Energy Policy 39 (2011) 1505–1513

Contents lists available at ScienceDirect

Energy Policy

0301-42

doi:10.1

$This

requirem

to than

Dr. Thom

at the N

Sharp f

supervi

omission Tel.:

E-m1 Cu

Econom

journal homepage: www.elsevier.com/locate/enpol

Negotiated settlements with a cost of service backstop: The consequencesfor depreciation$

G. Kent Fellows n,1

Department of Economics, University of Calgary, 2500 University Dr. N.W., Calgary, AB, Canada T2N 1N4

a r t i c l e i n f o

Article history:

Received 8 September 2010

Accepted 10 December 2010Available online 7 January 2011

Keywords:

Negotiated settlements

Regulation

Depreciation

15/$ - see front matter & 2010 Elsevier Ltd. A

016/j.enpol.2010.12.024

paper is adapted from an academic thesis co

ents of the Master of Arts program at the Univ

k the members of my thesis defence com

as Cottrell and Dr. Chris Auld for their valua

ational Energy Board, particularly Craig Rubie,

or their cooperation and comments. I would

sor Dr. Aidan Hollis for his assistance and supp

ns are my own responsibility as author.

+1 403 988 8078.

ail address: GKFellows@gmail.com

rrently a Ph.D. student at the University

ics.

a b s t r a c t

The movement from traditional regulatory hearings to negotiated settlements represents both a

departure from cost of service regulation and a relaxation of regulatory oversight. Under negotiation

parties are able to renegotiate inclusions in their cost of service while simultaneously creating a profit

margin for the regulated firm where none existed under the cost of service outcome of a traditional

hearing. This paper constructs a model to illustrate the existence of positive gains to pipeline and shipper

from the re-allocation of expenses through time in the regulated pipeline services market in Canada.

Behaviour consistent with the model is observable in anecdotal and econometric evidence gathered from

the library of the National Energy Board of Canada, responsible for pipeline toll regulation in Canada.

Empirical investigation by Littlechild (2009a) into settlement procedures in the Florida electricity market

reveals similar findings; however, this analysis represents the first attempt to model the behaviour

formally. The econometric analysis uses new data collected and compiled specifically for this exercise.

& 2010 Elsevier Ltd. All rights reserved.

1. Introduction

’’Cost of service’’ has historically been the standard for mucheconomic regulation, but its many faults have led to a variety of lessintrusive price-setting mechanisms. One of the most attractive – atleast superficially – is the ‘‘negotiated settlement’’ (NS) in whichthe otherwise regulated firm and a core group of customersnegotiate a price without direct intervention of a regulatoryauthority. This paper explores the appeal of these settlements,which have become increasingly common in certain previouslyregulated industries. While settlements obviously provide a ben-efit, relative to the regulated outcome, what is the source? Is itlower litigation costs? More efficient operations? What this papershows, using the Canadian pipeline market as an example, is thatthe most substantial part of the benefit appears to be deferral ofdepreciation. Deferral of costs, does not, of course, imply any

ll rights reserved.

nducted in part to satisfy the

ersity of Calgary. I would like

mittee; Dr. Robert Mansell,

ble input, as well as the staff

Jecielle Alonso and Dr. Karen

also like to thank my M.A.

ort. Any and all inclusions or

of Calgary, Department of

increased efficiency in the conventional manner: it simply transferscosts to future consumers.

Negotiated settlements have been used extensively for manyyears, but have attracted relatively little academic attention. TheUS Federal Power Commission (now the Federal Energy RegulatoryCommission, FERC), faced with a backlog of some 3000 rate cases inthe 1960s, began encouraging utilities and consumers to settle ratecases privately. At the state level, the Florida Public ServiceCommission (FPSC) pushed for settlement outcomes negotiatedbetween utilities and consumer groups (including the Office ofPublic Counsel created to represent general public consumerissues) since the 1970s. In Canada, the National Energy Board atthe Federal level and the Energy Utilities Board in the province ofAlberta began to facilitate similar negotiated outcomes beginningin the mid-1990s. There has also been significant discussion in theUK about moving from price cap regulation of telecom, water, gas,and electric utilities to a system in which the consumers wouldhave significantly more input into the pricing decisions (Littlechild,2008, 2009b).

In general, negotiated settlement procedures allow for incentiveschemes and other tradeoffs between contract consumers andproducers. In Canada, the positive responses and preference ofpipelines and shippers for NSs over more traditional methods hasbeen cited by the National Energy Board (NEB) as an indication ofthe achieved benefits over a pure cost of service model.

Contemporary examinations of settlement procedures inCanada and other jurisdictions have been made by Doucet andLittlechild (2006, 2009), Buchmann and Tongren (1996), Morgan(1978), Schultz (1999), and Walker (1986). These examinations

G.K. Fellows / Energy Policy 39 (2011) 1505–15131506

analyse the outcomes of NSs from the point of view of theregulatory authority and are largely occupied with legal issues ofjurisdiction (due to the existence of settlements as an alternativefor full litigation) or the reduction in burden and backlog on theregulatory authority.

Littlechild’s (2009a) examination of settlement outcomes in theFlorida electricity market identifies significant rate reductionsaccompanying a movement from litigation to NS. These ratereductions were found to be the result of a reduction in thedepreciation and amortization expenses allocated to those periods.The conclusion reached by Littlechild (2009a, pp. 32–33) relates theconsumers’ temporal preferences, in that ’’jam today in the form ofrefunds and rate reductions’’ is preferred by consumers to ’’lowerprices tomorrow’’.

Similar outcomes are observable in Canada whereupon adopt-ing NS-based tolls in 1996, TransCanada Pipelines (TCPL), itscontract consumers and other interested parties agreed to freezedepreciation rates at the 1996 levels. These rates remained frozenuntil 2001 despite significant demand reductions in the late 1990swhich led to a reduction in its forecast economic life. Since the NEBmodel of regulation is derived from a cost of service base, theimplied shorter lifespan combined with a current freeze ondepreciation implies a higher disparity between the immediatelow prices and the higher future prices. Depreciation rates willinevitably increase to ensure full depreciation over the shorterperiod. By freezing depreciation rates, the increase in prices isdeferred, not avoided.

Between 2001 (when TCPL defaulted to a litigated outcome) and2003 (just before the firm returned to a NS procedure) the effectivedepreciation rate used in its toll setting rose from 2.64% to 3.65%.2

This represents a reduction in the average anticipated economic lifeof assets from E38 years to E27.5 years. During the 2002 tollhearing, when the NEB eliminated some of the incentive schemesnegotiated by TCPL, there was another jump in composite depre-ciation rate (from 2.89% to 3.65%).

The 38 year average anticipated economic life imputed from theeffective depreciation rate prior to 2001 warrants further explana-tion. Under the Generally Accepted Accounting Principles inCanada, the longest allowed depreciation period for any class ofassets is 40 years. The pipeline’s small assets (computers or officeequipment) are depreciated over a period significantly less than 40years. These assets will therefore reduce the reported effectivedepreciation rate as it is an average of specific asset classdepreciation rates weighted by each class’ contribution to theasset base. This implies that, despite the fact that the effectivedepreciation rate imputed a 38 year economic life, in fact themajority of large pipeline assets were expected to last 40 years ormore in 1996.

Baumol (1971), using the framework developed by Littlechild(1970), illustrates that given socially optimal and profit-maximiz-ing depreciation rules, there is a positive relationship betweendepreciation expense and the shadow price on capacity. In the TCPLcase we observe the opposite. While the complexity of theregulated market violates some of the assumptions imposed bythe Baumol model, it is not intuitively apparent why TCPL wouldallow the depreciation expense to remain at 1996 levels whenpresented with contract non-renewals. Instead, these non-renew-als should have prompted an increase in the depreciation rate.One might be inclined to write this off as an anomaly, but theTCPL outcome is not isolated. Littlechild’s (2009a) survey of theFlorida electricity market also identifies several outcomes with a

2 NEB (2003), RH-1-2002, TransCanada PipeLines Limited, Tolls application,

July.

negotiated settlement leading to reduced depreciation leading to areduction in the utility’s cash flow.

Burness and Patrick (1992) model the issue of optimal depre-ciation under regulatory constraint as a linear control problemwith an upper and lower bound on the depreciation expense(defined due to regulatory restrictions). They find that, under a verybroad set of conditions, a back-loaded toll (where depreciationexpenses are set higher in later periods) is optimal. The model alsopredicts that, given an inflated cost of capital in its cost of service,a profit-maximizing regulated firm will choose a back-loadeddepreciation path.

This could be taken to imply that a regulator may wish to allowfor a positive margin on the cost of capital in order to entice thistype of behaviour; however, the result is not altogether convincingin the case of regulated firms or utilities in the energy or energytransportation sectors. The Burness and Patrick model implicitlyassumes that the costs of the regulated firm represent the entirecost of a finished product (or equivalently, that the demandfunction faced by the regulated firm is fixed across time periods).Transportation and other utility industries represent only a portionof the final cost of a delivered good or service, thus it is alsonecessary to account for the time path of other costs in deriving atotal surplus or policy implication (alternatively, one could take theview that the demand function faced by the pipeline is changingthrough time due to external costs faced by the consumers who payto transport on the system).

In much of the current theoretical work in the area of naturalresources it is generally assumed that firms extracting non-renew-able natural resources face an increasing marginal cost throughtime (Sweeney, 1993). Empirical work (mainly in the area of coalextraction) has been done to verify this assumption but the resultsare generally extensible (Harris, 1993). Given this it may be that thedeferral of depreciation exacerbates rather than mitigates tem-poral cost asymmetries in the final delivered good and would havenegative efficiency implications.

The remainder of this paper is laid out as follows. First a simplemathematical framework is constructed based on actual account-ing procedures used under National Energy Board (NEB) jurisdic-tion in order to identify the nature of possible pipeline and shippergains resulting from NSs. The implications of the theoreticalframework are then tested using econometric methods followedby some concluding remarks. While this examination is specific tothe NEB case the results should be generally consistent with manyof the other jurisdictions contemplating or having already movedfrom a litigated cost of service based outcome to a settlement orother light handed regulatory approach.

The contribution of this paper is its focus on the motives of thepipeline and shippers. With specific reference to depreciationmethodology the incentives to the pipeline and shippers bare closeexamination. Under a litigated outcome the NEB requires that adepreciation methodology be substantiated by evidence support-ing the assumptions on which the methodology is based while in ahearing, depreciation methodology is freely set and no informationto substantiate the effective rate is required. Given public interestconsiderations and the lack of representation by future consumersthe implications of deferring depreciation bare close examinationunder a settlement outcome where a depreciation rate may be setwithout providing supporting evidence.

2. An overview of NEB regulation

Both before and after the introduction of NSs, the NEB employeda formal litigation procedure (litigation is currently a defaultoutcome if negotiations fail to produce a satisfactory outcome),based on the established cost of service (COS) model with a focus on

G.K. Fellows / Energy Policy 39 (2011) 1505–1513 1507

cost of capital issues. Between 1995 and 2009, to reduce hearingdelay and uncertainty, the cost of capital calculations followed anestablished mathematical and accounting structure.3, 4 Every yearthe NEB set a fixed cost of capital for each pipeline meant to reflectits opportunity cost of capital. Other costs were then forecast andreported upon by the pipelines, scrutinized by hearing interveners(including shippers) and finally approved (or amended) by the NEB.

While it has been widely used in economic regulation there arewell-known criticisms of the standard COS structure. First, the useof COS pricing eliminates or significantly reduces incentives toreduce costs or improve product quality over time (Stigler, 1976;Sappington, 2005). Second, COS prices are difficult to implementsince the regulator (the NEB in this case) lacks perfect informationabout current or future costs and is therefore unable to directly setprices a priori such that profits are consistently equal to zero(Stigler, 1976).5

In addressing the latter issue the NEB has employed forecastingtechniques indexing future costs estimates to current costsadjusted using a public signal. The Return on Equity (ROE) formulahistorically used by the NEB to calculate the cost of capital isadjusted year to year based on bond rate forecasts.

In the context of these identified issues, Doucet and Littlechild(2006a, 2009) establish a hypothesis as to why NSs were developedas an alternative to litigation. Their explanation is based on cost andservice quality considerations. As Chairman of the NEB Ken Voll-man indicated, ’’Users of the pipelines had grown disenchantedwith a regulatory process that was costly, time-consuming, and atwhich they felt they could not win.’’6

References to a ’’costly’’ process and a ’’no-win’’ situation implythat both upstream and downstream parties are able to benefitfrom the increased flexibility inherent in NSs. Doucet andLittlechild (2006w),7 assert that negotiated settlements ’’providea more sensible way for the parties to do business’’ through theirability to ’’enable the wishes of customers to be more clearlyidentified and more closely aligned with the abilities of pipelines todeliver the services required.’’

The idea of moving from the perceived no-win game describedby Vollman (1996) to a setting in which parties can influence anddistribute total surplus via some interaction other than thetraditional rate/quantity relationships is the key to this discussion.Thus, if there is a possibility for a non-price/quantity improvement,the absolute gains in surplus must be realized from either a cost

3 This was defined in the multi-firm cost of capital hearing RH-2-94, NEB (1995).

RH-2-94 Her Majesty the Queen in Right of Canada as represented by the National

Energy Board.4 Although the practice has been discontinued from 2009 on, a base level of ROE

was historically calculated by the NEB every year since 1994. The figure is

distributed to both Pipelines and their Shippers. These reports can be found on

the NEB website at: http://www.neb-one.gc.ca/clf-nsi/rpblctn/ctsndrgltn/rrggnm-

gpnb/tll/tll-eng.html.5 The regulatory hearing process also presents other inefficiencies separate

from the effects on the producers and consumers. Additional costs in the form of

regulatory burden and decision lag are incurred. The federal regulator for pipelines

in the United States (the Federal Energy Regulatory Commission or FERC) has

identified formal regulatory hearings as an unacceptable burden in its decision to

employ Negotiated Settlements (NS) to resolve a backlog of cases that would have

taken decades to litigate rather than months to negotiate (Sappington, 2005).

Several informal sources also identify the additional burden on hearing participants

in contracting experts and legal council to represent their interests in formal

regulatory hearings. A complete discussion of the regulatory burden in the NEB case

can be found in Doucet and Littlechild (2009).6 Doucet and Littlechild (2009), Negotiated Settlements and the National

Energy Board of Canada Electricity, Policy Research Group Working Paper No. EPGR

06/29, page 17. Originally quoted from: Vollman (1996), Toward incentive regula-

tion of Canadian pipelines, Fifth Annual DOE-NARUC Natural Gas Conference, St

Louis, Missouri, April 28–May 1, available at NEB website, page 6.7 The w here references the fact that this specific quotation has been taken from

a working version of this paper. The quotation has been omitted in the final

published version; Doucet and Littlechild, 2009.

reduction, or an increase in the shipper or consumer valuation ofpipeline services.

The revealed preference by pipelines and shippers for nego-tiated settlements over litigation implies that the total surplusavailable for allocation between the pipeline and shippers for theperiods covered by the tolls must increase, and the distributionmust be such that the profits of no group fall. This in turn impliesthat in order to maintain credibility, any description of thisbehaviour must exhibit two outcomes. First increased surplus(higher willingness to pay by the shippers or lower costs to thepipeline) over the tolling period and second a mechanism todistribute this resulting surplus between pipeline and shippersuch that neither group is made worse off compared to a litigatedoutcome.

Two potential areas of negotiated settlement associated costreduction are the reduction in regulatory costs (Morgan, 1978;Doucet and Littlechild, 2006, 2009) and the change in depreciationmethodology. Littlechild (2009a, pp. 1) asserts that the purpose ofNSs ’’is not to save [regulatory] costs, which [he states] are orders ofmagnitude less than the revenues at stake.’’ Rather, he claims thatother means of rate reduction constitute the only material benefitfrom moving to NS. This analysis can be extended to the Canadianpipeline services market as well. A few figures illustrate theparallel.

First the average amount paid by a pipeline to the NEB for costrecovery is 2.2 million. Since the NEB is funded by the pipelineindustry rather than directly by the government this figurerepresents TCPL’s share of the total NEB operating costs. The TCPLmainline also recorded regulatory proceeding costs for 2003 of $2.4million. (This data is for the largest pipeline in the dataset in a yearin which a full toll hearing was conducted due to the breakdown ofnegotiations in 2002.)While this amount may be significant a 1%change in the average depreciation expense is roughly ten timeseither of these amounts and as shown in the econometric inves-tigation below a move to negotiated settlements is likely to resultin a 5–20% reduction in depreciation expense.

If the price (toll) is at or above the marginal cost of production,reductions in price resulting from a lower depreciation rate willunambiguously increase the total surplus in the market and anymechanism able to distribute this added surplus between pipelineand shipper will ensure a Pareto-optimal outcome. FollowingLittlechild’s (2009a) lead, the examination here starts with a closerlook at the depreciation methodology used under litigation and NS.

3. Depreciation methodology

Litigated NEB outcomes generally employ a straight line methodof depreciation, where the depreciation expense for a class of assetsis set equal in all periods. A simplified version of the depreciationexpense calculation for a pipeline with only one physical asset orasset class is given as follows:

Dt ¼F

Tð1Þ

where D is the depreciation expense allocated to period t; F is thefixed cost of the asset at the time of investment; and T is the lifespanof the asset (generally economic life for the pipeline itself).

In a toll hearing the NEB can and generally does request, as anelement of evidence to support tolls, periodic depreciation studiesfrom the pipeline in order to set the D based on appropriateexpectations of T. Under a NS, the NEB has not historicallyrequested these studies regularly since toll applications based ona NS are judged on negotiating process rather than outcome. If allinterveners and the pipeline support the NS, then the NEB will

G.K. Fellows / Energy Policy 39 (2011) 1505–15131508

generally approve it, otherwise a toll hearing is undertakenregardless of the specifics of the settlement itself.

Thus, under NS, the pipeline has private information on T which,unlike the private information the pipeline has on other costs, is notrevealed at the end of each period. The actual T need not be divulgedat all unless the NEB requests a depreciation study or the pipelinefinds it desirable to update depreciation expense (possibly near thefinal period of operations, to ensure full depreciation if thehistorical expense has been too low, or in the case of someincreased risk or deliverability issues). Since the NEB’s expectationsare based only on what is divulged in the toll application given anuncontested settlement outcome, the NEB will judge all compo-nents of the tolls as fair, including the depreciation expenseregardless of the effects on future parties who may be unrepre-sented in negotiations.

The lack of input from future shippers in current negotiations,combined with cost of capital considerations and the added free-dom inherent in NSs, is likely to produce an outcome that places ahigher share of costs (relative to the litigated outcome) on futuregenerations.

4. Mathematical framework

In either a NS or a litigated outcome the cost of capital isgenerally agreed to with respect to the NEB’s generic cost of capitalformula derived by the NEB in the RH-2-94 generic cost of capitalhearing.8 The generic formula uses bond yield forecasts to derive anadjustment to the rate of return on common equity every year. Thisadjustment is then combined with a deemed debt equity ratio,known debt service payments, and a risk premium decided for eachpipeline based on individual characteristics.9

Under a litigated outcome, this cost of capital formula has beenuniformly applied by the NEB to all regulated group 1 pipelines. NSproceedings, while they may base cost of capital decisions on the NEBgeneric formula have the ability to negotiate away from the NEB-mandated ROE. Known examples of this include a Trans-Northernpipeline NS in which Trans-Northern is able to earn 70.25% of the NEBcalculated ROR. Additionally other bonuses and incentives for servicequality and cost reductions are often applied to the ROR in practice.

In order to properly model the scope for profits mathematicallyit is important that the elements of the revenue requirement thatrepresent costs are distinct from those that represent profits. In thefollowing analysis, it is assumed that the return on capital includesthe opportunity cost of capital, and not the economic profits (thoserevenues above the opportunity cost of capital) realized fromincentive schemes which are allocated to equity holders in the formof profits. In the following functions, the excess ROE beyond what isrequired to offset the opportunity cost of capital is treated as anelement of profit rather than an item of costs.

