Decision No. 425-R-2011

• BACKGROUND
• AGENCY DETERMINATIONS OF COST OF CAPITAL
• DECISION FORMAT
• GENERAL COMMENTS AND POSITIONS OF PARTICIPANTS
• ISSUES AND AGENCY FINDINGS
• ISSUE RELATED TO NET RAIL INVESTMENT
  • Issue 1: Use of market values, instead of book values, to establish the asset base/net rail investment
• ISSUES RELATED TO CAPITAL STRUCTURE
  • Issue 2: Capital structure weights - book value versus market value
  • Issue 3: Treatment of deferred taxes
• ISSUES RELATED TO ESTIMATING COST RATES
  • Issue 4: Determining the cost rate of debt
  • Issue 5: Methodology for assessing the cost rate of common equity
  • Issue 6: The CAPM version to be used
  • Issue 7: Assessment and application of a grain risk premium
• ISSUES RELATED TO ELEMENTS OF THE CAPM METHODOLOGY
  • Issue 8: Appropriate market data
  • Issue 9: Risk-free rate
  • Issue 10: Averaging period for the market risk premium
  • Issue 11: Averaging methodology for the MRP
  • Issue 12: The use of income versus total returns for bonds in the MRP
  • Issue 13: Beta return interval and analysis period
  • Issue 14: Adjusting beta for mean reversion
  • Issue 15: Unlevering and relevering beta
• DECISION CONCLUSION
• APPENDIX A
• APPENDIX B

BACKGROUND

[1] In 2009, the Canadian Transportation Agency (Agency) initiated a review of the methodology it uses to determine the cost of capital for federally-regulated railway companies. The review was undertaken because more than five years had passed since the existing approach was reviewed, and the Agency considers it good practice for a regulatory body to periodically undertake such substantial reviews of the methodologies it uses.

[2] In this context, the Agency considered it timely to undertake a review of its existing cost of capital methodology to assess whether the methodology continues to employ the most reasonable, reliable and pragmatic approach possible for determining the cost of capital for federally-regulated railway companies. The goals set by the Agency were to determine whether there is a clearly superior methodology to, or if there are improvements that would clearly improve the Agency's existing cost of capital methodology and to adopt the cost of capital methodology that the Agency will use for a minimum of five years.

[3] The review has been conducted in two phases: a study phase and a hearing phase. In the study phase, after a consultative process involving a wide range of stakeholders to develop the terms of reference for the study, an independent consultant examined existing cost of capital methodologies and principles, and reviewed the Agency's current cost of capital methodology, as well as the cost of capital methodologies used by other economic regulatory bodies.

[4] This phase resulted in a report entitled Review of Regulatory Cost of Capital Methodologies, produced by the Brattle Group (the Brattle Report). The Brattle Report identifies and examines the strengths and weaknesses of a variety of existing cost of capital methodologies and models, and assesses potential issues and implications associated with the implementation of such models.

[5] In Decision No. LET-R-185-2010, the Agency initiated the hearing phase of the review and a consultation with a broad list of stakeholders, including railway companies, shippers, producers and others, on certain recurring issues identified by the Agency for discussion. Interested persons were given the opportunity to express their points of view on these issues and any other issues they thought pertinent to the Agency's examination, as well as to provide comments on the Brattle Report.

[6] The following made submissions:

Canadian National Railway Company (CN), with expert testimony provided by Ronald M. Giammarino, Ph.D. and Murray Carlson, Ph.D., both professors of Finance at the Sauder School of Business at the University of British Columbia. CN is a Class 1 federal railway company, involved in the movement of freight.

Canadian Pacific Railway Company (CP), with expert testimony provided by Bruce E. Stangle, Ph.D. and George Kosicki, Ph.D. from Analysis Group Inc. CP is a Class 1 federal railway company, involved in the movement of freight.

Canadian Canola Growers Association (CCGA) is a producer organization. It indicates that it represents approximately 50,000 canola farmers on national and international issues and policies that affect the profitability of canola growers.

Coalition of Rail Shippers (CRS), with expert testimony from Lawrence Kryzanowski, Ph.D., Full Professor of Finance and Senior Concordia University Research Chair in Finance at Concordia University. CRS is an affiliation of shipping industry associations that have been together since 2005. CRS indicates that its members account for over 80 percent of the revenues of CN and CP, and that it provides input to government on matters affecting Canadian rail freight transportation.

Western Canadian Shippers' Coalition (WCSC), with expert testimony from Lawrence I. Gould, Ph.D., Professor of Finance at the University of Manitoba. WCSC represents companies and associations involved in the transportation of Canadian natural resource-based products including: barley, cement, chemicals, coal, lumber, metals, newsprint, oilseed products, pulp and paper, sulphur, wheat, and wood pellets. WCSC indicates that its members provide over 320,000 direct and indirect jobs for Canadians in communities across the west and ship in excess of $35.5 billion worth of products annually.

Alberta Transportation, Province of Alberta (Alberta)

Manitoba Infrastructure and Transportation, Province of Manitoba (Manitoba)

Ministry of Highways and Infrastructure, Province of Saskatchewan (Saskatchewan)

AGENCY DETERMINATIONS OF COST OF CAPITAL

[7] Cost of capital is defined as an estimate of the total return on net investment that is required by debt holders and shareholders, so that debt costs can be paid and equity investors can be provided with a return on investment consistent with the risks assumed for the period under consideration.

Regulatory framework

[8] In regulatory environments, cost of capital is estimated for a broad range of industries and for an equally broad range of economic regulatory purposes, some very wide in scope, others more narrow. There are differing mandates and degrees of significance to determinations of cost of capital among various regulatory jurisdictions and industry contexts. These influence the methodological choices each regulator considers, as does the regulatory history and the prevailing legislative and economic environment unique to each jurisdiction.

[9] Section 5 of the Canada Transportation Act, S.C., 1996, c. 10, as amended (CTA), sets out the national transportation policy for Canada and states, in part, that regulation and strategic intervention are used to achieve economic outcomes only when they cannot be achieved satisfactorily by competition and market forces. Accordingly, the Agency makes annual cost of capital rate determinations for its regulatory purposes within a context of economic regulation of railway companies in Canada that relies primarily on market forces to govern relationships between railway companies and shippers. The mandate of the Agency is limited to setting cost of capital rates for only certain targeted purposes and the Agency, unlike some other regulatory bodies, is not required to set the cost of capital rates for the entire operations of the regulated party.

[10] More specifically, the cost of capital rates for federally-regulated railway companies factor into a number of prescribed statutory and regulatory applications under the CTA. The Agency calculates cost of capital rates for three main purposes: (1) the transportation of western grain; (2) interswitching; and, (3) other specified regulatory purposes.^{Note 1}

[11] Most significant of the three, from a practical standpoint, is the cost of capital rate calculated for the transportation of western grain. In that instance, the cost of capital rate establishes an appropriate return on railway companies' capital investments and forms part of the volume-related composite price index (VRCPI). The VRCPI is, in turn, a component in the annual calculation that establishes the maximum revenue entitlement for the movement of western grain by rail (referred to as the Revenue Cap).

[12] An annual cost of capital rate is also determined for use in the development of interswitching costs and rates and is applied in the Agency's costing model to determine the cost of interswitching activities or intermediate processes that employ capital, such as investments in infrastructure, rolling stock and maintenance and work equipment.

[13] A cost of capital rate is also calculated for regulatory purposes other than the transportation of western grain and interswitching. This cost of capital rate is one of the major inputs for the estimate of unit costs of railway activities and intermediate processes which, in turn, are used to determine the cost of a railway movement or service. The regulatory purposes that require such cost determinations include, among others, rate determinations for access and other rail services to be paid by a rail passenger service provider that uses the rail network, facilities and services of another railway company; the rate to be paid for running rights on another railway company's network; the establishment of competitive line rates and joint tariffs; cost apportionments for the maintenance and construction of road crossings; service and noise disputes where cost is a factor; and, provision of technical costing assistance to arbitrators in final offer arbitration (FOA) proceedings between a shipper and a carrier.

[14] With respect to its determinations of cost of capital, the Agency emphasizes the following points.

• There are fundamental differences between regulatory regimes, from unregulated, to deregulated, to heavily regulated. Some regulators determine capital project approvals and make rates for significant portions of a regulated company's operations, some regulate industries with a large number of diverse stakeholders, some regulate emerging markets and some forecast cost of capital rates several years into the future.
• The Agency's prescribed mandate with respect to cost of capital is narrow when compared to the full spectrum of regulators and their applications of cost of capital. For the most part, the Agency is regulating, for not more than one year in advance, small, well developed subsets of the operations of two mature companies, which have well established capital bases in a stable industry.
• The Agency's regulatory goal is to establish fair and reasonable rates of return on capital for federally-regulated railway companies for the sole purpose of Agency statutory and regulatory applications. Therefore, the Agency does not consider how else the rates may be used by railway companies, shippers or others, such as arbitrators, for their own purposes, for example in commercial negotiations or in arbitration proceedings.

Current methodology

[15] The principles currently used by the Agency in determining cost of capital rates for federally regulated railway companies were established in the following Decisions:

[16] Underlying all of these Decisions is a principle set out in the Railway Costing Regulations, which came into effect on December 10, 1980. Specifically, paragraph 7(b) of those Regulations requires the Agency, when setting rates for the carriage of goods, to apply the associated rate to the "variable portion of the net book value of the asset."

[17] The Agency's principal determinations of cost of capital rates are based on confidential submissions made by the two current Canadian Class 1 freight railway companies, CN and CP. The railway companies' submissions are based on their most recently completed fiscal year, which, for both railway companies, runs from January 1 to December 31.

Current cost of capital determination process

[18] The cost of capital determination process consists of four distinct steps:

1. Determination of net rail investment;
2. Determination of capital structure;
3. Determination of capital structure cost rates, which includes the cost rate of debt, deferred taxes and common equity; and,
4. Calculation of the cost of capital rate.

Net Rail Investment

[19]  The net rail investment is defined as the gross book value of all railway assets less accumulated depreciation. This first component defines the portion of the railway company's net assets that are providing railway transportation services and are under Agency jurisdiction. The net rail investment includes an amount for working capital. CN and CP make annual submissions regarding net rail investment, based on book values from their most recent financial statements, with certain approved adjustments. These submissions are verified by the Agency.

Capital Structure

[20]  The capital structure refers to the combination of the various sources of capital used to finance the net rail investment. In broad terms, funding can be achieved through borrowing, issuance of debt instruments, deferred taxes and shareholders' equity. Each year, CN and CP submit their actual capital structures, based on book values from their most recent financial statements, with certain Agency approved adjustments. These submissions are verified by the Agency.

Capital Structure Cost Rates

[21]  The cost of debt is taken as the actual cost of the debt, that is, the interest paid to financial institutions or bond holders for loans made to the railway companies, as recorded in the most recent financial statements of the railway companies and submitted to the Agency. These submissions are verified by the Agency. Deferred taxes are allocated a zero cost rate.

[22] The railway companies also make submissions proposing a cost rate for common equity. The Agency assesses the submissions, makes any changes necessary to adhere to its approved methodology and calculates the cost of common equity using three financial models; the Capital Asset Pricing Model (CAPM), the Discounted Cash Flow (DCF) Model and the Equity Risk Premium (ERP) Model. The Agency then assesses which model or combination of models best reflects the state of capital markets in that year. In each year since the 1997 Decision, the Agency has applied the results of the CAPM alone.

[23] An income tax allowance, based on the railway companies' submitted statutory federal and provincial income tax rates, is added to the cost of equity to establish the before tax value of the shareholders' return. No income tax allowance is applied to the interest paid on debt as it is income tax deductible.

[24] In conjunction with cost of capital rate determinations for the transportation of western grain, the issue of whether a risk adjustment for grain transportation should be included in the cost of equity is also assessed each year. Prior to the 1997 Decision, the Agency applied a grain risk adjustment of minus one percent to the cost of common equity. Subsequent to the 1997 Decision, no grain risk adjustment has been applied. However, the applicability of such an adjustment has been considered and determined annually.

Weighted Average Cost of Capital Rate

[25] The proportion of each type of funding in the capital structure is used to weight each cost rate and the sum becomes the cost of capital rate expressed in percentage terms. When this rate is applied to the net book value of the assets involved, it results in the cost of capital in dollar terms.

DECISION FORMAT

[26] There are three parts to this document, the decision and two Appendices, all of which together constitute the full Decision.

[27] The decision component consists primarily of the Agency's analysis and findings for the various aspects of determining cost of capital examined in the course of the review.

[28] Appendix A – Agency Cost of Capital Methodology provides an outline of the cost of capital methodology being adopted by the Agency pursuant to this Decision, including formulae and data sources.

[29] Appendix B – Summary of Methodological Issues provides the context and relevance for each of the issues under review, as well as a detailed summary of the participants' submissions. It also includes the practice of other Regulators and excerpts from the Brattle Report, where applicable.

GENERAL COMMENTS AND POSITIONS OF PARTICIPANTS

[30] In addition to responding to the specific issues raised in the Consultation Document that are addressed later in the Decision, some participants made certain general comments regarding the Agency's cost of capital methodology and the rates it has produced, and expressed concerns either about maintaining the existing methodology or the impact of potential future changes to the methodology. The following section summarizes these general points of view.

Canadian Pacific Railway Company

[31] CP is of the view that the Agency's current methodology produces a result that falls short of the standard laid out by the Agency in its consultation document, that "the cost of capital is the total return on net investment that is required by shareholders and debt holders so that debt costs can be paid and equity investors can be provided with an adequate return on investment consistent with the risks assumed for the period under consideration."

[32] CP submits that the regulated cost of capital rates issued by the Agency have historically been so low that they bear little resemblance to the rates the Agency is trying to approximate. Citing the rate set for 2009 as the latest and most dramatic example, CP submits that on an after-tax basis, this rate was well below the rate at which CP was even able to issue debt. It is CP's position that this disconnect is further demonstrated by comparing the cost of equity rates determined by the Agency over the last 10 years to the consistently higher rates set by the U.S. Surface Transportation Board (STB) for U.S. railroads.

[33] CP argues that this imbalance puts the sustainability of its infrastructure and its ability to serve its customers at risk. CP submits that it requires a return that is sufficient to attract capital for replacement, modernization and demand-driven capacity expansions. It states that inadequate returns impact its ability to properly invest in productivity, fluidity and reliability, which thereby reduces the competitiveness of Canadian commodities. CP maintains that ensuring a fair and reasonable return is the only way the national transportation policy's objective of a competitive, economic and efficient national transportation system can be met.

[34] CP makes particular reference to the cost of equity rates set by the Agency, submitting that they are essentially equal to the cost of debt, and that any methodology that yields such unreasonably low cost of equity cannot be consistent with the objective of providing a fair and reasonable return.

[35] CP states that cost of capital estimates below an appropriate level have the result of providing insufficient revenues and underinvestment in productive assets with respect to the revenue cap, final offer arbitrations, interswitching rates and other regulated activities. It submits that the consequences to underinvestment induced by inappropriately low cost of capital rates are decreased productivity, decreased service levels and longer shutdown periods due to inoperable infrastructure.

[36] To alleviate what it perceives to be an illogically low cost of capital rate, CP proposes an alternative methodology that it considers would result in a cost of capital estimate that is more reflective of the fundamental economic conditions of the Canadian railway industry and would result in greater incentives for railway investment and a more productive Canadian railway industry.

[37] In response to arguments made by other participants that the railway companies' "high profitability" implies that they are adequately compensated under the Agency's current methodology, CP indicates that between 2005 and 2010 it has spent, on average, approximately 17 percent of its revenues on capital programs, and in more than half of the last 20 years CP has experienced negative free cash flows.

Canadian National Railway Company

[38] CN makes full submissions for each of the issues identified by the Agency for discussion, suggesting alternatives which it considers more theoretically sound.

[39] CN submits that as cost of capital is the return investors would expect from a security of equivalent risk, it does not need to be modified out of deference to its impact on customers. CN submits that Courts have confirmed this, and it is CN's view that it would be inappropriate for a regulator to select a methodology or inputs into a methodology on the basis of this concern, or to resolve uncertainty in a methodology in favour of a lower return.

[40] CN also submits that the revenue cap regime provides none of the guaranteed revenue benefits associated with utilities while presenting all of the risk associated with a commercial undertaking, because revenues that fall below the revenue cap cannot be recovered but those above the cap must be returned and a penalty must be paid.

Western Canadian Shippers' Coalition

[41] WCSC describes the point of commonality for its group as being a reliance on market-dominant providers of rail freight transportation. WCSC submits that its member companies compete in world commodity markets against producers from the U.S., Asia, Europe, Scandinavia, Australia and South America and that rail freight transportation costs and reliability are major factors affecting whether its member companies succeed or fail.

[42] WCSC considers the cost of capital review to have potentially significant financial consequences for Canada's rail freight shipper community and the national economy. WCSC indicates that because of the problems that arise for captive shippers, its members are the largest users of the FOA process. While not ideal, it is at present the only reasonably effective remedy available to bulk commodity producers.

[43] WCSC submits that it is important that the benefits of this shipper protection measure are not undermined. WCSC further submits that negative changes to the cost of capital methodology will impact the efficacy of FOA and increase the market power advantage already enjoyed by the national rail carriers. WCSC also indicates that similarly, but on a larger scale, interswitching may be put at risk as a pro competitive measure for some shipments.

[44] WCSC argues that there is no obvious reason to change the Agency's current methodology. It provides economic theory and opinion to support the principles and technicalities underlying the methodology, and disagrees with the methodologies suggested by the railway companies. WCSC submits that the Agency's methodology is reasonable and fair to all and that the cost of capital rates determined by the Agency permit the railway companies adequate access to investment funds and provide financial viability. WCSC also indicates that, when the methodology is necessarily reviewed from time to time, adjustments to the established principles are preferable to revamping the established approach.

[45] WCSC submits that the review should concern itself with more than methodologies and take into account the impact changes may have on Canada's rail freight network in general. It cites the recent final report of the Rail Freight Service Review Panel's comments that: "...railways continue to have market power over some of their customers and ...there are sectors and regions where competitive alternatives are limited or lacking altogether. This railway market power results in an imbalance in the commercial relationships between the railways and other stakeholders."

[46] WCSC considers it important to look at this market power not only in the context of service levels but also as it applies to rate setting. Submitting that neither of the Class 1 railway companies is in financial difficulty, WCSC attributes this to their enjoying what the financial industry refers to as "pricing discipline." Further to this, WCSC describes pricing discipline as the ability resulting from market power to increase prices on a regular basis regardless of the general state of the economy or the impact that may have on their customers' competitiveness and rates of return.

