3.1.1. Time inconsistent optimality

Analyses of the problem of time inconsistency began with the seminal article by Kydland and Prescott (1977) in the arena of macroeconomic policy; they demonstrated the conditions under which policymakers have an ex post incentive to renege on a policy that may have been optimal ex ante. Time inconsistency is now a widely discussed issue in the literature on optimal pollution control, although there is still no unified opinion on its origin and direction. Some authors identify redistribution as the source of the problem (Marsiliani and Renström, 2000; Helm et al., 2003; Baldursson and von der Fehr, 2008), but others emphasize changes in marginal abatement cost curves (Biglaiser et al., 1995; Gersbach and Glazer, 1999). Some even predict higher ex post levels of the optimal carbon price (e.g. Marsiliani and Renström, 2000), while others the opposite (e.g. Abrego and Perroni, 2002).

Consider a benevolent and omniscient government whose sole objective is the maximization of social welfare. This government commits ex ante to tax carbon emissions, on the basis of which firms invest in low-carbon technologies. These investments generate sunk costs on the private sector side, because firms are committed by virtue of their investments in factories, power plants and so on, while the government is only committed by virtue of existing laws (many of which it can change). In pursuit of other goals, the government is tempted to exploit this difference by adjusting the policy ex post, for example by reducing the carbon tax to a level that permits low-carbon energy providers to only cover their variable costs (Blackmon and Zeckhauser, 1992). The appropriated quasi-rent of sunk investments is then redistributed to consumers in the form of lower energy prices. Utilities in general, and the energy sector in particular, are vulnerable to ex post appropriation, because plant value is highly dependent on location and use (Spiller, 1996).5 It is assumed here that the reason for the impartial and benevolent government's policy change consists not in resolved uncertainty or shifts in political preferences (these issues are discussed below) but rather in the maximization of social welfare once investments are made. Either the government re-optimizes its policy ex post (time-inconsistent) or maintains the now sub-optimal ex ante level (time-consistent). Hence, in this context, policy can be either optimal or time-consistent, but not both.

3.1.2. Investment hold-up

In order to achieve an optimal outcome, the government might try to ‘fool’ firms by first promising the ex ante policy before opportunistically adjusting the policy ex post. Of course, this strategy will only work with naïve agents. Rational agents who anticipate the government's incentive to renege might respond in several ways (Spiller, 1996); they can either postpone investment or require a higher rate of return to compensate for higher risk, or they can invest in areas where the payback period is relatively short. Moreover, firms may reduce maintenance expenditures or choose technologies that have a lower degree of specificity in order to reduce their exposure to governmental opportunism. This, however, reduces social welfare, because less specific technologies are less well suited for their task and hence less efficient.

The risk premium firm's demand, in the presence of regulatory risk, reflects the loss of the option value of being able to decide whether the investment should at a later stage be undertaken when additional information is available (Dixit and Pindyck, 1994). The value of that option increases with risk. For example, Blyth et al. (2007) have found that uncertainty regarding policy increases the carbon price required to stimulate investment in carbon capture and storage by 16–37% compared to a situation of perfect policy certainty.

3.1.3. Ratchet effects

Although having a flexible policy allows governments to pursue time-inconsistent strategies against firms, firms can also try to exploit regulatory discretion to their own advantage. So-called ‘ratchet effects’ can occur if firms' current performance is used as a criterion for setting future policy targets (Weitzman, 1980). For example, consider the periodic update of emission caps in the EU Emissions Trading Scheme (EU ETS). Flexible caps provide firms with an incentive to distort investment decisions in order to signal high compliance costs and prepare the ground for a more lenient cap in subsequent trading periods (Harstad and Eskeland, 2010). A firm's chance of winning with this strategy increases with its market power. Moreover, the use of grandfathering will further aggravate dynamic incentive problems if allowance allocation is based on historical emissions (Hepburn, 2006). Regulated entities can seek to delay abatement or even exacerbate carbon lock-in in order to maximize the number of allowances received.

Regardless of whether firms or governments pursue opportunistic strategies, flexible policy carries the risk of distorting the incentives for emissions abatement. The mere potential for strategic interactions between government and firms, with firms having incomplete knowledge about the direction and extent of policy change, generates risks and reduces investments below socially optimal levels.

3.4.1. Risk sharing

The dilemma of carbon policy is that society is forced to choose a balance between the different benefits and costs of commitment. Determining the appropriate degree of commitment is beyond the scope of our analysis (for an attempt in the context of utility regulation, see Blackmon and Zeckhauser, 1992). It should be emphasized, however, that regardless of the commitment level governments enshrine in their policies, the ultimate outcome of their decision is the allocation of risk between public and private actors. Under complete flexibility, the risk of policy change is fully borne by the private sector. Firms react by either postponing investment or increasing their required rate of return (which is not necessarily inefficient from a social perspective). Efficiency depends on who is best able to carry the risk in question. Optimal risk allocation requires that actors accept risk in proportion to their ability to bear it (Hepburn, 2006). IEA (2007) maintains that governments are better placed to underwrite some carbon policy risks for a number of reasons. First, the long-term and incalculable scope of climate change (and therefore carbon pricing policy) impedes insurability, and private agents might not be able to relocate and diversify policy risk. Moreover, if government faces a political incentive to change policy (e.g. pleasing voters by lowering carbon, and hence energy, prices), it can reap the benefits without bearing the costs. The costs of policy change will be more carefully taken into account if governments have a share in them.

3.4.2. The current commitment gap

Although risk sharing seems appropriate on efficiency grounds, carbon policy, as currently practised in most countries lacks credible commitment to long-term objectives. Politicians are quick to sign up to ambitious aspirations in the form of ‘reducing emissions of greenhouse gases in aggregate by 80% or more by 2050’ (G8, 2009, p. 19) but remain reluctant to enshrine necessary incentives in credible policies. The first commitment period of the Kyoto Protocol expires in 2012, with some signatories (e.g. Canada) steering towards explicit non-compliance. What happens after 2012 is still the subject of strenuous climate negotiations. With regard to domestic policies, the US is contemplating the introduction of long-term emissions caps, but prospects remain equally unclear. In Europe, the EU ETS extends until 2020 without specifying legal carbon constraints beyond this.9 In most countries, long-term reduction commitments only exist in the form of political declarations. Policy makers who are interested in triggering the profound economic transition, deemed necessary to limit global warming, might want to entrench their aspirational targets within more credible structures.

4.1.1. Adjustment rule and governance structure

Given the large uncertainties prevalent in carbon policy, the vehicle for commitment should be an adjustment rule: that is, a policy that is itself contingent on other factors. Ulph and Ulph (2009) point out that many authors somewhat artificially generate a commitment problem in their analyses by only allowing policy makers to commit to a fixed carbon price (tax rate) or quantity (emissions cap) that is supposed to remain constant. Instead, Jakob and Brunner (2011) demonstrate why policy makers should formulate a rule – rather than a rate – that sets the price or quantity of emissions as conditional on predefined parameters (e.g. new insights in climate science or new outcomes of international negotiations), thereby limiting the risk of opportunistic discretion and partly reconciling policy flexibility with commitment. Parameters should be chosen so that regulated entities have no influence over their development in order to avoid incentive distortions (Rodrik and Zeckhauser, 1988). For example, including abatement cost as one variable of the adjustment rule carries the risk of distortions if the private sector has influence over the cost itself.

There are two general limitations to built-in responsiveness. First, it is practically impossible to formulate rules that hold under all contingencies, a deficiency well known in contract law. One may therefore allow some freedom to respond to an unforeseen circumstance. Evidently, such force majeure provisions constitute loopholes for opportunistic policy adjustments (Dixit, 1996). Second, even if writing such contracts were possible, the relevant contingencies would need to be observed and assessed before they could be explicitly codified in a rule response. Legislation therefore needs to structure the processes under which adjustment rules are monitored and (in the case of emergency) triggered. Providing for a governance framework that structures these processes in a transparent and accountable way is thus a key element of credible carbon policy.

Of course, the legal quality of a rule matters for its level of credibility. Commitment by means of constitutional law presents a very high hurdle to policy change as parliaments usually need a qualified majority for constitutional amendments. Statutory law typically requires simple majorities. Although there is no case of constitutional carbon constraints, the UK was the first country worldwide to enact a fully comprehensive climate governance framework in statutory law. The UK Climate Change Act of 2008 stipulates ‘legally binding’ interim (2020) and long-term (2050) targets for emission reductions (UK CCA, 2008). It also set up a governance structure by defining the duties and powers of government, parliament and an independent advisory body in monitoring, implementing and (last but not least) updating targets when learning about changes in the relevant parameters (e.g. international climate politics, scientific insights).

Clarity over long-term carbon constraints is essential. Equally important, however, is breaking up long-term commitments into near-term incentives, the impacts of which can be monitored and evaluated in a timely manner. To this end, the UK CCA introduced a carbon budgeting system in which every single budget covers a period of five years and at least three budgets must be set in advance.10 Setting budgets not only takes into account the overarching long-term target but also a wide range of near-term factors such as economic and technological development (Hill, 2009). In addition, the UK government intends to break down accountability into departmental levels where each minister is responsible for meeting the sub-budget for their own economic sector (DECC, 2010). The downside of short-term budgets, however, is that they may hinder the intertemporal equalization of marginal cost. Efficient budgets require that the government knows the development of abatement cost over the entire time horizon.

4.1.2. Enforceability and public scrutiny

It is not obvious what the consequences of self-imposed ‘legally binding’ targets are at first sight. After all, every subsequent legislature will have the authority to change laws, and subsequent governments will be able to change the degree of enforcement. Legislation, however, raises the discursive hurdle for policy change. Targets can no longer be silently dropped when they become inconvenient. Changing laws entails a visible (and perhaps politically costly) process if constituencies are not convinced of the legitimacy of the action.

Governments sometimes choose to ignore legislated commitments instead of changing their legal basis. The ability to internally enforce targets against a non-complying government varies over jurisdictions. For example, consider unbalanced carbon budgets in the UK CCA. In theory, every stakeholder has the right to take the government to court if a carbon budget is not met. In practice, judicial review in this issue area is restricted to procedural misdemeanour, and case law demonstrates that such a challenge is likely to fail (Hill, 2009). The potential for judicial review is limited in the UK, but other legal systems allow a greater role for courts or citizens in ensuring effective enforcement. For example, the US legal system offers more opportunities to sue a non-compliant government. Sina et al. (2009) explore the potential of applying a CCA-like framework in Germany and highlight the power of the Federal Constitutional Court to enforce climate legislation. However, judicial procedures may be too time-consuming to be practical, and the deviating government may be long out of office before the courts sanction a judgement of non-compliance.

Plausibly, the main motivation for government to avoid non-compliance with the law is public scrutiny. If a governing majority anticipates that the political costs of pursuing a certain course of action will be a loss of public support, then taking this route is less attractive. Hence, carbon policy should be deliberately designed to encourage public scrutiny, for example by earmarking revenues from emissions trading for redistribution purposes. A large share of the proceeds from auctioning emission permits could be recycled back to consumers via annual lump-sum payouts. A congressional draft for climate legislation in the US included provisions on earmarking revenues from auctioning (Cantwell and Collins, 2010). By mitigating (for the poorest households potentially reversing) its regressive distributional impact (Burtraw et al., 2009), a ‘climate dividend’ could create long-term political support for carbon pricing. It would restructure the incentives for the ultimate enforcer, namely the electorate, to have the policy implemented over time.

4.2.1. Advisory

An advisory, or watchdog, is a government-independent monitoring entity11 that is delegated the authority to advise and monitor government's performance on a regular basis. The UK CCA institutionalized a regular reporting and monitoring cycle through the government-independent Committee on Climate Change. The Committee advises the UK parliament on carbon budgets and policies and monitors the country's progress along the transition path through annual progress reports. The merit of having an independent watchdog lies in forcing government to publically justify its own actions on a regular basis (Lazarus, 2009). However, the very existence of an organization does not suffice to increase policy credibility. Germany, for example, has three different government-independent advisory bodies for environmental policy, none of which can even closely match the Committee on Climate Change's high public profile in the UK.12 A legal obligation for government to regularly obtain and respond to the organization's advice, as laid down in the UK CCA, and the choice of figureheads with influence in politics and business, are critical to give it an authoritative standing. Time will show whether the political clout of the UK CCA suffices to effectively check political opportunism.

4.2.2. Agency

In contrast to an advisory, where policy control remains with the government, an agency is an institution that is delegated the authority to implement the policy on government's behalf. Two issues play a key role in the rationale for delegation: reputation and objectives. First, Barro and Gordon (1983) investigate how government reputation could mitigate the problem of time inconsistency in monetary policy. However, the transience of leaders and ruling majorities in politics leads to heavy discounting of the future value of reputation. Independent agencies that are insulated from political cycles have stronger incentives to build up and retain their reputation over longer time horizons (Persson and Tabellini, 1994). Second, in the context of commitment, agents can be most valuable when their objectives are distinct from those of the delegating principal (Schelling, 1956). Political opportunism often stems from governments facing multiple and at times conflicting objectives (e.g. low energy prices and climate protection). With delegation, the number of objectives per agency can be reduced. ‘Political unbundling’ (Brunekreeft and McDaniel, 2005) then eases the pressure to renege on past commitments and may help to increase policy credibility in investors' views.

The literature on the benefits of central bank independence inspired Helm et al. (2003) to transfer this device to carbon policy.13 Similar to monetary policy, where the central bank is assigned to meet an inflation target, a carbon agency or ‘carbon central bank’ could be delegated the duty and powers to meet a certain temperature or emission target by, for example, setting a carbon tax. If the initial staff has some conviction in the task, the evolving organizational culture and sense of mission will create a barrier to change (Wilson, 1989). The separation of politics and administration enables the government to commit to the original programme, while the rationale/legal constraints of Weberian bureaucracy insulate agents from efficiency-undermining political pressures (Moe, 1990). The agency retains the flexibility to react to changing circumstances, but it does so within clearly defined boundaries as laid down in its mandate. As such, delegation might be able to combine the advantages of credible commitment and flexibility.

4.2.3. Limits to institutional independence

If one argues that independent agents can solve the commitment problem in carbon policy, one needs to show when and why independence is costly to reverse. The central bank model has demonstrated how institutional independence can create credibility for time frames relevant for inflation targeting (Cukierman et al., 1992). If a government wants to regain influence over monetary policy (e.g. by lowering interest rates for an upcoming election, thereby boosting economic growth), the mere announcement of doing so may raise the level of expected inflation, undermining the political goal of price stability. Before government actually regains control over monetary policy to produce the benefits it seeks, it will pay the costs of its attempt, thereby acting as a deterrent to performing the action in the first place. Time frames in carbon policy, however, are considerably longer than in monetary policy. The inertia in the built energy system might allow governments to abolish carbon agency independence without immediately suffering the costs in the form of significantly higher emission levels.14 Note that an actual reversal of institutional independence is not even necessary. The mere threat to abolish independence suffices to influence agency decisions (McCubbins and Schwartz, 1984). Moreover, as Helm et al. (2003) point out, the trade-off between economic growth and monetary stability exists only in the short run. Carbon policy, in contrast, must balance the objectives of low energy prices and climate protection over longer time spans. To a certain extent, carbon pricing implies an intergenerational redistribution of welfare where the present generation bears costs (higher energy prices) to the benefit of future generations (lower damage cost from climate change). Delegating carbon policy to an independent agency may therefore face substantial resistance from those groups who argue that the task of resolving intergenerational distributional conflicts should remain within democratic institutions.

4.3.1. Contracts and carbon pricing

From an economic perspective, the legal characterization of commitment matters insofar as it impacts incentives. The less reliable a commitment is, the lower are the market participants' incentives to invest in its implications. In this context, it is worth considering feed-in tariffs for renewable energy sources. A major strength of feed-in tariffs is that they provide a legal guarantee of revenues. Lowering the tariff within the guaranteed period faces high legal hurdles, because it undermines contractually agreed payments and directly devalues investors' property.15 Carbon taxes and emission caps, in contrast, can be adjusted freely without violating any contractual agreements. In fact, the power to introduce and abolish taxes at their own discretion is constitutive of parliamentarian sovereignty, and political constitutions shield rather than confine this power. Moreover, taxes seem less suitable to establish credible commitment because, in contrast to tradable permits, they do not directly create a financial market constituency interested in the continuation of the policy.

