Incentive Regulation With Bounded Regulators 15 June2012
Download as PPT, PDF0 likes287 views
The document discusses the application of incentive regulation in the power sector, highlighting the limitations faced by real regulators compared to the assumptions of theoretical models. It proposes an analytical framework for aligning regulatory tools with the regulator's abilities and the tasks assigned to network operators, emphasizing the significance of controllability, predictability, and observability in choosing appropriate regulatory tools. The conclusion stresses the need for a practical approach that acknowledges the finite resources and capabilities of regulators while tailoring distinct tools for different regulatory tasks.
Incentive Regulation With Bounded Regulators 15 June2012
- 1. 15th June 2012 – EUI, Florence Incentive regulation with bounded regulators Jean-Michel Glachant, Haikel Khalfallah (FSR-EUI), Yannick Perez (LdP chair-EUI) Vincent Rious, Marcelo Saguan (Microeconomix, LdP chair-EUI) 1st Annual Conference on the Regulation of Infrastructure Industries 1
- 2. Do we really know how to apply incentive regulation in the power sector? The assumptions of the textbook model of regulation The reality for regulator –e.g. considering the national regulatory agencies for the power sectors but the rationale is also applicable to other sectors The regulator always has the She does not always have as many required powers, resources and powers, resources and abilities as abilities to implement any regulatory the textbook model assumes scheme The regulator applies distinct The regulator incentivises a TSO as regulatory tools to different TSO’s a whole with a single tool tasks 2
- 3. An analytical framework to choose in practice between the incentive regulation tools How to align the regulatory tools, the regulator’s endowment and the targeted tasks (“costs”)? The textbook model of incentive regulation proposes no solution to choose the regulatory tools considering – The regulator’s abilities to implement it – And the targeted network tasks (“costs”) and their characteristics We propose a way to align regulatory tools, endowment and tasks in practice, considering and combining – The actual bounded regulators’ endowment and abilities – And the actual characteristics of the network operator’s tasks 3
- 4. The real regulators are endowed with less abilities than the textbook assumes In the economic literature proposing and building regulatory tools, regulator is always thought to have all the desired cognitive, computational and judicial abilities to use any tool easily and efficiently – In particular, she knows ex nihilo how to choose the most efficient regulatory tools and she has all the desirable abilities to implement it But in reality, the regulators were endowed with tight resources (budget, staff, skills and judicial powers) which are likely to hamper their abilities to do their job “perfectly” Furthermore, regulators learn from experience how to use the different regulatory tools identified by our academic theory. How to: – To reduce their information asymmetry – To adapt tools to uncertainty and risk – To gain computational skills needed to design the regulatory tools 4
- 5. The regulatory tools require minimum abilities to be usefully implemented Regulator’s abilities Cost + Price cap PBR Menu Yardstick 5
- 6. The regulator regulates the network operator for various tasks and not for a single task The textbook regulator controls the TSO’s tasks (cost) as a whole while there are different tasks with different characteristics. The basis tasks for an El. TSO: – Operation of electricity system: Balancing + Reserves + Congestion + Losses + Market Operation – Maintenance of the existing grid – Investment and grid connection: Planning + Construction – Customer relationship The network operator may have to undertake new or renewed tasks because of new regulatory objectives from – The integration of massive renewables in electricity – Concerns about security of supply – Europeanization of markets with a key TSO role in market building + RD&D in infrastructures and services (“smart” everything) 6
- 7. Regulating a task (a cost) is betting on its controllability, predictability, & observability to choose appropriate regulatory tool Considering the diversity of tasks, systems and environments that TSOs may encounter, they should be targeted with distinct regulatory tools in a building block approach Other things being equal, that is to say with a regulator having all the desired abilities to use any tool, the appropriate regulatory tool to choose for a given task/cost should depend on the tasks’ regulatory characteristics being – (Task outcome) Controllability – (Task outcome) Predictability – (Task outcome) Observability 7
- 8. 1° The regulator incentivises the TSO on tasks/costs that the TSO can control Controllability measures the TSO’s ability to control a cost/task or a combination of costs/tasks for a given output Input A NO Output A Controllable? AND/OR internal Input B process Output B Controllable? If the task/cost is not controllable, the regulator should implement a cost plus scheme If the task/cost is controllable, the regulator could incentivise the TSO – Under the constraints relative to predictability and observability 8
- 9. 2° The regulator can only incentivise the TSO on tasks/costs that are predictable Predictability measures the possibility to foresee the influence of external factors on costs/tasks and the relationship between the costs/tasks and the outputs Input A NO Output A Predictable? internal Predictable? Input B process Output B Predictable? If the task/cost and its relationship with the outputs are not enough predictable, the regulator should implement a cost plus scheme Otherwise the regulator can implement an incentive scheme whose risk for her and the network companies depends on the degree of predictability – “Low predictability implies high risk” versus “High predictability implies low risk” 9
- 10. 3° The regulator can only incentivise the TSO on tasks/costs that are observable Observability measures the quantity of available information to the regulator about efficiency gains on tasks, either in terms of tasks themselves, or inputs or outputs Input A NO Output A Observable? internal Observable? Input B process Output B Observable? The regulatory tool should then be chosen depending on the level of observability – When there is no observability, cost plus should be implemented – When input is observable, price cap or a menu of contracts should be implemented – When output is observable, PBR or a menu of contracts should be implemented – When information is available from several network operators, one should benchmark them 10
- 11. A decision tree to align regulatory tools with the tasks’ No characteristics … Controllability? Cost + Yes Price cap No Predictability? PBR ility r vab to bse No u utp Menu Yes ho servab ility Hig t) ob pu t or outpu High (in Observability? Data from several o Yardstick perators 11
- 12. … and alignment with the regulator’s abilities No Regulator’s Controllability? Cost + abilities Yes Price cap No Predictability? PBR ility r vab to bse No u utp Menu Yes ho servab ility Hig t) ob pu t or outpu High (in Observability? Data from several op Yardstick e rators 12
- 13. Examples of regulatory tools on … No Regulator’s Controllability? Cost + abilities Yes Price cap No Predictability? PBR ility r vab to bse No u utp Menu Yes ho servab ility Hig t) ob pu t or outpu High (in Observability? Data from several o Yardstick perators 13
- 14. … example #1 Transmission maintenance Regulator’s Controllability? Cost + abilities Yes Price cap Predictability? PBR ) t put y (ou u alit No le q vab bility Menu Yes s er lity) observa Ob sts or qua inten a nc e c o Hi gh (ma Observability? Data from several o Yardstick perators 14
- 15. … example #2a Transmission losses volume in an isolated system Regulator’s Controllability? Cost + abilities Yes Predictability? PBR t utpu ble o No se rva Ob Menu Yes ility servab High ob Observability? Data from several o Yardstick perators 15
- 16. … example #2b Transmission losses volume in an interconnected system No Regulator’s Controllability? Cost + abilities Predictability? Observability? 16
- 17. … example #3 RD&D e.g. Meshed DC grid Regulator’s Controllability? Cost + abilities Yes Price cap No Predictability? PBR ility r vab to bse No u utp Menu Yes ho servab ility Hig t) ob pu t or outpu High (in Observability? Data from several op Yardstick e rators 17
- 18. Conclusion Textbook regulation assumes an “unlimitedly endowed” regulator targeting a single type of TSO’s task (cost) – Yes: the practical successes of incentive regulation are maximised when the regulator is able to mimick the expected theoretical behaviour However reality is not with unlimited regulatory power or resources – Regulator may have tight resources and only limited abilities – Distinct regulatory tools have to be applied to different targeted costs/tasks Regulatory tools should be aligned with – The regulatory characteristics of the targeted tasks (costs): controllability, predictability and observability – And the regulator’s endowment 18
- 19. 15th June 2012 – EUI, Florence Incentive regulation with bounded regulators Thank you for your attention! Comments and questions are welcome Jean-Michel Glachant, Haikel Khalfallah (FSR-EUI), Yannick Perez (LdP chair-EUI) Vincent Rious, Marcelo Saguan (Microeconomix, LdP chair-EUI) 1st Annual Conference on the Regulation of Infrastructure Industries 19
- 20. Have a look the in at ternat journ ional I am c al hief-e ditor of!!
