Optimal liability apportionment in programmatic credit-based emissions trading

Abstract

The inclusion of Programmes of Activities (PoAs) within the Clean Development Mechanism (CDM) has been limited by the fact that third-party project validators, who determine the eligibility of a CDM Project Activity (CPA), are currently held liable for any certificates that are erroneously issued. As such, validators must replace any credits issued for the relevant CPA. Moreover, the risk associated with the validation of small-scale CPAs is considerably higher than that associated with traditional CDM projects. Using a simple game-theory model to model the interactions between project validators and coordinators, it is shown that shifting liability for certificates that are erroneously included – from the former to the latter – is never optimal, does not provide a strong enough incentive to enforce first-best levels of due care in CPA selection and inclusion, and can induce overprovision in validation efforts. The main problem with such a simple proportionate liability regime is that an increase in incentives for one player automatically leads to a decrease in incentives for the other. Two additional instruments are also considered that would both rectify this problem and improve the environmental integrity of the CDM mechanism.

Résumé

L'inclusion de programmes d'activités (PoAs) au sein du mécanisme de développement propre (MDP) a été limité par le fait que les validateurs de projet tiers, qui déterminent l'éligibilité d'une activité de projet MDP, sont actuellement tenus responsables pour l'émission de tout certificat erronné. De ce fait, les validateurs sont dans l'obligation de remplacer tous crédits issus pour l'activité en question. De plus, le risque associé à la validation d'activités de projet de petite échelle est considérablement plus élevé que celui étant associé aux projets MDP traditionnels. A l'aide d'un modèle simple de théorie des jeux pour modéliser les interactions entre validateurs et coordinateurs de projets, il est démontré que le transfert de la responsabilité pour ces certificats inclus par erreur – du validateur au coordinateur – n'est jamais optimal, ne fournit pas d'incitations assez fortes pour renforcer les niveaux suffisamment élevés de diligence dans la sélection et inclusion des activités de projet, et peut engendrer un approvisionnement excessif de services de validation. Le problème principal avec un régime d'apportionnement de responsabilité si simple est qu'une hausse des avantages pour une partie entraînerait automatiquement une baisse d'avantages pour l'autre partie. Deux instruments supplémentaires sont également pris en compte qui rectifieraient ce problème et amélioreraient l'intégrité environnementale du MDP.

Keywords:

• Clean Development Mechanism
• credit-based mechanisms
• programmatic CDM
• validation
• verification

Mots clés:

• Mécanisme de développement propre
• mécanismes de crédit
• MDP programmatique
• validation
• vérification

Notes

The underlying incentives within the seller–buyer relationship differ significantly from those in the project coordinator–validator context, and thus the insights from the aforementioned research cannot be applied to the present case of PoAs and CPAs. Similarly, insights from the literature on financial auditing, which is a matter of ex post control, are only of limited use. The inclusion of CPAs within the CDM PoAs is made before any certificates, and hence revenues, are generated.

See, for example, the first registered PoA, ‘CUIDEMOS Mexico’, available at http://cdm.unfccc.int/ProgrammeOfActivities/registered.html.

In its 55th session, the CDM Executive Board decided to limit the timeframe within which the inclusion of a CPA can be questioned: ‘within 1 year after the inclusion of the CPA into a registered PoA or renewal of the crediting period of the CPA, or within six (6) months after the first issuance of certified emission reductions (CERs) for that CPA, whichever the latter’ (UNFCCC, 2010, §26). Hence, while the liability burden in the first crediting period is limited to the replacement of certificates issued for the first year, this burden can increase considerably with each renewal of the crediting period.

The rules regulating the programmatic and standard CDM are laid out in UNFCCC (2010) and UNFCCC (2002), respectively.

Note that the exposition of the model is formulated for the inclusion of a single CPA into a PoA. However, the model can also be thought of as representing the inclusion of several different CPAs. In this case, variables p, q, and parameter λ (introduced later) represent aggregated probabilities.

Note that the assumption that the players’ choice variables are the respective probabilities represents a shortcut in modelling, which is often used within information economics (e.g. Laffont and Martimort, 2002). In practice, an agent chooses a level of due care e, which influences the probability p(e), with p → 1 for e → ∞. As this relationship is generally concave, it can be assumed, first, that this concavity is also captured in the degree of convexity of the cost function C(p), with C(p) → ∞ for p → 1, and second, that p is chosen directly. The presumption that the probabilities in these functions will never both be equal to one (i.e. p = 1) is based on the plausible assumption that other factors beyond the control of the agent also affect the probability.

Alternatively, t_i could be interpreted as renewed crediting periods instead of years.

With increased learning of the regulator, the probability of discovery might change over time. However, the choices of both players in our context are solely driven by their conjecture of the level of this probability. Hence, λ can also be interpreted as both players’ expectations of the probability of discovery.

Non-additional certificates can crowd out good-quality certificates on the carbon market (Ohndorf, 2010).

In order to entirely focus on the incentive effects, F is assumed to be exogenous. In reality, market participants will only take the audit fee as given if competition on the market for validation is large enough.

It is possible to calculate the level of s implementing the second-best optimum for the above-presented game. However, this result is sensitive to the specifications of cost functions C _v(q) and C _c(p). As the main purpose of this article is to identify incentive effects that hold generally, it is beyond the present scope to present these results here.

Furthermore, the additional sanctions generate funds, as they represent monetary fines. These could be used, for example, to cover the regulator's costs for spot checks or any other measures to increase the environmental integrity of the mechanism.