Regulatory tools for CAPEX review and approval: overview of European approaches

Transcription

1 Regulatory tools for CAPEX review and approval: overview of European approaches Prepared by Dr Srini Parthasarathy Principal, Oxera www. erranet.org

2 Overview what do regulators try to assess: objective and definitions issues with benchmarking CAPEX ways of measuring CAPEX menu of tools: top-down and bottom-up case studies key takeaways 2

3 What regulators try to assess: optimal level of spend to meet outputs and outcomes Exogenous factors L, OPEX Key Y K L outputs/outcomes capital labour M materials (or all other inputs) MM K CAPEX Y input minimisation what is the minimum level of input that the unit could operate at, given the level of output? output maximisation what is the maximum level of output that the unit can produce, given the level of input? 3

4 Regulatory cost model: assessing efficiency and the potential for cost reductions A production function describes the transformation relationship a black box which converts inputs into outputs X = Inputs L = labour K = capital M = materials and supplies C, cost = w.x w = input prices Transformation Factors outside management control Z: country- or region-specific factors (e.g. customer density, topography, regional wages) Outputs Y = total energy delivered, passengers transported, water delivered, track replaced, etc. production frontier output-oriented (OO) technical inefficiency: output loss due to techncial inefficiency inefficient input-oriented (IO) technical inefficiency: the amount by which inputs can be reduced without reducing output 4

5 Regulatory cost model: notions of efficiency focused on cost or expenditure as a function of outputs and outcomes Technical efficiency what quantityof output can be produced, given the inputs? how few inputs are required, given the outputs? Allocative efficiency whether the observed input (i.e. resources) combination is the optimum given the relative prices of inputs Other types of efficiency: dynamic, revenue and profit efficiency dynamic efficiency: what is the potential rate of improvement in economic efficiency over time? Scale efficiency are firms operating at the optimal scale size i.e. are they exploiting economies of scale in production Economic efficiency (technical + allocative) what is the lowest cost of production, given the outputs? generally, the focus of utility regulation 5

6 Illustration of static and dynamic efficiency gains Frontier shift vs catch-up Frontier shift what is the potential for best practice to improve? new technology/management practices target applied to every company in the industry Catch-up how relatively inefficient is the company? by how much of this gap can the company catch up to current best practice in a given timeframe? individual target set for each company Cost Inefficient Efficient Catch-up Frontier shift Frontier shift Overall scope for efficiency Some regulators set a general or a combined productivity target 0 5 Time 6

7 Overview what do regulators try to assess: objective and definitions issues with benchmarking CAPEX ways of measuring CAPEX menu of tools: top-down and bottom-up case studies key takeaways 7

8 What do regulated companies spend their money on? operations costs are generally predictable and well-explained by cost drivers maintenance expenditure on maintaining existing assets; largely predictable renewals expenditure on replacing assets that have reached, or are nearing, the end of their useful lives enhancements expenditure on new assets that expand the capacity or capability of the network ring-fenced funds Most regulated companies are capital-intensive intensive Therefore, capital costs typically account for a large proportion of the cost base 8

9 Forecasting the efficient level of capital investment is difficult enhancements determining the efficient cost of large capital projects is inherently difficult they may be one-offs and company-specific projects take a long time to deliver large inefficiencies can be baked into the high-level specification and/or the detailed design of the project market or technical innovations may not be pursued history is a poor indicator of future costs forecasting uncertainty how much allowance should be made for risk? 9

10 Forecasting the efficient level of capital maintenance could be equally difficult maintenance and renewals capital maintenance and Renewals expenditure is more predictable than enhancement expenditure regulators still face difficulties in assessing the optimal level of capital maintenance and renewals examples of asset failure in most regulated sectors: Railtrack, air traffic control systems failures (Ireland and UK), Northeast blackout 2003 (USA and Canada), Herne Hill floods The costs of failure can be high! Difficulties for regulators the impact on outputs and the probability of failure of changing maintenance expenditure may be unclear (particularly in the short term) companies typically have much better knowledge of asset base/health possible bias towards higher-profile enhancements projects what is the value of reliability and thus the optimal level of asset risk? 10