[47] WCSC underscores its concern over the methodology changes being proposed by CN and CP, submitting that, absent operational changes, efficiency increases or growth or expansion of any sort and without any benefit to customers, the cost of rail freight transportation could increase by as much as $504 million/year. WCSC does not explain how it arrived at this figure but WCSC considers that this would be a gratuitous windfall, "resulting literally from the stroke of a pen," which would come directly from the pockets of the producers that are driving Canada's economic recovery.

[48] WCSC indicates that it would be difficult to justify, as well as to afford, any inappropriate changes to a stable and effective methodology for determining the cost of capital that already produces windfall profits, in excess of two times the cost of equity, to financially healthy railway companies.

[49] With specific reference to the decline in the cost of equity rates determined by the Agency in the last ten years, WCSC submits that CP has overestimated the decline and has not properly attributed it to the decline in interest rates. WCSC asserts that there is general agreement among financial experts that interest rates are a basic component of the cost of equity capital.

[50] WCSC also cites inconsistencies and errors in CP's claim that cost of equity rates are lower than the cost of debt and unusually low when compared with U.S. Class 1 railroads. WCSC further notes that there is no reason to expect the cost of equity rates based on a composite of four U.S. railroads to be the same as CN or CP, because the railroads are different, the capital markets are different, Canadian railway companies have not faced the same types of problems as the U.S. railroads, and CN and CP have not had difficulty in raising capital and are financially viable. WCSC provided data which it considers clearly demonstrates that CN and CP have consistently achieved returns that exceed the cost of capital determined by the Agency and are currently earning far in excess of the cost of equity capital.

Canadian Canola Growers Association

[51] CCGA submits that, in general, it views the Agency's current methodology as reasonable and that it meets the Agency's three criteria for evaluating cost of capital methodologies.

Alberta Transportation, Province of Alberta

[52] Alberta indicates that after reviewing the Brattle Report, as well as previous decisions made by the Agency on cost of capital issues, it does not see evidence of a clearly superior methodology to the existing CAPM that would warrant a change in the existing cost of capital methodology.

[53] Alberta highlights its concern that a departure from the exclusive use of the CAPM, to a DCF methodology, would unreasonably and adversely impact western grain shippers while unreasonably enhancing the earnings of CN and CP from the hauling of western grain.

Manitoba Infrastructure and Transportation, Province of Manitoba

[54] Manitoba expresses concern that the cost of capital review will result in adopting alternative cost of capital methodologies that will potentially increase costs to farmers and rail shippers at a time when the major Canadian railway companies are highly profitable and continue to grow. Manitoba supports its assertion regarding profitability with evidence for the years 2005 to 2011 of consistent double digit returns earned or expected to be earned by the railway companies, and the extent to which the railway companies' stocks have outperformed both the Toronto and New York stock exchanges.

[55] Manitoba considers that the revenues earned by the railway companies in the carriage of grain represent a direct deduction from the revenues received by grain producers, because the price of grain, unlike the rates charged by the railway companies, is not determined by, or even related to, the cost of producing grain, but rather by the interplay of demand and supply for the respective crops worldwide.

[56] Recognizing that the Agency has a statutory obligation to provide an adequate rate of return for the railway companies, Manitoba submits that the Agency also has a public interest obligation to ensure that the returns allowed are the very minimum required to meet the railway companies' capital return requirements. For this reason, Manitoba further submits that any ambiguity or uncertainty in the cost of capital determination should be decided in favour of a lower return.

[57] Manitoba also submits that because the Agency's overall costing model is not adjusted for rail productivity increases, it consistently overstates rail costs. Contrasting this with STB's quarterly adjustment of its rail costing metric for productivity increases, and the consistent record of increased rail productivity and declining costs demonstrated by that adjustment, Manitoba submits that this highlights a major issue thus far not addressed by the cost of capital review.

[58] It is Manitoba's view that because increasing rail productivity is a significant contributor to the railway companies' profitability from grain transportation, its effects should be recognized. Manitoba submits that the use of high throughput elevators, loading 50+ car blocks, and rail line abandonments have reduced costs for the railway companies. Manitoba further submits that these costs have been transferred to provincial governments, through road maintenance and short-line railway support, to farmers, through increased trucking costs, and to grain companies, through the construction of high throughput facilities. Manitoba urges the Agency to bear these issues in mind in making its capital cost assessments.

[59] Manitoba supports the Agency's current cost of capital methodology and submits that changes to that methodology should only be considered upon a more comprehensive costing review that would implement productivity adjustments in estimating the grain revenue cap. It considers a costing review to be a greater priority for government, to adjust the Agency's cost-based revenues to account for the extensive productivity gains and changed market conditions in railway operating costs over the last 18 years.

Ministry of Highways and Infrastructure, Province of Saskatchewan

[60] Saskatchewan recognizes that the federal government has established a variety of regulatory regimes to equalize the market power between shippers and railway companies, the most recent for western grain being the revenue cap. Saskatchewan also acknowledges that, while rail rate regulation has been established to protect shippers, there is a responsibility to ensure that the railway companies are adequately compensated for services being provided, which includes provision of an adequate rate of return to investors.

[61] Saskatchewan submits that, having reviewed and assessed the various options provided in the Brattle Report, it has not been clearly demonstrated that any other methodology has met the standard of being clearly superior. It therefore sees no reason for the Agency to change its methodology.

[62] Saskatchewan considers that one of the most effective ways of determining if a company is generating adequate rates of return for its investors is to examine the overall profitability of the company and the price its shares are trading at in comparison to past performance. If investors are pleased with the profitability of a company, it will be reflected in a rise in the share price of the company. On the other hand, if investors are disappointed in the company's profit levels, share prices will decline. In a manner similar to Manitoba, Saskatchewan submitted evidence which indicates that consistent double digit returns have been earned or are expected to be earned by CN and CP for 2005 to 2011.

[63] Saskatchewan views it important to ensure that the rate of return for a particular company is comparable to the rates of return of other companies in the same sector. In that regard, Saskatchewan submitted evidence which demonstrates that the rates of return for Canadian railway companies are generally superior to those of U.S. railroads for 2005 to 2009.

[64] Saskatchewan submits that the cost of capital rate is primarily used for the calculation of rates that can be charged for the transportation of western grain under the revenue cap, which it indicates is a relatively small portion of CN and CP's total revenue, 6.4 percent for CN and 10.6 percent for CP in 2009. Saskatchewan considers it important to examine the financial results and the rate of return of the revenue cap to determine if it compares favourably to the overall financial results of the railway companies. Saskatchewan submits that the findings found in the October 15, 2007 Hopper Car Consultation Document indicate that for crop year 2007-2008, the estimated level of contribution to constant costs for grain shipped under the Revenue Cap is in excess of 45 percent with an adjustment to hopper car maintenance included and 60 percent without an adjustment. Based on this, Saskatchewan considers it reasonable to infer that the cost of capital model being used by the Agency is providing a rate of return in excess of the aggregate rate of return of each railway company.

[65] Saskatchewan considers the impact of proposed changes to be important. It therefore provided its own financial impact assessment of changing from the CAPM to a DCF Model, or to an average of the DCF Model and the CAPM. Based on this assessment, and assuming that each percentage point increase in the final weighted cost of capital rate increases the revenue cap by approximately $14.1 million, Saskatchewan submits that for crop year 2009/2010, use of the CAPM would provide approximately $91.1 million in the revenue cap for cost of capital. It further submits that the use of the DCF Model for the same period would increase the revenue cap by approximately $88 million, to provide approximately $179.1 million for cost of capital, and simple averaging of the two models would increase the revenue cap by approximately $44 million.

Agency analysis

[66] The Agency has limited its analysis of the general commentary of the participants to a few key statements made by the participants that the Agency considers important to clarify or confirm.

[67] CP's statement that the Agency methodology yields cost of equity rates no different from the cost of debt is incorrect. Examining both on an after-tax basis, the approved unweighted cost rates for equity for CP for crop years 2005-2006 to 2011-2012 are, in all cases, greater than the cost rates for debt.

[68] CN submits that the Courts have confirmed that regulators must set the cost of capital as the return that investors expect to earn when compared to what they could earn on alternative investments that are equivalent in risk to their investment in railway companies. On this basis, CN advances that, in setting cost of capital, regulators do not have to give deference to the impact on customers, and that to do so would be inappropriate.

[69] With respect to the contrary comments made by participants, including Saskatchewan, regarding the potential impacts of possible Agency component decisions in regards to the formulation of the cost of capital, the Agency reiterates that its objective is to establish a fair return based on a methodology that is reasonable, reliable and pragmatic.

[70] Manitoba has commented that the Agency should only undertake a review of its cost of capital methodology or change its methodology after having conducted a cost-based review that would account for the productivity gains of railway companies. The Agency notes that it does not have the mandate to carry out the cost-based review requested by Manitoba, but does have an ongoing legislative mandate and responsibility to establish the cost of capital for federally-regulated railway companies, which requires periodic assessment of methodology, as has been done in the current consultation process.

[71] Manitoba and Saskatchewan also argue that the cost of capital model set by the Agency is generating returns in excess of the aggregate rate of return of each railway company. The Agency notes that this argument does not acknowledge that the rate set by the Agency is only applicable to certain legislative remedies and that it is only in respect of these remedies that the Agency must set the cost of capital rate. The Agency is not mandated to regulate the returns generated by federally-regulated railway companies. It is mandated to set a cost of capital rate that generates a fair return and that is reflected in the statutory rates and revenue ceilings established under the CTA.

ISSUES AND AGENCY FINDINGS

[72] The Agency established in the Consultation Document the following three criteria for its assessment of an appropriate methodology to develop cost of capital rates for federally-regulated railway companies:

1. That it be reasonable. That is, the methodology should be: (1) consistent with the objective being pursued, namely, to provide federally-regulated railway companies with a fair and reasonable return; and, (2) transparent by relying as much as possible on a formula/structured methodology and by minimizing the use of judgemental factors.
2. That it be reliable. That is, the methodology should be: (1) based on auditable information; (2) able to produce consistent results for like conditions; and, (3) robust and reasonably sensitive to a broad range of economic/financial conditions.
3. That it be pragmatic. That is, the methodology should be: (1) based on readily available information or information that can be obtained with minimal costs; (2) simple to implement for both the regulator and regulated parties; and, (3) compatible with the regulatory context and legislative requirements in which the Agency is exercising its responsibilities (i.e., timeframe for issuing decisions, nature of regulated parties, context in which the cost of capital is being applied).

[73] In applying these criteria, the Agency's analysis and findings with respect to the issues identified in the Consultation Document and other issues raised by participants in the consultation are framed by: (1) as noted above, the regulatory context in which the Agency makes cost of capital determinations; (2) finance and economic theory as it relates to the issues being examined; (3) the submissions of the participants; (4) the Agency's own expertise; as well as, where appropriate, (5) the practice of other Regulators; and, (6) the Brattle Report.

ISSUE RELATED TO NET RAIL INVESTMENT

Issue 1: Use of market values, instead of book values, to establish the asset base/net rail investment

Should the Agency continue using an asset base (net rail investment) that is based on book value, or instead adopt a market value asset base?

[74] This is an issue raised by CN in the course of the consultation.

Issue 1: Overview of the positions of the participants

[75] CN advocates the use of market values (replacement cost) for the asset base (net rail investment), in connection with its submission in favour of a market value capital structure. It asserts that an investor's opportunity costs would not be met unless both capital structure and asset base are market based.

[76] WCSC disagrees with CN's position, arguing that the illustration set out in CN's submission proves the opposite, that the market value of assets should not be used as the asset base. WCSC also argues that CN makes the mistake of trying to apply in a regulatory context, financial theories applicable to unregulated companies.

[77] A more detailed summary of the position of each participant that made a submission with respect to this issue can be found in Appendix B, Sections 4.1 and 4.2.

Issue 1: Agency assessment against methodology criteria

[78] The use of book values for net rail investment is at the very foundation of the Agency's costing methodology and is prescribed in both legislation and regulation. In the determination of rates to be paid to a railway company by a public passenger service provider, paragraph 152.2(2)(b) of the CTA requires cost of capital rates established by the Agency to be applied to the net book value of the capital assets of the host railway company used by the public passenger service provider. With respect to the VRCPI, the asset base is established and implicitly embedded in the program's legislated formula, as set out in section 151 of the CTA. Furthermore, paragraph 7(b) of the Railway Costing Regulations requires the Agency, when setting rates for the carriage of goods, to apply the associated cost of capital rate to the "variable portion of the net book value of the asset".

[79] The Agency has many concerns with respect to moving to a market value asset base that are unrelated to cost of capital rates. Moving to a market value asset base would have significant, substantive implications for the development of railway unit costs, interswitching rates and various inputs to the western grain revenue cap determinations. Developing unit costs based on the market value of assets would result in regulatory costs departing from reflecting the incremental costs already incurred by a railway company over many years in support of the provision of rail service, and instead reflecting the hypothetical costs that would be incurred if the railway company were to build its asset base today. Clearly, that would be unreasonable.

[80] A move to a market value net rail investment would also not meet the tests of pragmatism or reliability. CN does not describe how market values for these accounts would be calculated. The Agency has reservations about whether the railway companies could produce market values for these accounts and, if they were able to do so, how volatile they would be. The Agency is also concerned that extensive resources would have to be devoted by the Agency to verify those values.

[81] During the original 1960's hearings that established the Agency's costing methodology, in responding to a similar proposal by CN to use the current value of assets for all costing purposes under the requirements of the Railway Act, the RTC stated in Reasons for Order No. R-6313 Concerning Cost Regulations dated August 1969, page 351:

We agree with the opinion of EBS (EBS Management Consultants, Inc) that this proposal would not be useful for regulatory purposes. The use of current dollars may be acceptable for corporate planning purposes, but it would inject new elements of uncertainty in the regulatory process and would be no less subject to criticism as to relevancy than investment calculated on the present historical basis. Carried to its logical extreme, this procedure should involve a calculation of reproduction cost for the entire existing plant. The resultant railway investment would be of little relevance to Canadian National as it actually exists.

[82] The Agency also notes that STB has recently rejected a proposal from the Association of American Railroads (AAR) to consider using replacement cost (market value) rather than historical cost to calculate net investment used to determine return on investment in STB's annual revenue adequacy proceedings.

[83] In its Decision^{Note 2}, STB identified three main challenges in the implementation of the market value approach: (1) the need to estimate the replacement cost of rail assets such as bridges, tunnels, land, track and grading; (2) the need to estimate the "real" cost of capital to avoid double-counting the effects of inflation; and (3) the need to identify the rail assets that would not be replaced as it would be inappropriate to provide a return on the replacement costs on assets that will not be replaced.

Issue 1: Agency conclusion

[84] The Agency finds that the concept of using a market value asset base fails to meet any of its three methodology assessment criteria. Therefore, the Agency rejects the use of market values for determining net rail investment.

ISSUES RELATED TO CAPITAL STRUCTURE

Issue 2: Capital structure weights - book value versus market value

Should the Agency use book values or market values when determining the relative weights of long-term debt and common equity in the railway companies' capital structure?

[85] The Agency currently determines an actual capital structure for each railway company using accounting book values to determine the relative weights of long-term debt and common equity. Capital structure can also be derived by determining the market value of the long-term debt and common equity components. The market value of common equity is calculated by multiplying the price per share of equity by the total shares of equity outstanding. The market value of long-term debt can be estimated based on several factors: the book value of long-term debt, the duration of the debt, the coupon rate on the debt, the current market interest rate for comparable debt and the weights attached to debt with different maturities. Calculating the market value of long-term debt requires complex data analysis, calculations and assumptions and the market values of some types of debt instruments are not directly observable.

Issue 2: Overview of positions of the participants

[86] CP takes the position that market value weights should be used because they are consistent with the use of market value inputs in the CAPM and the DCF Model for determining cost of equity, and that they are more relevant as new debt and equity have to be raised in the market at prevailing prices.

[87] CN submits that market values and estimates of market values should be used because book values do not sufficiently cover the opportunity cost of investors if book values depart from market values.

[88] WCSC submits that book values are the correct measure for determining capital structure and that as long as a firm is expected to earn its required return on the book value of its assets, it will be able to raise capital for new investment and maintain its financial viability. WCSC also submits that due to their lack of stability, market value weights would dramatically increase the variability in the weighted average cost of capital.

[89] CCGA submits that the Agency's current approach seems reasonable and that determining the market value of thinly traded debt instruments requires extensive analysis.

[90] CRS recommends the use of book values. CRS examines the issue through the price (P) to book value (BV) ratio, where price is the market value of the stock. If a firm earns its fair return on new assets, then the P to BV ratio will be equal to one as price will still be equal to book value. CRS submits that the notion that each regulated entity should maintain a market value above book value is contradictory, as it suggests that each regulated entity should plan to earn a return on new investment above the required rate of return.

[91] Alberta submits that book values should continue to be used because they are more transparent and because of difficulties and compromises associated with calculating the market value of long-term debt.

[92] Manitoba submits that book values should continue to be used because of circularity issues that can arise through the use of market values.

[93] A more detailed summary of the position of each participant that made a submission with respect to this issue can be found in Appendix B, Sections 2.2 and 2.3.

Issue 2: Agency assessment against methodology criteria

Conformance with the asset base

[94] In considering whether or not to develop the capital structure weights based on a book or market value approach, the Agency is mindful that its cost of capital rates are applied to a defined regulatory asset base, as discussed in Issue 1.

[95] The CTA requires the Agency to use net book value to establish certain rates, although it does not prescribe the methodology the Agency must use for establishing the capital structure weights. However, the Agency recognizes that there are obvious advantages to having an approach that fully reconciles the financing structure to the net rail investment.

Market value capital structure

[96] The Agency's examination of this issue is based on some basic understandings related to the market value of equity and debt.

[97] When common share equity is issued, the amount of the initial share offering received by the company is reflected in the company's book value of equity and is the amount of money the company actually has at its disposal to finance investment in new assets.

[98] In the case of publicly traded companies, as both CN and CP are, once this initial transaction has taken place, trading on a secondary market(s) subjects the share to fluctuations in its value based on the performance of the company as well as many other market factors, which ultimately affect the levels of supply and demand of the share, and thus its price at any point in time.

[99] Similarly, some long-term debt issued in the form of bonds is traded on a secondary market. The market price and yield of those debt instruments are subject to fluctuations, depending on their individual terms, as well as general market conditions, including prevailing interest rates, the bond issuer's credit rating and other factors.

[100] In general, when interest rates fall, prices of outstanding bonds with coupon rates higher than the prevailing interest rate rise, or trade at a premium to their face value. The inverse also holds true: when interest rates rise, prices of outstanding bonds with lower coupon rates fall, or trade at a discount to their face value. This brings the market yield (the rate of return, based on coupon payment/price) of those bonds into line with lower or higher interest bearing new issues.

[101] In the case of both share equity and bonds, none of the market value added or lost in the secondary market(s) accrues directly to the company. That is, regardless of the market value of its shares or its bonds, the company has no more or no less capital at its disposal than it received in the initial transaction. Furthermore, in the case of bonds, the original repayment and annual interest obligations of the company remain unchanged.

[102] The issue then becomes the relevance of these market values to the capital structure used to calculate the cost of capital input into the Agency's cost-based regulatory determinations.

[103] Examining the use of market values from the perspective of reasonableness, the Agency observes that the total market value of a firm's equity, calculated by multiplying its current share price by the number of shares outstanding to arrive at its market capitalization, can at any point in time, be substantially different from the book value of its equity, the difference being measured as the price-to-book ratio. The higher the ratio, the higher the demonstrated premium the market is willing to pay for the company above the book value of its assets.

[104] WCSC and CRS submit that the expected return on the book value of common equity is a fundamental determinant of share price, in that a price to book value of 1 demonstrates that a firm has earned its fair return on new assets, and any return above what investors required would be reflected in a higher share price. Mindful of the fair return standard, the Agency agrees with the submissions made by WCSC and CRS that it is contradictory in a regulatory setting to set a cost of capital rate that allows a regulated entity, with respect to the activities being regulated, to earn a return above the required rate of return on the funds that have been invested.

[105] It is also important to note that the rates set by the Agency affect only a relatively small proportion of the revenues collected by railway companies. To the extent that railway companies are able to exercise some market power in their unregulated activities, which would be reflected in the price of their shares, the Agency, by adopting a market-based approach to evaluating stock prices, would be extending the effects of that unregulated market power into its regulated activities, possibly undermining the objectives of specific legislative provisions and regulations. The Agency finds that this would not be reasonable.

[106] The Provinces expressed concerns regarding volatility and circularity with respect to the use of the market value of equity. The Agency shares these concerns. In an environment in which a firm's stock price is fluctuating significantly due to market volatility, the market value of equity will also be volatile. This causes the proportion of equity to debt capital to become erratic, potentially inflated or deflated, which could result in a weighted cost of capital that is inappropriately inflated or deflated over the short term. In terms of circularity, the Agency agrees that a circular relationship emerges when cost of capital rates are based on a market value capital structure. The Agency finds that deliberately introducing such volatility and circularity into its regulatory determinations would also be inconsistent with the reliability criterion and therefore unreasonable.

[107] The Agency also observes that, other things being equal, rising (declining) market yields on long-term debt produce an overall increase (decrease) in the cost of capital rate. However, regardless of what has occurred in the secondary bond market, the original contractual obligation surrounding the debt has not changed. From this, it can be inferred that in an environment of rising (declining) interest rates, using the market price and market yield for debt overstates (understates) the cost of capital relative to the use of book values and embedded cost rates. The Agency finds that with respect to debt, the use of market value capital structure also does not meet its reliability criterion.

[108] Pragmatic implementation issues related to debt in the use of a market value capital structure are another major consideration. Estimating the market value of debt is a complex exercise. While the difficulties are not insurmountable, the estimate requires the application of significant subjective judgement and assumptions, and raises transparency issues. Debt capital is obtained from a variety of instruments (bonds, notes, debentures, commercial paper, revolving credit facilities, capital leases, purchase agreements, etc.), for some of which the market values are not directly observable. Estimating the market value of debt requires intricate and complicated data analysis, calculations and assumptions, to arrive at a value that is at best an estimate, which cannot be readily replicated. This raises issues of transparency.

[109] In this connection, the Agency notes the transparency issues that have arisen with respect to STB's market value of debt methodology, as evidenced by recent complaints raised by the Western Coal Transportation League that the cost of debt calculations submitted by the AAR could not be replicated.

[110] Estimating the market value of debt may be considered necessary and appropriate for regulators who are modelling a composite railway, or those trying to develop a deemed capital structure applicable to a large number of regulated companies, or even those that set cost of capital rates for several years into the future. However, the Agency makes annual determinations, which are mainly applicable to only two well-documented Class 1 freight railway companies. The imprecision, distortions and degree of difficulty associated with using market values for debt in the capital structure make it, in the Agency's opinion, an inappropriate choice from the perspective of reasonableness, reliability and pragmatism.

[111] Different approaches to determining capital structure are used in different applications. Market values may be preferred to book values in the financial markets for company valuation, project evaluation and estimation purposes. However, the Brattle Report notes that the return on equity for most regulated companies is based on the book value of investment in the company, not the market value of the invested assets^{Note 3}. WCSC also submits that companies commonly use book values to raise incremental funds and maintain target capital structures, because of the uncertainty of market values, and as incremental funds are based on book values, the earnings to cover capital costs must be computed using book value weights as well.

[112] CN and CP present arguments for a market value capital structure, in the context of raising new funds in order to buy a firm today and of covering the opportunity costs of investors when book values depart from market values.

[113] When determined for these purposes, a cost of capital rate derived from a market value capital structure may be appropriate. However, in setting a cost of capital rate for regulatory purposes, the Agency is not seeking to exclusively determine the cost of new capital, but is instead annually determining the cost rate of capital required by each railway company to both service the investment that underlies its existing asset base (financial viability) and to attract new capital (to replace and grow the capital stock).

Issue 2: Agency conclusion

[114] Observing that determinations for regulatory purposes are similar to accounting practices, in that they favour the stable over the volatile, the known over the speculative, the empirical over the arcane, the Agency finds several substantive shortcomings associated with adopting a market-based capital structure.

[115] It involves a departure from the alignment of the capital structure with the book-based net rail investment the Agency is mandated to use for setting of a number of regulatory rates. It also involves the need to develop a sound methodology to estimate market values and yields, a methodology which would require numerous assumptions and complex data analysis as the market values of some types of debt instruments are not directly observable. Further, it creates the likelihood of a more volatile, less reliable cost of capital.

[116] The alignment of the capital structure with net rail investment ensures that the railway companies are compensated for the capital used to invest in rail assets. In the absence of such an alignment, the railway companies may be under or over compensated for the investment in rail assets. Capital structures derived from market values can be useful for certain valuation and business purposes, but it must be recognized that the resulting total of a market value capital structure is volatile and does not necessarily correlate to the net book value of assets. Market values are not relevant to the amount of capital actually received by a railway company and its obligations with respect to that capital, which are properly the Agency's focus in determining cost of capital rates. For the Agency's regulatory purposes, maintaining the link between the capital structure and the net Canadian asset base is integral in that the cost of capital rates developed by the Agency are used for Canadian rail traffic programs (i.e., western grain revenue cap, interswitching rate regulations, domestic rail/shipper service complaints, etc.).

[117] The Agency therefore concludes that the adoption of a market-based capital structure would not be clearly superior to its existing approach in terms of the criteria of reasonableness, reliability and pragmatism. Accordingly, the Agency will continue to rely on a capital structure based on book values.

Issue 3: Treatment of deferred taxes

Should the Agency continue to include and give weight to deferred taxes in the capital structure and if so, what rate should be assigned?

[118] The Agency currently includes and gives weight in the capital structure to the book value of deferred taxes and assigns it a zero cost rate.

Issue 3: Overview of positions of the participants

[119] CP submits that using market weights, as it advocates, eliminates the need to include deferred taxes as a component of capital structure because any economic effects associated with the amortization policies that generate deferred tax liabilities will be captured in the market value of equity. Where book values are used, CP submits that neither operating liabilities nor its equity equivalents, such as deferred tax liabilities, should be considered part of the capital structure, but should instead be included in equity. CP further submits that the Agency's current treatment of deferred taxes contributes to an unreasonably low cost of capital and counteracts the investment incentives associated with the government's accelerated depreciation policy.

[120] CN submits that if book values used to determine capital structure are to be an unbiased estimate of market value, deferred taxes must be included in the total asset value and allowed a return that ensures the weighted average cost of capital is earned on the entire book value of assets.

[121] WCSC argues that in establishing rates for regulated business, the rates charged to customers are set so that income tax falls on the company's customers and not on the company and its shareholders. A rise (fall) in taxes raises (reduces) customer charges but leaves the return in investor capital unchanged. WCSC submits that under normalization, which it describes as the recognized proper and correct accounting procedure for deferred taxes, rates paid by customers are not reduced by the reduction in taxes paid, forcing on customers periodic loans equal to the provision of deferred taxes.

[122] CCGA's position is that deferred taxes, being temporary, should be assigned a zero cost rate.

[123] Alberta and Manitoba share the same opinion that deferred taxes are a non-cash expense that provides railway companies with a no cost source of free cash flow.

[124] A more detailed summary of the position of each participant that made a submission with respect to this issue can be found in Appendix B, Sections 3.2 and 3.3.

Issue 3: Agency assessment against methodology criteria

[125] Deferred income tax balances arise because a company may, through the claiming of capital cost allowances, depreciate fixed assets for income tax purposes at a faster rate than it depreciates the same assets for regulatory (accounting) purposes. Regulatory reporting is based on Generally Accepted Accounting Principles (GAAP), of which a fundamental principle is to match revenues and expenses in the period in which they occur. This means that assets should be expensed over a period approximating their useful lives. However, because of the difference between the reporting requirements of income and expense for regulatory purposes, as set out in the Canada Business Corporations Regulations, 2001, SOR/2001-512 (CBCR) and for income tax purposes, as set out in the Income Tax Act, R.S.C., 1985, c. 1 (5th Supp.), a company may recognize an item (income or expense) for one reporting purpose and the same item may not be required to be reported in the same manner for the other.

[126] The difference in the reporting requirements produces different figures for taxes payable depending on whether the financial statements were prepared for regulatory or for income tax purposes. As a result, in order to properly reflect their financial position, businesses are required to create an account for accumulated deferred income taxes.

[127] Because of the magnitude of their investment, amortization (depreciation) significantly affects deferred income tax for railway companies.

[128] Two accepted approaches are used to determine amortization expense. One is the straight line depreciation approach, which aligns with the requirements of the CBCR (and the Uniform Classification of Accounts administered by the Agency). The other is the accelerated depreciation approach, which is permitted under certain terms by the Income Tax Act. The straight line approach amortizes the capital cost of an asset evenly over the useful life of the asset, while the accelerated depreciation approach allows the capital cost of an asset to be amortized at a higher rate at the start of its life than at the end. For tax purposes, accelerated depreciation provides a way of deferring corporate income tax by reducing taxable income in current years, in exchange for increased taxable income in future years.

[129] The Brattle Report summarizes the implications of the two amortization approaches for railway companies as follows:

Generally, the difference between the two approaches involves difference in the "timing" of the depreciation expense on the company's financial statements and depreciation expense on the company's income tax return. Tax depreciation is often higher than regulatory depreciation in the early years of the asset's life and lower in the later years, though total depreciation under the two methods is equal over the life of the asset. As a result, the railways report low taxes in the early years of the asset's life and higher taxes in the later years thus enjoying a tax saving in the early years of the asset's life because of the mismatch in cash flows resulting from the two depreciation approaches^{Note 4}.

[130] With respect to capital structure, several approaches can be employed for the treatment of deferred taxes: (1) the normalization approach, which can be implemented in one of two ways; and, (2) the flow-through approach.

[131] Under the normalization approach, the deferred taxes are either included in the rate base with a zero cost rate (normalization type 1) or deducted from the rate base/net rail investment (normalization type 2). Under the flow-through approach, the deferred taxes are applied dollar for dollar to reduce the income tax expense in the year or years in which the tax effects are realized. Normalization approaches are advocated by WCSC, CCGA, Alberta and Manitoba. The flow-through approach, which CN and CP support, reduces the current period tax expense, increasing net income and shareholders' equity, in turn.

[132] A study by Brigham and Nantell discusses the implication of the two approaches^{Note 5}. According to the study, companies that use the flow-through approach have higher cost of capital than those that use the normalization approach. In addition, the study indicates that flow-through accounting is inconsistent with the traditional theory of rate-making because it benefits current customers at the expense of future customers by increasing current period net income. Instead, "normalization involves establishing a deferred tax reserve, which is treated as "costless" capital, and all the firm's customers, both present and future, benefit from the existence of this "costless" capital".

[133] Because they are interconnected, another consideration with respect to applying the flow-through approach, is the need to simultaneously consider the tax adjustment currently applied to the cost of equity. At present, the Agency applies an adjustment, based on the railway companies' statutory tax rates, (i.e., maximum rate payable) and not the effective tax rate (i.e., the rate actually paid) to the cost of equity rate produced by the CAPM, in order to adjust the cost rate of common equity to a before-tax basis. This ensures sufficient revenue to achieve the allowed after-tax return on equity, after all corporate income taxes are paid.

[134] However, including deferred taxes in equity and applying the statutory rate tax adjustment to that total would result in a cost of equity rate that over-compensates for income taxes, in that it would apply the same statutory tax rate applied to shareholders' equity to the amount of tax neutral capital arising from deferred tax. If the Agency were to adopt such an approach, issues related to all three of the Agency's criteria could be raised with respect to its cost of equity determinations. Arguments that the effective tax rate, not the statutory tax rate should be used, could give rise to pragmatic issues regarding the methodology to be used and the timing of any necessary adjustments, as well as more fundamental issues regarding the basic reasonableness and reliability of the methodology.

[135] Based on these considerations, the Agency finds that the flow-through approach does not meet its methodology assessment criteria.

[136] The two techniques of normalization: inclusion of deferred taxes in the capital structure with a zero cost application (type 1) and exclusion of deferred taxes from the rate base and the capital structure (type 2), have opposing impacts on the cost of capital. The inclusion of deferred taxes in the capital structure at a zero cost rate puts downward pressure on the cost of capital rate. The removal of deferred taxes from the capital structure puts relatively upward pressure on the cost of capital rate.

[137] In theory, although the exclusion of deferred taxes results in a higher cost of capital rate, this rate is applied to a smaller asset base and there is no difference on the cost of capital in dollar terms between the two approaches. The cost of capital in dollar terms is exactly the same for the two types of normalization approaches.

[138] However, it is important to note that for the purpose of the Agency's annual VRCPI determination, the two approaches are not equivalent to each other. In the VRCPI determination, the Agency is not mandated to set the cost of capital in dollar terms but rather a cost of capital rate as an input into revenue entitlement. Any changes to the cost of capital rate results in a different revenue entitlement. To be more specific, for VRCPI purposes the cost of capital rate is applied to grain related capital assets. This asset base (which is a further subset of the net rail investment used for cost of capital purposes and which was established and implicitly embedded in the legislative formula when the program was first established) cannot be adjusted to exclude the value of any assets funded by deferred taxes. As such, normalization approach type 2 is not pragmatic, as it cannot be implemented by the Agency in its regulatory context.

[139] The railway companies advance that deferred tax amounts should be included in shareholders' equity and allowed an equivalent rate of return, and that to do otherwise eliminates the benefit of the government's tax incentive. The shippers and Alberta, Manitoba and Saskatchewan disagree, considering deferred tax amounts to be a source of free cash flow supplied by the customer through revenues, at no cost to the company.

[140] To reconcile these conflicting viewpoints, the impact of deferred taxes on free cash flow was examined. From this, the Agency observes that deferred taxes provide a dollar-for-dollar increase to the amount of free cash flow by reducing the taxable profit of the company and as a result the amount of tax currently payable. It is only a company's profit that attracts corporate income tax obligations and profit is a measure of the excess of revenue over expenses. A company's revenue does not come from loans or shareholder investment, but rather from its customers.

Issue 3: Agency conclusion

[141] Deferred taxes are a source of funds generated from profit, which, as a business incentive, current tax law permits a company to retain for its use until those taxes become payable at some future time. These funds do not carry either an interest obligation to a debt holder or a rate of return obligation to a shareholder. Therefore, by assigning a zero cost of financing for these capital structure elements, the Agency is acting in a manner consistent with its mandate. To do otherwise, that is, to allow the railway companies to earn a rate of return on a portion of capital structure that bears no financing costs, would not be reasonable.

[142] Further, while its mandate is not to interpret tax codes or their objectives, it is not apparent to the Agency how the federal government's intended incentive is diminished by not allowing a rate of return to shareholders on these funds. In particular, by returning these lower tax obligations in the form of lower rates for customers, (an outcome that competitive forces should encourage when they are sufficiently present), demand for the activities using those resources subject to accelerated tax depreciation would also be encouraged.

[143] The Agency's current treatment of deferred taxes was determined in the 1985 Decision, which viewed deferred taxes as a cost-free source of capital for the railway companies. The Agency concluded in the 1985 Decision:

...from a cost viewpoint, accumulated deferred taxes are in essence an interest-free loan and as such must be considered as a zero cost source of capital, since the objective is to determine a fair level of compensation....to allow the cost of equity rate on these balances, would provide excess returns to the shareholders....

[144] The Agency finds that the considerations in this Decision with respect to deferred taxes are the same as those at the time of the 1985 Decision. Tax incentives provide increased working capital generated from profit, which is intended to be used in a productive way and which does not warrant a return to investors. Therefore, the Agency determines that giving weight to deferred taxes in the capital structure and assigning to them a zero cost rate is the most appropriate approach for the treatment of deferred taxes in terms of the methodology assessment criteria, and the Agency will continue this practice.

ISSUES RELATED TO ESTIMATING COST RATES

Issue 4: Determining the cost rate of debt

Should the Agency determine the cost rate of long-term debt by using the historic cost of debt in the railway companies' financial statements for the most recently completed fiscal year? If not, how should the Agency determine the cost of long-term debt?

[145] The Agency determines the cost rate of long-term debt based on the historic cost of debt in the railway companies' financial statements for the most recently completed fiscal year (i.e., January 1 to December 31). It has been suggested by CN and CP that in the current climate of rising financing costs some method of projecting future debt costs should be used.

[146] In addition to the issue noted above, the examination of a market value capital structure also led some participants to raise the issue of market-based measurements of the yield on long-term debt.

Issue 4: Overview of positions of the participants

[147] CP argues that market values should be used to assess the cost of debt, maintaining that they are a better estimate of CP's current borrowing costs. Specifically it recommends using the yield to maturity on the average yield for an index of similarly rated corporate bonds with a weighted average time to maturity consistent with CP's existing debt instruments.

[148] The essence of CN's argument is similar to CP's that, in the context of rising interest rates, the historical cost of debt will not suffice to raise funds in the market. CN recommends that the cost of debt should be based on current yields extracted from recent trades in CN debt issues or issues that are comparable in terms of default risk and duration.

[149] WCSC recommends that cost of debt be determined by the actual embedded interest cost in the railway companies' financial statements. WCSC submits that because the contractual obligations with respect to annual interest and repayment are not affected by market yields, using market rates to calculate bond yields would result in a company's shareholders capturing windfall profit.

[150] CCGA views the use of the historic (i.e., book value) cost of debt published in the railway companies' annual reports as reasonable. It submits that a company is more likely to issue debt in a low interest environment, that projecting debt costs in volatile markets would be difficult, and that the amount, the term and even the currency in which the debt is issued in a particular year may not be known.

[151] CRS recommends determining the cost of long-term debt based on the historic cost of debt in the railway companies' financial statements. CRS agrees with WCSC that using market yield to determine cost of debt would create windfall gains for shareholders because when forecasted yields rise, no additional payments are made by the company, causing any increase in the cost of debt measure to be captured by equity holders.

[152] Alberta submits that the yield on long-term debt should be calculated using the yield-to-maturity method.

[153] Manitoba suggests that cost of debt should be measured using the most recent available embedded cost of debt. Manitoba is opposed to the use of forecasted interest rates to represent the cost of debt, because they may not be relevant to the actual debt expense and will be less transparent and more variable.

[154] A more detailed summary of the position of each participant that made a submission with respect to this issue can be found in Appendix B, Sections 5.2 and 5.3.

Issue 4: Agency assessment against methodology criteria

Measuring Bond Yield

[155] Some of the participants suggest that the Agency alter its current practice of determining the cost of long-term debt using the embedded cost of long-term debt as shown in the railway companies' books. They propose instead that a market-based approach to assessing debt costs, through the use of bond yields, be adopted, in order to either capture cost rates current in the debt market or to project future debt costs.

[156] To assess the issue, the Agency first examined the three models commonly used to measure the yield on bonds, namely: (1) using the coupon rate of the bond, (the Agency's current approach); (2) calculating the current yield on the bond; and, (3) calculating the yield-to-maturity of the bond. All three models are applied extensively in finance and economics to assess the return on bond issues, albeit for different purposes. They have also been proposed separately by the participants, in the context of how the cost rate of debt should be determined.

[157] The coupon rate, also known as the nominal yield or coupon yield, is the interest rate the bond issuer is obligated to pay to the bondholder each year until maturity relative to the face value of the bond. It is most commonly used to determine a company's existing debt obligations. In its simplest form, it is expressed as a ratio of the annual coupon payment to the bond's face value.

[158] The advantages of using the coupon rate to estimate the cost of debt are that it is relatively straightforward to calculate, and the results are not dependent on speculative estimates of illiquid secondary debt markets, where current price signals may be lacking and volatility is common. It is also well understood and easy to implement and audit. Data for calculating the coupon rate is readily available and published in a variety of publicly available reports ranging from company financial statements (for embedded rates) to industry reports and investment brokerages.

[159] The main disadvantage of using the coupon rate is that, because the calculation is based on the stock of existing debt, it is backward looking. There are no current price signals built into the calculation and it does not reflect current market conditions for issuing new company debt or capture any capital gains or losses that might accrue to investors purchasing the bond in a secondary market.

[160] The current yield is the ratio of the annual coupon payment in dollar terms to the current market price. It will be less than the coupon rate when the market price of a bond is greater than its face value (i.e., selling at a premium), and will be greater than the coupon rate when the bond value is less than face value (i.e., selling at a discount). In essence, the current yield reflects the inverse relationship between a bond's price and its yield. It is generally used to provide a quick estimate of the current market assessment of a bond's value at a point in time.

[161] While also easy to calculate, well understood and auditable, current yield is dependent on a liquid market to provide price signals to properly reflect market conditions, and price variability over time may result in more volatile debt estimates. As long as a long-term debt instrument is traded in secondary markets, data for estimating the current yield is readily available. In the absence of price data (or in the case of debt that is not traded at all), an estimation model is required to estimate the returns of debt instruments of similar risk to infer the price. Current yield relates the dollar value of the annual coupon payments to the actual market price (not the face value) of the bond issue, but does not take into account other returns to the bondholder (capital gains and reinvested interest payments).

[162] Yield-to-maturity represents the interest rate that equates the present value of cash flows from a long-term bond with its market price, plus any accrued interest. The most complete method for calculating market yield in the bond market, yield-to-maturity captures not only the coupon income but also the capital gain or loss that the bondholder will experience if they hold the bond to maturity. It can be expressed as the discount rate that returns the market price of the bond. Of the three models, the estimation of the yield-to-maturity is the most complicated because of the cash flow internal rate of return formula, the data requirements, and the same problems related to illiquid markets cited in the discussion of current yields.

[163] Because of the direct inverse relationship between the market value of a bond and its yield, in the context of determining cost of capital, any consideration of adopting a measurement of market yield to estimate the cost rate of existing debt must also take into consideration the issue of whether the capital structure is based on market or book values of debt. The Agency observes that, unless there has been no market fluctuation from its face value in the price of a debt instrument, neither current yield nor yield-to-maturity accurately reflects the cost rate that applies to the actual existing debt obligations of the company reflected in the company's financial reports. Whereas, on the other hand, the coupon rate method, based on embedded rates, provides a highly transparent, easy to calculate and accurate measure of the railway companies' existing debt obligations.

[164] Both current yield and yield-to-maturity may be suitable for use as market-based determinants of the cost rate of debt and could be compatible with the use of a market-based capital structure or to project future debt costs. But, as they are both measures that calculate the yield or cost rate of a bond relative to its current market price, the Agency finds neither current yield nor yield-to-maturity to be suitable or accurate measures for calculating the embedded cost of existing debt in a book-value based capital structure. In the context of assessing the cost rate of existing debt in a book based capital structure, they do not meet the Agency's methodology assessment criteria and are not clearly superior to the coupon rate.

Embedded costs versus current or projected debt costs

[165] There is also the substantive question of whether market conditions should play a role in the determination of the value and cost rate of long-term debt. To the extent that the objective may be to calculate the current opportunity cost of long-term debt, the answer is yes. However, other than as an estimate of the interest rates currently available to the company, the market value and market yield of existing debt do not appear directly relevant to the cost of debt of the company responsible for paying the debt.

[166] In this context, the Agency considers that measuring the market yield of a company's existing debt and assigning this amount as the cost of debt is not theoretically consistent with the regulatory goal of providing a fair and reasonable return on capital to regulated railway companies. Applying market values and market yields to the entirety of a company's debt would in fact present a distorted view of a company's actual debt obligations. The use of market value yields would be an appropriate consideration only if a practice of projecting future debt costs were adopted.

[167] CP claims that in recent crop years, historical rates have not been indicative of actual market conditions and that the cost of issuing new debt was higher than the historical debt rate. In its submission for the 2010-2011 crop year, CP argued that Agency-determined cost of debt rates were not indicative of the 2010 interest rate environment. CP raised this issue again during this cost of capital review consultation. Comparing Agency results to CP's market value estimates, CP claims that debt was available to CP at a higher rate than the weighted average cost of capital determined by the Agency in the 2010-2011 crop year.

[168] CP relates both of these claims to a specific debt instrument that was issued in 2009. The Agency finds that this claim, as framed by CP, (that its cost of debt financing was higher than the cost of capital set by the Agency), does not have merit. The weighted average cost of capital for the 2009-2010 crop year (the appropriate year for comparison with debt issued in November 2009) was determined by the Agency as 6.58 percent, an amount higher than the interest rate on the debt instrument cited by CP. Further, the face value of this debt represented only a portion of CP's existing total debt, and was factored into the Agency's cost of debt calculations in accordance with its appropriate book value weight and rate.

[169] Further to this argument, in its current submission, CP provides calculations indicating that the market-based cost of debt was in fact higher than the Agency's calculated cost of debt during 2008 and 2009. With specific reference to 2009, CP submits that its market-based calculations indicate a cost of debt of 7.71 percent, compared to 5.74 percent as determined by the Agency.

[170] In considering this argument, the Agency notes that CP's submitted market-based estimate (7.71 percent) is higher than CP's own prior submission for the cost of a debt instrument actually issued. The Agency further notes that CP's market-based calculations indicate that the Agency's numbers were higher than market-based values in the remaining four years for which CP provided this calculation.

[171] This indicates to the Agency that general market information (in this case, month-end 10 and 15-year corporate bond maturity yields from Bloomberg for companies rated BBB – CP's rating category) is too broad a generalization for its regulatory context.

[172] The Agency acknowledges that the embedded costs of debt can differ from the costs of issuing debt in the coming year and that this raises a valid issue from a regulatory perspective when forecasting a rate of return. However, this concern only becomes material to companies that, in a climate of rising borrowing costs, intend to renew existing lower interest debt or make large debt-financed capital investments in the next year.

[173] The Agency's current embedded cost method, equivalent to dividing the bond's coupon rate by its face value, provides an accurate measure of the companies' actual long-term debt cost rate, albeit one year in arrears. It is not, however, a gauge of the cost of debt financing in the coming year. Thus, the Agency's current approach is more suited for evaluating a company's current obligations on the existing stock of debt, rather than the interest rate investors would demand if the company were to issue additional debt, or when a company wishes to renew or "roll over" existing debt as it matures. In those instances, the cost of new debt in the coming year, if it could be reliably estimated, would, in principle, better reflect the debt financing costs faced by the company.

[174] In the Agency's annual determination of cost of capital, any debt issued in a given year will be captured in Agency calculations the following year. Further, a comparison of the Agency-approved forecast of the cost of debt rate for a given period to the actual cost of debt for that period, as shown in Exhibit 1, indicates that the one-year lag in recording new debt has, on average, only resulted in a minor overstatement of the cost of debt during the last ten years.

Exhibit 1: Forecasted and actual weighted average cost of debt rate, CN and CP

[175] It is apparent from this that during the last 10 years, the effect of the cost of debt rate incurred within one year has been marginal, leading the Agency to the conclusion that in the context of a mature regulated industry, issues regarding anticipated future debt are not particularly material to the cost of capital calculation. This conclusion is important when the Agency considers the potential loss of transparency and reliability in adopting any change to reflect anticipated future debt.

[176] If the Agency were to attempt to estimate the cost of issuing debt in the coming year, it would first have to define a methodology for determining this rate. CP suggests that debt rates being charged for corporate bonds with 10 and 15 year maturity periods to companies with a similar bond rating to the railway company represent an appropriate projection of future debt costs. However, the fact that CP's own market-based calculations overstated the cost of debt in 2009 compared to actual rates paid by CP raises significant concerns about the reliability of this approach. The Agency is also concerned that a process of evaluating market values of debt and verifying party submissions about valuations, as occurs in the process employed by STB, would be resource-intensive and not pragmatic. As would some method of calculating and/or verifying estimates based on interest rates currently being charged to like-rated corporations.

[177] Similarly, estimating the capital base to which a market based projected cost of debt would be applied could require the Agency to assess the investment and financial plans of the railway companies and possibly to create a deemed capital structure, all of which would entail significant speculations, controversies, and debates. Much of the information would likely be confidential, and thus the process would not be transparent. The auditing of such projections would be challenging and any adjustments made by the Agency to correct inaccurate submissions would prove to be extremely difficult to implement. Overall, the Agency considers that the ultimate outcome of such an exercise, if indeed it could be implemented, would not be clearly superior to the current approach.

[178] Further, applying projected financing costs to an actual book value financial structure would serve to exacerbate the original problem. Because over a one-year period most of the debt obligations of the railway companies are fixed, this approach would result in applying a potentially artificial debt rate to the capital structure. Any inaccuracies in the projected rate relative to the realized rate would result in windfall gains accruing either to shareholders or shippers. It would also unnecessarily add an element of uncertainty to the process and could lead to issues of prediction bias.

[179] Another option would be to apply the projected cost of debt to only the renewed debt, that is limiting the review of future debt costs to debt obligations that could be renewed or "rolled over" in the next year, (i.e., debts expected to be renewed within the term of the Agency's cost of capital determination), while retaining the application of the embedded cost of debt to all other existing debt.

[180] This would provide accurate results only if the maturing debt was always reissued as new debt and not fully or partly refinanced with equity. The same problems that were noted earlier with attempting to project a future capital structure would be encountered. In addition to those drawbacks and implementation difficulties, this approach could be subject to manipulation for maturing debt resulting in a potential loss of transparency. Clear and objective criteria to determine when and how to measure projected debt would be required, as would a methodology to properly project future costs of debt and an adjustment mechanism where necessary. The additional effort may result in little improvement as the new or renewed debt within one year may not be significant enough to materially affect the overall cost of debt.

Issue 4: Agency conclusion

[181] To the extent that the method for measuring the yield on long-term debt is attempting to reflect the actual financing cost of existing debt, the Agency finds that the coupon rate method is the most reasonable, reliable and pragmatic of the three models examined.

[182] The Agency finds that projecting future debt and future debt costs and the new issues arising from identified related problems is not a clearly superior approach to the one currently in place.

[183] The Agency determines that it will also calculate the cost of debt rate based on the financing rates recorded in the financial reports of each company, and account for only existing debt and debt costs.

Issue 5: Methodology for assessing the cost rate of common equity

What model or combination of models should the Agency use to estimate the cost rate of common equity?

[184] The Agency assesses the cost rate of common equity using three models of calculation (CAPM, DCF Model and ERP Model). Each year since the early 1990's, the Agency has determined that the CAPM alone provided the best reflection of the state of capital markets. It has been suggested that weight should also be given to the DCF Model, and that alternative models should be considered.

Issue 5: Overview of positions of the participants

[185] CP proposes that the Agency rely on the consistent use of an average of the cost of equity estimates from the CAPM and the multi-stage DCF Model, for which it submits a detailed methodology. It also advances that the Agency's CAPM has major shortcomings and adopting the methodology it proposes would produce a more stable cost of equity estimate.

[186] CN submits that the CAPM best meets all of the Agency's criteria, but that the DCF Model is conceptually sound and should be used to provide a consistency check on the results from the CAPM. CN also indicates that it would be comfortable with a simple average of the two methodologies, although it concedes that the consensus earnings growth rates used in the DCF Model are not sustainable in perpetuity, and that the direct use of even the five-year growth rate will lead to an overstated cost of equity.

[187] CCGA supports the use of the CAPM, but submits that reviewing the results of the DCF Model and the ERP Model provides an additional means of reassurance.

[188] CRS submits that a convincing case has not been made for why the Agency should not retain its current flexibility with reference to its practice of reviewing three models and using its judgement to select the best model.

[189] WCSC submits that the Agency should continue to estimate the cost rate of common equity using the CAPM, without giving weight to the DCF Model. WCSC also questions the accuracy of the estimated growth rates used in the DCF Model.

[190] Alberta submits that the Agency should maintain the CAPM, submitting that, based on the Brattle Report, a clearly superior model does not exist and no improvements could be made to the CAPM.

[191] Manitoba views the CAPM as reasonable, reliable and pragmatic, and considers the DCF Model problematic because the current forecasted growth rates are unsustainable in the long run.

[192] Saskatchewan submits that no methodology has met the high standard of being clearly superior to the CAPM currently used by the Agency and advocates continued use of the CAPM.

[193] A more detailed summary of the position of each participant that made a submission with respect to this issue can be found in Appendix B, Sections 6.3, 6.4 and 13.

Issue 5: Agency assessment against methodology criteria

[194] The Agency's analysis of the approach to be used to determine the cost rate of common equity is undertaken in two steps. First is a review of each of the various cost of equity methodologies to determine their appropriateness in estimating the cost rate of common equity and second, the Agency considers the question of using one model alone or using a combination of models.

Appropriateness of individual cost of equity models

[195] The Agency set out specific criteria by which to analyze the suitability of any cost of capital methodology, stipulating that the methodology be reasonable, reliable and pragmatic for determining cost of capital in its regulatory environment. Based on its review of the CAPM, the DCF Model and the ERP Model, the Agency has come to the following conclusions, relative to its stated criteria, for each cost of equity methodology.

Reasonableness

[196] The CAPM is reasonable. It has a strong theoretical foundation in financial economics and it makes intuitive sense. For an investor to make an investment in a risky asset, the asset must earn the risk-free rate of return, plus a premium to compensate for the risk associated with the investment. This premium is reflected by both the market risk premium and the non-diversifiable risk associated with a specific equity. As such, it is consistent with the objective being pursued in that it provides a fair and reasonable return. Its formula is transparent and if time periods for the variables of the model are pre-determined, the judgemental factors are minimized.

[197] The DCF Model can be considered somewhat reasonable. It can potentially provide regulated railway companies with a fair and reasonable return. However, the reasonableness of the return will depend on judgemental determinations as to the growth rate(s) of earnings assumed in its calculation, the time period over which the growth rate(s) are assumed to apply and on whether a single stage, two stage or multi-stage DCF Model is used.

[198] In the case of the single-stage DCF Model, it is assumed, unrealistically, that dividends will grow in perpetuity at a constant rate. The two-stage and multi-stage DCF Models assume that dividend growth will converge to a terminal rate and that once that rate is achieved, dividends will continue to grow at that rate in perpetuity. Varying the assumptions about dividend growth rates and the periods over which these rates should apply can have a significant impact on the cost of equity calculated using the DCF Model.

[199] The ERP Model is somewhat reasonable in that it can provide railway companies with a fair return. It relies on a formula and is therefore very transparent. Judgemental factors for the ERP Model are minimized if its parameters, the risk-free asset and the time period over which the market risk premium is estimated, are pre-determined. However, the return that is provided by the ERP Model is not company specific. Without a company-specific risk factor, a market beta of 1.0 is assumed in the calculation - an assumption which has been shown to be incorrect in the cases of the regulated railway companies.

Reliablility

[200] The CAPM is considered reliable. It is based solely on auditable information and will produce fairly consistent results for like conditions. Its sensitivity to a broad range of economic/financial conditions will depend on the periods used to estimate the parameters and variables of the model. It is very sensitive to current risk-free rates. However, the degree to which it is responsive to current equity market conditions depends on the length of the period used to calculate the market risk premium.

[201] The DCF Model is somewhat reliable. The stock price and the current dividend yield are known and auditable. However, growth rate estimates, which are key variables in the DCF formula, are not auditable. In the case of the multi-stage DCF Model, more than one growth rate must be estimated and growth rates may need to be estimated farther into the future. The model responds to market conditions only indirectly, through company-specific stock prices, existing earnings (or cash flows) and assumed growth rates.

[202] The ERP Model is somewhat reliable. It is based on auditable information. It will produce similar results for similar conditions and it is somewhat sensitive to economic/financial conditions. Its responsiveness to broad range market factors will depend on the length of time over which its parameters are averaged and, being entirely market-based, it lacks in sensitivity to company specific economic-financial factors.

Pragmatism

[203] The CAPM is based on readily available information. It is simple to implement both for regulators and regulated parties and is suitable for the regulatory environment in which it is being applied. Given the time frame over which it is being applied, it can easily be estimated. As such, it is considered pragmatic.

[204] The DCF Model is somewhat pragmatic. Most of the data that it relies on is readily available at no cost, particularly for the single-stage model. Analysts' forecasts of five years of dividend growth rates are both free and easily accessible. However, forecasts of longer than five years are not available, nor could they be expected to be very accurate. The DCF Model is easy to implement for both regulators and regulated entities, as long as the source of the growth rates and periods for each stage of growth are pre-determined.

[205] The ERP Model is somewhat pragmatic. It is based on readily available information that can be obtained at a minimal cost. Of the three models, it is perhaps the simplest to implement for both regulators and regulated parties. However, the ERP Model is not the most compatible with the regulatory context in which the Agency is exercising its responsibilities. The Agency is charged with calculating the cost of capital for regulated railway companies and there is no component of the ERP Model that is specific to them.

[206] In evaluating these models, the Agency finds that the CAPM meets all three Agency criteria of reasonableness, reliability and pragmatism. The DCF Model performs less well against all three because of issues related to the need to assume company growth rates into perpetuity. The ERP Model also falls short in all three because of its assumption that the beta of each company is the same as that of the market as whole (i.e., equal to unity), which has been shown to be incorrect in the cases of the regulated railway companies.

[207] The Brattle Report summarizes the comparison of reasonableness and reliability between the CAPM and the DCF Model as:

The Capital Asset Pricing Model (CAPM), for example, has a transparent and well-explored economic theory underlying it. Its results can be replicated easily, since the data required is widely available from many public sources. Implementing the CAPM, however, requires a number of subjective decisions – decisions which can be hotly contested and can lead to significantly different results. Conversely, the Discounted Cash Flow (DCF) model can be relatively objective to implement in its simplest form, although required data on growth rates may be difficult to cross-check in publicly available datasets. Moreover, the DCF model is highly sensitive to growth rate estimates, which can vary widely among analysts – and that variation may increase in times of greater economic uncertainty. As such, the reliability of DCF methods can be questionable in times of economic turmoil or when an industry is in transition. These reliability concerns are further exacerbated by the extent of simplification underlying the constant growth version of the DCF model. For example, assuming that cash flows will grow at a constant rate into the infinite future is a gross simplification, and makes the model highly sensitive to the growth rate assumption. If five-year growth rate forecasts are used as the constant growth rate, as is often the case, then the reliability of the model can be significantly reduced in periods of abnormally high or low growth. Moreover, the results of applying the methodology can be unstable over time, leading to rapid shifts from high cost of capital estimates to low ones.

[208] From this analysis it is apparent that the CAPM presents an appropriate methodology for estimating the cost of equity for the railway companies for regulatory purposes, as it meets all three of the Agency criteria. The Agency finds that neither the DCF Model nor the ERP Model meets the criteria to the same degree as the CAPM does. The DCF Model requires assumptions about growth rates over future periods and the projection of an assumed future growth rate into perpetuity. The ERP Model lacks a company-specific factor and assumes company risk to be equal to that of the market as a whole, an assumption that is known to be incorrect.

Single model or combination of models

[209] Given that the Agency finds one model to be clearly superior, in that the CAPM satisfies all its identified criteria for an appropriate cost of equity model, while the DCF and ERP Models fall short in some areas, reliance entirely on CAPM needs to be given strong consideration. However, the Agency also notes the suggestion of the Brattle Report that there is some merit to looking at evidence from a number of models:

Many regulators review estimates from multiple models before arriving at a decision on which cost of capital to allow. Some regulators, e.g., the Surface Transportation Board, have explicitly determined which weight to assign to each model, while others use a range of estimates to guide their decision. Looking at evidence from a number of models continues to be best practice, because different models may be better at capturing different aspects of pricing.

[210] On the other hand, the Agency sees merit in CRS's contention that:

Adding Cost of Equity ("COE") estimates derived from inferior estimation methods to those from superior estimation methods, as advocated in the BG Report (page 5), only increases estimation error and potential bias. It does little to produce a more accurate (fair) estimate of the cost of equity.

[211] In assessing these two conflicting viewpoints, the Agency recognizes that the major differences between them lies in two mutually exclusive potential benefits to determining cost of equity: retaining discretion in the choice of approach versus precision and transparency in the determination.

[212] Adopting the use of the CAPM as a single model, as advocated by four of the eight industry participants (Alberta, Manitoba, Saskatchewan and WCSC), would reduce uncertainty in the regulatory environment in which the rates are being applied because all stakeholders would know, unequivocally, which methodology would be used to calculate cost of equity.

[213] Another alternative, one favoured by CN and CCGA, would be to rely on the CAPM, but to review estimates provided by the DCF and ERP Models, in order to monitor the continued reasonableness of the CAPM. However, the lack of objective criteria by which the Agency can assess the relative reasonableness of the estimates from the three models makes this option problematic. Because of inherent differences in their methodologies, the models might provide widely divergent estimates and there are no recognized guidelines for determining that the differences have passed beyond what should be expected. Furthermore, even if it were decided that the differences between the CAPM and the DCF and ERP Models were somehow abnormal, it is unclear how and on what basis the results would be adjusted.

[214] A third alternative is to adopt a combination of two or more models, with specified weights for each clearly defined. This is the approach advocated by CP, which proposes that the Agency average cost of equity estimates provided by the CAPM and multi-stage DCF Models. This third approach is also, to some extent, suggested by CN when it states that "it would be comfortable" if the Agency adopted a simple average of cost of equity estimates from the CAPM and the DCF Model. With this option, it would be necessary to specify which model's estimates are to be combined and the weight to be assigned to each model. As there are no theoretical guidelines for combining estimates from cost of equity models, the determination of weights becomes entirely judgemental.

Issue 5: Agency conclusion

[215] CAPM is the only cost of equity model that was (at least partially) accepted by all stakeholders. The CAPM also has theoretical support, is widely used in regulatory settings, has been systematically chosen by the Agency in each of the last 19 years, and has an intuitively rational way of characterizing risks (risk-free asset; equity market risk; company-specific non-diversifiable risk). Its three components react in different ways to market information (rapidly for the risk-free rate; moderately rapidly for the company-specific risk; and slowly for the equity market risk), providing both responsive and stable elements in the estimation of the cost of equity. Finally, relying solely on the CAPM would reduce uncertainty in the regulatory environment in which the Agency's cost of equity estimates are applied.

[216] Accordingly, the Agency determines that, in the interest of providing greater certainty and transparency, it will use the CAPM alone to estimate the cost rate of equity for federally-regulated railway companies, and the practice of annually assessing the results of three models and applying judgement regarding the appropriate weight to assign to each will be discontinued.

Issue 6: The CAPM version to be used

[217] In the course of the consultation, certain participants proposed alternative versions to the traditional CAPM for consideration.

Issue 6: Overview of positions of the participants

[218] CN suggests that, while the traditional CAPM meets the Agency's objectives, because CN competes for capital in international markets, the International CAPM (ICAPM), an international version of the CAPM, could be used to provide additional estimates to complement the Agency's results.

[219] CN also asserts that the traditional CAPM represents a static framework which posits a fixed investment horizon and does not reflect the dynamic nature of the investment environment in which the railway company operates. In this regard, CN proposes that the Inter-temporal CAPM is a more appropriate version to use.

[220] Manitoba recommends use of the Fama-French Three-Factor CAPM over the traditional CAPM, because it enhances the CAPM by avoiding the implicit assumption that the beta is the sole measure of a company's risk relative to the market.

[221] A more detailed summary of the position of each participant that made a submission with respect to this issue can be found in Appendix B, Sections 8.2.

Issue 6: Agency assessment against methodology criteria

[222] The theoretical foundations and implementation requirements of these alternative versions of the CAPM were reviewed and assessed against the traditional (or unconditional) CAPM, in terms of the Agency's established criteria that any model used must be reasonable, reliable, and pragmatic.

[223] The alternative versions of the CAPM examined, with the possible exception of the Fama-French Three-Factor Model, have a solid grounding in economic theory, yet each has weaknesses because of a number of simplifying assumptions that may not be realised. The ICAPM could be considered appropriate for estimating the returns that must be offered by companies that operate and raise capital in open capital markets, but the true functional forms of the relationship are extremely complex and, for practical purposes, require further simplifying assumptions to arrive at tractable parameters. The Inter-temporal CAPM tries to account for the fact that investors have tastes, preferences and fears, across which they hedge over time. However, the nature and types of hedges are currently not adequately defined. The Fama-French Model is criticized as lacking a clear economic foundation for the SMB (small minus big) and HML (high minus low) factors, though the authors suggest those factors act as proxies for hedging variables that are absent in the traditional CAPM and are present as additional factors in the Inter-temporal and International versions. Based purely on financial economic theory, it is difficult to say that any one version of CAPM is more reasonable than another.

[224] Data needed to estimate most parameters of the traditional CAPM are easily available from public sources, or directly from the railway companies and are auditable. By contrast, the ICAPM requires index returns on world markets, which are not readily available, and the Inter-temporal CAPM has additional terms for which acceptable functional forms are not agreed upon, even if data to estimate the unknown additional terms were available. The Agency concludes, therefore, that the ICAPM and Inter-temporal CAPM do not meet its reliability criteria.

[225] Each of the alternative CAPM versions can only be implemented if methodological problems can be overcome. The ICAPM, in its standard form, is extremely difficult to estimate as there is no consensus in academic literature on the appropriate functional forms to estimate the international market risk premium and to estimate sensitivities of returns to exchange rate and inflation risks. Therefore, the standard ICAPM does not meet the pragmatism test. Currently, the Inter-temporal CAPM is mainly a subject for academic study, as no consensus exists about what factors to include in the hedging portfolio. CN suggests that the additional terms in the Inter-temporal CAPM, which account for the hedging portfolio could be approximated by the difference between short-term and long-term risk-free rates, but provides no theoretical or empirical support for this proposition. For the above reasons, the Inter-temporal CAPM also does not meet the pragmatism test.

[226] The Fama-French Three-Factor CAPM also presents some problems in estimation. Creation of the SMB and HML by an analyst would be time-consuming and exhaustive, as all companies in the market index would have to be classified in terms of the small versus big and high versus low book value parameters. Alternatively, the parameters can be obtained free of charge on the Internet, but the Model would lose transparency with respect to its development, and the Agency's methodology would be dependent on the continued availability of the parameters from a third party source. In view of this, the Fama-French Model also does not meet the tests of reasonableness and pragmatism.

Issue 6: Agency conclusion

[227] The Agency concludes that only the traditional or unconditional version of the CAPM meets the Agency tests of reasonableness, reliability and pragmatism. The alternative versions of the CAPM suggested during the course of the consultation, while holding promise possibly in the longer term, have not yet evolved to the point where they can be considered mature enough for regulatory applications. The Agency, therefore, rejects their use.

Issue 7: Assessment and application of a grain risk premium

Should the Agency continue to make an annual assessment of whether the cost of common equity should be adjusted to reflect the risk of carrying grain and if so, on what basis should it be established?

[228] In accordance with the 1997 Decision, the Agency makes an annual assessment with respect to the appropriateness of applying a grain risk adjustment to the cost of common equity. Furthermore, it has determined each year since that time that none should apply.

Issue 7: Overview of positions of the participants

[229] CP comments that many agronomic and geopolitical factors play a role in the volatility of grain volumes produced and shipped from year to year, but acknowledges that much of the uncertainty and risk identified in the 1997 Decision is no longer applicable today. CP also indicates that "the inclusion of a grain risk adjustment would be contradictory to the Agency's objective to have a methodology that is reasonable, reliable and pragmatic." CP also disagrees with any suggestion that a grain risk discount is appropriate, and submits that the risks and uncertainties faced by the grain community translate into risks for CP.

[230] CN submits that there have been no changes in the situation that led to the Agency's 1997 Decision on this matter, and any suggestion that the risk is different would be inconsistent with the Agency's criteria, in that as the systemic risk of transporting grain cannot be properly measured it would be totally subjective and discretionary.

[231] WCSC also submits that it is practically impossible to measure any difference between the systemic risk of transporting grain and the overall systemic risk faced by the railway companies because market prices do not exist for the non-traded investment in carrying grain.

[232] CCGA submits that it is not warranted to include a risk premium for grain, as the cost of capital is determined on an enterprise basis.

[233] Alberta and Manitoba submit that the Agency should continue to monitor the matter of a risk adjustment for the carriage of grain, and also submits a case for viewing the carriage of grain as a risk reducing factor.

[234] A more detailed summary of the position of each participant that made a submission with respect to this issue can be found in Appendix B, Sections 7.2.

Issue 7: Agency assessment against methodology criteria

[235] In the 1985 Decision, after hearing evidence and examining the question of from what perspective the risk associated with transporting grain should be assessed, the RTC determined that the risk of grain movement relative to other specific types of freight traffic was the relevant point of comparison. This has remained the intent of any of the subsequent annual examinations of the appropriateness of a grain risk adjustment.

[236] In the 1997 Decision, the Agency determined that, with the repeal of the Western Grain Transportation Act (WGTA) and the end of a subsidy regime, the carriage of grain was no less risky than the carriage of other commodities. However, the Agency considered it prudent to continue monitoring grain risk due to concerns stemming from potential changes in the grain industry and the grain handling system brought about by this legislative change.

[237] Some of these potential changes related to volume, others to increased competition and the available supply of rail cars. In the 1997 Decision, the Agency found that uncertainty, in and of itself, constitutes risk, and acknowledged the potential for reduced grain volume resulting from elimination of the subsidy regime. As almost fifteen years have passed since the end of the WGTA regime, the Agency finds sufficient time has elapsed to allow an examination of whether these concerns have indeed materialized and to assess the reasonableness of continuing to monitor the situation.

[238] Exhibit 2 summarizes the results of the Agency's analysis into whether the grain volumes have fallen since the repeal of the WGTA. The Exhibit depicts statutory grain volumes moved by rail and primary elevator shipments of grain for Western Canada for the years 1995 to 2010. It shows that, while there appears to have been some reduction in the volumes of grain shipped by rail after the 1997 repeal of the WGTA, volume declines are shown to have been temporary (much of the decrease can be explained by issues of weather and related crop volumes) and volumes in the past five years are comparable to those of the years prior to the 1997 Decision.

Exhibit 2: Volume of Regulated Grain Movement and Primary Elevator Shipments

Period	Volume of Regulated Grain Movement (Millions of Tonnes)	Crop Year	Primary Elevator Shipments (Millions of Tonnes)
1995	29.600	1995/1996	29.994
1996	28.400	1996/1997	35.748
1997	32.900	1997/1998	33.649
1998	26.300	1998/1999	29.781
1999	25.100	1999/2000	32.498
2000-2001	29.200	2000/2001	33.488
2001-2002	22.100	2001/2002	26.331
2002-2003	16.400	2002/2003	19.503
2003-2004	24.500	2003/2004	27.737
2004-2005	24.300	2004/2005	27.242
2005-2006	28.400	2005/2006	30.355
2006-2007	28.600	2006/2007	32.019
2007-2008	26.800	2007/2008	31.415
2008-2009	31.200	2008/2009	33.584
2009-2010	31.900	2009/2010

[239] Regarding concerns that under the new regime there would be a shift from rail to trucking for grain transportation, the risk of competition from trucking is not unique to the movement of grain, though it is reasonable to suppose that with the former rate structure for grain movements guaranteed by the WGTA, shippers lacked incentive to move away from using rail service for grain shipments. However, it must be recognized that rail transport continues to be a more cost-effective mode of transportation for bulk commodities moving over long distances and a large portion of any increase in trucking services for the carriage of grain could be attributed to the railway companies' efforts to increase efficiency under the CTA. With the elimination of uneconomic branch lines and the resulting closure of grain elevators, more producers require trucking services to carry grain to the closest high-throughput elevator. However, any increase in trucking stemming from the closure of grain elevators does not detract from the core business of the railway companies, as they had already abandoned that service.

[240] Any uncertainty (risks) surrounding a change of ownership of the federal government's grain car fleet has also been resolved for the time being, with the Government's decision in 2006 to retain ownership of the fleet. Further, should there be a change of ownership with respect to the government fleet that would affect the railway companies' costs, there are mechanisms in the CTA for addressing the matter.

[241] Based on statistics submitted, Alberta and Manitoba argue that the cost of equity should be discounted to reflect that investors regard regulated monopoly companies as significantly less risky than competitive non-regulated companies, and the same risk differential applies to monopoly versus competitive segments of the railway companies' transportation market. They also reason that the railway companies face reduced risk because, unlike the revenue from non-grain commodities, the volume of grain and grain revenues are not tied to the general economy.

[242] While there may be some merit to these arguments, it is unclear to what extent the beta ascribed to both railway companies does not already capture some of the effects being advocated to be recognized here, nor is it clear how to isolate the non-diversifiable risk of the regulated grain-related business. The railway companies have also argued that regulation itself, with its associated uncertainty and constraints, increases risk to a regulated company.

Issue 7: Agency conclusion

[243] The Agency has not found compelling evidence to suggest that the movement of grain is more or less risky than the movement of other commodities, nor has it found a practical, transparent methodology for assessing such risks and translating such information into a reasonable cost of equity estimate specific to the movement of grain.

[244] The Agency's analysis of the concerns that gave rise to the commitment made in the 1997 Decision clearly indicates that the new regulatory regime under the CTA is well established and much of the potential risks and feared consequences of the repeal of the WGTA have not materialized.

[245] Accordingly, the Agency determines that the carriage of grain is deemed to be risk neutral as it relates to the risk of the carriage of other commodities. The Agency further concludes that it is consistent with its three methodology assessment criteria to no longer annually monitor and give consideration to any adjustment to the cost of equity for the risk of grain.

ISSUES RELATED TO ELEMENTS OF THE CAPM METHODOLOGY

Issue 8: Appropriate market data

Should the Agency continue to use Canadian data or use some combination of Canadian and U.S. data to establish the variables for each of the models used to calculate the cost of common equity?

[246] To establish the variables for each of the models used to calculate the cost of common equity, the Agency relies exclusively on Canadian data. It has been suggested that U.S. data or a combination of Canadian and U.S. data should be used.

Issue 8: Overview of positions of the participants

[247] CP submits that as both railway companies trade on the New York Stock Exchange, operate significant parts of their networks south of the border, and compete for equity with other North American Class 1 railroads, a combination of Canadian and U.S. data could be used to establish the variables used to calculate the costs of debt and equity.

[248] CN advocates that international opportunities must be included in cost of capital estimates and suggests that the North American market, represented by the U.S. capital market, is best suited to accomplish this.

[249] WCSC submits that the Agency should continue to use only Canadian data to establish the variables used to calculate the cost of common equity because there are major differences in the two markets, there are other countries that should be considered in an expanded market, and the Agency regulates and determines cost of capital exclusively for Canadian rail operations.

[250] CCGA suggests that, insofar as the Agency is regulating companies incorporated and conducting business in Canada, it is appropriate that the Agency rely on Canadian data to calculate the cost of equity. It also suggests that Canadian capital markets are sufficiently large and liquid to facilitate effective arbitrage with global financial markets.

[251] CRS submits research by its expert that has concluded that the U.S. market does not make a statistically significant contribution to explaining the portion of the return of Canadian utilities that is not explained by the Canadian market. It further submits that some U.S. regulators refuse to include Canadian companies as proxies because of their reasoning that the comparability of Canadian to domestic companies is impacted by significantly different regulatory structures.

[252] Alberta indicates that, because the stocks of the two companies are traded on both the Toronto and New York exchanges, and investors' requirements on both sides of the border should be reflected, that either a combination of Canadian and U.S. data or separate analyses of the two should be used.

[253] Manitoba submits that the use of Canadian data should be continued because the investments in grain transportation subject to Agency regulation are entirely within Canada.

[254] A more detailed summary of the position of each participant that made a submission with respect to this issue can be found in Appendix B, Sections 9.2 and 9.3.

Issue 8: Agency assessment against methodology criteria

Suitable market proxy

[255] In principle, the market portfolio to be captured in the CAPM is not merely assets traded on a given stock exchange, but includes all risky assets that can be held by investors. It includes all assets in all countries, such as bonds, private companies, real estate, human capital, precious metals, fine art and more. With the market portfolio being so extensive as to be beyond measurement, a proxy is always used for measuring the market return. The first issue then becomes determining what is the best representative market proxy.

[256] The argument that the Agency should consider CN and CP's risk/return tradeoffs on markets outside of Canada has merit. However, the Agency finds a "whole world" international perspective impractical due to the complexity of the required calculations and data availability issues. Further, there is no evidence that world markets are integrated enough to make such an exercise meaningful as, apart from mutual funds, it is difficult or expensive for individuals in North America to purchase stocks traded on world markets.

[257] It has been argued that the U.S. market, as the largest and most liquid capital market in the world, would be the best representative proxy for an investor's choices (in particular a Canadian investor) in a world market. Under such a scenario, the cost of equity would be developed using only the risk-free rates and market returns available in the U.S. capital market, which would serve as a proxy for all capital markets.

[258] As a general comparison of the different market indices, the U.S. S&P 500 Index (S&P) is a much broader index than the Canadian S&P/TSX Composite Index (TSX). Where the TSX is heavily influenced by the financial and energy sectors (over 50 percent of the index), the S&P is more diverse as the financial and energy sectors make up about 28 percent of the index^{Note 6}. As such, the S&P is influenced by a wider range of industries which, depending on the economic realities at the time, can have a significant impact. During periods of economic volatility, examining the results of a broader index may provide a more complete picture.

[259] Nevertheless, the Agency does not consider it reasonable, nor does it believe it would be considered reasonable by most observers, regardless of any possible theoretical argument, that a cost of capital developed for Canadian regulatory purposes would use exclusively U.S. data and ignore the Canadian capital markets.

[260] The next question to address is whether the Canadian market alone is adequate for the purpose of estimating an appropriate cost of equity for CN and CP. The Agency's current position on the use of the TSX as a market proxy is based on the 2004 Decision:

With respect to an appropriate data source to be used for estimating beta for CN and CP, the Agency concurs with CN on the use of a broad value-weighted index. As a reflection of the Toronto Stock Exchange, the seventh largest stock market in the world and the third largest market in North America, the Agency finds the S&P/TSX Composite Index, which covers approximately 95% of Canadian equity market capitalization, to be an acceptably broad market measure, suitable for calculating beta values for Canadian regulatory purposes.

[261] Also, several participants have observed that CN and CP appear to have no difficulty raising capital on both sides of the border, with the implication that the current cost of equity for either company, calculated using solely Canadian data, does indeed reflect investor preferences on both sides of the border.

[262] The cost of capital expert, Morningstar/Ibbotson (Morningstar), takes the position that the market proxy should be based on where the capital is deployed, regardless of where the capital was raised, and the example in support of this argument is proposed in its SBBI Yearbook^{Note 7}. The Agency finds the notion that an investor, when considering investment in a foreign market, would be concerned only with the risks and returns available in the foreign market and would not take into consideration the risks and returns available in his or her own domestic market to be contrary to all known investor behaviour. Thus, the Agency is not persuaded by this hypothetical example, and the conclusion drawn from it that the market where the capital is raised should be excluded from consideration.

[263] The cost of equity rate is intended to reflect the return on equity invested in assets of equivalent risk. The major argument against the use of solely Canadian data is that it is not fully representative of the choices available to a potential investor in CN and CP. These companies are listed on both the Toronto Stock Exchange (TSE) and the New York Stock Exchange (NYSE). As such, they compete for capital in both the Canadian and U.S. markets.

[264] It is argued that using only Canadian markets in the CAPM calculation does not adequately capture an investor's options, which include foreign markets, and that as CN and CP are in competition for funds in foreign markets, the required rate of return should account for markets outside of Canada. In this regard, the Agency recognizes that, when setting a rate of return for a regulated activity, a regulator must consider the fact that a company needs to meet the return expectations of its investors, given the alternative investment opportunities with comparable levels of risk available in the market.

[265] In the course of its examination of this question in the 2004 Decision, the Agency noted that:

• there are significantly different regulatory regimes, tax structures, monetary policies and market climates between the U.S. and Canada;
• the cost of equity rate established by the Agency is applied respectively to CN and CP's Canadian assets;
• according to Ibbotson Associates^{Note 8} cost of capital specific to the country in which equity is used should be determined regardless of where the equity was raised, because factors such as the equity market, political system and market regulations distinctive to the country in which the investment is deployed are those that effect the risk associated with the investment, and differ from the risks associated with the same equity deployed elsewhere;
• an informed investor would recognize that CN and CP are Canadian companies, subject to Canadian income tax law and regulations, as are the Canadian commodity industries that comprise a major portion of their customer base and would recognize these railway companies are intrinsically linked to the Canadian economy, no matter where their equity capital is raised; and,
• the costs of equity forecasts determined by the Agency are used in the development of cost of capital rates primarily for Canadian rail traffic programs (i.e., western grain revenue cap, Interswitching Rate Regulations, domestic rail/shipper service complaints, etc.).

[266] While much of the above is still the case, there is a key difference between the applicability of this rationale then and now. In seeking a model for assessing a rate of return on equity consistent with the comparable investment opportunities available to investors, the Agency must acknowledge that today there is evidence, as shown in Exhibit 3, to suggest that both CN and CP rely considerably more on U.S. investors than at the time of the 2004 Decision.

Exhibit 3: Proportion of CN and CP Shares Traded in Canadian Markets and U.S. ^{Note 9}

[267] Exhibit 3 shows that up to 2008 a greater proportion of CN and CP shares were traded in Canadian markets, but that the U.S. market has since overtaken the Canadian market for CN and CP shares in terms of the volumes traded. The current market proportions are 52 percent U.S. and 48 percent Canadian. Therefore, it is reasonable to consider that the U.S. capital markets are currently at least as important as the Canadian capital markets to CN and CP. The Agency, therefore, accepts that, in principle, it seems reasonable to consider a combined capital market that includes both Canada and the U.S. in the development of the cost of equity for CN and CP.

[268] In the context of the comparability of risk between the Canadian and U.S. markets, arguments have been made by WCSC, CCGA and Alberta, pertaining to major structural differences affecting the capital markets of the two countries. Differences in regulations, tax structures and exchange rate risks between the two countries make the U.S. market not completely barrier free for Canadian investors, and vice versa. These differences, it is argued, alter the commercial operating environment, and therefore the risks, for companies doing business in the different jurisdictions.

[269] On the other hand, regulatory barriers to Canadian companies trading their shares on U.S. stock exchanges have been relaxed. There is also increased Canadian investment in foreign markets, partly due to retirement savings tax incentives that have removed restrictions to the allowable proportion of an investor's foreign portfolio. This includes all foreign markets and is not specific to U.S. markets.

[270] However, it is commonly accepted that the Canadian and U.S. financial markets have not fully converged and the degree to which the Canadian market is integrated with the U.S. market has not yet been reliably measured. In considering this issue, the Agency observes that while the degree of integration of the Canada and U.S. markets is still unknown, it is possible to examine whether they are reasonably correlated with each other.

[271] Exhibit 4 shows the year-over-year percentage change in market returns for the TSX and the S&P, for the years 1957 to 2010. The Exhibit illustrates that the two markets are reasonably correlated. Generally, their peaks and valleys occur at the same time and have approximately the same duration, although their highs and lows do not necessarily match each other in terms of magnitude, which is indicative of different degrees of risk in the two markets.

Exhibit 4: Year over Year Percentage Changes in the TSX and S&P Market Indices^{Note 10}

[272] To examine the relative risk of one market to the other, a beta estimate for the relative risk of the movement of the TSX as regressed against the movement of the S&P was developed by the Agency, using weekly observations from 2006 to 2011. Based on this calculation, Canada's beta, adjusted for convergence, is .88. This signifies that the Canadian market is less volatile than the U.S. market, and accounts for the observation that Canadian data produces a lower cost of equity result than U.S. data. It could also be argued that this lower volatility incorporates the effect of all the country-specific factors that come to bear on the assessment of comparable risk for the Canadian operations of Canadian companies, and may better reflect the risk/return reality of the Canadian commercial environment than the U.S. experience does.

[273] Based on the balance of these considerations, specifically: (1) that the cost of equity is an estimate of the returns expected of a company by investors relative to the company's risk/reward profile; (2) both CN and CP compete for capital in both Canadian and U.S. markets; (3) capital and trade flows freely between Canada and the U.S.; and, (4) investors in the two countries can enter the capital markets in both countries with minimal restrictions, the Agency is persuaded that limiting the cost of capital estimate solely to the risk/return tradeoffs on the domestic Canadian market is no longer appropriate.

[274] The Agency finds that, while it is true that the degree to which the Canadian market is integrated with the U.S. market has not been reliably measured, it is evident that it is significant. Many barriers to the flow of capital that existed at the time of the 2004 Decision no longer exist, and capital now moves easily between the two counties. In this environment, CN and CP must compete for capital with companies across Canada and the U.S. Therefore, the Agency determines that the cost of equity determination for the two railway companies should take into consideration the risk/return experiences of both the Canadian and U.S. markets.

Implementation issues

[275] While the principle of determining a cost of equity for CN and CP that reflects the risk/return expectations on both sides of the border seems reasonable, its implementation presents several challenges.

[276] To address these, the Agency must first revisit the full ICAPM and restate its rejection of that model. To use the ICAPM to develop a cost of equity estimate that would reflect investor expectations in a combined Canadian and U.S. equity market, it would be necessary to account for differences in regulations, tax structures and exchange-rate risks between the two countries. A methodology would need to be developed to quantify and adjust for the effect of these regulatory differences and foreign exchange exposures. This would be difficult, probably requiring extensive use of judgemental factors, and may not reflect the Agency's goal for the cost of equity model to be pragmatic.

[277] To address the pragmatism criterion, ICAPM, under some restrictive assumptions, can be adapted into a simplified form similar to the traditional CAPM. That simplified North American CAPM (NA-CAPM) may be stated as:

where:

R_e is the expected return on equity;

R_f,DC is the risk-free rate in the domestic market;

R_m,NA is the return provided by the combined North American market; and

ß_NA is the systemic risk in the company relative to the North American market.

[278] However, in the context of the individual components of the CAPM, even the implementation of a simplified NA-CAPM presents several practical problems. Two issues need to be decided in selecting representative and compatible CAPM variables for the combined North American market: 1) determining the instruments from each country that are the most appropriate to represent each variable; and, 2) determining the weight for each country-specific variable to arrive at a combined input.

[279] With reference to the risk-free rate, because CN and CP are traded in capital markets on both sides of the border, a single risk-free rate, R_f,DC in the simplified ICAPM, seems inappropriate. The risk-free rate for investors in Canada is that of a Government of Canada bond, whereas the risk-free rate for investors in the U.S. is that of a U.S. Government bond.

[280] The calculation of the market risk premium in an NA-CAPM is complicated by the question of how to measure the total return provided by a combined North American stock market (R_m,NA). No widely-accepted market index that reflects the choices available to investors on both sides of the border currently exists.

[281] The MSCI North American Index (MSCI) could be considered as representative of the North American market. Morgan Stanley, the creator of that index, describes the MSCI as a size and liquidity screened market-cap weighted index comprising 500 diverse large and mid-cap companies, that include 91 percent U.S. and 9 percent Canadian companies. However, the Agency rejects the use of the MSCI index on the grounds that it does not adequately reflect the investment choices faced by the Canadian investor. It also does not appear to be a total return index and thus understates the expected market returns. Finally, the historical values of the index only go back as far as 1969, which may not be a sufficiently long period to adequately capture investor expectations covering a broad range of market situations. These problems have unknown impacts on the annual index measures and call into question the reasonableness of the index in terms of representing the market returns in a combined North American market.

[282] An alternative to using the existing MSCI index to represent the combined North American market would be to construct a new market index that meets the representativeness objective. However, construction of such an index is not a trivial exercise and would involve numerous judgemental factors and be subject to disagreements and disputes. The Agency rejects such an approach as impractical.

[283] Based on these difficulties, the Agency concludes that it is not feasible to estimate a cost of equity for either railway company that adequately reflects the risk/return expectations of both Canadian and U.S. investors, using either the full ICAPM or a simplified NA-CAPM.

[284] Rather, the Agency finds that a solution more consistent with its established criteria would be to determine separate cost of equity estimates in the Canadian and U.S. markets, and weight the two. The cost of equity determined for the Canadian market would use risk-free rates derived from yields on Government of Canada bonds, market returns derived from the TSX, and betas derived from regressing company-specific returns against the TSX returns. Similarly, the cost of equity determined for the U.S. market would use risk-free rates derived from yields on U.S. Government bonds, market returns derived from the S&P, and betas derived from regressing company-specific returns against the S&P returns.

[285] With respect to weighting factors, the Agency rejects the use of the GDP of the two countries or the market capitalization of the TSE and NYSE on the grounds that both measures may overweight the combined rate in favour of the U.S. market. The proportion of each railway company's stock owned by residents of the two countries is also rejected as a weighting factor because the data is not readily available. A more reasonable, reliable and pragmatic option is to weight the cost of equity using each country's data based on the annual volume of each railway companies' shares traded on the TSE and NYSE. As the volume of shares traded represents actual investor choices on both sides of the border, and the data is readily available, the Agency considers that this weighting approach best meets its methodology assessment criteria.

Issue 8: Agency conclusion

[286] The Agency acknowledges that North American financial markets are increasingly integrated and that the railway companies raise capital on an enterprise-wide basis in both the Canadian and U.S. markets, with no distinction being made in the raising of that capital that it will be used exclusively in one market or the other. Further, the Agency recognizes that in this environment CN and CP have to respond to the return expectations of investors in both markets. Therefore, the Agency finds that the concept of incorporating the use of both Canadian and U.S. market data into the CAPM calculation is justified and reasonable, and allows for a better assessment of comparable risk.

[287] The Agency will implement this concept in a transparent way that can be applied consistently and meets the Agency's three criteria. Accordingly, the Agency will determine a separate cost of equity for the Canadian and the U.S. markets, and combine the two using a weighted average based on the relative proportions of the volume of each company's shares traded on the TSE and NYSE, in the full calendar year preceding the forthcoming crop year, for each of CN and CP.

Issue 9: Risk-free rate

Bearing in mind the stability and responsiveness implications and given that the Agency develops cost of equity rates annually, should the Agency develop risk-free rates using short-term or long-term bonds, or a combination of both and what time horizon should it use for each?

[288] The Agency uses the yields from a combination of short-term (1-3 year) and long-term (10+ years) Government of Canada marketable bonds as the proxies for risk-free rates of return in the CAPM. The bond choice influences the cost of equity result. In terms of stability and responsiveness, both short-term and long-term bonds have certain advantages and disadvantages, as does the use of a combination of both.

Issue 9: Overview of positions of the participants

[289] CP proposes that a long-term maturity period be used to develop the risk-free rates as it is consistent with the railway industry's long-term investment horizon, and that combining short term and long-term rates underestimates the cost of equity.

[290] CN submits that neither the short-run nor long-run CAPM should be expected to provide an appropriate cost of capital in an environment with uncertain future interest rates. CN indicates that recent research suggests that long-duration cash flows warrant a risk premium in excess of that predicted by the short-run CAPM, which justifies adding a "term premium" to the short-run CAPM. CN concludes that the CAPM, with the current risk-free rate estimated from long-run Government of Canada bond yields and the historical risk-free rate provided by historical short-run Government of Canada bond returns, can be implemented to provide costs of capital while accounting for the additional risks associated with the long-run cash flows generated by CN's regulated and unregulated operations.

[291] WCSC suggests that the risk-free rate should match the period of regulation, but recognizing the regulatory goal of rate stability, considers the Agency's current practice reasonable.

[292] CCGA considers a blended rate more appropriate and better reflective of the railway companies' capital financing activities and related costs and submits that the Agency's current practice is appropriate.

[293] CRS submits that, based on empirical findings for the tests of the CAPM, accuracy problems of the CAPM related to allocational efficiency have occurred. CRS suggests that using long Canada bonds as the risk-free proxy accounts for such accuracy problems.

[294] Alberta and Manitoba both advocate using the yield on 1-3 year Government of Canada bonds. They submit that it is the best indication of a risk-free rate, over unstable shorter terms and longer terms influenced by future inflation.

[295] A more detailed summary of the position of each participant that made a submission with respect to this issue can be found in Appendix B, Sections 10.2 and 10.3.

Issue 9: Agency assessment against methodology criteria

Maturity periods

[296] Short-term and long-term risk-free assets typically have different return characteristics. The rates of return of short-term bonds are lower and more volatile than long-term bonds, being more responsive and communicative of recent changes in the market. By contrast, the rates of return on long-term bonds are higher and more stable, making the choice between the short-term and the long-term CAPM a trade-off between responsiveness of the rate of return to market fluctuations and stability of the rate. The Agency has examined the effect of differing maturity periods in terms of stability and responsiveness, using the historical cost of equity for CN and CP to illustrate, as shown in Exhibit 5.

[297] Exhibit 5 shows that the cost of equity closely follows the yield curve. For the years up to 2006 and after 2008, longer maturity bonds would have resulted in higher costs of equity. The Exhibit also suggests strongly that the yield curve inverted in 2006, with longer maturities resulting in lower costs of capital. An inverted yield curve has historically been a leading indicator of a major recession, which occurred in 2008. The line labelled Actual represents the rate of common equity calculated using the Agency's existing methodology of an average of short-term and long-term bond yields. The line lies almost halfway between the 1-3 year (short-term) and 10+ year (long-term) lines and closely matches the 3-5 year line. This suggests that 3-5 year bond yields could be an appropriate alternative proxy for the risk-free rate, if the Agency were to continue to consider a balance between responsiveness and stability of the cost of equity rate to be important.

Exhibit 5: Effects of Differing Maturity Periods on Cost of Equity - CN and CP^{Note 11}

Notes

1. For the U.S., the 1-3 yr bond is an average of 1-year and 3-year bonds
2. For the U.S., the 3-5 Year bond is an average of 3-year and 5-year bonds
3. For the U.S., the 10+ year bond is simply a 10-year bond
4. For the U.S., the Agency's current practice was used replacing Canadian data with U.S. data

[298] Exhibit 6 shows the standard deviations of the cost of equity estimates for the years 2002 to 2010 using the four maturities of debt instruments depicted in Exhibit 5. The standard deviation of a series of observations measures how far the individual observations can differ from the average value. Higher standard deviations reflect more volatile observations, while lower standard deviations reflect more stability.

Exhibit 6: Standard Deviations of Cost of Equity Estimates for 2002 to 2010

Maturities	Standard Deviation COE - Canada/U.S. - Weighted Average
	CN	CP
3-Month	0.63	0.65
1-3 yr	0.48	0.51
3-5 yr	0.42	0.44
10+	0.52	0.55

[299] The Exhibit shows that, as expected, 3-month T-bills have the highest standard deviation and are therefore to be considered the maturity term that produces a cost of equity rate most responsive to current conditions. The 10+ year bonds produce more stable cost of equity rates than either the 3-month T-bills or the 1-3 year bonds, but the Agency notes that the 3-5 year bonds have produced the most stable results of all the maturities.

[300] The Agency disagrees with CP's assertion that, because railway companies have a long-term investment horizon for their assets, the long-term rate is the most appropriate in the CAPM. The theory behind the traditional CAPM formulation requires that the debt instrument be risk-free, not necessarily that it matches the investment horizon of the company. Theoretically, the yield from a long-term debt instrument, in this case the Government of Canada marketable bonds with maturities of 10 years or more, is the farthest away from a truly risk-free rate, as it includes inflation and capital appreciation effects.

[301] In theory, the yield on the shortest term debt instrument, in this case the Government of Canada 3-month Treasury bill, is considered the closest to a truly risk-free rate of return in the economy, as it includes the least amount of inflation effects in the yield. As such, it is considered by some to be the most appropriate risk-free rate to use in a CAPM estimate. As noted in Exhibit 6, it also results in a cost of equity estimate that is the most responsive to economic conditions. The Agency observes that the use of this instrument could potentially discourage complaints about the cost of equity rates being too high or too low, as the rate would always reflect the current economic conditions. On the other hand, highly responsive rates also mean less stable rates, which the Agency recognizes is not conducive to planning by either railway companies or shippers and producers.

[302] The 1-3 year Government of Canada bond, which represents the second best measure of a true risk-free rate, was also considered. It more closely matches the regulatory period for which the cost of equity is being applied. It is less responsive than the 3-month T-bill, but still lacks stability relative to a longer term bond.

[303] Alternatively, the status quo using an average of the yields from the Government of Canada 1-3 year and 10+ year bonds as the risk-free rate balances the advantages of the short-term rate (responsiveness) with those of the long-term rate (stability). However, the Agency also notes that the Brattle Report states (page 23):

Whether the short-term or long-term version of the CAPM is selected, at a minimum, it is advisable to maintain consistency across the various parameters. For example, it is inappropriate to mix a MRP estimated using a long-term estimate of the risk-free rate with a short-term risk-free rate (i.e., avoid two distinct risk-free rates in the CAPM equation). Similarly, from an economic theory point of view, it is preferable to determine the CAPM using a short-term and a long-term risk-free rate rather than to average the two risk-free rates and the associated MRPs. This is because the average of two risk-free rates generally is not a risk-free rate available to investors. From a practical perspective, the calculation of both a short-term and a long-term CAPM estimate allows the analyst to examine both results, and then make a decision regarding the weight to assign to each version of the CAPM. [Emphasis added]

[304] A further alternative, using the Government of Canada 3-5 year marketable bonds provides the same desired balance between responsiveness and stability in the cost of equity rate as does using the average of the 1-3 year and 10+ year marketable bonds. It also provides more stable cost of equity estimates than the average, and avoids the problem with theoretical consistency that arises with combining a short-term and a long-term bond series.

Estimating the forward-looking risk-free rate

[305] Along with selection of the appropriate debt instrument, consideration was given to the specific observations that may form the estimate of a risk-free rate. The Agency annually determines three different cost of equity estimates for each Class 1 railway company: 1) for the determination of the revenue caps for the transportation of western grain, 2) for interswitching rate development, and 3) for all other regulatory purposes. With the exception of the risk-free rate of return, all of the elements necessary for the development of all three cost of capital rates are those determined annually in the cost of capital rate for the transportation of western grain.

[306] For both the western grain and interswitching determinations, where the cost of equity is determined prospectively, the Agency must forecast the risk-free rate for a future period. These are unlike the determinations for all other regulatory purposes, where the cost of equity is determined retrospectively and the actual risk-free rate for the past period can be determined directly from published data. To estimate the risk-free rates to be used in the prospective estimates of cost of equity for the western grain revenue cap and for interswitching, the options are to use either currently published yields or forecasted yields.

[307] With the first approach, a current observed risk-free rate is used as a proxy for the risk-free rate in the future period for which it is applied, on the assumption that the current observed rate would be a good representation of what the rate will be in the near future. The current Agency practice is to monitor the daily rates over a defined period preceding the determination and select a rate from within that period that, in the judgement of the Agency, best reflects the potential interest rate over the prospective period. Alternatively, the prospective rate may be estimated as the average of daily rates over a fixed period preceding the determination. The fixed period approach reduces the application of judgemental factors and allows for greater transparency. Agency analysis of these two approaches indicates that there are minor differences between the flexible and fixed-period rates in individual years, but that over time the two approaches provide very similar risk-free rates.

[308] With respect to a forecasted yield approach, a standard methodology, accepted in finance theory for forecasting the interest rate in a future period is the so-called "forward rate" method. Under certain assumptions, the forward rate is considered to represent the market's consensus of future interest rate expectations. It is calculated from the zero coupon bond yield curve published by the Bank of Canada (also known as the spot rate curve), using a standard method of calculation based on arbitrage theory, as described in Fixed Income Analysis for the Chartered Financial Analyst Program by Frank J. Fabozzi.

[309] The three methodologies for forecasting the risk-free rate were compared with respect to the western grain revenue cap cost of capital determination. Using Government of Canada 3-5 year marketable bonds, the average of the daily yields assessed and selected from a flexible period preceding the crop year, an average of the daily yields for the month of January preceding the crop year, and the forward rate covering the crop year period were compared with the actual bond yields for the specific crop year, as published by the Bank of Canada.

[310] From this comparison of the three methodologies, as set out in Exhibit 7, the Agency notes the following. Using the current rate to forecast the risk-free rate generally provides closer approximation to the actual risk-free rate than using the forward rate, with an average deviation of 0.17 basis points of the current rate from the actual rate as compared to an average deviation of 0.69 basis points for the forward rate. The results also show that on average the rates averaged over a fixed January period perform the same as the rates averaged over a flexible selection period. Similar results were obtained using the Government of Canada 1-3 year and 10+ year marketable bonds as the debt instruments.

Exhibit 7: Comparison of Current and Forecasted Risk-Free Rates with Actual

Average of Daily Yields From Government of Canada 3-5 Year Marketable Bonds

Crop Year	Actual	Estimated Future Risk-Free Rates			Differences From the Actual Rate
	Rate (1)	Flexible (2)	January (3)	Forward (4)	Flexible (2)	January (3)	Forward (4)
2002-2003	4.00	4.50	4.56	5.49	0.50	0.56	1.49
2003-2004	3.70	4.34	4.15	4.47	0.64	0.45	0.78
2004-2005	3.54	3.52	3.52	4.48	-0.02	-0.02	0.94
2005-2006	3.98	3.59	3.37	4.07	-0.39	-0.61	0.09
2006-2007	4.14	4.05	3.95	3.90	-0.09	-0.19	-0.23
2007-2008	3.61	4.03	4.08	3.93	0.43	0.48	0.33
2008-2009	2.25	2.99	3.45	3.89	0.74	1.20	1.64
2009-2010	2.36	1.96	1.96	2.19	-0.40	-0.40	-0.17
2010-2011	2.12	2.26	2.21	3.42	0.14	0.09	1.30
Average	3.30	3.47	3.47	3.98	0.17	0.17	0.69

Notes

(1) Average of actual daily rates published by the Bank of Canada for the crop year period

(2) Rate based on flexible periods preceding the crop year using daily yields published in the Globe and Mail

(3) Average of daily yields published by the Bank of Canada for the month of January preceding the crop year

(4) Agency calculation of forward rate for the crop year period, estimated using the previously mentioned "forward rate" methodology as described by Fabozzi

[311] The forward rate calculation is widely accepted in academic literature, can provide a forecast covering the exact periods of the regulatory applications and is based on the use of publicly available data. However, based on the results of the above comparison, it cannot be considered to be clearly superior to the currently published yield approach in terms of reasonableness and reliability.

[312] Of the other two approaches for forecasting the risk-free rate, while both are acceptable, the Agency finds the fixed period approach to be preferable because, in using daily yields that are specified and publicly available to all participants, it eliminates the use of judgemental factors and therefore imparts a greater degree of transparency and predictability to the cost of equity determination.

Issue 9: Agency conclusion

[313] The Agency finds that yields from a mid-term debt instrument, specifically marketable government bonds with maturities in the range of three to five years, produce an acceptable balance of responsiveness and stability. This approach also provides a greater degree of theoretical correctness than the one currently in place. Therefore, determining that it best meets its criteria, the Agency will adopt the use of Government of Canada 3-5 year marketable bond yields as its proxy for the risk-free rate applied in the CAPM calculation for determining the Canadian cost of common equity rate. In the case of its determination of the U.S. cost of equity, the Agency will use U.S. Government Treasury 3-year and 5-year marketable bonds to develop separate cost of equity estimates and take the simple average of the two.

[314] With respect to the forward-looking risk-free rate in the Canadian CAPM, for determining the cost of capital rate for the transportation of western grain, the Agency will rely on the average of daily yield observations of Government of Canada 3-5 year marketable bonds for the month of January immediately preceding the crop year, as published by the Bank of Canada. For the cost of capital rate for the development of interswitching costs and rates, the Agency will rely on the average of daily yield observations for the month of September immediately preceding the interswitching year, as published by the Bank of Canada. For the cost of capital rate for other regulatory purposes, the value used will be based on the average yield for the calendar year for which the determination is being made, as published by the Bank of Canada.

[315] For the U.S. CAPM, two distinct forward looking risk-free rates will be determined using the yields on U.S. Government Treasury 3-year and 5-year marketable bonds. For determining the cost of capital rate for the transportation of western grain, separate averages will be calculated of daily yield observations on U.S. Government Treasury 3-year and 5-year marketable bonds for the month of January immediately preceding the crop year, as published by the Federal Reserve. For the cost of capital for the development of interswitching costs and rates, the averages of daily yield observations for the month of September immediately preceding the interswitching year, as published by the Federal Reserve will be taken. For the cost of capital rate for other regulatory purposes, the values will be based on the average yields for the calendar year for which the determination is being made, as published by the Federal Reserve.

Issue 10: Averaging period for the market risk premium

What averaging period should the Agency use to develop the market risk premium used in the CAPM?

[316] Since the 1997 Decision, the Agency has applied a 45-year moving average to develop the market risk premium (MRP) in the CAPM. It has been suggested that a longer averaging period should be used.

Issue 10: Overview of positions of the participants

[317] CP submits that a long period should be used to calculate the MRP, viewing the long-run MRP as a more stable estimate that best represents an estimated future performance based on many types of historical events recurring over time. CP advocates a risk premium that incorporates all readily available market data, suggesting that data from 1926 to present be used and that the Agency rely on Morningstar's annually published Canadian Risk Premia Over Time Report for the MRP^{Note 12}.

[318] It is CN's position that more observations provide more confidence, as long as the data used is from a period where the fundamental risk determinants were no different from today's. CN submits that as there is no evidence of any such structural breaks to the fundamentals, as much data as is available should be used. CN also submits that the use of a moving period may result in volatile and inaccurate estimates.

[319] WCSC suggests that the averaging period should be a long period for which high quality data is available and expresses confidence in the breadth and reliability of data available for the last 54 years.

[320] CCGA views the 45-year MRP currently used by the Agency as a statistically valid sample size that is sufficiently long to reduce the impact of any single year and submits that there is no evidence that a longer averaging period would produce a materially better measure of risk premium.

[321] CRS is of the opinion that the method of measuring historical realized risk premiums using long periods produces an overstated estimate, and advocates the use of survey evidence from estimates of investment professionals and economists for prospective market returns.

[322] In connection with using the equity risk premium approach to determine the cost of equity, Alberta and Manitoba express concern over the use of a prospective, rather than historic approach to determining risk premium.

[323] A more detailed summary of the position of each participant that made a submission with respect to this issue can be found in Appendix B, Sections 11.1.2 and 11.1.3.

Issue 10: Agency assessment against methodology criteria

[324] Estimating the MRP through a historical average of past MRPs is a methodology that is well accepted and described in academic literature, including discussion of the advantages and disadvantages of short-term and long-term periods.

[325] A longer period reduces the volatility in the time series and takes into account more past market behaviours. Investors assess risks by considering market behaviours from as far back as available data permits and a longer period provides more information about previous business cycles and their impacts on yields. As such, it is thought to provide a more accurate estimate of the risk premium demanded by investors.

[326] On the other hand, it is also argued that a longer period gives too much weight to distant market events that may have no bearing on current market conditions. As suggested in the Brattle Report (page 25), "returns over more recent periods are likely to be a better measure of investor expectations going forward, because the economy and capital markets have evolved so much over time."

[327] In examining this issue, the Agency is also mindful of its criteria of reliability and pragmatism in the methodology, which leads to considerations of the availability and quality of data to be used in the Agency's estimation.

[328] The Agency continues to be of the opinion that the MRP should be based on a period that has sufficient length to incorporate many business cycles, periods of low and high performance, periods of volatility and stability, as well as to reflect the impact of unusual events and significant changes that the world has undergone.

[329] The Agency also agrees with arguments put forward that accounting for structural shifts should be an important consideration in selecting analysis periods for use in the CAPM, a structural shift being evidence of a change in the fundamental determinants of risk in the market. Further, the Agency observes from an examination of the historical total returns for the TSX and S&P 500, shown in Exhibits 8a and 8b, that it is not evident that any structural shifts have occurred since 1924 for the TSX and since 1936 for the S&P 500.

Exhibit 8a: Annual Total Returns from the S&P/TSX Composite Total Return Index^{Note 13}

Exhibit 8b: Annual Total Returns from the S&P 500 Total Return Index^{Note 14}

[330] The Agency also considers arguments put forward by CP that the Agency's current practice of focusing on a 45-year period excludes several periods when the Canadian equity market generated unusually low or high returns and that it does not accurately reflect the MRP.

[331] At the time of the 1997 Decision, the Agency adopted the use of a 45-year analysis period and in the years since has continued to use a 45-year period, in the interest of consistency. There were limitations to the availability of reliable data beyond 45 years at the time of the 1997 Decision. However, the passage of time has now provided a greater span of reliable data. In consideration of this and the arguments in favour of the use of the longest period possible, the Agency has revisited the use of a 45-year moving average.

[332] The option to maintain the status quo is advocated by CCGA and Alberta. However, it may be argued that, in the absence of a structural break in the market risk determinants, a moving 45-year analysis period represents an arbitrary cut-off point that has little or no objective economic support. Every year in their annual submissions, the railway companies have raised the point that this practice represents selective use of historical data by the Agency and imparts some measure of inaccuracy and a downward bias to the cost of equity estimate.

[333] The railway companies instead advocate using the complete historical data set available. Specifically, CP suggests the use of market return and bond yield data from 1936 to the current year, as used in the Morningstar estimates of the Canadian MRP.

[334] CP's suggestion, however, has shortcomings relative to the Agency's criteria because reliable data on the government marketable bonds needed to provide the risk-free return comparisons in each year is not available for the same length of period as the market returns, both for Canada and the U.S. Specifically, data that the Agency is satisfied is reliable is only available from 1951 on the 3-5 year Canada bond series and from 1954 for the 3-year and 5-year U.S. Government Treasury bonds.

Issue 10: Agency conclusion

[335] Given the absence of any conclusive evidence of a structural break in the market premium time series and in order to satisfy its three criteria, the Agency will use as much historical return data as possible, subject to the availability of reliable data. Accordingly, the Agency determines that it will adopt the use of an averaging period that comprises return data from 1951 to the current year in the calculation of the MRP in the Canadian CAPM, and from 1954 to the current year for the MRP calculated with respect to the U.S. CAPM.

Issue 11: Averaging methodology for the MRP

An issue raised during the consultation is whether the MRP should be calculated using arithmetic or geometric averages.

Issue 11: Overview of position of CRS

[336] CRS was the only participant to comment on this issue, submitting that long-term bond yield is a geometric, not arithmetic, type of average yield, and using arithmetic yields can lead to misleading estimates.

[337] A more detailed summary of CRS's position with respect to this issue can be found in Appendix B, Section 11.2.2.

Issue 11: Agency assessment against methodology criteria and conclusion

[338] Estimates of the MRP calculated using both arithmetic and geometric averaging for the years 2000 to 2009 were examined, and a material difference was noted in the estimates. A further analysis was performed to determine whether or not the annual stock market returns and the equity premiums (stock returns less bond returns) are each serially correlated, using returns data from 1924 to 2009 in the Ljung-Box Q test of serial correlation. Based on this analysis, the Agency determined that neither the stock market returns nor the equity premiums are serially correlated.

[339] Because the use of a geometric average is warranted only when returns are serially correlated, the arithmetic average better aligns with the Agency's methodology assessment criteria and the Agency will maintain the use of the arithmetic average.

Issue 12: The use of income versus total returns for bonds in the MRP

An issue raised during the consultation is whether the historical risk-free return used in the MRP should be the income return or the total return on the bond instruments examined.

[340] The Agency's current practice is to use the total return.

Issue 12: Overview of positions of the participants

[341] CP submits that income returns instead of total returns should be used as the basis for assessing government bond returns to develop the historical risk-free rate of return in the MRP calculation. CP views the Agency's use of total returns as one of two major shortcomings in its MRP methodology. CP argues that the income return is the truly risk-free portion of a bond return and proposes that the Agency rely on Morningstar's annually published Canadian Risk Premia Over Time Report for the MRP.

[342] WCSC counters that the expected [market] risk premium is taken to be an average of the difference between the realized stock index return and the realized government bond return from each period, and that it is the realized total return, not just the income return, on these bonds that should be used.

[343] A more detailed summary of the position of each participant that made a submission with respect to this issue can be found in Appendix B, Section 11.3.2.

Issue 12: Agency assessment against methodology criteria

Bond risk determinants

[344] Analysis of this question first involved an examination of the risk attributes of a risk-free instrument. In theory, the risk-free rate is the rate of return on an investment that all investors perceive as being risk-free. Because no such investment vehicle exists, the risk-free rate must be estimated using a proxy. For the best proxy, two key sources of uncertainty will be minimized or eliminated altogether:

1. Default Risk - The risk that the bond issuer will not honour repayment of interest and principal as per the coupon schedule and bond maturity.
2. Reinvestment Risk - The risk associated with the uncertainty around the expected rate of return for coupons paid on the bond.

[345] Of the two risk factors, default risk is the more significant. Theorists and practitioners alike agree that the return on Treasury bills or bonds issued by a developed country with a well-developed financial market and a strong sovereign debt rating minimizes or eliminates default risk. Such bills or bonds are therefore commonly used as proxies.

[346] Reinvestment risk is the risk that future proceeds will have to be reinvested at a lower potential interest rate. This reinvestment risk is especially evident during periods of falling interest rates where the coupon payments are reinvested at less than a bond's yield to maturity at the time of purchase.

Composition of bond returns

[347] In assessing the return on an interest bearing instrument, the total return on a bond may be decomposed into three basic components:

1. Income Returns - The returns resulting from periodic cash flow paid by the bond issuer to the bond holder (coupon payments).
2. Reinvestment Returns - The returns resulting from the reinvestment of the income returns.
3. Price Returns – The returns resulting from capital appreciation or depreciation that occurs when bonds are purchased and sold in secondary markets.

[348] When a decision is made to purchase a bond, the investor can rationally assess the anticipated yield that they will receive. At the time the investment decision is made, the market has priced this anticipated yield into the price of the bond. However, the only portion of the total return that can be anticipated with certainty (assuming no default risk) is the income return. Price returns, which represent the change in the price of the bond over the investment period, cannot be anticipated conclusively at the time of bond purchase. Accordingly, price returns have uncertainty associated with them. Similarly, potential returns from reinvesting the bond's coupon payments are uncertain. Considering this, the Agency is of the view that the income return more reasonably represents the most risk-free measure of return and is consistent with the theoretical foundation of the CAPM.

Estimation of income returns

[349] To properly assess all the factors associated with the possible adoption of the use of income returns, the Agency considered the formulae, data sources and implementation issues that exist for calculating the income return on both the Canada and U.S. government securities used to determine MRP.

[350] A common specification for the income return is the current bond yield using the market price of the bond at the beginning of the period. This specification, suggested in the Brattle Report (page 21) and by others,^{Note 15 Note 16} can be expressed as:

[351] In this equation, t represents the point in time at which the yield is observed, which can be the beginning, middle, or end of the year. When t is taken as the beginning of the year, the income return is equivalent to the current yield. Using this formula, the annual income return on, for example, a $1.00 bond with an 8 percent coupon selling at $0.95 at the beginning of the year would be 0.08/0.95, or 8.42 percent.

[352] To analyze this issue, an income return test series for Bank of Canada 10+ year bonds was calculated by subtracting Morningstar's published MRP from the average return on common stock from the TSE. Then the beginning of year specification was used to develop a second income return series for the same bond term using the yields for January published by the Bank of Canada. A comparison of the two demonstrates that the income return series track each other extremely well with a simple correlation of 0.996 (perfect correlation is 1.0).

[353] The two series were also compared with each other using cumulative moving averages of bond returns. Under this analysis even smaller differences were found between the two. The correlation between the two series is almost perfect (0.99996). Therefore, the Agency is confident this methodology for estimating the income return produces reliable results for estimating the income return.

Comparison of income and total returns

[354] Exhibit 9 presents the annual income returns and total returns for 10+ year Government of Canada bonds between 1936 and 2009, and illustrates the relative volatility of the total return estimate. This volatility primarily reflects the impact of period-to-period price effects and the resultant capital appreciation (or depreciation). Annual intervals where capital depreciation overwhelms the income effects are noticeable as those periods where the total returns dip below zero. The income return estimate, however, is much less volatile, and cannot, by definition, be negative (in nominal terms). The volatility difference between the two types of return measurements is clearly demonstrated from the analysis of Canadian 10 + year bonds. The same holds true for Canadian 3-5 year bonds or U.S. bond returns.

Exhibit 9: Historical Income and Total Returns for 10+ Year Canada Bonds^{Note 17}

[355] Based on a cumulative moving average, Exhibit 10 illustrates the comparison between historical total returns with the corresponding cumulative moving average of income returns. The comparison shows that of the two estimates the income return produces a much smoother and more stable result.

Exhibit 10: Cumulative Moving Averages of Total and Income Returns on 10+ Year Canada Bonds^{Note 18}

Issue 12: Agency conclusion

[356] The income return is considered to be a better indicator of the truly risk-free portion of a bond return and is, therefore, more aligned with the underlying CAPM. Further, Exhibit 10 shows that the use of income returns results in a more stable estimate of historical risk-free returns. The Agency also finds that the methodology is easier to understand and apply, using calculations that are simpler and more transparent than those for calculating total return, and the data required (in its equivalent form as the current return) is easily accessed and publicly available. Based on the criteria of reasonableness, reliability and pragmatism, the Agency considers income returns a clearly superior methodology.

[357] Accordingly, in assessing the rate of return from government bonds used to calculate the historical risk-free rate in its estimations of the MRP, the Agency will adopt the use of income returns rather than total returns.

Issue 13: Beta return interval and analysis period

An issue examined by the Agency during the review was the appropriate return interval and analysis period to be used in estimating beta.

[358] While this issue has not been disputed by any participant and the current approach is not inconsistent with current financial practices, this review offered an opportunity to further examine this question.

Issue 13: Agency assessment against methodology criteria

[359] In the 2004 Decision, the Agency directed the use of five years of return observations (when available), measured at weekly or monthly intervals to calculate beta for CN and CP.

[360] In its annual cost of capital determinations the Agency has consistently relied on beta estimates based on weekly rather than monthly data.

[361] The Agency considered the Brattle Report's literature review and analysis of the betas for a portfolio of U.S. railroads, and its conclusion (page 36) that:

While there is less concern for lack of sufficient observations in daily returns, these are sometimes deemed to be too noisy to provide reliable estimates. There are also concerns that market microstructure effects [the fact that daily prices reported can be affected by whether the price quoted is a bid price or an ask price] can bias daily beta estimates... The additional sampling error for monthly betas (since they are based on significantly fewer data points) seems to dominate other sources of variation, and produce a less stable estimate than at the weekly horizon.

[362] In order to assess the continuing validity of the Agency's current choices with respect to the return interval and analysis period used to estimate beta, alternative return intervals and analysis periods for beta estimates were compared for CN and CP, using daily, weekly and monthly returns for the railway companies against market returns represented by the TSX, over one year, three year and five year analysis periods.

[363] The analysis over five years indicated that daily returns sometimes provide negative betas, suggesting the counterintuitive result that expected returns for CN and CP would decrease with increasing risk, and have statistical significance values that are close to zero. Therefore, the use of daily returns is rejected. The analysis also demonstrated that betas estimated using weekly returns almost invariably have higher statistical significance than betas estimated from monthly returns. The weekly betas also better conform with a priori expectations and display less volatility. These results suggest the use of weekly observations to estimate beta is clearly superior to the use of monthly observations.

[364] Similar results are obtained when a three-year analysis period is used. Analyses for periods greater than five years were not undertaken, due to a lack of sufficient trading data for CN and CP.

[365] With respect to a one-year analysis period, counterintuitive beta estimates are sometimes produced, due perhaps to too few observations. Both the three-year and five-year analysis periods produce stable beta estimates, and the statistical reliability tests do not point to a definite advantage of one period over another, suggesting that either period could be used.

Issue 13: Agency conclusion

[366] The Agency's analysis confirms that beta estimates using weekly return intervals are more stable and have higher statistical significance than the daily or monthly return intervals, and are therefore preferable. With respect to the analysis period, while the Agency's analysis confirms that both the three-year and five-year analysis periods using weekly data provide stable beta estimates, the Agency will maintain the current practice of estimating beta using five years of weekly return observations.

Issue 14: Adjusting beta for mean reversion

Two issues concerning beta raised during the consultation are whether beta estimates should be adjusted for the mean-reverting tendency, and if so, what is the appropriate adjusting formula.

Issue 14: Overview of position of CRS

[367] Only CRS filed comments on the beta adjustment and calls for the Agency to revisit its practice of adjusting beta for a mean-reverting tendency, suggesting that there is little evidence that the betas of CN and CP exhibit mean reversion to a market beta of 1.0. CRS also submits that if an adjustment is applied, the Vasicek adjustment is preferable to the Blume adjustment currently used by the Agency.

[368] A more detailed summary of CRS's position with respect to this issue can be found in Appendix B, Section 12.2.2.

Issue 14: Agency assessment against methodology criteria

Whether or not to adjust beta

[369] Calculations of Canadian and U.S. beta estimates for CN and CP for the years 2001 to 2010 were reviewed in an attempt to assess their mean-reverting tendencies. Data availability issues prevented the examination of historical Canadian estimates before 2006, leaving insufficient Canadian data to assess the existence of a tendency. However, the results using U.S. data, as shown in Exhibit 11, suggest that such a tendency may exist.

Exhibit 11: CN and CP Betas – U.S. market – 2001-2010^{Note 19}

Note: CP betas from 2001 to 2005 contain varying degrees of data preceding the October 2001 split of Canadian Pacific Limited (CP conglomerate) into five separate companies. Specifically, the beta for 2001 contains almost exclusively CP conglomerate data, whereas the 2005 beta contains mostly CP Railway Company data.

[370] In addition, both CRS and the Brattle Group suggest mean reversion to 1.0 is not the only reason for adjusting beta, other reasons include interest rate sensitivity of regulated companies, conformance with the methodologies used by analysts and commercial data services, and beta estimation errors. The Agency determines that maintaining its practice of adjusting beta for mean reversion satisfies the Agency's three consultation criteria.

Method of adjustment

[371] The beta parameter is traditionally estimated using simple linear regression, where the returns of the asset are regressed on market returns. A number of problems with this methodology have been discussed in academic literature. First, beta determined on the basis of historical market data might not meet the theoretical requirements of a forward-looking CAPM. More importantly, individual company beta estimates contain considerable sampling errors, with standard errors typically exceeding 0.2. As a consequence, very low or high beta estimates tend to underestimate or overestimate the true beta.

[372] A variety of models have been developed to deal with these problems, the most common being the Blume (1971, 1973) and Vasicek (1973) adjustments. Blume found that betas tend to revert toward their mean value, or the market beta of 1.0. High historical betas (excess of 1.0) tend to overestimate betas in future time periods, and low historical betas (under 1.0) tend to underestimate betas in future time periods. Blume's analysis regressed estimates of beta in one period against estimates in the previous period. By performing this analysis over different periods, Blume was able to develop a convergence tendency that could be measured by the following formula:

where:

ß₁is the prospective beta

ß₀ is the historical beta

[373] The Blume equation has the impact of lowering high historical betas and increasing low historical betas, as well as dealing with order bias. Blume suggests that all betas using historical regression techniques should be adjusted in this fashion. The closer that beta is to 1.0, the less the magnitude of the adjustment. The Blume equation is often simplified by using a 1/3 + 2/3 adjustment. The Agency uses this simplified Blume adjustment.

[374] In response to the problems associated with the raw beta estimation, Vasicek proposed a technique that considers the statistical accuracy of the beta estimation. His adjustment seeks to overcome one weakness of the Blume adjustment, namely, applying the same adjustment to every security. Rather, Vasicek puts forth a security specific adjustment that is dependent on the statistical quality of the regression. It focuses on the standard error of the beta estimate; a company with a high standard error should have a greater adjustment than a company beta with a low standard error. The following formula describes the Vasicek adjustment:

where:

ß_s1 is the Vasicek adjusted beta for security s

ß_s0 is the historical beta for security s

ß₀ is the beta of the market, industry, or peer-group

(s_ß0)² is the variance of betas for the market, industry or peer-group

(s_ßs0)² is the variance of the historical beta for security s

[375] The weight given to the company's historical beta depends on the statistical significance of the company beta statistic. If a company beta has low standard error, then it will have a higher weight in the Vasicek formula. If a company beta has high standard error, then it will have a lower weight in the Vasicek formula. In all cases, the Vasicek weights will sum to one.

[376] To examine this question, both the Blume and Vasicek adjustments were applied to beta estimates for CN and CP from 2005 to 2010. The Agency observes that, on average, the Blume and Vasicek adjustments provide estimates similar to each other as well as to the unadjusted betas, supporting the Brattle Group's contention that, because railway companies' betas tend to be close to 1.0, their adjusted betas generally do not depart significantly from the unadjusted betas.

[377] The Blume adjustment is the simpler, more easily understood, and easier to implement of the two methodologies. The Vasicek adjustment, while it could be considered theoretically more accurate than the Blume, presents serious implementation difficulties. Calculation of the variables ß₀ and (s_ß0)² can be challenging on many fronts. First the analyst must decide whether to use a market, industry or peer-group beta for comparison purposes. This choice depends, in part, on the availability of data for the three groups. Intuitively, the market beta of 1.0 seems to be the ideal choice because of its simplicity but, in practice this would prove to be close to impossible as it would require calculations of beta for all companies in the market index (300 in the S&P/TSX Composite Index) and a determination of the variance of the company betas ((s_ß0)²). From a practical standpoint, the use of a market beta is not feasible.

[378] A peer-group beta was tested. This was also not without its problems. Calculating a peer-group beta (unlevered) for CN and CP requires estimation of the debt to equity ratios of all railway companies in the peer group. For the purposes of this analysis, a railway industry beta for 2011 published by New York University was used, which provided an average unlevered beta of 14 railways, including CN and CP. Because similar industry betas were not available for previous years, the same values had to be assumed for the 2005 to 2010 calculations, and the future availability of the industry beta estimates is uncertain. Therefore, the Vasicek adjustment does not meet the Agency's pragmatism test.

[379] Most financial data providers, such as Bloomberg, and Value Line, report adjusted betas using Blume's methodology as their default beta. According to the Brattle Report, from a decision-theory point of view, this estimate is generally inferior to Vasicek's adjustment, which computes a specific weighting tailored to the information content of the data. However, the Brattle Report adds that in general the Vasicek adjustment requires more computations and has not performed significantly better than the simplified Blume adjustment, which may be the reason the Vasicek adjustment is rarely employed by analysts in business and regulatory applications, and is not a standard reported by most data providers.

Issue 14: Agency conclusion

[380] As the Agency, for the most part, uses beta to forecast a cost of capital rate for a future period, the Agency considers it appropriate to employ a methodology that is used by the majority of financial data providers, insofar as investor expectations are formed by financial analysts. Also there is greater transparency with the Blume adjustment as compared to the Vasicek method, which better meets the Agency's reasonableness criteria. Accordingly, the Agency will continue to use the Blume methodology to adjust beta.

Issue 15: Unlevering and relevering beta

An issue raised during the consultation is whether the Hamada equation for unlevering and relevering beta estimates was valid.

Issue 15: Overview of position of CRS

[381] CRS raised this issue and is the only participant that submitted comments. It submits that using the Hamada equation is fraught with problems.

[382] A more detailed summary of CRS's position with respect to this issue can be found in Appendix B, Section 12.3.2.

Issue 15: Agency assessment against methodology criteria

[383] Unlevering is a process for separating the financial risk of a company from its business risk. Only in special circumstances, where a company is not publicly traded and beta cannot be directly calculated, does unlevering and relevering beta become an issue for the Agency.

[384] Because CN and CP are publicly traded companies, the Agency calculates company-specific market betas, which by their very nature are levered betas, in that market beta is the beta of the company and reflects its existing debt to equity ratio. Therefore, unlevering and relevering betas is not applicable to CN and CP.

[385] Despite CRS's criticism of the practice, the Hamada equation to unlever and relever betas is widely used among financial economists and experts, in part because it is the only known way to estimate beta for a non-publicly traded company.

Issue 15: Agency conclusion

[386] The Agency has not found an alternative approach for estimating beta in cases where a company is not publicly traded, either in practice or in academic literature. Therefore, considering that it is appropriate in terms of its criteria, the Agency will continue to employ this estimation methodology, as required, when determining the cost of equity for a private company.

DECISION CONCLUSION

[387] The methodology that will be used by the Agency, within its legislative mandate and effective with the determination for Crop Year 2012-2013, to determine cost of capital rates is set out in Appendix A.

[388] The Appendix A methodology will be applied by the Agency until at least 2018. The Agency retains the discretion to make adjustments to that methodology in the event of extraordinary circumstances that warrant such adjustments. Where such circumstances apply, any adjustments will be made by the Agency in a transparent manner.

• APPENDIX A
• APPENDIX B