- 21. Appendixes 21
- 22. A reminder of the 5 standard regulatory tools Cost + – The network operator is then paid based on its cost-of-service Price cap – The network operator has then a maximum allowed tariff level Performance (Output) regulation – The network operator has then an efficiency target and is rewarded or penalised depending on its over- or under- performance Menu of contracts – The regulator proposes different regulatory contracts to the network operator with different degrees of incentives Yardstick or benchmarking techniques – These techniques can only be applied if the regulator controls the cost of several homogeneous network companies – The regulator sets the efficiency target to a network company as a function of its performance relative to the other network companies’ performance 22
- 23. The real regulators are not always endowed with a full range of judicial powers The European example before the 3rd directive 5 An example of 4 or 4½ evaluation of the European regulators’ 3 or 3½ abilities 0 or 2 Ex ante Regulation (=1) Independency from government (= 1 or ½ or 0) TPA setting (=1) Ability to solve conflicts (=1) Ability to acquire information (=1) Source : EU, 2004. 3rd Benchmarking Report on implementation of electricity and gas internal market. N.B.: Information for Germany is up to date and taken from the German regulator’s website 23
- 24. The real regulators are not always endowed with the highest amount of ressources The example of budget and staff in 2009 (for 100 TWh) Legend Sources: Own calculus and •Budget & staff from www.iern.net •Annual load from http://epp.eurostat.ec.europa.eu/portal/page/portal/energy/data/main_tables# •Power Purchase Parity from http://data.worldbank.org/indicator/PA.NUS.PRVT.PP 24
- 25. The regulator might be unable to distinguish between the effect of the network operator’s effort and the effect of uncertainty Effect on Effect on congestion cost congestion cost of the effort by of the effort by the network the network operator NOT operator detectable by detectable by the regulator in - - Without any NO’s effort the regulator in presence of _ With NO’s effort presence of low high uncertainty uncertainty 25
Editor's Notes
- #2: Ladies and gentlemen, here is a piece of work about incentive regulation we realised both at the Florence School and at the Loyola de Palacio chair in collaboration with Microeconomix. We wondered if we all of us really know how to apply incentive regulation in reality while the regulators are far from their supposed theoretical characteristics.
- #3: Reality of regulation is fundamentally different from its theoretical framework. In the textbook model of regulation, the regulator is always assumed to have all the required abilities to do her job efficiently But in reality, a lot of regulators undergo a limitation of their abilities to implement efficiently any of the regulatory tools. Besides in theory, it is generally assumed that the regulator incentivises a TSO as a whole with a single regulatory tool. But in reality the regulator faces a TSO performing multiple tasks and she applies distinct regulatory tools to these different tasks Considering these two major discrepancies between theory and reality of regulation,
- #4: How to help the regulator to choose the most adapted regulatory scheme considering her own limited endowment and the charateristics of the TSO’s costs/tasks she targets? The textbook model of regulation gives no answer to this question, in particular because theory assumes a regulator perfectly able to perform her job. We propose here a way to choose the regulatory tools in practice taking into account the real regulators’ abilities and the characteristics of the targeted network operator’s tasks.
- #5: First of all, why are the real regulators generally so different from their textbook model? The theoretical model of regulator is assumed perfect with all the required cognitive and computational abilities to do her job efficiently, knowing for sure how to choose the good, the most appropriate regulatory tools. But the government and the legislator endow the regulators with tight resources which are likely to limit their abilities to regulate efficiently the network operators. And even the best endowed regulators are learning with experience how to use regulatory tools, that is to say to reduce asymmetry of information, to adapt these tools to uncertainty and to increase their computational abilities.
- #6: That is why each regulator will not easily implement all of the 5 standard regulatory tools. Of course, any regulator is able to implement cost+ regulation. But it is more difficult to set a price cap. It requires notably to compute the efficiency factor. Difficulty is again increasing for the regulator when she wants to apply performance-based regulation (because she must first define the outputs she wants to target). Even if the regulator relies on the participation of third party to define network outputs, this task remains difficult. The two last tools, menu of contracts and yardstick are the most difficult ones to implement because the regulator must have in mind that there exist different types of network companies intrinsically more or less efficient. For yardstick competition, she must also have big computational abilities to treat and compare big sets of database and to estimate the relative efficiency of the different regulated companies.
- #7: Beside her own endowment, the regulator must also have in mind that she does not regulate the TSO’s cost as a whole. The TSO performs different tasks with different characteristics. Classically the TSO operates the system, managing balancing, reserves, internal congestion, losses and market operation. She maintains her network too. She also plans and builds her network to upgrade it and connect new users. And a last classical task she performs is customer relationship management. The TSO may also have to realise new or renewed tasks (these tasks may be renewed) because of new regulatory objectives from the climate change policy with the integration of renewables in the electricity sector, from the concerns about security of supply mainly in the gas sector and from the Europeanization of market building with the role for the TSOs of market architects. All this of course requires a revival of RD&D in both the domains of infrastructures and of services.
- #8: Considering this diversity of tasks and costs, the regulator must use a building block approach implementing distinct regulatory tools on the different network operator’s tasks/costs. And the regulatory tool to choose for a given task is determined by its characteristics of controllability, predictability and observability.
- #9: Controllability measures the TSO’s ability to act on a cost/task or a combination of costs/tasks for a given output Of course, if the cost is not controllable, the regulator should pass it through to consumer implementing a cost plus scheme However if the TSO can control the targeted cost, the regulator could incentivise the TSO under the following constraints of predictability and observability
- #10: Predictability catches the influence of external factors on costs/tasks and their relationship with outputs Of course, with no predictability the regulator should pass the cost through to consumer with a cost plus regulation Otherwise the regulator should implement an incentive scheme whose risk for her and the regulated company is inversely related to the degree of predictability
- #11: The last characteristics of tasks/costs is observability. It measures the quantity of available information to the regulator about efficiency gains on the TSO’s tasks. The regulatory tool should then be chosen depending on the level of observability With no observability of efficiency gains, cost plus should be implemented When efficiency gains are observable on input, price cap alone or in a menu of contracts should be implemented And PBR alone or in a menu of contracts should be implemented when efficiency gains are observable on output At last, when the regulator has information from a sufficiently high number of network operators, she could compare them and implement yardstick competition
- #12: We end up with the following decision tree to choose a regulatory tool With no controllability, no predictability or no observability, cost plus should be implemented Otherwise observability determines the regulatory tools to implement
- #13: Under the constraints of course that the regulator is able to implement it.
- #14: We illustrate briefly the use of this framework combining the regulator abilities and the characteristics of the targeted costs on three TSO’s tasks
- #15: First maintenance, which is well known to be well fitted to incentive regulation. This is because it has all the required characteristics in terms of controllability, predictability and observability. And the tools that the regulator should implement will once again depend on her regulatory experience and her endowment/abilities.
- #16: Then we wonder what regulatory tools could be applied on the losses volume. We first consider the situation of a non interconnected system. Controllability and predictability are then high. And observability will depend on the regulator’s experience in regulating the cost of losses. With little experience, she has few historical data and faces low observability. Otherwise adapted incentive regulation tools could be implemented.
- #17: In an other extreme case, for instance a network company whose network is only used for transit from abroad, she cannot control losses and cost plus should be applied
- #18: A last example is RD&D, for instance supergrid with meshed DC grid. It is obviously the most difficult TSO’s task to deal with for the regulator because predictability and observability depends on the technology maturity. A low maturity of the technology implies a low predictability and a low observability because even the network company is not sure of the possible interaction of the innovation with the rest of the power system and of the outcome and benefits of his research. Predictability and observability will then increase with the technology maturity.
- #19: To conclude The practical successes of incentive regulation were realised when the regulator was endowed with abilities close to the theoretical perfection But reality is not always so perfect and the regulators may have been endowed with limited abilities and tight resources while they should applied distinct regulatory tools to different TSO’s tasks The regulatory tools should then be adequately adapted to the characteristics of the targeted costs (in terms of controllability, predictability and observability) and to the regulator’s endowment
- #20: Ladies and gentlemen, here is a piece of work about incentive regulation we realised both at the Florence School and at the Loyola de Palacio chair in collaboration with Microeconomix. We wondered if we all of us really know how to apply incentive regulation in reality while the regulators are far from their supposed theoretical characteristics.