11 Possible CAPEX bias Separate CAPEX allowance or TOTEX approach? instead of separate treatment of CAPEX and OPEX, regulators may focus on cost assessment and recovery based on total expenditure (TOTEX) key aim is to combat a perceived bias towards capital investment company given freedom to determine whether OPEX or CAPEX solution is more appropriate in the long term also to mitigate issues with accounting treatment of cost items, balancing incentives across cost categories and avoiding setting an inequitable benchmark by assessing costs separately How the TOTEX approach works a proportion (Y%) of TOTEX is added to the RAB and recovered over an assumed asset life (1-Y)% of TOTEX is recovered as it is incurred since Y% is determined ex ante, the company should be indifferent between CAPEX and OPEX solutions also allows for a single efficiency assessment and incentive rate 11

12 Overview what do regulators try to assess: objective and definitions issues with benchmarking CAPEX ways of measuring CAPEX menu of tools: top-down and bottom-up case studies key takeaways 12

13 Measuring CAPEX Cash cost and capital stock approaches Cash cost approach CAPEX Capital stock approach Cash cost approach simple to compute as based on outturn data may not take into account the long-lived nature of assets requires possible steady-state assumptions often smoothed to account for differences in investment profiles across comparator units (peers) Smoothed CAPEX Unsmoothed CAPEX Annuity Depreciated historical cost Capital stock approach long-lived nature of the asset is recognised could be a more realistic approach to measuring CAPEX requires a range of assumptions and adds to the regulatory burden outcome of assessment can be sensitive to the approach taken and assumptions made 13

14 Measuring CAPEX Cash cost approach CAPEX A Year CAPEX Industry average CAPEX C Year CAPEX Industry average CAPEX B Year CAPEX Industry average CAPEX D Year CAPEX Industry average To smooth or not regulators tend to smooth lumpy patterns of CAPEX to reduce fluctuations over time that may be down to company need or managerial decision period over which to smooth (5 years, 10 years?) is debatable as it may need to correspond with asset lives smoothing can reduce degrees of freedom by reducing sample size can introduce internal inconsistency between inputs (smoothedcapex) and outputs (outturncost drivers) can result in different treatment of OPEX and CAPEX that may not be consistent with the objective of TOTEX benchmarking 14

15 Measuring CAPEX Capital stock approach Annuity approach gross value of initial investment x annuity factor Depreciated historical cost approach = depreciation + depreciated asset value x WACC CAPEX Age of asset Return Depreciation 15

16 Overview what do regulators try to assess: objective and definitions issues with benchmarking CAPEX ways of measuring CAPEX menu of tools: top-down and bottom-up case studies key takeaways 16

17 Assessing CAPEX efficiency Top-down approaches C A Unit cost Cost Regression (OLS) adjusted for noise A Ordinary least squares (OLS) Inefficiency OLS shifted to the upper quartile Inefficiency C C C Benchmark unit cost = C C /Y C B Noise? C Corrected OLS (COLS) Y C Output, Y Output Cost Stochastic Frontier Analysis (SFA) Standard regression model is extended Inefficiency A Expected cost c = a + by + inefficiency + noise ; Stochastic frontier Cost Data Envelopment Analysis (DEA) F A E Inefficiency D B Noise C B V C Efficiency frontier Output, Y O 17

18 Assessing CAPEX efficiency Bottom-up approach basket of works combine a group of activities in a basket of works with the associated expenditure compared with an implied level of CAPEX for the complete basket of works avoid comparing on an activity-by-activity basis, which could be distorted by companies allocation policies and definitions of works the basket of works approach compares the unit expenditure on similar capital projects with standardised unit (replacement) costs the standardised unit cost for each project can be set by using the lowest unit cost, or average unit cost, or other benchmark. Some regulators apply adjustments to the unit cost to normalise for exogenous differences (e.g. fixed costs, regional factors) stringency of benchmark should be proportionate to data comparability and quality considerations and availability of suitable comparator data the rate of catch-up set by the regulator could be: For company and capital project 18

19 Overview what do regulators try to assess: objective and definitions issues with benchmarking CAPEX ways of measuring CAPEX menu of tools: top-down and bottom-up case studies key takeaways 19

20 Case studies Ofgem Office of Gas and Electricity Markets Ofwat Office of Water Services UK rail, health and telecoms regulators Council of European Energy Regulators Norwegian Water Resources and Energy Directorate Energiavirasto (Finnish Energy Authority) 20

21 UK Ofgem s approach to assessing CAPEX prior to RIIO Ofgem used a mix of regression modelling and engineering expert analysis to review CAPEX activities for example, in GDPCR1 (gas distribution price review before RIIO): engineering consultant reviewed the GDNs forecast costs to understand whether they were based on appropriate assumptions, including justification for their workload forecasts, assumptions for real price increases and productivity assessment of GDNs efficiency for specific CAPEX and replacement expenditure by benchmarking costs against workload across GDNs using regression bottom-up analysis by PB Power to consider the appropriate costs for particular activities, based on information submitted by the GDNs and its own engineering experience comparison was limited to the GB networks, which are subject to similar licence conditions and obligations, to avoid comparability issues that might arise from using data outside a specific data set Ofgem considered relevant network-specific factors that could be driving differences in the reported costs of companies to enable an equitable comparison 21

22 UK Ofgem s approach in RIIO Estimation approach, model specification and selection Aggregating results Efficiency targets Company-specific factors Model inputs Models Predicting future costs Triangulation Efficiency challenge Reverse cost adjustments Baseline RIIO-ED1 and RIIO-GD1 historical and forecast data from the companies pre-adjusted for company-specific issues (one- and twosided adjustments) RIIO-ED1 2 top-down TOTEX models (different cost drivers) 1 disaggregated result from a number of unit cost and econometric models RIIO-GD1 RIIO-ED1 data RIIO-GD1 1 top-down TOTEXmodel (historical and forecast data) 3 middle-up models (OPEX, CAPEX, REPEX) (but not ultimately used) 7 activity-level disaggregated models (forecast and historical data) Special factors Level of aggregation modelled business plan data directly alongside historical apply coefficients from historical data and modelling first 2 years of business plan data to forecast cost drivers Use of forecast (BP) data RIIO-ED1 RIIO-ED1 and RIIO- fast-track track: 12.5% GD1 weighting on each of upper quartile (UQ) the top-down models, efficiency and 75% weighting on adjustment the disaggregated cost threshold model based on 75:25 slow-track track: 25% weighting on weighting on each of Ofgem s view and the top-down models, companies forecast and 50% weighting on respectively the disaggregated immediate catch-up model set and no time- RIIO-GD1 profiled efficiency 25% weighting on each targets of historical and forecast data, top-down and disaggregated TOTEX models middle-up results ignored RIIO-ED1 and RIIO-GD1 pass-through of special factors allowance Frontier shift and RPEs overlay Frontier shift Predicted future efficient cost level 22

23 UK Ofgem s approach to assessing CAPEX in RIIO Ofgem used a combination of top-down (regression, unit cost) and bottom-up (unit cost, basket of works) approaches to assess CAPEX of energy networks to account for different investment profiles across the modelling period, 7-year smoothed CAPEX was used CAPEX cost drivers (workload volume) were normalised over similar periods to ensure comparability, apart from smoothing, certain costs were excluded where: these costs could not be sufficiently explained by the cost drivers used in the CAPEX models. Such costs were either assessed separately or added back as an adjustment there was a substantial change in the nature of the activity between price reviews these were company-specific and not comparable across companies composite scale variables (CSVs) were often used to simplify the models to assess expenditure these were weighted geometric means of various scale and activity drivers costs were normalised for differences in operating circumstances and regional labour prior to modelling density/urbanity: applied a 15% productivity adjustment to the labour cost element of REPEX and CAPEX mains reinforcement and connections, based on the proportion of work that is carried out within the M25 23

24 UK Ofgem s approach in RIIO-GD1 Ofgem used a toolkit approach to assess the costs submitted by the GB GDNs in their business plans, with relatively greater emphasis placed on econometric modelling Cost area Sub-cost category Drivers Top-down TOTEX OPEX +7-year smoothed CAPEX CSV ofmeav,repex workload, mains reinforcement and connections workload, external conditions reports, maintenance MEAV and emergency servicing CSV. and time trend Middle-up TOTEX (ultimately not used) OPEX CAPEX REPEX Bottom-up regressions Work management Emergency service Repairs Maintenance Mains reinforcement Connections Tier 1 REPEX CSVof MEAV, external condition reports, maintenance MEAV and the emergency CSV CSV of MEAV, connections s workload and mains reinforcement workload Weightedaverage REPEX workload MEAV CSV based on customer numbers and external condition reports External condition reports Maintenance MEAV Weighted averageworkload (calculated by multiplying the work volume for each pipe diameter by an industry average unit cost for each diameter) Weighted average workload Weighted average tier 1 workload 24

25 UK Ofgem s approach in RIIO-ED1 Cost area Top-down TOTEX (macro model) Top-down TOTEX (bottom-up model) Sub-cost category OPEX + 7-year smoothed CAPEX OPEX + 7-year smoothed CAPEX Drivers CSVofMEAV andcustomer numbers using a statistical approach and time trend CSV ofmeav,units distributed,network length, customers, faults, spans cutandcustomer numbers using industry spend proportions and time trend Bottom-up regressions Tree cutting Trouble call Closely Associated Indirects(project management, network policy, vehicles & transport, system mapping, etc.) Business support costs (finance & reg, IT, CEO and management, etc.) Non-op op CAPEX (property, IT&T, small tools & equip, etc.) 20+ other disaggregated models: connections, reinforcement, asset replacement, I&M, etc. Spans cut, spans inspected Faults MEAV and value of network additions Ratio benchmarking using MEAV Ratio benchmarking using MEAV Ratio benchmarking or regression 25

26 Case studies Ofgem Office of Gas and Electricity Markets Ofwat Office of Water Services UK rail, health and telecoms regulators Council of European Energy Regulators Norwegian Water Resources and Energy Directorate Energiavirasto 26

27 UK Ofwat s approach in the 2004 price review (PR04) Econometric models Maintenance expenditure was split into four categories, three of which were modelled using statistical analysis. For example on water, the following models were developed: water distribution infrastructure log unit cost model with length of mains as the scale driver and properties per length of mains as a key cost driver water distribution non-infrastructure log unit cost model with booster pumping station capacity as the scale driver and ratio of storage capacity to pumping station capacity as a key cost driver water management and general log cost model with total number of billed properties as the scale driver and proportion of billed properties that are non-household as a key cost driver Adjustments to costs were made after modelling to account for leakage expenditure, company-specific factors and deviations in future expenditure Cost base (basket of works) Ofwat compiled a dataset of the unit costs for 120 standardised projects and used the basket of goods approach to compare cost predictions. The benchmark was set at the most efficient company Setting allowance Results from the two approaches were weighed equally. In determining the allowance, Ofwat assumed that companies would close 40% of the gap to the econometric benchmark and 50% to the cost base benchmark over the price control period 27

28 UK Ofwat s approach in the 2009 price review (PR09) Econometric models were dropped from the efficiency analysis and the focus was instead on the cost base approach The benchmark was relaxed from the frontier company to the median company. Where data comparability might have been an issue, a company only had to close 50% of the gap The Capital Incentive Scheme (CIS) was introduced to encourage robust business plans and incentivise target outperformance. This involved: setting the baseline level of expenditure comparing a company s forecast to the baseline and using this to calculate an expenditure allowance providing an incentive for further outperformance which declines as the ratio of a company s forecast to baseline increases calculating ex post rewards/penalties as the difference between allowed and outturn expenditure multiplied by the incentive rate Ofwat noted that CAPEX menu regulation increased the quality of the data available for the analysis and limited the need to set a stringent, one-company-based benchmark 28

29 UK Ofwat s approach in the 2014 price review (PR14) TOTEX = OPEX + 5-year average CAPEX Base TOTEX = OPEX + 5-year average maintenance CAPEX Refined TOTEX RE Refined TOTEX OLS (50%) (50%) Triangulate 1 Refined TOTEX top-down (33%) 2 Full TOTEX COLS (33%) Triangulate Water TOTEX Refined base RE Refined base COLS Enhancements unit costs Enhancements unmodelled costs (50%) (50%) Triangulate Base Add Add Add 3 TOTEX bottom-up (33%) Source: Ofwat (2014), Final price control determination notice: policy chapter A3 wholesale water and wastewater costs and revenues, December. 29

30 UK Ofwat s approach in the 2014 price review (PR14) Ofwat modelled three types of capital enhancement expenditure. Certain exclusions/adjustments (e.g. grants and contributions) were removed from data prior to modelling Enhancement expenditure Enhancements to the supply demand balance (e.g. building a new reservoir to cope with increased peak demand) Lead reduction New developments Volume measure Totalenhancement to the supply demand balance Number of lead pipes replaced Number of new connections Four different ways to model unit costs were used: average of the unit costs for each company weighted average of the unit costs for each company OLS regressions of costs against volume OLS regression of log costs against log volumes A significant proportion of enhancement expenditure could not be modelled adequately, so an uplift of 8.4% was considered. This was calculated by: a) finding the industry total modelled enhancement expenditure b) finding the industry total unmodelled enhancement expenditure c) finding the ratio of b) to a) Source: Ofwat (2014), Basic cost threshold model Appendix C Enhancement modelling, December. 30

31 Case studies Ofgem Office of Gas and Electricity Markets Ofwat Office of Water Services UK rail, health and telecoms regulators Council of European Energy Regulators Norwegian Water Resources and Energy Directorate Energiavirasto 31

32 UK Other regulators approach to assessing CAPEX (I) Office of Rail and Road (ORR), 2013 price review PR13 econometric analysis used a subset of the Lasting Infrastructure Cost Benchmarking (LICB) data set developed and maintained by the International Union of Railways (UIC) for 14 European rail infrastructure managers (IMs), including Network Rail, for the period 1996 to 2008 CAPEX (renewals expenditure) is measured as annual reported costs plus a steady-state adjustment to ensure comparability of historical costs across international IMs the adjusted cost is the renewals expenditure that would have been incurred had Network Rail renewed 2.5% of its track. This was necessary as there had been systematic underinvestment in renewals historically the analysis considered a total of 21 econometric models, including a variety of approaches (simple to advanced) 32

33 UK Other regulators approach to assessing CAPEX (II) National Health Service Improvement (NHSI), Monitor top-down econometric models with a particular focus on advanced econometric models to assess CAPEX TOTEX modelled based on reference costs. The following factors were controlled for in the model: hospital outputs (case mix-adjusted hospital activity, the degree of specialisation in Trusts, and the quality of service based on NHS staff surveys); Trust-specific specific drivers (including local disease prevalence, demographics of the patient population, the proportion of emergency admissions, and a market forces factor); location and type of Trust (that control for differences in trust categorisation and former Strategic Health Authority regions) Ofcom used advanced econometric models to assess TOTEX of BT s Openreach activities using publicly available information, BT was compared with the US local exchange carriers (LECs) by normalising BT s data to be comparable with that of the US companies CAPEX was measured using the depreciated historical costapproach (depreciation and cost of capital) 33

34 Case studies Ofgem Office of Gas and Electricity Markets Ofwat Office of Water Services UK rail, health and telecoms regulators Council of European Energy Regulators Norwegian Water Resources and Energy Directorate Energiavirasto 34

35 Council of European Energy Regulators Pan-European benchmarking of gas and electricity TSOs (I) TSOs have a natural and legalised monopoly tariffs need to be based on efficient rather than incurred cost no or limited domestic peer companies in most countries input is TOTEX = OPEX + annuitised CAPEX Source: Sumicsid, Swiss Economics (2016), Benchmarking European Gas Transmission System Operators, Project E2Gas, June. 35

36 Council of European Energy Regulators Pan-European benchmarking of gas and electricity TSOs (II) Restricting the costs/functions of a TSO can help in comparability across TSOs, but can also introduce errors OPEX grid planning, grid construction, grid maintenance, gas transport and metering, and administrative support included costs: labour costs of direct personnel, cost of third-party services, indirect cost of management and other costs labour costs are adjusted to reflect country-specific input prices (wages) CAPEX investment streams are annualised using a standard annuity factor all costs are subject to inflation adjustment (CPI) and currency conversion the average exchange rate is used and national currencies are converted into euros. PPP is used as a sensitivity additional discussions on cost construction underway e.g. a depreciated historical cost approach to measuring CAPEX could be considered and benefit of doubt (i.e. higher of resultant efficiency values) could be set on TSOs 36

37 Council of European Energy Regulators Pan-European benchmarking of gas and electricity TSOs (III) cost driver selection based on engineering expertise and statistical analysis the core model includes only three cost drivers: normalised grid grid length adjusted by various complexity factors to account for land use, sub-surface features, topography and soil humidity normalised grid is clearly an important variable, but there does not appear to be any sensitivity analysis with respect to its definition total number of connection points grid capacity additional discussions on cost drivers underway can three variables capture the key differences between TSOs? What is missing? How can this be mitigated? does the normalised grid adequately take into account firm-specific factors relating to the environment? modelling approaches: top-down methods, in particular, optimisation approaches such as DEA and econometric techniques such as SFA, are the main methods used 37

38 Case studies Ofgem Office of Gas and Electricity Markets Ofwat Office of Water Services UK rail, health and telecoms regulators Council of European Energy Regulators Norwegian Water Resources and Energy Directorate Energiavirasto 38

39 Norway Three-stage approach separate regulation of networks between 33kV and 132kV (86 companies regional distributors), and local distributors from 0,230kV to 22kV (136 companies) three-stage process for arriving at allowed revenue revenue cap method Benchmarking (DEA) models C*** Adjustment for operating environment C** C* Calibration process Cost norm 39

40 Norway Modelled cost benchmarking analysis C*** top-down approach (DEA) constant returns to scale (CRS) the frontier is estimated using 5-year-averaged data the modelled cost C t (input input) is: Where: CPI C t = CPI t ( OM t 2 + VOLLt 2 ) + ( Lt 2 Pt ) + DEPt 2 + RABt 2 WACCt t 2 OM t-2 = operatingandmaintenancecostsinyeart-2 VOLL t-2 = valueoflostload(customerinterruptioncosts)inyeart-2 CPI t = consumerpriceindexintheyeart CPI t-2 = consumerpriceindexintheyeart-2 L t-2 xp t = valueofnetworklossesinmwh,yeart-2,consideringlocalareapricesatnordpoolspotinyeart DEP t-2 = depreciationinyeart-2 RAB t-2 = regulatoryassetbase,consideringbookvalueinyeart-2 WACC t = weightedaveragecostofcapitalinyeart 40

41 Norway Treatment of CAPEX RAB since 2007, there has been strong criticism of using the book value of the RAB as the measure of CAPEX two companies with equivalent assets, maintenance costs and efficiencies will have different DEA scores if the timing of investment differs Replacement cost in 2012, NVE considered moving to a replacement cost measure of CAPEX while it may offer a more correct measure of CAPEX, it was deemed inapplicable in a regulatory framework as it gave too strong an incentive to reinvest Reducing the age effect while maintaining incentives elements of the third stage of calculation changed in order to reduce the age effect in the first stage (DEA), a higher RAB punishes a company, while in the third stage (cost calibration) a higher RAB rewards a company 41

42 Norway The DEA model Outputs of the DEA model the weighted values take into account different costs of each type of grid component, while the costs of transformers, switches and compensation devices include the value of substations there is recognition that some of these outputs may technically be under management control, but they are practically exogenous since many decisions are determined by other regulations Local distributors Length of HV network (km) Number of network substations Number of subscriptions (customers) Regional distributors Weighted value of overhead lines Weighted value of ground cables Weighted value of sea cables Weighted value of substations 42

43 Norway The Norwegian Water Resources and Energy Directorate calibration reaching C* further calibration for deviation between actual and expected costs (with two-year lag): at this stage the Z factor-corrected DEA scores are calibrated to set the aggregate of cost norms equal to the sum of the observed costs in the sector. The savings in the sector come through the distributors incentives for cost reduction the calibration is as follows: Where: = + Ci*= cost norm of distributor i, considering the cost of capital defined by NVE Ci** = cost norm adjusted for operating environment RAB i = regulatory asset base of distributor i final calibration between proposed costs and cost norm to accommodate possible modelling limitation and balance trade-off between cost-reduction incentive and rent transfer to the end-customer there is no specific requirement of efficiency. This is because if a DSO improves its position relative to the average, it will have a higher rate of return. This is the main feature of the model that aims to encourage efficiency in operation, usage and development of the distribution network 43

44 Case studies Ofgem Office of Gas and Electricity Markets Ofwat Office of Water Services UK rail, health and telecoms regulators Council of European Energy Regulators Norwegian Water Resources and Energy Directorate Energiavirasto 44

45 Finland Energiavirasto Up until 2011 the efficiency score was calculated as an average of scores from two models: an NDRS DEA model and a linear SFA model The input to the models was: TOTEX = OPEX + CAPEX + DCO Where: OPEX = controllable operating costs CAPEX = straight-line depreciation from the replacement value of t network DCO = a monetary disadvantage caused by outage for electricity customers for , an advanced econometric method, stochastic non-smooth envelopment of data (StoNED), was used to assess TOTEX CAPEX was dropped from the efficiency target and investment was monitored separately only half of DCO was added to OPEX to generate TOTEX 45

46 Finland Energiavirasto in the fourth ( ) and fifth ( ) regulatory periods, the efficiency challenges will continue to be estimated using the StoNED method, as informed by the regulator the input output variables and the operating environment variable are modified as follows: DCO has moved as an undesirable output modelled cost (input) is limited to OPEX only Source: Energiavirasto (2015), Regulation methods in the fourth regulatory period of 1 January December 2019 and the fifth regulatory period of 1 January December 2023, October. CAPEX (based on replacement value) is used as a fixed input in that an efficiency challenge is only set on OPEX, however the impact of investment on OPEX (i.e. the trade-off) is still captured the Z variable is still intended to capture differences between urban and rural networks using a connection density measure defined as number of connections per metering point 46

47 Overview what do regulators try to assess: objective and definitions issues with benchmarking CAPEX ways of measuring CAPEX menu of tools: top-down and bottom-up case studies key takeaways 47

48 High-level messages (I) CAPEX construction: plethora of approaches but with limited understanding of the gaps UK energy and water regulators tend to use a cash cost approach where outturn CAPEX is smoothed or left unsmoothed over a sufficiently long period to mitigate volatility and differences in investment profiles among EU regulators, the capital cost approach seems to be preferred. Two approaches are taken to measuring CAPEX annuity and depreciated historical cost all approaches appear to have limitations. For example, asset condition/health, as well as service quality measures, are often ignored, while the inflation measure and the WACC assumptions affect the capital cost approaches cash cost approach will also require careful consideration of the state of the networks, investment profile and base year of analysis 48

49 High-level messages (II) CAPEX assessment: no agreed approach among European and UK regulators CAPEX assessment is inherently difficult and regulators tend to be relatively conservative in a number of regulatory applications, assumptions underlying the assessment approach do not appear sufficiently justified (e.g. the correspondence between CAPEX and its drivers) moreover, there could be inconsistencies in the treatment of CAPEX in the benchmarking exercises and how a TSO is remunerated, which governs its investment decisions different approaches to measuring and treatment of CAPEX should be examined, to ensure outcomes are not skewed by assumptions or limitations with particular approaches similarly, more disaggregated, bottom-up or operational evidence is often used as a cross-check in a regulatory context lastly, any benchmarking exercise rests on certain simplifications and assumptions and due allowance must be made for a margin of error 49

50 THANK YOU FOR YOUR ATTENTION! Srini Parthasarathy W Web: