systems more generally, rather than setting the example that it should.  Further reform is needed, including further adjustments to supply and preferably auction reserve prices.

The advantages of cap-and trade systems remain.  Quantity limits are in line with the international architecture set by the Paris Agreement.  They also provide a clear strategic signal that emissions need to be reduced over time.

However there is little evidence to date that emissions trading systems can produce adequate prices. The EU, with by far the most experience of running an ETS, should be taking the lead in substantially strengthening its system.  At the moment this leadership is lacking.  Wider efforts to tackle climate change are suffering as a result.

Adam Whitmore – 23rd January 2017


[2] World Bank State and Trends in Carbon Pricing 2016.  See link in reference 1.

[3]  Canadian provinces with volume based schemes such as Quebec with its ETS must achieve emissions reductions equivalent to these prices.

[4] $40/tonne in $2007, see, escalated to about $50 today’s dollars.

[5] See this recent discussion:

[6] Ref: Tvinnereim (2014)


[7] The South Korea ETS may be a partial exception to the pattern.  However it is unclear due to the lack of liquidity in the market.

When carbon pricing works

The UK’s carbon price floor mechanism has proved very effective at securing cost-effective emissions reductions.  It offers lessons for other carbon pricing schemes.

I have previously pointed out some of the flaws with carbon pricing schemes, notably the EUETS, so it seems appropriate for balance also to look at an example of carbon pricing working well.

There are few more successful examples of carbon pricing, or indeed of climate change policy generally, than the UK’s carbon price support (CPS), which in effect tops up the carbon price set by the EUETS for UK power generation.  It has been an essential element in securing large scale, cost effective emissions reductions over the last few years, mainly by incentivising a switch from coal to gas.

Emissions from coal generation in the UK have fallen by around 80% in the last four years (see chart).  The net reduction in emissions less than that, as much of the generation from coal has been replaced with gas, but this still amounts to a net reduction of around 10-15% of total UK emissions of greenhouse gases.  (Gas generation remained almost constant between 2012 and 2015, as demand fell and renewables grew rapidly, although it may increase this year.  However it would have almost certainly fallen in the absence of the CPS, with coal declining much less rapidly than it has.)

Emissions from coal generation in the UK have fallen rapidly …


Source:  EUETS data, 2016 estimate by Sandbag.

The reduction in emissions from coal has been due to a range of factors, including the age of the UK’s coal plant, almost all of which dates from the 1960s, regulation of other pollutants, falling demand, the growth of renewables and low gas prices.  But a good deal of the reduction has been due to the CPS, especially in the last two years, when its level has been increased significantly.

The carbon price floor (EUETS plus carbon price support) has led generation from gas to be cheaper than that from coal.  This is shown in the chart below from Cornwall Consulting.  The red crosses show the costs of generation without carbon price support.  Coal is cheaper than gas (around £30/MWh vs. £38/MWh).  However with the carbon price in the order is reversed.  Gas becomes cheaper than coal, so tends to run more.  Coal stays on the system due to the capacity price, providing system security, but runs much less.


Source: Cornwall Consulting  

This has been achieved with a carbon price that remains quite moderate.  It is little more than half the cost of the damages caused by emissions (the social cost of carbon, see here and reference therein), and in line with or lower than a range of other carbon taxes (see the recent World Bank Report p.27 ).

The carbon price floor also has a number of other advantages.  For example, but increasing the price of fossil generation it makes onshore wind more competitive.  Onshore wind is the cheapest form on low carbon power in the UK, and receives no subsidy at present.

It has been argued that the emissions reductions due to carbon price support are not retained due to the presence of the EUETS cap, but this is not true in practice (see previous post here), with the vast majority of the reductions retained in full.  However it would be much better still if there were an EU wide floor price, and the EUETS price were at or above the current UK carbon price.

This UK experience shows a number of things.  Firstly carbon pricing can be effective even at fairly moderate levels.  Second, fuel switching can indeed offer substantial, low cost abatement in the short to medium term even if it is not enough in the long term.  Third, a lower bound on price is useful in an ETS – these reductions would not have happened as a result of the EUETS alone.  Fourth, system security is consistent with emissions reductions if coal is kept on the system for a while, but does not run much.

The UK government is understood to be reviewing the role of the carbon price floor at the moment.  In view of its effectiveness and the example this sets for other jurisdictions, the government needs to ensure that the price is maintained at a minimum of its current level, and extended beyond 2020.

Adam Whitmore – 9th November 2016


Additional actions in EUETS sectors can reduce cumulative emissions

It is often claimed that additional actions to reduce greenhouse gas emissions in sectors covered by the EUETS are ineffective because total emissions are set by the level of the cap.  However this claim is not valid in the current circumstances of the EUETS, and is unlikely to be so even in future.  Additional emissions reduction measures in covered sectors can be effective in further permanently reducing emissions.

This post is longer than usual as it deals with a very important but relatively technical policy issue.

The argument about the effectiveness of additional actions to reduce emissions …

Many additional actions are being taken to reduce greenhouse gas emissions in sectors covered by the EUETS.  These include energy efficiency programmes, deployment of renewables, replacing coal plants with less carbon intensive generation, and national carbon pricing.

It is often argued that such additional actions do not reduce total emissions because the maximum quantity of emissions is set by the EUETS cap, so emissions may remain at the fixed level of the cap, irrespective of what other action is taken (see the end of this post for instances of this argument being used publicly).

However, this argument does not stand up to examination.

Assessment of the argument needs to take account of the current circumstances of the EUETS.  Emissions covered by the EUETS were some 200 million tonnes (about 10%) below the cap in 2015.  This year emissions are likely to be 13% below the cap.  The EUETS currently has a cumulative surplus of almost three billion allowances, including backloaded allowances currently destined for the Market Stability Reserve (MSR), and the surplus is set to grow as emissions continue to be less than the cap.

In these circumstances emissions reductions from additional actions will mainly increase the surplus of allowances, with almost all of these allowances ending up in the (MSR).  These allowances will stay there for decades under current rules, and so not be available to enable emissions during this time.

Indeed, in practice these allowances are unlikely ever to enable additional emissions.  The argument that they will assumes that the supply of allowances is fixed into the long term.  In practice this is not the case.  Long term supply of allowances is determined by policy, which can and does respond to circumstances.  Additional surpluses and lower prices are likely to lead to tighter caps than would otherwise be the case, or cancellation of allowances from the MSR or elsewhere.

The remainder of this post looks at these issues in more detail, including why the erroneous view that additional actions don’t reduce cumulative emissions has arisen.

Why current circumstances make such a difference

The argument that additional actions to reduce emissions will be ineffective reflects how the EUETS was expected to operate when it was introduced. It was assumed that demand for allowances would adjust so that the quantity of allowances used would always equal to the cap, which was assumed to be fixed.

This is illustrated in stylised form in the diagram below.  The supply curve is vertical – perfectly inelastic supply.  Demand for allowances without additional actions leads to prices at an initial level.  Additional actions reduce demand for allowances at any given price, effectively shifting the demand curve to the left by the amount by which additional actions reduce emissions.  This leads price to fall until the lower price creates sufficient additional demand for allowances, so that total demand for allowances is again equal to the supply set by the cap.  Because the supply curve is fixed (vertical) the equilibrium quantity of emissions is unchanged, remaining equal to the cap[1].

Chart 1: A price response to the change in demand for allowances can lead to emissions re-equilibrating at the cap when allowances are scarce …


However, at present, large increases in emissions (such that emissions rise to the cap) due to falling prices are clearly not occurring, and they seem unlikely to do so over the next few years.  As noted above, the market remains in surplus both cumulatively and on an annual basis.  The price would be close to zero in the absence of banking of allowances into subsequent phases, because there would be a cumulative surplus over Phase 3 of the EUETS, and so no scarcity[2].

If demand were further reduced in the absence of banking there would be no price fall, because prices would already be already close to zero.  Correspondingly, there would be no increase in demand for allowances to offset the reduced emissions from additional actions.  The emissions reductions from additional actions would be retained in full. This is again illustrated in stylised form in the diagram below. 

Chart 2: With a surplus of allowances and price close to zero (assuming no banking) any reduction in demand for allowances will be retained in full …


In practice the potential to bank allowances and the future operation of the MSR supports the present price.  It is expected that in future as the cap continues to fall allowances will become scarce.  There is thus a value to allowances set by the cost of future abatement.

Additional actions now to reduce emissions increase the surplus, and so postpone the expected date at which the market returns to balance.  This reduces current prices.  This will in turn lead to some increase in emissions.  However, this increase will be small – much smaller than if the market were short of allowances now.

Quantifying this effect 

Modelling indicates that if additional actions are taken over the next 10-15 years, then the increase in demand for allowances due to falling price will be less than 10% of the size of the reduction in emissions[3].  Correspondingly more than 90% of the emissions reductions due to additional actions are retained, adding to the surplus of allowances which, which end up in the MSR.  Modelling parameters would need to be in error by about an order of magnitude to substantially affect this conclusion.

This effect arises in part because of the low level of prices at present.  This means that even a large percentage change in price leads to a small absolute change, and thus a small effect on demand for allowances.  Even a 50% price fall would be less than €3/t at current price levels.  It also reflects that the shape of the Marginal Abatement Cost curve, with price falls only increasing abatement by a small amount.  This means that even if prices are higher than current levels the effect of price falls on demand for allowances is still relatively small.

The relatively small response to price changes is consistent with the current market, where there is a lack of sufficient increase in demand to absorb the current yearly surplus (or even to come close to doing so).

The 90%-plus of the allowances freed up by additional actions are added to the surplus end up over time in the MSR.  They then stay there for several decades.  This is because even without additional actions, and even with some reform of the current proposals for Phase 4 (which covers 2021 to 2030), the MSR is likely contain at least three billion allowances by 2030, and perhaps as much as five billion.  This will take until 2060 to return to the market, and perhaps until the 2080s, at the maximum rate written into the legislation of 100 million per annum.

Any additional surplus will only return after this.  Even if the return rate of the MSR were doubled the return time for additional surplus would still be reckoned in decades from now.

This will be even more the case if proposals for the EUETS Phase 4 are not reformed, and the surplus of allowances being generated anyway is correspondingly greater.

The implications of the very long delay in the return of allowances

It seems unlikely that allowances kept out of the market for so long would ever lead to additional emissions.  It would require policy makers to allow the allowances to return and enable additional emissions.  This would be at a time when emission limits would be much tighter than they are now, and indeed with a commitment under the Paris Agreement to work towards net zero emissions in the second half of this century.

There are several policy mechanisms that could prevent the additional surplus allowances enabling emissions.  Subsequent caps tighter as unused allowances reduce the perceived risk of tighter caps, and additional actions now set the economy on a lower carbon pathway.  Furthermore, with a very large number of allowances in the MSR over several phases of the scheme, allowances may well be cancelled.  Indeed, over such long periods the ETS itself may even be abolished or fundamentally reformed, with allowances not carried over in full.  Or a surplus under the EUETS may persist indefinitely as additional actions succeed in reducing emissions.

As the market tightens towards 2030 it is likely that a higher proportion of any additional emissions reductions will be absorbed by the market via a price effect, but it still seems unlikely to be as much as 100% given the long term trend to lower emissions and the lack of additional sources of demand, especially in the event of large scale additional actions[4].  Some of the policy responses described would still be expected to reduce the supply of allowances.


The argument that emissions will always rise to the level of the cap manifestly does not hold at present, when emissions are well below the cap. and there is a huge cumulative surplus of allowances.

In future, it seems likely that more than 90% of reductions in emissions from additional actions will simply add to the surplus, and eventually end up in the MSR.  They at least stay there for several decades, because of the very large volume that will anyway be in the MSR.

While there is in principle a possibility that they will eventually return to the market and allow additional emissions this appears most unlikely in practice.  Policy decisions will be affected by circumstances and this can readily prevent additional emissions, through some combination of tightening of the cap and cancellation of allowances.

Even when the market returns to scarcity these policy responses are likely to hold to a large extent, for example with lower prices enabling more stringent caps.  The hypothesis of no net reductions in emissions from additional actions thus seems unlikely ever to hold true.

Spurious arguments about a lack of net emissions reductions should not be used as a pretext for failing to take additional actions to reduce emissions now.

Adam Whitmore – 21st October 2016


Note:  A more detailed review of the issues raised in this post, and the accompanying modelling can be found in this report.


Examples of statements invoking the idea of fixed total emissions

For example, in 2015 RWE used such arguments in objecting to the closure of coal plant:

“The proposals [to reduce lignite generation] would not lead to a CO2 reduction in absolute terms.   [The number of] certificates in the ETS would remain unchanged and as a result emissions would simply be shifted abroad.” [5]

Similarly, in 2012 the then Chairman of the UK’s Parliament’s Energy and Climate Change Select Committee, opposed the UK’s carbon price support mechanism for the power sector arguing that:

“Unless the price of carbon is increased at an EU-wide level, taking action on our own will have no overall effect on emissions”[6]

Neutral, well-informed observers of energy markets have also made this case.  For example, Professor Steven Sorrel of Sussex University recently argued that:

“Any additional abatement in the UK simply ‘frees up’ EU allowances that can be either sold or banked, and hence used for compliance elsewhere within the EU ETS[7]



[1] This is analogous to the well-established rebound effect for energy efficiency measures.  Improved domestic insulation lowers the effective price of energy, so consumers take some of the benefits as increased warmth, and some as reduced consumption.  The argument here is that in effect there is a 100% rebound effect for emissions reductions under the EUETS.

[2] Such a situation occurred towards the end of Phase 1 of the EUETS (2005-7), which did not allow banking into Phase 2.  Towards the end of the Phase there was a surplus of allowances and the price fell to close to zero.

[3] The price change is modelled by assuming the price is set by discounting future abatement costs, with a later date for the market returning to balance leading to greater discounting and so a lower price.  The increase in demand for allowances is modelled based on a marginal abatement cost curve and consideration of sources of additional demand.  See report referenced at the end of this post for further details of the modelling.

[4] There are likely to be path dependency and hysteresis effects in the market which prevent a full rebound.

[5] See RWE statement, “Proposals of Federal Ministry for Economic Affairs and Energy endanger the future survival of lignite”, 20 March 2015.



The EUETS and the need for price floors (and maybe soft ceilings)

Standard objections to introducing price containment mechanisms into the EUETS carry little weight.  It’s time to give price containment more serious consideration.

With the price of allowances in the EUETS currently down at around €4/tCO2 the question of whether direct price containment (price floors and ceilings) should be introduced has naturally been the subject of renewed debate, especially in the light of the French proposal earlier this year to introduce a price corridor.

The debate tends always to feature a standard set of objections to price containment.  Most of these lack validity when applied to well-designed mechanisms.  Here I take a look at why this is so, in the hope that the debate can become more realistic and constructive, focusing on the benefits and design challenges around price containment.

The broad themes underpinning the rationale for price containment are as follows:

  1. All emissions of GHGs are damaging, not just those above the cap. Reducing emissions below the cap and further tightening the cap thus have benefits.
  2. The financial cost of damages emissions (the social cost of carbon – SCC), although uncertain, is well above current prices[1]. This implies that further emissions reductions with a cost between the current price and the cost of damages have a net benefit.  However these are not currently being incentivised by the carbon price.  This is one reason why a floor prices is beneficial.
  3. The market structure and parameters are set by regulatory decisions. These decisions are inevitably taken under uncertainty, and market design is about optimising outcomes under uncertainty.  Design is more robust to uncertainty with both price and quantity targets than with either alone.
  4. Supply adjusting in response to price makes the EUETS more like a normal market.
  5. It is essential for reasons of international obligations and environmental integrity that the cap is maintained[2], so moving to a pure carbon tax is not a good idea.

Based on these premises the following responses to standard objections to price management can be made.

“Price management is interfering in the market”

The form of the market is a politically determined construct. Modifications to this construct are appropriate to correct shortcomings in the current design, where supply is too rigid to accommodate uncertainties. The cap does succeed in limiting the total emissions but fails to produce adequate signals for additional abatement.   Modification is required to reduce supply of allowances if prices become too low, in order to retain efficient price signals.

Allowing the supply of allowances to respond to price is not interfering with the day-to-day operation of the market. On the contrary, it is designing it to function more like a normal market.  In most markets supply varies with price (elasticity of supply is not zero in most markets[3]).

 “There is no environmental benefit to a floor price because the cap does not change” or “it does nothing to reduce supply or increase ambition towards targets in the Paris Agreement”

The critical question here is what happens to unsold allowances. There are many possibilities for dealing with unsold allowances, including cancelling them at the end of a phase, cancelling a proportion at the end of a phase, or cancelling them on a rolling basis.

Provided that there are appropriate provisions for cancelling unsold allowances, total emissions over time can be reduced, and so there is a clear environmental benefit.  Even if this is not the case, allowances may simply stay in the reserve, or caps may be tighter in future due to emissions reductions achieved, also creating an environmental benefit.

“If the EU is meeting its target at low cost the price should be correspondingly low”

No it should not.  The low price signals that the target is not stringent enough to adequately reflect damages.  All emissions are damaging, even those within the cap, and if more abatement can be achieved at lower cost than the damage caused this is what should happen.

Measures which further decrease emissions in response to lower cost of abatement also help reinforce the EU’s international leadership position on climate change.

“It goes against the quantity based nature of the EUETS” or “it’s introducing a carbon tax”

Prices can managed by automatically adjusting supply in response to price, for example by putting a reserve price in auctions.  This is entirely consistent with the quantity based nature of the EUETS, in that it works by adjusting quantity.  Indeed, as noted, it makes the EUETS more like almost all other markets where the quantity of supply varies in response to market prices.

It is possible to use tax-based measures to impose a floor, as the UK does and France will do from January 2017, but it is not necessary to do so.

Characterising price floors  as a tax appears often to be used as a way of creating political momentum against the idea.  An EU tax requires unanimity among Members States and attempts to introduce a carbon and energy tax in the 1990s were notably unsuccessful, and similar efforts would doubtless prove challenging.  Characterising floors as a tax may also help develop political opposition to a floor.  Branding the Australian ETS as a tax (which it was not) was successful in helping build opposition there, with eventual repeal of the scheme.  Price management through adjusting quantities should not be misrepresented in this way to artificially discredit it.

“It reduces market efficiency”

This confuses efficiency of trading with efficiency of the price signal.  If you were never to change the number of allowances, trading alone might indeed remain the most efficient way of meeting the cap.  However this has created prices which failed to adequately signal efficient abatement (in effect the market is telling you that the current cap is too loose).  There is thus a misallocation of resources towards to many emissions and too little abatement.

“The price may be set at the wrong level”

Having both price and quantity limits increases robustness to the unexpected.  If the cap has been set at appropriate levels then prices will anyway lie within the range of any  price containment, and price limits will not bind.  However the existing EUETS cap has been set at a sub-optimal level –too many allowances have been issued and the price is too low.

Limiting the price simply recognises that future demand for allowances may be mis-estimated, or the level of the cap may be subject to biases, for example due to asymmetries of political risk from setting the cap too high or too low.

 “It will never be possible to agree a price”

Price will doubtless be contentious but there are several reference points, notably the following:

  • estimates of the SCC, which represents the financial cost of damages, although calcualtions typically exclude important costs of damage. The SCC is highly uncertain, but well above the €4/tonne currently prevailing in the EUETS under almost any reasonable set of assumptions.
  • prices under other schemes, especially those with price management;
  • prices consistent with those needed to signal abatement sufficient to reach climate targets.

This gives a framework of negotiation.  The level of the cap, which is always set with a view to abatement costs and prices, is anyway contentious.

There are many difficult issues to resolve in designing appropriate price containment mechanisms for the EUETS and setting price boundaries at appropriate levels.  Spurious objections such as the ones discussed here should not be allowed to form an obstacle to much-needed debate about the best way forward.

Adam Whitmore – 14th September 2016

Note:  The advantages of hybrid price quantity instruments have been extensively reviewed in the environmental economics literature, going back to the original paper on the subject by Roberts and Spence Effluent Charges and Licenses Under Uncertainty (1976).  Understanding the need for prices to fully reflect the cost of environmental damages goes back further, to Pigou “The economics of welfare” (1920).  See standard texts on environmental economics for a fuller treatment.  These conclusions are not uncontentious, in particular because some observers continuing to favour a carbon tax.  My own view remains that including a cap on emissions is preferable, and that many of the advantages of a carbon tax can be realised by a well-designed floor price.

[1] Furthermore there are other non-priced damages which imply the benefit of abatement is greater than implied by the SCC.

[2] Also, any ceiling should be soft to allow prices to rise above the ceiling rather than allowing emission to go above the cap, for example with allowances in price containment reserve taken from within the cap.

[3] Almost the only markets with completely fixed supply are the markets for tickets to major sporting events and for authentic works by dead artists.  For example the number of tickets to the men’s final at the Wimbledon tennis championships is limited by the number of seats, and the number of authentic Picasso’s cannot now increase with price (although the number of fakes can).


Reflecting reality in the EUETS Phase 4 cap

The cap for Phase 4 of the EUETS, which runs from 2021 to 2030, needs to start at a level that matches the reality of emissions in 2020, rather than starting where the Phase 3 cap finishes.   

The EUETS surplus will continue to grow through under current proposals …

The surplus of allowances in the EUETS looks to set to get worse with the Commission’s current proposals for the Phase 4 cap, which covers the period 2021 to 2030.   In 2015 emissions covered by the EU ETS were already below the level of the cap for 2020[1].  Emissions are expected to continue falling through the remainder of this decade, driven mainly by increasing deployment of renewables and weak electricity demand.  By 2020 emissions look likely to be over 10% below the cap at the end of Phase 3 (see chart).   This will lead to additional surplus allowances generated from the start of Phase 4, continuing through all or most of Phase 4.   This will in turn lead to the EUETS remaining weak even in the presence of the Market Stability Reserve (MSR).

Chart 1:  Currently proposed cap against emissions forecasts and 2020 gap to cap

Chart 1

This problem arises in large part because the starting point for Phase 4 cap is out of date.   It was effectively set in 2010 as part of the cap for Phase 3, because the EUETS Directive implicitly assumes that the Phase 4 cap will simply continue from where the Phase 3 cap finishes.   However the Phase 3 cap was set before many subsequent trends were known, including the growth of renewables and the length and depth of the economic recession.  Consequently it does not form a suitable starting point for Phase 4, and now looks far too loose.

This problem can be mitigated by changing the starting point of the Phase 4 cap …

The Phase 4 cap needs to start at a level that reflects actual emissions (if this is, as expected, below currently proposed level, which would act as an upper bound in any case).   Rebasing the cap in this way would lead to a much more effective EUETS that delivers effective signals for emissions reductions and investment.  Without this sort of reform the EUETS risks being reduced to little more than an accounting tool, with a chronic surplus and individual Member States increasingly taking their own action to ensure the necessary investment.

This increases the robustness of the mechanism and is more effective than changing the Linear Reduction Factor ….

Rebasing to actual emissions increases robustness of the system by making it dependent on actual outcomes.  Aligning the cap with actual emissions also tightens the cap more quickly and more effectively than changes to the Linear Reduction Factor (LRF – the amount of annual emission reductions built into the EU ETS during the phase).   This is shown in the chart below.

The LRF would need to approximately double from the currently proposed value of 2.2%, to 4.2%, to have the about the same effect on cumulative number of allowances over Phase 4 as rebasing the cap, even in a high emissions case.  And an even greater LRF would be needed to match the effect of rebasing if emissions by 2020 are low.  Even then, changing the LRF reduces the level of the cap more slowly than changing the starting point of the cap.  However increasing the LRF in addition to rebasing the cap helps ensure that surpluses are eroded and do not re-emerge through Phase 4, and so increasing the LRF remains a useful complement to rebasing the cap.

Chart 2.  Decrease in the total Phase 4 cap relative to the current proposal

Chart 2

Rebasing the cap is consistent with a range of precedents …

This approach of adjusting caps to reflect the reality of actual emissions, where these diverge from earlier expectations, has been applied elsewhere.  For example, in the Regional Greenhouse Gas Initiative (RGGI) in the USA, the cap was reduced from 165 million short tons in 2012-3 to 91 million short tons in 2014 to more closely reflect actual emissions[2].  As a result, prices have moved away from the auction floor price, where they were had previously been stuck.

Looking beyond carbon markets, incentive-based regulation of electricity, gas and water network charges in the UK in the 1990s imposed price caps typically lasting five years.  In practice, costs fell more rapidly than was expected when the price cap was set, leading to high margins of price over cost.  One-off cuts in the level of prices, referred to as P0 cuts, were implemented at the start of the next phase of the price control to realign the price cap with outturn costs, and thus capture the benefits of efficiency gains for consumers.

The new starting point for Phase 4 would also be closer to that which was envisaged under the December 2008 European Council Conclusions[3] in case an international agreement was reached and the EUETS would start from a reduction of 30% from 2005 levels by 2020.  A 30% reduction from 1990 would, assuming the EU ETS cap to have been reduced in line with the reduction in other sectors, have led to a starting point for the Phase 4 EU ETS cap of approximately in line with emissions now expected.  This was made conditional on action by other countries.  Commitments to such action have now been made under the Paris Agreement.


With the cap proposed by the Commission the EUETS seems likely to continue providing a largely ineffective signal for abatement well into the 2020s and possibly beyond.   This would mean that by 2030 the  EUETS will have been in existence for a quarter of a century, but will have provided an effective price signal for only a short period in the early part of Phase 2 (around 2009).

A simple adjustment to bring the cap at the start of Phase 4 into line with the reality of emissions would go a long way towards solving this problem by reducing the Phase 4 cap, likely by around 2 billion tonnes or more over the 10 years of the phase.   There are few easier and more natural adjustments to the scheme which could have such an impact.

Adam Whitmore – 20th  June 2016

Thanks to Boris Lagadinov for providing the analysis shown in this post.   This post is based on a recent paper Boris and I wrote for Sandbag – see

[1] The cap for 2020 is 1816 MtCO2 excluding the effects of backloading.  Emissions were 1802 MtCO2 in 2015.



How fast could the market for electric vehicles grow?

Various policy driven scenarios show electric vehicles gaining market share over the next few decades but with the turnover of the vehicle stock taking longer.

I recently argued that BP’s projections showing almost no take-up of plug-in vehicles[1] by 2035 was unrealistic in view of several convergent trends.  There is increasing pressure to reduce CO2 emissions, there is large and growing concern about urban air quality,  and electric vehicles are likely to prove attractive to consumers in many respects.  In line with these drivers, sales are growing very quickly and many new models are coming on line, while battery technology is improving rapidly, with costs falling sharply and energy density rising.

However while these factors suggest that electric vehicles will gain substantial market share it does not say how much how soon[2].  So how fast might the market for plug-in vehicles grow if policy drivers are strong and matched by favourable economics?  Here I consider how quickly electric vehicles could gain market share on that sort of optimistic view.

Market share gains for new technologies

The transition to electric vehicles is in its early stages, so extrapolating historical trends offers only limited guidance.  Similarly, highly detailed modelling may not offer robust insights, because too many assumptions are required.  Instead it seems appropriate to look at some broad indicators.

A good starting point is to look at adoption other new technologies.  The chart below shows the rates of penetration of new technologies in the USA over the 20th and early 21st centuries.  It shows variants on a characteristic s-curve shape, with most technologies reaching eventual penetrations of 80-100%.  The typical time to reach about 80% penetration following the first 1% or so of deployment (about where plug-in vehicles are now) is around 20-30 years, although some modern highly scalable technologies have become nearly ubiquitous faster than this, and other technologies have taken as long as fifty years or so to reach high penetration.

For example, cars themselves experienced rapid growth between around 1910 and 1930, reaching 60% of households, before experiencing hiatus and decline during the Great Depression and Second
World War, before growing steadily again through the to the second half of the 20th Century.

However these timings are for the USA, and, even in increasingly homogenous, world global adoption may take a little longer.

Chart:  Transitions of major technologies

 new technology chart

The chart mainly shows technologies that fulfil a new function, rather than those that replace existing technologies, as plug-in vehicles do.  However replacement technologies can also gain market share quickly.  Digital cameras replacing film almost completely over a period of around 15-20 years, and DVDs replaced VHS in less than 10 years.  In these cases the new technology brought clear advantages.  For plug in vehicles a combination of some advantages plus regulatory drivers could play a similar role.

Modelling the transition

EVs are rather different from many of these cases in that there is a large existing capital stock which is expensive to replace – a new car is much more costly than a new camera.  This limits the rate of change of the stock.   I have therefore applied the sorts of timescales shown above to a rough and ready model representing the potential rate of gain market share of new vehicles, rather than changes to the stock.  The model uses a standard s-curve (logistic function).  Changes in the stock are then calculated considering stock turnover.

I have developed three scenarios representing respectively strong policy drivers, more moderate policy drivers, and a delayed transition representing either weaker policy or greater practical or economic obstacles.  The strong policy case fits better with the historic data, but this may not be a reliable marker as the history is so short and there are a number of particular circumstances at work.

I have assumed plug-in vehicles will eventually account for 80%-90% of the market for light vehicles, with markets for internal combustion vehicles likely to remain where consumers seek low capital costs or they need long range with poor infrastructure.  There will doubtless also be small niches for car enthusiasts seeking experience of the internal combustion engine, just as there are for taking photographs on film.  However these are likely to play only a small role.

The rate at which the stock of vehicles is replaced depends on how long vehicles last.  I have assumed this to be 15 years, although there is obviously a distribution around this.  If this were to lengthen further it would slow the change in the stock, or could be shortened by incentives to scrap older vehicles.  The life of new electric vehicles is unproven (although battery guarantees of typically around 8 years are available), but in any case I have assumed buyers replace their battery packs, or replace their EVs with other EVs rather than returning to internal combustion engines.

Growth of the vehicle fleet leads to a faster proportional changeover of the stock, assuming plug in vehicles gain the same share of the larger market, because current sales are a greater proportion of the historic stock.  I’ve here assumed a 2.5% p.a. global growth rate for car sales[3].

The results of this analysis are shown in the chart.  Annual sales of EVs reach 20-60% of the market by 2030, expected to be over 100 million vehicles p.a. by then.  They by then account for around 7-22% of the vehicle stock, or around 100-330 million vehicles.  By 2050 electric vehicles account for a majority of light vehicles on the roads in all the scenarios.

Global market share of plug in light vehicles

EV growth chart

So do  these projections make sense, and what might stop them?

Cost competitiveness.  Analysis by a variety of commentators show EVs becoming economically competitive in the early to mid-2020s, varying between geographies depending on factors such as driving patterns and petrol prices.  This timing corresponds with the period when vehicles begin to gain market share much more rapidly in the above model, especially in the first two cases, which appears consistent.

China.  A large proportion of vehicle sales in the coming years will be in developing countries, especially China.  Concerns around urban air quality, development of the indigenous automotive industry, infrastructure development, and oil imports look likely to tend to favour EVs in China.  Driving patterns based around lots of shorter distance urban driving are also compatible with EVs.  For these reasons government policy in China strongly favours EVs.  Again this seems consistent.

Growth rate.  The compound annual growth rate for annual sales over the period to 2030 ranges from 25% to 33%, both well below current growth rates of around 60% p.a.

Scale-up.  The need to produce tens of millions of additional EVs by 2030 is a formidable challenge.  However the international car industry increased production by about 35 million units p.a. over the two decades between the 1990s and 2015, and added 20 million units p.a. in the last decade alone[4].  Replacing models with electric equivalents at this scale does not seem like an insuperable barrier, at least in the lower two scenarios.  However the challenges of achieving this for the stronger policy scenario are formidable, and policy drivers would need to be correspondingly strong to overcome these barriers.

Battery production would also need to be scaled up by orders of magnitude.  There don’t appear to be any fundamental barriers to supply of the vast quantities of lithium that would be needed, but it may take time to develop the infrastructure.

The need to ramp up production of both new models and batteries may act to slow growth, at least for a while and especially in the strong policy case, but do not seem likely to act as a fundamental longer term constraint.

Grid constraints.  EVs are likely to require reinforcement of grids, but again this does not look like a major barrier given the timescales involved.

Other projections

These projections show much faster growth than analysis by BNEF, which suggests 35% market share by 2045[5].  However the reasons that BNEF sees growth being so restricted are unclear, and there appear to be few examples where the penetration of a new technology has been so slow.  It seems a more likely estimate for a share of the stock by that date, though even then looks to be towards the low end of the range.

Goldman Sachs estimates 22% of the market being EVs by 2025[6].  This includes conventional hybrids, with the share of plug-in vehicles being only about a third of this, closer to the moderate case.  However it would not seem to require a fundamental change to the market’s development for a greater share of hybrids to be plug-in, so Goldman’s analysis seems at least potentially consistent with the strong regulation case shown here.

Other scenarios show something close to the moderate case shown here.  The IEA 450 scenario and Statoil’s reform scenario both show EV sales reaching around 30% of the market by 2030[7].

Outturn will doubtless differ from these projections.  But they do highlight the extent to which policy might succeed in stimulating a major transition in car markets in the next two or three decades.

Adam Whitmore – 24th May 2016


[1] All estimates here refer to pure electric vehicles and plug in hybrids, which get much or all of their energy from externally generated electricity.  Depending on driving patterns, a PHEV may typically get 70% of its energy from external electricity.  I exclude conventional hybrids, which are essentially a variant of internal combustion engines with increased efficiency, in that still get all their energy from petrol.


[2] Some have made  the case that on pure resource cost grounds internal combustion engine vehicles will continue to predominate.  See  This is potentially true in the absence of any policy drivers due to emissions or other factors, but this seems unrealistic.

[3] For comparison, BP assume a doubling of the vehicle fleet by 2035, about a 3.5% p.a. growth rate (see there 2035 outlook).




[7] See Lost in transition, Carbon tracker p. 102 for plots of these projections

The constrained role of biomass

The role of biomass in the world energy system looks likely to be constrained, so there will be a need to focus on high value applications where there are few low-carbon alternatives.

This is the second of two posts looking at the role of biomass.  Here I focus on potential resource constraints.

A wide range of possibilities

The amount of biomass available to provide energy depends a lot on the amount of land available to grow energy crops, and how much that land can yield.   Different assumptions on these variables produce quite different estimates of the total resource, and numerous studies over the years have produced a wide range of results.    The amount of waste biomass available also matters, but potential availability from this source is smaller.

A comprehensive review of estimates of the biomass resource was carried out two years ago by researchers at Imperial College[i] (see chart).  It showed a variation in estimates of a factor of around 40, from of the order of 30 EJ to over 1000 EJ (1EJ =1018 J, or a billion GJ, or 278 TWh).  This compares with total world primary energy demand of just under 600 EJ, transport demand of around 100 EJ, and at least 250 EJ to produce present levels of electricity, assuming biomass combustion to remain relatively inefficient[ii].

Estimates of available biomass resource

biomass chart processed

Source: Slade et. al. (2014)

The authors examine reasons for differences in estimates, which I’ve summarised in the table below.  The differences are largely assumption driven, because the small scale of commercial bioenergy at present provides little empirical evidence about the potential for very large scale bioenergy, and future developments in food demand and other factors are inevitably uncertain.

Reasons for variation in estimates of total biomass supply

Range Typical assumptions
Up to 100EJ Limited land available for energy crops, high demand for food, limited productivity gains in food production, and existing trends for meat consumption.  Some degraded or abandoned land is available.
100-300 EJ Increasing crop yields keep pace with population growth and food demand, some good quality agricultural land is made available for energy crop production, along with 100-500Mha of grassland, marginal, degraded and deforested land
300-600EJ Optimistic assumptions on energy crop availability, agricultural productivity outpaces demand, and vegetarian diet
600 EJ + Regarded as extreme scenarios to test limits of theoretical availability


Reasons for caution

In practice there seem to me to be grounds for caution about the scale of the available resource, although all of these propositions require testing, including through implementation of early projects.

Land Availability

  • There will rightly be emphasis on protection of primary forest on both carbon management and biodiversity grounds, with some reforestation and rewilding.
  • There is little evidence of a shift away from meat consumption. With the exception of India, less than 10% of people in  most countries are vegetarian despite many years of campaigning on various grounds[iii].   In China meat consumption is associated with rising living standards.
  • Demand for land for solar PV will be significant, although a good deal of this will be on rooftops and in deserts


  • The nitrogen cycle is already beyond its limit, constraining the role of fertiliser, and water stress is a serious issue in many places (agriculture accounts for 70% of current fresh water use). The UN Food and Agriculture Organisation has projected fairly modest increases in future yields.

Policy support

  • Difficulties in limiting lifecycle emissions from biofuels are likely to lead to caution about widespread deployment.
  • Concerns about food security may limit growth of biofuels.

Small scale to date, despite many years of interest

  • There has been little progress to date compared with other low carbon technologies. Though traditional biofuels remain widely used, modern biofuels account for a very small proportion of demand at present.  World biofuels consumption currently accounts for only 0.2% of world oil consumption[iv] .  Many biofuels programmes have had subsidies cut and there is still limited private sector investment.

In this context some estimates of the potential for biomass to contribute to energy supply seem optimistic.  For example, Shell’s long-term scenarios (Oceans and Mountains) show biomass of 74 EJ and 87 EJ respectively for commercial biomass, 97-133 EJ including traditional biomass by 2060[v].  These totals are towards or above the more cautious estimates for the resource that might ultimately be available (see table above).  A recent review article[vi]  suggested that by 2100 up to 3.3 GtCp.a. (around 12 billion tonnes of CO2) could be being removed, and producing around 170EJ of energy.  However the land requirements for this are very large at about 10% of current agricultural land.  The authors suggest instead a mean value for biomass potential of about a third of that, or 60EJ.

On balance it seems that biomass is likely to account for at most less than 10% of commercial global energy (likely to be around 800-900EJ by mid-century), and potentially much less if land availability and difficulties with lifecycle emissions prove intractable.

It thus seems likely that biomass energy will be relatively scarce, and so potentially of high value.  This in turn suggests it is likely to be mainly used in applications where other low carbon alternatives are unavailable.  These are not likely to be the same everywhere, but they are likely often to include transport applications, especially aviation and likely heavy trucking, and perhaps to meet seasonal heat demand in northern latitudes.  For example, according to Shell’s scenarios aviation (passengers + freight) is expected to account for perhaps 20-25EJ by 2050, and biomass could likely make a useful contribution to decarbonisation in this sector.

None of this implies that biomass is unimportant, or has no role to play.  It does imply that policies focussing on deploying other renewable energy sources at large scale, including production of low carbon electricity for transport, will be essential to meeting decarbonisation targets.  And the optimum use of biomass will require careful monitoring and management.

Adam Whitmore  – 11th April 2016


[i] Slade, Global Bioenergy Resources, Nature Climate Change February 2014

[ii] Data on final consumption and electricity production from Shell and IEA data.  35% efficiency for biomass in electricity is assumed, which is likely to be somewhat optimistic, especially if CCS is employed.


[iv] BP Statistical Review of World Energy

[v]  These totals include biofuels, gasified biomass and biomass waste solids, and traditional biomass.

[vi] Smith et. al., Biophysical and economic limits to negative CO2 emissions, Nature Climate Change, January 2016.  The paper estimates land requirement for 170 EJ of 380-700 Mha, around 10% of total agricultural land area in 2000 of 4960Mha.