The introduction of contracts in carbon pricing therefore rests on the use of quantity-based instruments. The conventional assumption among economists is that tradable emission permits represent property rights. Legal experts, however, maintain that permits only grant a limited authorization rather than a private property right per se (Cole, 1999; Woerdman, 2005). Allowance holders do not own disposal space in the atmosphere. Rather, regulators recognize property rights in allowances, and market participants can receive, hold and transfer allowances. The value of allowances of course depends on regulation, and regulators can modify or terminate trading regulations without necessarily infringing property rights. Hence, the establishment of a tradable permit system where participants hold emission ‘rights’ does not provide sufficient protection against opportunism.

4.3.2. Enhanced tradable permit systems

Enhanced tradable permit systems build on current allowance markets but add some additional features. One means to strengthen the commitment to trading systems is to prolong the validity of permits (McKibbin and Wilcoxen, 2007). Long-dated permits allow their owners to emit one unit of emissions each year over the lifespan of the permit (e.g. 30 years). Permit owners have an interest in keeping the system running and the cap tight, because these increase the scarcity value of their asset. They form a countervailing constituency against political attempts by subsequent governments to dilute commitments. To allow for flexibility, a fixed supply of long-dated permits could be combined with a variable supply of annual permits that can be adjusted in response to new information and preferences. However, long-dated permits may be prone to strategic investment hold-up (Biglaiser et al., 1995) as permit owners with market power might be inclined to strategically under-invest in abatement, thereby driving up permit prices and increasing the market value of their assets.

With put-options on emission allowances, investors could hedge their low-carbon assets against the downside of a carbon price risk while also providing government with a contractual incentive to honour its commitment (Laffont and Tirole, 1996; Kemp and Swierzbinski, 2007; Ismer and Neuhoff, 2009). Put-options may be partly seen as an equivalent to feed-in tariffs in terms of incentive- and risk-structure, because they too ensure that part of the costs of learning effects are borne by tax payers. The government, having sold the option, has a financial incentive to adhere to set targets and keep permit prices above the committed minimum. If the volume of outstanding options is sufficiently large, a price floor for emission allowances will emerge in the carbon market. As financial contracts, put-options are protected by constitutional provisions and can be enforced against government through independent courts. Although they represent a powerful commitment device, regulators can freely choose the overall level of aspired commitment via three parameters: (i) strike price, (ii) the number of options issued and (iii) the duration of options (Ismer and Neuhoff, 2009). Increasing any of these strengthens the commitment.

4.3.3. Carbon bonds and contracts

Carbon bonds and carbon contracts are both devices that can be implemented as complements to existing cap-and-trade (C&T) systems. First, consider a government that issues a bond whose coupon payment negatively correlates with the market price of allowances in an ETS (Mainelli and Onstwedder, 2009). Project developers investing in low-carbon technologies could buy this bond in order to hedge against carbon price risk. The carbon bond would transfer part of the risk to government, which then faces a financial incentive to keep prices in the permit market high. The device leverages on the credibility that governments seek on international bond markets, because breaking with this kind of commitment imposes high costs in the form of higher risk premiums for public finance. The degree of commitment can be chosen via pay-off structure, duration or the aggregate volume of outstanding bonds.

Second, the possibility that governments auction off long-term procurement contracts over the supply of emission reductions is complementary to the institutional choice of C&T (Newbery, 2003; Helm and Hepburn, 2007). Firms bid their prices for a specified quantity of emission reductions, and the government closes a contract with the lowest bidder. Tenders can be made technology-blind. Carbon contracts provide firms with a forward revenue stream with long-term price certainty. The government, in contrast, could resell emission credits into C&T systems such as the EU ETS. However, carbon contracts generate emission reduction credits relative to an assumed baseline (an analogue to credits stemming from the Clean Development Mechanism of the Kyoto Protocol). As such, their use is confined to sectors and installations not covered by C&T systems.