The corresponding pipeline profit function for an NEB regulatedpipeline in a single period (t) can be given as:

pt ¼ ðwL*tÞþðrþgÞKtþDtþ It

h i|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

Period t revenue requirement

� ðwLtÞþrKtþDtþ It½ �|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}Period t cost

ð2Þ

where p is the reduced form regulated profit in period t; w is thewage rate (unit cost of non capital means of production); L is theforecast quantity of labour (quantity of the non-capital means ofproduction); Ln is the incurred quantity of labour; r is the rental rate

8 NEB (1995), RH-2-94, reasons for decisions, multi-pipeline (Cost of Capital),

March.9 As of late 2009 the RH-2-94 formula has been repealed. Pipelines must now

either negotiate a cost of capital or apply to the NEB to calculate an individual cost of

capital for them.

of capital; g is an allowed margin on the rental rate of capital; Kt isthe capital stock in period t; It is the income taxes in period t; and Dt

is the depreciation expense in period t.The above equation is vulnerable to criticism in that it is

representative of ’’regulated profit’’ rather than economic profit.This distinction occurs since the true path of depreciation (i.e. theactual physical deterioration of the asset) is not the same as thereported depreciation expense. For the equation to be representa-tive of economic profit, the entry for depreciation in the secondterm (period t cost) would not be identical to the entry fordepreciation in the first term (period t revenue requirement).The latter entry would be the ‘‘true’’ value of depreciation while theformer entries would be the reported value. The entries for capitalstock would follow the same pattern since, over time, the capitalstock is partially a function of the depreciation expense. While thisis an important distinction this exercise will continue using Eq. (1)as an abstraction of profits in general.

Due to the treatment of depreciation expense as the singleperiod allocation of the pipeline’s fixed costs, the single periodexpense under the standard straight line method of depreciation,assuming smooth intergenerational depreciation rates, is definedsimply by Eq. (1).

In practice the depreciation calculation is more complex sincethe fixed costs and finite periods of pipeline assets are variable inthe long run due to expansions and/or retirements/replacement ofphysical assets. Similarly the pipeline’s income tax calculations arevery complex in practice. For the purposes of this discussion,income tax can be simplified to a function of the equity componentof the cost of capital, the depreciation expense and a few fixedparameters.

Since the debt equity ratio of the firm is fixed, the return onequity is directly related to the return on rate base, so taxes can besimplified to be a function of the return on capital, depreciation, themarginal effective tax rate and some fixed inclusions. Defining I asincome taxes

It ¼ tðmðrþgÞKtþDtÞ ð3Þ

where the newly introduced variables t is the marginal effectiveincome tax rate and m is the proportion of return on rate baseaccrued to equity holders (debt/equity ratio).

There are elements included in the calculation of taxable incomenot expressed by the model, but these are largely invariant withrespect to the regulatory structure. As such their inclusion in themodel is not required and would needlessly complicate the analysis.10

Substituting in the functional form for taxes into Eq. (2)

pt ¼ ðwL*tÞþðrþgÞKtþDtþtðmðrþgÞKtþDtÞ

h i|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

Period t revenue requirement

� ðwLtÞþrKtþDtþtðmðrþgÞKtþDtÞ� �|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

Period t cost

ð4Þ

If a toll hearing is able to accurately identify the cost of capitalthen the margin on the opportunity cost of capital is zero (g¼0).This is obviously a strong assumption almost certainly violated inpractice, but it serves to illustrate the point that if an accurate costof capital is combined with accurate forecasts on the non-capitalcosts (Ln) then Eq. (4) will reduce to zero. It should be noted thatwhile not realistic in practice this is consistent with the theory ofcost of service regulation.

10 Once such example is the use of a capital cost allowance (CCA) by TCPL for tax

purposes. This does not have any effect on the rate base, and due to the relative

invariance of the CCA allowed under tax law, with respect to regulatory regime

(litigation vs. NS) there is no need to complicate the analysis with its inclusion in the

model.

G.K. Fellows / Energy Policy 39 (2011) 1505–1513 1509

From Eq. (4), the total revenue collected from the shippers in anyperiod can be given as

TRt ¼ ðwL*t Þþ½ðrþgÞðKtÞ�þDtþtðmðrþgÞKtþDtÞ

h i|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

Period t revenue requirement

ð5Þ

Changing notation slightly, the cost of the assets (the fixed cost)will be denoted from here on as K (capital) rather than F (as it was inEq. (2)). It is important to note the abstraction here, since, in realityK and F are not equal (the reason for using F to define depreciation inthe first place). The money invested in building the plant is not theeconomic measure of the value of the capital stock. This is similar toa point made above, where it is noted that the profit functiondeveloped is not directly indicative of the economic profits, butrather a measure of the reduced regulated profits. With this in mind,the act of switching notation to K is not detrimental, and is useful asit reinforces the idea that rK is a measure of the cost of capital.

In a NS outcome, the return on equity, and by extension the rateof return restrictions imposed by the RH-2-94 formula are relaxed,and pipelines are allowed to earn a margin above the opportunitycost of capital. The implications and incentives of a shipperagreeing to this are not immediately clear. In a ceteris paribussetting, allowing an increase in ROE implies higher tolls. Allowinghigher tolls is not a rational action for profit maximizing shippers;however, scope for tradeoffs between ROE and other cost areas(specifically the depreciation expense) exist. These tradeoffs canlead to increased NPV of the pipeline and shipper profit streams.

Outside a NS it is also probable that in litigation the NEB may errand allow a ROR above that necessary to cover the cost of capital;however, since the NEB is assumed to have more accurate knowl-edge of the economic life of the pipeline under full litigationrelative to NS, this issue is sidestepped and it is assumed that the NScost of capital will be no lower than the litigated outcome and nocloser to the actual cost of capital.

A two period NPV calculation is given for the pipeline below. Toensure full depreciation D1¼ZK and D2¼(1�Z)K given 0oZo1. Inthe model Z represents the proportion of depreciation expenseallocated to period 1. The NPV of the model can be given as

NPVðpÞ ¼�KþðwL*

t Þ�ðwLtÞþðrþgÞKþZK

1þr

þðwL*

tþ1Þ�ðwLtþ1ÞþðrþgÞðK�ZKÞþð1�ZÞKð1þrÞ2

ð6Þ

The derivative of this NPV calculation with respect to Z impliesthat the total NPV increases as Z falls. This implies, under theassumption of a positive g, that the pipeline has an incentive todefer depreciation expenses into the future

@fNPVðpÞzg@Z ¼

�Kgð1þrÞ2

o0 ð7Þ

ðwL*t ÞþðrpþgÞKþZKþtðmðrpþgÞKþZKÞþ

ðwL*tþ1ÞþðrpþgÞð1�ZÞKþð1�ZÞKþtðmðrpþgÞð1�ZÞKþð1�ZÞKÞ

1þrs|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}TR with inflated cost of capital ð0ogÞ and deferred depreciation ðZ40:5Þ

rðwL*t ÞþrpKþðK=2ÞþtðmrpKþðK=2ÞÞþ

ðwL*tþ1ÞþrpðK=2ÞþðK=2ÞþtðmðrpþgÞðK=2ÞþðK=2ÞÞ

1þrs|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}TR without inflated cost of capital ð0 ¼ gÞ and smooth depreciation ðZ ¼ 0:5Þ

ð8Þ

This result is not surprising, and intuitively very simple andstraightforward. As long as the pipeline is assured that lifetimedepreciation will be sufficient to cover its fixed costs, the pipelinewill elect to retain capital for longer if it can earn a margin on thecost of capital g40. Given no margin on the cost of capital (g¼0)the derivative collapses to zero and there is no incentive toreallocate depreciation expense through time.

The behavioural implications are similar to those expected viathe Averch–Johnson effect (Averch and Johnson, 1962). The differ-ence is that here pipelines are borrowing from the future (bydeferring depreciation expense) in order to inflate their rate basewithout actually employing more capital. In the Averch–Johnsonmodel, the firm is actually employing more capital and failing toequate marginal rates of factor substitution to the ratio of factorcosts; however, in both cases the actions are undertaken to inflateincoming revenues through the margin on the cost of capital.

Another notable departure from the Averch–Johnson effect isthat, due to the deferral of depreciation, the shippers find itpreferable to allow for the inflated cost of capital. In the AJ model,an inflated cost of capital raises the costs to the downstream(consumers) whereas, under this model, the deferral of deprecia-tion reduces these costs in the short term.

From the shipper perspective there are two cases under which itwould be rational to participate in a NS and allow the pipeline anincreased positive margin on the cost of capital. First if the shipperplans to stop shipping on a pipeline before the pipeline retires itsassets, or second, if it has a higher cost of capital compared with thepipeline. In either of these cases the shipper will support a deferralof depreciation expense and make the requisite transfer describedabove via a margin on the cost of capital or an incentive envelope.

Using a two period model, a shipper contracting in only the firstperiod will unambiguously benefit from a deferral of depreciation.The deferral remains beneficial to any period one shipper even if aconcession is made to the pipeline in the form of increased cost ofcapital or revenue sharing. The transfer from shipper to pipelineprofits need not be greater than the reduction in depreciationexpense in order to be beneficial to the pipeline. In this case newshippers in period two unambiguously lose since they end uppaying a portion of the depreciation that the now absent period1 shipper should have paid.

The tax burden also shifts, since depreciation is not considered adeductible expense when calculating taxable income for federalCanadian pipelines. In effect deferring depreciation expense alsodefers the income tax associated with the revenue generated bythis non-deductible pass through cost. In this manner, all parties tothe period one negotiation (the pipeline and period one shipper)are better off. New period two shippers unambiguously lose as aresult which is not surprising as new shippers have had no say inthe earlier negotiations.

The costs incurred by shippers when contracting for pipelinecapacity directly correspond to the total revenue of the Pipeline.Using Eq. (5) as a base, we can modify the revenue function(shipping costs) for the NS case and the litigated case andinvestigate the conditions under which the NS outcome representsa lower cost compared to the litigated outcome.

The Pareto improving condition for the long lived shipper in asimple two period model is

Simplified and adding the Pareto improving condition for thepipeline

0rgr ðð1=2Þ�ZÞ½rsð1þtÞ�rpð1þtmÞ�

ð2þrs�ZÞð1þtmÞð9Þ

The implications of this constraint are again reasonable andintuitive. Given an equal distribution of depreciation between the

G.K. Fellows / Energy Policy 39 (2011) 1505–15131510

two periods (Z¼0.5) the margin on the rate of return is restricted toless than or equal to 0 in order to maintain a Pareto improvement(or, more accurately Pareto equivalence). As the proportion ofdepreciation expense falls in period 1(Zk), the Pareto improvingmargin increases, consistent with the increased margin being atrade-off for the lower depreciation expense in the earlier period.This result is extensible to multiple period (finite) models.

Another necessary condition for the existence of a Paretoimproving incentive agreement illustrated by Eq. (9) is

0orsð1þtÞ�rpð1þtmÞ ð10Þ

There are logical arguments as to why the cost of capital of theshipper would be higher than that of the pipeline. Shippers face higherrisks as many of them are exploratory companies, and pipelines aregenerally very long lived. In addition, m (the equity thickness) isobserved to be between 0.2 and 0.4 across the NEB’s group 1 pipelines.This constitutes ample evidence to support satisfaction of theseconditions, and further substantiates the existence of an NS that isPareto improving for the current shippers and pipeline, based ondeferral of depreciation and a revenue sharing scheme.

Through this framework it is shown that negotiated settle-ments, if they make feasible both a deferral of depreciation and amargin on the cost of capital are able to produce an outcome thatsimultaneously increases the net present value of the pipeline’sprofits and minimizes the net present value of transportation costsfaced by long and short lived shippers who are parties to thesettlement.

5. Econometric evidence

The financial data available for econometric modelling andestimation of the effects of NSs on the speed of depreciation comesfrom surveillance reports filed by the pipelines with the NEB. Theaccounting values reported to the NEB are maintained by thepipelines under the federal Toll Information Regulations,11 the GasPipeline Uniform Accounting Regulations,12 and the Oil PipelineUniform Accounting Regulations13 (the last two for Gas and Oilpipelines, respectively).

It is not possible to reconstruct an accurate measure ofcomposite depreciation rates from the available dataset due to alack of data on plant additions, retirements and working capital forall pipelines in the dataset. Due to this, the dependant variable (Y)in the model is a measure of the depreciation expense (in dollars)charged by each pipeline in each period.

The simple empirical model to be estimated can be written as

Yit ¼ b0iLitþb1

Lit Ratebaseitþb2Lit Periodit

þd bNS0i �b

Lit0i þ bNS

1 �bLit1

� �Ratebaseitþ bNS

2 �bLit2

� �Periodit

h iþuit

ð11Þ

where Y is the depreciation expense; b0i is the intercept; and b1 isthe effect on depreciation expense of a change in rate base;

b2 ¼CT¼

CThe expected economic life of the pipeline ðthe planning horizonÞ

� �

C is the relationship between the time trend and accrued depre-ciation; i is the index of pipelines in the dataset; and t is thetime index.

11 Online: http://laws.justice.gc.ca/en/showdoc/cr/SOR-79-319/bo-ga:s_3/20-

090804/en#anchorbo-ga:s_3, last accessed November 2009.12 Online: http://laws.justice.gc.ca/en/showdoc/cr/SOR-83-190/sc:3/200907-

30/en#anchorsc:3, last accessed November 2009.13 Online: http://laws.justice.gc.ca/en/showtdm/cr/C.R.C.-c.1058, last accessed

November 2009.

The fixed cost being depreciated can be broken into constituentelements, which are linear in parameters

F ¼ Rate baseþAccumulated depreciation�Working capital

A time trend is included in the empirical model to be estimatedas a proxy for accumulated depreciation since it is expected that,even with periodic additions and retirements, the amount ofaccumulated depreciation to the accounts of in service assets willincrease with persistence through time. For clarity, the accumu-lated depreciation variable cannot be reconstructed from theobserved depreciation expenses in the dataset for two reasons.First the depreciation expense is only observed after 1993 andmuch of the sample has an in service date before 1993 and second,adding up the observed depreciation ignores the effect of retire-ments on the aggregate fixed costs of in service assets.

A proxy for accumulated depreciation is necessary to maintainunbiased estimation. Given that the model is constructed toidentify changes in a depreciation expense set via the straight linemethod, accumulated depreciation is expected to have a positiveeffect on the depreciation expense. This occurs since the effect ofaccumulated depreciation in the model exists to offset the persis-tent drawing down of the rate base through time in constructing aproxy measure of the book value of the investments (fixed cost)currently being depreciated.

Since the treatment effect occurs in later time periods, failure toaccount for accumulated depreciation would place its effect in theerror term, and the corresponding omitted variable bias wouldaffect the other coefficient estimates on the treatment effect.

The rate base and a time trend together act as a proxy for fixedcosts without adjustment for working capital. Unfortunately noproxy is available for working capital, and so it becomes acomponent of the error term. This does not represent significantdamage (bias) in the empirical results. Working capital is not likelyto be related to other explanatory variables for a single pipeline so itis not expected that its omission represents a significant omittedvariable bias.

Note that the intercept b0i is indexed by i. The practical modelemployed below accounts for fixed effects by allowing the inter-cept to vary across pipelines. This is necessary to maintainconsistency in the coefficient estimates due to the existence ofpipeline level fixed effects in the data.14 The use of a pooled OLS orrandom effects model would produce inconsistent estimates underthe conditions explained to this point. Both procedures would failto capture elements of firm heterogeneity beyond those explainedby the rate base (these can be significant due to regional differencesbetween pipelines in Canada).

Since working capital is an element of the rate base but not anelement of the fixed cost being depreciated a certain amount ofnoise is expected in the signal from the rate base variable. Asexplained by Griliches and Hausman (1986) a fixed effects model inthis type of application will lower the signal to noise ratio due to thestrong serial correlation of the explanatory variable and may causea bias in the coefficient estimate. Regardless, a fixed effects model isstill the most appropriate, as the parameter of interest is the effectof treatment on the treated and not the coefficient on rate base.

In Table 1, all financial figures (that is the coefficients on ratebase and period) are in $1000s. Recall that the dependant (right-hand side) variable is the reported depreciation expense of thepipelines.

As expected given the discussion above, the estimated treat-ment effect (the effect of a movement to NS) is negative across all

14 The Hausmann test for fixed effects vs. random effects illustrates the

existence of the former. The statistic is not shown here.

Table 1OLS estimates with fixed effects and Driscoll Kraay standard errors.

1 2 3 4 5

Depreciation Depreciation Depreciation Depreciation Depreciation

NS (treatment dummy) �13176.57 [0.375] �15024.90 [0.032]n �13761.51 [0.033]n �17000.13 [0.022]n

Rate base 0.009 [0.419] 0.008 [0.502] 0.010 [0.414] 0.011 [0.257] 0.004 [0.675]

NS� rate base 0.00083 [0.864] 0.00036 [0.936] 0.00087 [0.856]

Time trend 3486.91 [0.028]n 3930.13 [0.007]nn 3410.87 [0.005]nn 3362.18 [0.005]nn

NS� time trend �213.01 [0.879] �1470.69 [0.040]n

constant 28,299.19 [0.169] 26,567.93 [0.168] 28,625.89 [0.157] 27,211.31 [0.165] 78,465.50 [0.000]nn

corr(U_i, xb) 0.905 0.8807 0.9101 0.9213 0.4521

Within R2 0.2784 0.2692 0.2782 0.2757 0.3926

Observations 101 101 101 101 101

# of groups 8 8 8 8 8

p-values in brackets.

– All financial figures in $000’s.

– The model in column 5 includes dummy variables for each period.

– Coefficient estimates for time period dummies are not given in the table.

n Significant at 5%.nn Significant at 1%.

Table 2Summary statistics for model variables.

Variable Obs Mean Std. dev. Min Max

Depreciation 102 63438.12 116885.8 2911.154 421,971

NS 103 0.592233 0.4938225 0 1

Rate base 103 1,434,750 2,748,297 21,403 9,404,498

G.K. Fellows / Energy Policy 39 (2011) 1505–1513 1511

the models estimated (a reduction in depreciation expense result-ing from the movement from litigation to NS).

The coefficient estimates on the treatment dummy variable areconsistently negative as are the coefficient estimates on the interac-tion terms between the treatment and the time trend variable(d�period). The model in column 4 (qualitatively the most appro-priate model, explained below) predicts a reduction in the deprecia-tion expense of $14 million (22%) in any year where tolls are set via anegotiated settlement rather than a traditional hearing.

While the coefficient estimates on the interaction term on ratebase (d� rate base) are positive they are small by comparison to othercoefficient estimates. The positive effect of the rate base interaction ismore than offset by the treatment dummy and period interactions foreven the largest pipeline in the dataset (TCPL). Additionally, thecoefficient estimates for the interaction between d and rate base arenot statistically significant (denoted by the high p-values). Given theestimated negative effect of treatment on the depreciation expensethere is econometric support for the theoretical assertion that theincreased freedoms granted by NSs over litigation allow for, andprovide incentive to, defer depreciation expense.

The time trend (recall, it is a proxy for accumulated depreciation)carries an estimate which is consistent with expectations. Thepositive estimate is expected since accumulated depreciation on inservice assets is growing in each period which is a result of thephysical pipe being the longest lived and largest fixed cost. Thisgrowth will have a positive effect on depreciation expense insofar asit balances the draw of accumulated depreciation from the rate base.

Table 2 presents some summary statistics for the variablesincluded in the econometric estimation, useful for interpreting themagnitude of the predicted treatment effect.

Comparing the values in Table 2 to the estimates in Table 1, theeffect of NS is in the order of an approximate average 22% fallin depreciation expense. However, it is important to note thatthese results are strongly influenced by the inclusion of the TCPLmainline in the dataset. Dropping each unit individually and

re-estimating the model in column 4 from Table 1 with the other7 units, the influence of the TCPL mainline is illustrated in contextin Table 3.

Removal of the TCPL mainline both increases the p-value on theNS coefficient and reduces the estimated treatment effect by asignificant margin compared with the removal of any other unit.This severely limits the strength of these results as evidence of theeffect of negotiated settlements on depreciation rates.

Note that dropping the TCPL mainline has an effect on thesummary statistics from Table 2 such that the approximate averagefall in depreciation expense is estimated at about 2.75% (p-value of0.152) absent TCPL from the dataset. The estimation results(statistically significant or not) are still generally supportive ofthe above hypothesis (the coefficient estimates on the treatmentvariable are unambiguously negative in all cases). While not ideal,the results of this estimation are nonetheless encouraging given thesevere limitations of the dataset.

A graphical depiction of the data used for estimation furtherillustrates the reasons for the sensitivity of the estimation to theinclusion of TCPL. Graph 1 suggests that much of the result is beingdriven by the spike in TCPL’s depreciation expense concurrent witha movement back to litigation from NS in 2003.

In some ways TCPL is a special case; in 2001 when TCPL wasunable to negotiate an unopposed settlement the NEB convened ahearing to review the contested settlement. As explained above,one result of the hearing was an increase of 0.20% (2.64% in 2000 to2.84% in 2001) in the composite depreciation rate, with a furtherincrease of 0.15% to occur in 2002. This depreciation increase waslargely prompted by reductions in firm-contract renewals in thelate 1990s and as stated by TCPL, changes in the West CoastSedimentary Basin (gas fields that comprise the pipeline’s feedstock) supply and in the end market demand. These firm-contractnon-renewals also led to general revenue shortfalls which were tobe paid for 2/3rds by shippers (via an increase in the next periodtolls) and l/3rd via the pipeline. The disposition of the pipeline’s

TCPL Mainline Depreciation Expense

0

50000

100000

150000

200000

250000

300000

350000

400000

450000

1993 2000Year

$000

s

Denotes a year in which the pipeline tolls wereset via NS.

Denotes a year in which the pipeline tolls wereset via traditional litigation.

1994 1995 1996 1997 1998 1999 2001 2002 2003 2004 2005 2006 2007 2008

Graph 1. TCPL mainline depreciation expense.

Table 3Dropped panel OLS estimates with fixed effects.

Dropped TCPL main TCPL BC M&NP TQM Zone 6 Zone 7 Zone 8 Zone 9

NS �403.9 [0.152] �19446 [0.076] �14419 [0.028]n �22012 [0.001]nn �13371 [0.061] �13294 [0.060] �13569 [0.048]n �13303 [0.062]

Time trend 312 [0.001]nn 4162 [0.021]n 3504 [0.006]nn 4584 [0.000]nn 3532 [0.060]nn 3575 [0.006]nn 3519 [0.007]nn 3560 [0.006]nn

Rate base 0.031 [0.000]nn 0.012 [0.214] 0.011 [0.264] 0.013 [0.152] 0.011 [0.258] 0.011 [0.247] 0.011 [0.253] 0.011 [0.252]

Constant 3532 [0.004]nn 26135 [0.250] 28559 [0.161] 23204 [0.246] 30983[0.145] 32091 [0.140] 31554 [0.150] 30577 [0.153]

Observations 85 88 93 101 87 87 87 87

# of groups 7 7 7 7 7 7 7 7

p-values in brackets.

All financial figures in $000’s.

n Significant at 5%.nn Significant at 1%.

G.K. Fellows / Energy Policy 39 (2011) 1505–15131512

l/3rd was to be agreed upon by the tolls task force with the NEBhaving final say.

The modest increases in depreciation rates in 2001 and 2002were followed by larger increases in the RH-1-2002 hearing whichdid not include any settlement elements and increased thedepreciation rate to 3.65%. In addition to allowing this increase,the NEB chose to eliminate some (but not all) of the incentiveschemes negotiated in prior hearings. These schemes are difficult tomodel formally, but in general they served to provide cost savingincentives to TCPL through channels such as fuel usage and costs.Absent the profits from these incentive schemes it appears thepipeline lost incentive to defer depreciation expense, and began tore-accelerate depreciation in order to ensure full depreciation overthe life of its assets and avoid additional risk.

While conceptually difficult to model or estimate empirically,the regulatory outcomes over the early 2000s also produced otheroutcomes beyond altered depreciation schedules. Most notably theTCPL’s return on equity was re-examined at length and the pipeline’sdebt/equity ratio was increased from 30% to 36% reflecting an in-crease in business risk from the reduced contract renewals and theperceived decline in West Coast Sedimentary Basin supply (Doucetand Littlechild, 2009). In light of the breakdown of negotiations theNEB also implemented alternative dispute resolution mechanismsand released a new set of guidelines for negotiated settlements

that specifically allowed for the possibility of contested settlementoutcomes.

6. Concluding remarks and additional considerations

The results found here are nearly identical to those fromLittlechild’s (2009a) investigation into the Florida electricity sector.In both cases rate reductions based on deferred depreciation arefound to be a significant source of dominance of NSs in a regulatedutility sector.

With regards to the firms’ choice of depreciation path; themodel presented independently reaches much the same conclusionas Burness and Patrick (1992); however, the notable addition madehere is that deferred depreciation is shown to be beneficial to thecurrent consumers as well as the regulated firm (whether or not theconsumers plan to contract with the regulated firm in the future).

The existence of deferred depreciation expense in the NEB case issuggested by the econometric analysis and anecdotal evidenceabove. The theoretical model also identifies deferral as a cause ofrate reductions while simultaneously identifying the conditions(which are both plausible and probable given anecdotal evidence)necessary for this argument to be logical in the explanation of therevealed preference of pipelines and shippers for NSs over litigation.

G.K. Fellows / Energy Policy 39 (2011) 1505–1513 1513

The mathematical framework here deals with only two periods,and is constructed to show, as simply as possible, the preference ofpipeline and shippers to defer depreciation under the assumption ofa margin on the cost of capital. The comparison drawn in thisexercise is between the prevalent straight line depreciation rate thatis standard in litigated outcomes, and the modifications to thisdepreciation rate resulting from the move to negotiated settlements.

More work is needed to completely evaluate the policy implica-tions of this result. One of the NEB’s own guidelines for negotiatedsettlements prohibits a settlement from infringing upon any publicinterest consideration beyond the immediate concerns of thenegotiating parties; however, the deferral of depreciation expenseis doing just that by increasing the burden on future consumersrelative to the litigated outcome. Costs are shifted onto futureparties not currently involved in a negotiation.

An additional consideration in this context that is not discussedhere is the probable lack of protection of the interests of smallerparties. It is expected that interests of smaller and less wellorganized parties (like household consumers) would be protectedby NEB oversight in litigated outcome. In moving to a negotiatedsettlement these parties may be disadvantaged by a lack ofbargaining power or may not have an opportunity to be includedin the negotiation procedure at all. This area is a prime candidate forfuture examinations related to existing literature on bilateralmarkets and bargaining power.

From an environmental standpoint further investigation intonegotiated settlements may also provide interesting policy impli-cations. The NEB has become more interested in end of life issuesand the magnitude of terminal negative salvage (dismantlement)costs. The risk of stranding assets and being unable to safely andenvironmentally decommission a pipeline due to a lack of fundsresulting from incomplete depreciation will unambiguouslyincrease with any deferral of depreciation. The issue of alteringthe depreciation path is also more generally related to the socialassessment of future environmental damage through both theeffect on the consumption path (through resulting changes in tolls)and the risk of the pipeline in general.

There is also the possibility that deferring depreciation mayincrease the opportunity cost of capital incurred by the pipeline,due to the relatively weaker perceived financial position and theincreased risk. A further investigation of differing investmentincentives and the implications for maintenance schedules (eco-nomically efficient decay vs. overspending on upkeep) under thetwo regulatory regimes may produce interesting results as well.This would likely follow from Rogerson’s (1992) investigation ofregulatory lag’s effect on the incentive to invest in inputs thatminimize the present value of costs allocated to the period of lag.

Despite the lack of a comprehensive efficiency discussion,through examination of an additional industry in a differentcountry using new data, the above stands as an expansion ofLittlechild’s (2009a) thesis. Also, despite the differences in themodel structure, this paper expands on the Burness and Patrick(1992) conclusion by illustrating the ability of a deferred deprecia-tion schedule to be beneficial to current consumers as well as theregulated firm.

References

Averch, Johnson, 1962. Behaviour of the firm under regulatory constraint. TheAmerican Economic Review 52, 5.

Baumol, 1971. Optimal depreciation policy: pricing the products of durable assets.The Bell Journal of Economics and Management Science 2 (2), 638–656.

Buchmann, Tongren, 1996. Nonunamimous settlements of public utility rate cases:a response. Yale Journal on Regulation 13, 337–345.

Burness, Patrick, 1992. Optimal depreciation, payments to capital, and naturalmonopoly regulation. Journal of Regulatory Economics 4 (1), 35–50.

Doucet, Littlechild, 2006. Negotiated settlements: the development of economic andlegal thinking. Utilities Policy 14 (Dec), 266–277.

Doucet, Littlechild, 2006w. Negotiated settlements and the National EnergyBoard of Canada Electricity Policy Research Group Working Paper No.EPGR 06/29.

Doucet, Littlechild, 2009. Negotiated settlements and the National Energy Board inCanada. Energy Policy 37, 4633–4644.

Griliches, Hausman, 1986. Errors in variables in panel data. Journal of Economics 31(1), 93–118.

Harris, 1993. Mineral Resource Stocks and Information. Handbook of NaturalResource and Energy Economics vol. III.

Littlechild, 1970. Marginal cost pricing with joint costs. The Economic Journal 80(318), 323–335.

Littlechild, 2008. Some alternative approaches to utility regulation. EconomicAffairs 28, 3.

Littlechild, 2009a. The bird in hand: stipulated settlements and electricity regulationin Florida. Utilities Policy 17, 3–4.

Littlechild, 2009b RPI-X Regulation: Ofgem’s RPI-X@20 Review and the Scope forMore Customer Involvement, Network, the Australian Competition and Con-sumer Commission for the Utility Regulators Forum. 34 (Dec).

Morgan, 1978. Toward a revised strategy for ratemaking, Law Forum, University ofIllinois, vol. 1, pp. 27–78.

Rogerson, 1992. Optimal depreciation schedules for regulated utilities. Journal ofRegulatory Economics 4 (1), 5–33.

Sappington, 2005. Regulating service quality: a survey. Journal of RegulatoryEconomics 27 (2), 123–154.

Schultz, 1999. Light-handed regulation. Alberta Law Review 37 (2), 387–418.Stigler, 1976. The Xistence of X-Efficiency, The American Economic Review, Vol. 66,

No. 1 (Mar).Sweeney, 1993. Economic theory of depletable resources: an introduction, Hand-

book of Natural Resource and Energy Economics, vol. III.Vollman, 1996. Toward incentive regulation of Canadian pipelines. In: Proceedings

of the Fifth Annual DOE-NARUC Natural Gas Conference, St Louis, Missouri, April28–May 1, available at NEB website.

Walker, 1986. Settlement practice at the FERC: boom or bane. Energy Law Journal 7,343–359.

Canadian Government Publications

Gas Utilities Act: Default Gas Supply Regulation, Alberta Government Publication.NEB, 1987. RH-3-86, Tolls application. TransCanada PipeLines Limited May.NEB, 1991. RH-3-91, Tolls application. Trans Mountain Pipe Line Company Ltd.

June.NEB, 1992. RH-1-91, Tolls application. TransCanada PipeLines Limited September.NEB, 1993. RH-2-92, Tolls application. TransCanada PipeLines Limited February.NEB, 1994. RH-1-93, Application for Certain Orders Respecting Tolls and Tariffs.

Foothills Pipe Lines Ltd. November.NEB, 1995, RH-2-94, reasons for decisions, multi-pipeline (Cost of Capital),

March.NEB, 1995. RH-2-95 Summary, Compilation of Key Documents Related to RH-2-95.

TransCanada PipeLines Limited June.NEB, 1996. RH-2-95, Tolls application. TransCanada PipeLines Limited. December.NEB, 1998. RHW-1-98, Framework for Light Handed Regulation Summary.

Westcoast Energy Inc. June.NEB, 2002. RH-1-2001, Tolls application. TransCanada PipeLines Limited November.NEB, 2003. RH-1-2002, Tolls application. TransCanada PipeLines Limited July.NEB, 2009. File OF-Tolls-Group 1 -M124-Fin Aud 02, Board Letter and Regulatory

Audit Report 2007 to 2008. Maritimes & Northeast Pipeline Management Ltd.September.

NEB, 2006, 2007, 2008, 2009. The Canadian Pipeline Transportation system; Queen’sPrinter, Government Publication.

National Energy Board, 2002, revised. Guidelines for negotiated settlements oftraffic, tolls and tariffs, Online at: /https://www.neb-one.gc.ca/ll-eng/Livelink.exe/fetch/2000/90463/157025/ 208496/ A0E4C1_-_Letter_Decision.pdf?nodeid=208497&vernum=0&redirect=3S (last accessed December 2009).

National Energy Board Cost Recovery Regulations, 1991. Queen’s Printer. CanadianGovernment Publication.

The National Energy Board Act, 1958. Queen’s Printer. Canadian GovernmentPublication.