Reflecting on the Treasury’s latest carbon tax proposals

The Minister of Finance on Wednesday reiterated in his budget speech that the need for carbon pricing in South Africa is accepted. This should not have surprised anyone – the fact that a carbon tax was on the cards was clearly signalled in last year’s budget and a discussion paper dealing with carbon taxes in South Africa was released in December 2010. What was more surprising, however, was that the implementation date for a carbon tax, its proposed design features, and its suggested levels were included in the Budget Review. The 2011 Budget Review did mention that the design features and implementation schedule for a carbon tax would be announced in the 2012 Budget, but this was scheduled to follow the publication of an updated carbon tax policy paper for public comment in November 2011. The revised carbon tax discussion paper will now only be published in 2012. Until the Carbon Tax Policy Document is released uncertainty remains on how exactly the proposed carbon tax design features mentioned in the 2012 Budget Review will be implemented. This blog post aims to provide some initial reflections on the latest proposed carbon tax design features and highlight some of the uncertainties that remain.

The proposed carbon tax will be implemented in a number of phases. The first phase will last from 2013/2014 to 2019/2020 and the second from 2020 to 2025. The carbon tax will be levied on actual carbon dioxide equivalent (CO2e) emissions calculated using ‘agreed’ methods. Currently there is still a lot of uncertainty about the preferred methodology to measure direct CO2e emissions for a number of processes, and in recent years, switching between methodologies have seen large swings in the total emissions reported by firms participating in the Carbon Disclosure Project. There are also still significant methodological issues and data gaps regarding the measurement of emissions from difficult to measures sources of emissions that have been included in the proposed carbon tax framework, like waste, land use and fugitive emissions from coal. It will thus be interesting to see what processes are put forward in the forthcoming carbon tax discussion document to reach a consensus around these measurement issues.

A percentage-based emissions ‘threshold’ will be set below which the tax will not be payable. In the case of an emissions threshold (whether defined on a relative/intensity or absolute basis) [1], emissions below a threshold would have a marginal and average tax rate of zero percent. Once the threshold is breached, the marginal cost of each unit of CO2e will be 100% of the carbon tax rate. In the proposed South African carbon tax design, both the marginal and average cost of each unit of CO2e will be 60% of the carbon tax rate. The Treasury is thus proposing tax relief for 60% of a firm’s basic carbon tax liability in the first phase of implementation, irrespective of what its actual level of emissions is. The level of support provided in the second phase will be reduced, and the percentage-based threshold may be replaced with an absolute tax-free emissions threshold.

The main concern with a move to a true threshold approach (whether it is a benchmark/intensity threshold or an absolute emissions threshold) is that it effectively undermines the carbon pricing mechanism. The reason for this is that a price (or at least an opportunity cost) [2] is not placed on every unit of CO2e emitted. This runs against one of the primary objectives of using carbon pricing to mitigate climate change, namely the creation of a broad-based carbon price that ensures that prices accurately reflect the internalised carbon externality across a broad range of sectors, fuels and processes/applications.  Keeping the carbon price signal intact throughout the economy avoids the unintended consequences which may arise when individual producers and consumers are faced with differential carbon costs. A threshold approach, in practice, can thus more accurately be described as a regulatory (command-and-control) instrument where firms are forced to face a penalty if they do not meet a minimum production standard.

The tax-free emissions threshold will be adjusted based on how carbon-efficient a firm is. The carbon-efficiency of a firm will either be measured relative to an industry benchmark or a base year. The rationale for including a mechanism that adjusts the basic tax-free emissions percentage is to encourage firms to reduce their carbon intensity of production. The implicit assumption here seems to be that carbon prices initially will not be high enough to encourage behaviour change, and that the additional leveraging effect obtained by adding an efficiency ‘kicker’ to the basic tax-free allowance may incentivise firms to reduce their emissions faster. While this may be the case for firms with easy and cheap mitigation options open to them, the efficiency adjustment could significantly reduce the relief provided to firms that are relatively emissions-intensive with few mitigation options. It is presumably these firms that require more time to transition to lower-emissions production processes that the basic tax-free allowance was designed to assist.

So the carbon-efficiency adjustment (in the first phase of the carbon tax in particular) could work against the objective of the general relief provided in the carbon tax design. Also, it could create a level of uncertainty regarding the actual carbon price that firms will face in future. Since production processes are typically configured to operate most efficiently at full capacity, changes in output can significantly impact on the carbon-intensity of a given process. What’s more, since reductions in output tend to increase carbon-intensity, the carbon-efficiency adjustment could lend a procyclical dimension to the carbon tax. In times of depressed demand firms may find that their carbon tax liabilities (as a percentage of costs) increase at the same time as their revenues decline.

Negotiating industry benchmarks (or efficiency metrics in base years) in sectors where multiple products are produced, or where a number of different production processes are used to produce only marginally differentiated products, is likely to be a challenge. A mechanism for dealing with new entrants would also need to be created. It will be interesting to see how these issues will be addressed in the carbon tax policy paper.

A distinction is made between energy-related and industrial process emissions within the proposed carbon tax design, with a higher tax-free allowance (10 percent) being applicable to sectors that will find it difficult to reduce their process emissions in the short term. Given this distinction, it is not clear whether one benchmark or emissions metric in a base year (i.e. emissions per unit of output) will be set for an industry when the efficiency adjustments to the  basic carbon tax-free allowance mentioned in the previous paragraph are calculated, or whether separate process and energy emissions benchmarks will be set. From the information provided in the Budget Review, it would seem that the maximum additional allowance for process emissions will be calculated as a percentage of total emissions. Given that process emissions are seen as distinct from energy emissions, it is difficult to see the rationale for using total emissions as the denominator when defining relief. The typical proportion of total emissions made up by process emissions varies widely for the sectors singled out as having significant process emissions, so it will be interesting to interrogate the rationale behind providing a standard level of relief (as a percentage of total emissions). It is also interesting to note that one of the stock examples of an industry with significant process emissions that are difficult to mitigate, namely the coal-to-liquid sector, has not been afforded an additional allowance for process emissions.

Additional relief is suggested for firms that are considered trade-exposed in the form of an additional 10 percent tax-free allowance. How ‘trade-exposed’ is defined, and whether emissions-intensity enters into the formula, will become evident when the carbon tax policy paper is released. Based on international experience, the proposed additional support provided to trade-exposed firms relative to non-trade exposed firms seems relatively low. Also, it is not immediately clear how relief will be provided to firms that are vulnerable to competitiveness issues due to carbon pricing increasing the cost of grid-electricity. In the short term the carbon-efficiency adjustment to the much larger basic tax-free allowance may also negate the additional support provided to trade-exposed firms.

Companies will be allowed to use offsets to reduce a portion of their carbon tax liabilities. Firms with a total potential tax-free allowance of 80 percent will be allowed to offset up to 5 percent of their emissions, while firms with a total potential tax free allowance of less than 80 percent will be allowed to offset up to 10 percent of their emissions. It will be interesting to see what projects and/or mechanisms qualify as allowed offsets. Recognising offsets does allow a way that future local trading mechanisms can be integrated with the carbon tax. The use of offsets in the electricity sector in particular is interesting. Given the important role that the electricity sector has in facilitating the transition to a low-carbon South African economy, and that the need to effectively manage the transition to a low-carbon South African economy is explicitly mentioned in the Budget Review, it is not clear what the rationale for allowing offsets in this sector is. The use of offsets may divert resources and expertise away from energy-related mitigation activities in this sector.

An initial carbon tax at R120 per tonne of CO2e emitted that increases with 10 percent per annum during the first phase of operation is proposed. Given the basic 60 percent tax-free allowance, this equal to an initial escalating carbon price of R48 per tonne of CO2e before the efficiency adjustment is taken into consideration. Given the number of additional allowances, this initial carbon tax could be as low as R24 per tonne of CO2e in some sectors (excluding activities that will receive 100% tax-free allowances). The use of offsets will reduce the effective carbon price further. Offset projects or mechanisms will involve costs, so it is not clear by how much individual firms’ carbon tax liabilities will be reduced. 

The use of a carbon-efficiency adjustment to the basic tax free allowance further weakens the link between the published carbon tax rates and the effective carbon tax liabilities of individual firms. The main benefit of a carbon tax relative to other carbon pricing mechanism, namely a clear, credible and transparent carbon price that creates certainty for planning purposes, is thus unlikely to be realised in the proposed carbon tax design. Apart from practical issues like how volatile and uncertain carbon prices are dealt with in the pricing decisions of firms or the setting of regulated prices in the electricity or liquid fuels sectors, the tenuous link between official carbon prices and the actual carbon liabilities of firms creates opportunities for firms to generate windfall profits in relatively concentrated markets. Furthermore, based on the relative carbon efficiency of firms, the significance of their process emissions, whether or not they are trade exposed, and the cost of offset options available to them, the same activity (say the combustion of coal in boilers or even for onsite electricity generation – depending on whether the tax is levied on an activity or sector basis) may be subject to very different carbon prices. This could distort firm’s energy choices, and may lead to unintended consequences as firms try to minimise their overall energy costs (which will now include their carbon tax liability). Without accounting for the efficiency adjustment to the basic tax-free allowance and offsets, the carbon price Eskom will face when emitting one tonne of CO2e from the combusting of coal will be R48. This is a third more than the carbon price faced by firms that are trade exposed (R36 per tonne) and double that of firms that are trade exposed and have significant process emissions (R24 per tonne – assuming that the additional 10 percent tax-free allowance is applicable to a firm’s total emissions as the Budget Review seems to indicate). Depending on how the carbon tax costs of Eskom and IPPs are passed on into regulated electricity prices, this may influence firms’ choices between using grid electricity and other energy sources.

Finally, in holding with the Treasury’s earlier policy stance, revenues from the tax will not be explicitly earmarked, but soft (on-budget) earmarking is implied by stating that “consideration will be given to spending to address environmental concerns”. In keeping with good mitigation policy practice, supporting policies like the proposed energy-efficiency tax incentive and measures to assist low-income households to deal with possible cost-of-living increases, will be implemented in addition to the carbon tax.

The Budget Review mentions the need to align a carbon tax with the carbon budget approach adopted in the National Climate Change Response White paper. Given the differences in carbon prices between sectors, and the possibility of unintended consequences as firms’ energy choices are skewed, this may be difficult. Some mechanism for adjusting the effective carbon price faced by sectors when they risk exceeding their carbon budgets would probably need to be included in the carbon tax. An absolute tax-free threshold would easily integrate with a carbon budget approach. But as mentioned earlier, a carbon tax that incorporates an absolute tax-free threshold should be viewed as a regulatory rather than an economic instrument, and as such is unlikely to lead to least-cost abatement. From a timing perspective, it is unfortunate that the carbon tax is scheduled to be implemented at around the same time as the National Climate Change Response White Paper indicates carbon budgets should be in place. A carbon tax preceding the implementation of carbon budgets would have generated useful information that could have simplified the carbon budgeting process [3]. It is not clear which institution or agency will be responsible for negotiating the “agreed” industry benchmarks or base year emissions efficiency metrics. This information will also be central to the development of a carbon budget, so it would make sense that the same institution/agency/unit that will handles the carbon budget process on behalf of the DEA also deals with this aspect of the carbon tax administration.

The Carbon Tax Discussion Document released in 2010 showed a strong preference for keeping a carbon tax as simple as possible. The proposal of a broad-based low and escalating carbon tax was considered theoretically sound, but many commentators lamented the fact that local conditions were not sufficiently considered when the skeleton carbon tax design was created [4].  The proposed carbon tax design included in the 2012 Budget Review shows a significant reversal in the Treasury’s thinking. What has been proposed is, based on international experience, a relatively complicated carbon tax design. The Treasury has attempted to address a number of important issues in this latest proposed design, and has obviously taken the concerns of the private sector and other stakeholders to heart. While the Treasury should be lauded for taking an innovative approach to the design of a carbon tax, the current proposed design may err on the opposite end of the complexity scale from the skeleton design put forward in 2010. While most of the individual components of the carbon tax design make sense, combined as the overall carbon tax design package they may be too unwieldy to create a broad, stable and transparent carbon price. The debate around the optimal design of a South African carbon tax is expected to continue.

End notes

[1] If the threshold is defined on a relative or intensity basis, a tax-free level of emissions is determined per unit of output. Firms thus effectively face a threshold above which they pay a carbon tax on every additional unit of CO2e emitted, and this threshold is directly related to their level of production. In the case of an absolute threshold, the threshold above which emissions are priced is independent from the level of production. Firms are thus exempted from paying a carbon tax if they meet a predetermined absolute CO2e emissions path, irrespective of how much they produce.

[2] Theoretically a threshold approach is closely related to a “grandfathered” emissions trading scheme in which a set number of emissions permits is initially allocated free to participants based on their historical emissions. If firms’ emissions exceed that allowed by the free emissions permits they have received, they need to purchase additional permits or incurring penalties for breaching their emissions caps. The difference, and the reason why the use of a threshold of free emissions within a grandfathered ETS is a form of carbon pricing, is that emission allowances are allocated to cover all the permitted emissions of the firms in the scheme, and the ability to sell grandfathered emissions certificates creates an opportunity cost for every unit of CO2e. Even under a grandfathered ETS every unit of emissions (represented by an emissions certificate) thus has a real value since the options exists to reduce production and sell the certificates rather than offset them against actual emissions

[3] For more details on the information required to construct carbon budgets, and a possible approach to generating this information, please see a recent WWF report that DNA Economics contributed to available at:

[4] See for instance presentations by stakeholders at the National Treasury Carbon Tax Workshop in March 2011 available from: