HELPING TO

ELIMINATE POVERTY

THROUGH PRIVATE

INVOLVEMENT IN

INFRASTRUCTURE

TRENDS AND POLICY OPTIONS

How Profitable AreInfrastructure Concessions in Latin America?Empirical Evidence and Regulatory Implications

No. 2 • January 2005

Sophie SirtaineMaria Elena PingloJ. Luis GuaschVivien Foster

37568

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

© 2005 The International Bank for Reconstruction and Development / The World Bank1818 H Street, NWWashington, DC 20433Telephone 202-473-1000Internet www.worldbank.orgE-mail feedback@worldbank.org

All rights reserved.

1 2 3 4 07 06 05 04

The findings, interpretations, and conclusions expressed herein are those of the author(s) and do not necessarilyreflect the views of the Board of Executive Directors of the World Bank or the governments they represent.

The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors,denominations, and other information shown on any map in this work do not imply any judgment on the part of the World Bank concerning the legal status of any territory or the endorsement or acceptance of suchboundaries.

Rights and PermissionsThe material in this work is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The World Bank encourages dissemination of its work and will normally grant permission promptly.

For permission to photocopy or reprint any part of this work, please send a request with complete information tothe Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400,fax 978-750-4470, www.copyright.com.

All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@worldbank.org.

How Profitable Are Infrastructure Concessions in Latin America? iii

Acknowledgments vii

Abstract viii

Executive Summary ix

1. Objectives of the Study 1

2. The Sample 4

3. The Methodology 5

The Capital Asset Pricing Model 5

Measures of returns 6

Hurdle rates–Cost of capital 7The cost of equity 7The weighted average cost of capital 7Interpretations of results 9

4. Issues in Measuring Returns and Hurdle Rates: Cost of Capital 10

Data related issues 10Data consistency, quality, and availability 10Hurdle rates’ time sensitivity 10Concession data’s time sensitivity 10Concessions versus industry-specific data 11Estimating reasonable terminal values 11

Adjustments to financial returns 12Management fees 12Investments (transfer pricing) 12Other possible adjustments 13Acquisition value 13Pre-acquisition asset value 13Transfer pricing (other than for investments) 13Depreciation 13

TABLE OF CONTENTS

5. Computation of the Hurdle Rates: Cost of Capital 14

Computation of the cost of equity 14

Computation of the weighted average cost of capital 14

Variability during the concessions’ lifetime 16

6. Concession and Shareholder Returns 18

Concessions’ returns 18Overall concession returns 18Concession returns by sector 19Concession returns by country 21Concession returns by concession maturity 21Individual concession returns 22Concession return, a conclusion 22

Shareholders’ returns 23Overall shareholder returns 23Shareholder returns by sectors 25Shareholder returns by country 26Shareholder returns by concession maturity 27Individual shareholder returns 28Volatility of returns across concessions 28Volatility of returns from year to year 29Shareholder returns, a conclusion 29

Concession and shareholder returns, a conclusion 30

Sensitivity analysis 31Future concession growth 31Future concession dividend policy 33Future concession growth and dividend policy 34Acquisition price 35

7. The Impact of Regulation on Profitability 37

Conceptual framework 37

Measuring regulatory quality 38

Simple differential (myopic consumer-protection perspective) 39

Absolute differential (protecting both consumers and investors) 40

8. Conclusions and Policy Implications 42

Appendix 1: Measures of shareholders’ effective returns 44

The Shareholder Internal Rate of Return 44

The Return on Equity 45

The Project Internal Rate of Return 46

The Return on Capital Employed 47

iv Table of Contents

How Profitable Are Infrastructure Concessions in Latin America? v

Appendix 2: Computation of the cost of equity and the weighted average cost of capital 49

Computation of the cost of equity 49

Computation of the weighted average cost of capital 55

References 57

List of TablesTable 1: The sample of concessions used 4Table 3: Interpretations of various measures of return 7Table 2: Measures of return used 8Table 4: Guide to interpretation of results 9Table 5: Variation in the cost of equity and WACC over the concessions’ lifetime 17Table 6: Volatility of profitability by sector 21Table 7: Investment levels by sector 21Table 8: Investment levels by country 22Table 9: Historical concession returns by concession maturity 23Table 10: Dispersion of returns across concessions 25Table 11: Pay out ratios by sectors 26Table 12: Management fees by sector 27Table 13: Pay out ratios by country 28Table 14: Relation between return and concession maturity 29Table 15: Dispersion of adjusted RoE and Shareholder IRR across concessions 30Table 16: Construction of regulatory quality indices 38Table 17: Summary of regression results 39Table 18: Summary of regression results 40Table 19: Summary of second regression results 41Table 20: Unleveraged betas by sector 51Table 21: Typical leverage by sector 52Table 22: Nominal corporate income tax rates by country 52Table 23: Leveraged betas by sector and by country 53Table 24: Returns on stocks compared to government bonds 53Table 25: Country risk premiums 54Table 26: Cost of debt by country 55

List of BoxesBox 1: The required rate of return on an asset 5Box 2: Definition of the cost of equity 7Box 3: Definition of the weighted average cost of capital 9Box 4: Definition of the shareholder internal rate of return 44Box 5: Definition of the shareholders’ internal rate of return with a terminal value 45Box 6: Definition of the return on equity 46Box 7: Definition of the project internal rate of return 46Box 8: Definition of the project internal rate of return with a terminal value 47Box 9: Definition of the return on capital employed 47Box 10: Definition of the cost of equity 49Box 11: Leveraged and unleveraged betas 51Box 12: Definition of the weighted average cost of capital 55

vi Table of Contents

List of FiguresFigure 1: Private investment in infrastructure, 1990–2002 2Figure 2: Estimated cost of equity by country, May 2004 14Figure 3: Estimated cost of equity by sector; United States and Argentina, May 2004 15Figure 4: Estimated weighted average cost of capital by country, May 2004 15Figure 5: Estimated WACC by sector: the example of the USA and Argentina, May 2004 16Figure 6: Comparison of the estimated cost of equity and the estimated WACC 16Figure 7: Overall concession return 19Figure 8: Average annual profitability of sample concessions 19Figure 9: Long-term concession returns by sectors 20Figure 10: Average operational profitability by sector 20Figure 11: Average concession returns by country 22Figure 12: Average annual concession profitability by country 23Figure 13: Financial RoCE and Project IRR by concession 24Figure 14: Overall long-term shareholder returns 25Figure 15: Average annual return on equity of sample concessions 26Figure 16: Overall shareholder returns by sector 27Figure 17: Shareholder returns by country 28Figure 18: Financial RoE and shareholder IRR by concession 29Figure 19: Evolution of annual RoE by sector 30Figure 20: Evolution of the annual RoE by country 30Figure 21: Project IRR sensitivity to concession growth rates 31Figure 22: Shareholder IRR sensitivity to concession growth rates 32Figure 23: Shareholder IRR sensitivity to dividend payout 33Figure 24: Shareholder IRR sensitivity to concession growth rates and dividend payout 34Figure 25: Project IRR sensitivity to concession acquisition price 35Figure 26: Shareholder IRR sensitivity to concession acquisition price 36Figure 27: Evolution of the risk-free rate, 1960–2004 50Figure 28: Evolution of country risk premiums over time 54

How Profitable Are Infrastructure Concessions in Latin America? vii

The authors are all at the Finance, Private Sector andInfrastructure Department, Latin America andCaribbean Region (LACFPSI), the World Bank. J. LuisGuasch is, in addition, professor of Economics,University of California, San Diego. The authors aremost grateful for comments and suggestions from IanAlexander, Soumya Chattopadhyay, Antonio Estache,Danny Leipziger, Isabel Sanchez Garcia, and IliasSkamnelos. Partial funding from the Public PrivateInfrastructure Advisory Facility (PPIAF) is gratefullyacknowledged. Contact e-mail address jguasch@world-bank.org.

The findings, interpretations, and conclusionsexpressed in this paper are entirely those of the authorsand should not be attributed in any manner to the Public-Private Infrastructure Advisory Facility (PPIAF) or to theWorld Bank, to its affiliated organizations, or to mem-bers of its Board of Executive Directors or the countriesthey represent. Neither PPIAF nor the World Bank guar-antees the accuracy of the data included in this publica-tion or accepts responsibility for any consequence of theiruse. The boundaries, colors, denominations, and otherinformation shown on any map in this report do notimply on the part of PPIAF or the World Bank Group anyjudgment on the legal status of any territory or theendorsement or acceptance of such boundaries.

ACKNOWLEDGMENTS

viii Abstract

This report estimates the returns that private investorsin infrastructure projects in Latin America really madeon their investments, and assesses the adequacy of thesereturns relative to the risks taken—the cost of capital—and the impact that the quality of regulation had on thecloseness of alignment between returns and the cost ofcapital. This is done by estimating both historical andprojected future returns earned by a sample of privateinfrastructure concessions, across a variety of LatinAmerican countries and infrastructure sectors, and com-paring them against expected returns given the level ofrisk taken—the cost of capital. In this way, it is possibleto evaluate whether private investors did indeed earnabnormally high returns on their investments. Thereport develops a quality of regulation index and exam-ines the extent to which the quality of the regulatoryframework contributed to maintaining a closer align-

ment between rates of return and cost of capital, or didallow for the capture of excessive rents by the investorsor of excessive benefits by the users at the expense of theinvestors. The findings of this report are that contrary togeneral public perceptions, the financial returns of pri-vate infrastructure concessions have been modest andthat in fact for a number of concessions the returns havebeen below the cost of capital. On average telecom andenergy concessions have fared better than transport andwater. It also shows that the variance of returns acrossconcessions and countries is considerable; that the vari-ance of returns across concessions can be partiallyexplained by the quality of regulation; and that the bet-ter the quality of regulation the closer the alignmentbetween financial returns and costs of capital, as is desir-able. Thus this report shows and validates the claim thatregulation indeed matters.

ABSTRACT

How Profitable Are Infrastructure Concessions in Latin America? ix

Background and objectiveDuring the 1990s many countries in Latin Americaimplemented broad privatization and concession pro-grams for infrastructure services, with the aim of raisingfiscal revenues and improving sector performance. Adecade later many are questioning whether private sec-tor participation yielded the anticipated benefits, andwhether those benefits were equitably distributedamong the different stakeholders to the privatizationprocess. A frequent complaint is that investors may havecaptured a disproportionate share of the benefits in theform of excess profits over and above what was neces-sary to attract private capital into these sectors.However, to date there has been very little empiricalevidence against which to assess this claim.

The objective of this study is to estimate the returnsthat private investors in infrastructure projects in LatinAmerica really made on their investments, to assess theadequacy of these returns relative to the risks taken,and the impact that the quality of regulation had onthose returns relative to the cost of capital. The studydoes not attempt to evaluate the overall impact of pri-vatization and concession programs, but simply focus-es on the narrow aspect of profitability. This is done byestimating both historical and projected future returnsearned by a sample of private infrastructure conces-sions across a variety of Latin American countries andinfrastructure sectors and comparing them to expectedreturns, given the level of risk taken. In this way it ispossible to evaluate whether or not private investorsearn abnormally high returns on their investments. Inaddition, the study examines the extent to which thequality of the regulatory framework put in place at thetime of privatization contributed to maintaining a clos-er alignment between rates of return and hurdle rates,or the cost of capital.

Sample and methodologyThe study is based on a sample of 34 concessions thatare representative of global privatization trends in LatinAmerica from close to 1,000 infrastructure concessionsin the region. It includes companies from nine countrieswith widescale privatization programs: Argentina,Bolivia, Brazil, Chile, Colombia, El Salvador, Mexico,Peru, and Venezuela. The number of concessions in eachcountry has been chosen to be representative of the rela-tive importance of privatizations in that country. Thesample includes companies in four sectors: telecommuni-cations, water, electricity (generation and distribution),and transport. On average these concessions have beenin operation for seven years. It must be noted the dataused run to 2001. They are thus largely exempted fromthe impact of the recent crisis in Latin America, and it islikely that returns would have looked significantly worsehad 2002 and 2003 been included in the analysis.

To ensure sufficient quality of information, onlyaudited financial statements and official company pressreleases were used. The study does not attempt to adjustfinancial statements for differences in accounting stan-dards. It is recognized that regulation by return may cre-ate incentives for concessionaires to dress up theiraccounts to present the lowest profitability or returnpossible. As a consequence the profitability resultsimputed here ought to be construed as lower bound esti-mates of the true profitability of those regulated firms.However, the scope for such accounting distortions arelimited because 56 percent of the sample concessions arelisted companies or part of listed groups and financialstatements are audited.

Recognizing that private investors can be remuner-ated in various ways, two sets of returns are comput-ed. First, the financial returns resulting from the distribution of dividends from the concession to the

EXECUTIVE SUMMARY

x Executive Summary

concessionaires’ parent companies (mostly abroad)are computed. Second, the adjusted returns are com-puted by attempting to include indirect forms of divi-dends. The most common of these are managementfees, based on the assumption that all the explicitmanagement fees paid by concessions were in fact div-idends to their strategic shareholders. Another adjust-ment is made for the possibility of investment costmarkups, which arise when intragroup purchases arepriced above cost, thereby implicitly transferring dividends out of the concession toward the parentcompany.

The study is built on the Capital Adequacy PricingModel (CAPM), which formalizes the observation thatexpected returns are related to risk. A two-prongedapproach is used. The first step is to measure the over-all return which shareholders in each selected projectearned on the capital they invested in that project. Thesecond step is to determine whether those returns werecommensurate with the risk taken. Thus, the expostreturns that investors effectively earned on the asset orproject they invested in (effective returns) are comparedwith the threshold minimum return given the risk pro-file of the project-cost of capital (hurdle rates).

The study uses four measures of the effective returns:the shareholders’ internal rate of return (ShareholderIRR), the return on equity (RoE), the project internalrate of return (Project IRR) and the return on capitalemployed (RoCE). The first two are measures of thereturns earned by equity investors; the last two aremeasures of the profitability of the concessions overall,independent of their financing structure.

The measure of returns chosen dictates the natureof hurdle rates one needs to use. The Shareholder IRRand the RoE, both of which measure returns earnedover equity capital, must be compared to the appro-priate cost of equity (CoE), which is a measure of thereturn investors require on equity investments, giventhe level of risk of such investments. The Project IRRand the RoCE, which measures returns earned on theconcession’s overall capital structure, must be com-pared to the weighted average cost of capital(WACC), which represents the expected return on allof a company’s securities. Importantly, the appropri-ate benchmark value for each hurdle rate varies foreach project depending on the country and sector ofinvestment, reflecting that market risks also varyacross countries and sectors.

Conclusions and implicationsThe analysis shows that concessions are capable of gen-erating adequate returns in the long term, and arepotentially interesting business proposals. Concessionsin the water sector appear relatively the least attractive,while concessions in the telecommunications sectorappear to be the most profitable overall. On average,concessions seem to become profitable after about 10years of operation. However, about 40 percent of thesample concessions do not seem to have the potential togenerate attractive returns, with this number climbingto 50 percent in the energy and transport sectors.Concessions are thus risky businesses.

Low dividend distribution ratios have, however, nottranslated this overall profitability into adequate returnsfor shareholders to date. In fact, on average, concessionshareholders have so far earned negative returns ontheir investments, even including management fees, esti-mated accumulated capital gains, and potential invest-ment markups.

With historical growth maintained into the future,only telecom concessions would seem to have an inher-ent profitability high enough to generate adequatereturns to their shareholders in the long term, this, pro-vided they can capture annually the capital gains accu-mulated in their concessions over all years of operationand that the full value of their management fees corre-spond to dividends. In all other sectors, shareholderscan hope to earn long-term returns commensurate tothe risk taken only if the sectors consistently and signif-icantly outperform historical market growth. This con-clusion would not change if the concessionaires hadpaid up to 20 percent less for their concessions. Theimplication is that to build an adequate return, share-holders must rely both on various sources of remunera-tion (including dividends, management fees, and capitalgains), and on outperforming historical market growthconsistently, over the entire length of their concession.

These results suggest that concessionaires operatewith long-term perspectives, giving priority to growth-enhancing investments in the early years (at the cost ofdepressing returns in the short term), and relying on theentire concession period to build an adequate return.This may be driven by their contractual obligations,which usually require high investments in the earlyyears. It implies that early breaks of concession con-tracts may have a highly negative impact on expectedreturns.

How Profitable Are Infrastructure Concessions in Latin America? xi

The results also highlight that management fees maybe needed to build adequate returns, but that their treat-ment—from an accounting stand point they ought to betreated more like dividends than costs—ought to bemore transparent. In addition, allowing concessionshareholders to be fairly compensated at the end of theperiod for the capital gains accumulated during the lifeof the concessions is also an important component oftheir return.

The relatively low returns earned so far by conces-sion shareholders also suggest that either regulatorshave been tough at setting tariffs or that concession bid-ding processes have been successful in creating strongcompetition (aggressive bidding) among bidders, bring-ing their offered price to the limits of what made con-cessions interesting investments for them. That old con-cessions are on average more profitable than youngones suggests that returns may be depressed in earlyconcession years by inadequate prices, corrected afterthe first price control period (high investments in thefirst years of operation may also have a toll on youngconcession returns). This squares with the argumentsand data presented in Guasch (2004), where it appearsthat a significant number of concessions were won byaggressive bidding, perhaps too aggressive, and thatshortly afterward the contracts were renegotiated, oftengranting better terms to the operators. That would atleast partially explain why old concessions tend to bemore profitable than young ones.

The analysis also highlights that returns (in particu-lar shareholder returns) are highly volatile across sec-tors, concessions, and from year to year. Thus infra-structure concessions, in Latin America are a high-riskinvestment proposal, which explains why the requiredrates of return on such investments are high.

Given that virtually all the concessions included inthis study are regulated monopolies, their profitability is

not only a consequence of market conditions and man-agerial skills, but also partly a reflection of regulatorydecisions on service tariffs. A good regulator should aimto maintain alignment between a company’s rate ofreturn and its cost of capital in the medium term. Thisis because a rate of return in excess of the cost of capi-tal inappropriately penalizes consumers, while a rate ofreturn beneath the cost of capital inappropriately dis-courages further investment.

An evaluation of the quality of the regulatoryregimes faced by concessionaires in the study samplefinds that these do not score very high on average, andthat there is a high variance in the quality of regulato-ry frameworks across concessions. Furthermore, thequality of regulation is found to be a significant deter-minant of the divergence between the overall prof-itability of the concession and its corresponding hurdlerate, explaining around 20 percent of the variation.Thus regulation does matter. However, regulatoryefforts seem to be more closely associated with mini-mizing the simple IRR-WACC differential (and therebykeeping tariffs as low as possible for current con-sumers), than with minimizing the absolute IRR-WACC differential (and thereby keeping profitabilitywell aligned with hurdle rates of return). A striking fea-ture of the results is that regulatory quality variablesseem to have overall significance, more than individualsignificance, in determining IRR-WACC differentials.This is in fact consistent with the fact that performancealong different dimensions of regulatory quality is nothighly correlated, and that the benefits of high regula-tory quality along one dimension can be completelyoffset by low regulatory quality along another dimen-sion. Thus, for regulation to be effective, one needs thewhole package of regulatory characteristics. If some ofthe key ingredients are missing the effectiveness of reg-ulation is highly diminished.

During the 1990s many countries in Latin Americaimplemented broad privatization and concession pro-grams for infrastructure services. In aggregate, privateparticipation in infrastructure in less developed andemerging countries amounted to US$690 billion duringthe 1990s (World Bank 2003). The Latin America andthe Caribbean Region (LAC) proved to be the investors’preferred destination, receiving 50 percent (US$345 bil-lion) of worldwide private capital flows to the infra-structure sectors during the same period (figure 1a).Within LAC, these flows were predominantly channeledto the telecommunications and electricity sectors (Figure1b), and, moreover, heavily concentrated in a handful ofthe larger economies: Brazil, Argentina, Mexico, andChile (figure 1c).

LAC’s ability to attract such an inordinate share ofinfrastructure investment flows can be explained by theregion’s early opening of its infrastructure sector to pri-vate sector participation, the existence of substantiallevels of unmet demands in practically all infrastructuresectors, and perspectives of macroeconomic stabilityand reasonably high growth. In addition, to a muchgreater extent than in other regions, LAC went aheadwith major divestitures of public enterprises. Thus, it is estimated around 60 percent of these capital flowswere captured by the state as fiscal revenues associatedwith asset sales, while the remaining 40 percent wereinvested directly within the infrastructure sectors

Latin America’s private sector participation in infra-structure programs was generally part of a broader set ofpolicy reforms. The reforms were expected to improvemuch needed sector performance, increase levels of serv-ice coverage, and attract private sector financing forlong-delayed investments in infrastructure expansionand upgrading, thereby enabling scarce public funds tobe used for investment in the social sectors and for thecreation of fiscal benefits by creating sale revenues and

reducing ongoing subsidies. After a decade of reform,popular support for privatization around the region hasdwindled, and public debate increasingly questions theextent to which these reforms delivered the anticipatedbenefits, and (if so) whether these benefits were equi-tably distributed among different stakeholder groups.

With any privatization process, there are a number ofdistinct stakeholder groups whose interests are likely tobe affected. First, the state has major fiscal interests inprivatization transactions, standing to gain from priva-tization proceeds, as well as from any reductions in sub-sidy or increases in tax revenues, often made possible asa result of privatization. Second, the interests of currentconsumers will be affected by the resulting changes inthe price and quality of the services provided, while newconsumers may be incorporated as service areas expand.Third, the interests of employees will be affected as aresult of potential layoffs and changes in the pattern andconditions of employment. Fourth, the extent to whichtransactions are designed to generate benefits for theother stakeholder groups, as well as the quality of sub-sequent regulatory decisions, will affect the residualprofitability of the enterprise to the private investors.

The huge complexity of privatization transactions, aswell as their major ramifications for the economy’s gen-eral equilibrium, make it difficult to generalize as tohow the costs and benefits of privatization will play outacross the different stakeholder groups in any particularcase. However, a relatively new, but growing, literatureaims to document the economic and distributionalimpact of privatization (Andres, Foster, and Guasch2004; Birdsall and Nellis 2002; Nellis 2003; McKenzieand Mookherjee 2004; Ugaz and Waddams-Price 2003;and Chong and Lopez-de-Silanes 2005). The emergingconclusions of this literature are that the efficiency gainsand increases in quality in the provision of infrastruc-ture services and fiscal payoffs of privatization have

How Profitable Are Infrastructure Concessions in Latin America? 1

OBJECTIVES OF THE STUDY

1.

been substantial; that while the layoffs of workers havebeen large relative to the size of the industry but smallrelative to the workforce as a whole, overall sectoremployment levels have increased on average after a fewyears from the transactions; that existing customershave generally seen quality improve but have sometimeshad to pay higher prices in return; and that serviceexpansion has accelerated bringing major benefits tothose previously unserved.

So far this literature has had relatively little to sayabout the extent to which private investors have bene-fited from the privatization process. There are popularperceptions that investors have profited excessivelyfrom privatization transactions, repatriating dividendsto their countries of origin instead of reinvesting themin the host country. However, there had been no sys-tematic empirical evidence from which to evaluatesuch a claim.

2 Objectives of the Study

Source: World Bank 2003.

Figure 1: Private investment in infrastructure, 1990–02

(c) For Latin America, by country(b) For Latin America, by sector

(a) Across regions

Private investments in infrastructure in less developed countries

0

20

40

60

80

100

120

140

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2002

US$

bn

Latin America & Carribean East Asia & Pacific Other

25%

46%

28%

21%

5%

35%

16%

7%

17%

TelecomEnergyTransportWater

BrazilArgentinaMexicoChileOthers

The objective of this study is, therefore, to estimatethe returns that private investors in infrastructure proj-ects in Latin America really made on their investmentsand to assess the adequacy of these returns relative tothe risks taken. The study does not attempt to evaluatethe overall impact of privatization and concession pro-grams, since this, as mentioned, has already been under-taken elsewhere, but simply focuses on the narrowaspect of concession profitability. The study estimatesthe historical and projected future returns of a sample of

private infrastructure concessions, across a variety ofLatin American countries and infrastructure sectors,and compares them with expected returns given thelevel of risk taken—the cost of capital. In this way, itevaluates whether private investors earned abnormallyhigh or low returns on their investments. In addition,the study examines the extent to which the quality ofthe regulatory framework put in place at the time of pri-vatization was a factor in aligning rates of return andcost of capital.

How Profitable Are Infrastructure Concessions in Latin America? 3

As of 2003, there were more than 1,200 infrastructureconcessions in Latin America with private sector partic-ipation. From that universe of private contracts, a sam-ple of 34 concessions was selected, using the followingcriteria: (a) to include most Latin American countrieswith meaningful privatization programs; (b) to includecompanies from all main infrastructure sectors; (c) tofocus on companies with at least five years of operation(to have a time series of data of adequate duration forthe analysis); and (d) to focus on companies publishinggood quality financial statements.

The resulting sample of 34 companies are represen-tative of global privatization trends in Latin America. Itincludes companies from nine countries with wide-scale

privatization programs in the region: Argentina, Bolivia,Brazil, Chile, Colombia, El Salvador, Mexico, Peru, and Venezuela. The number of concessions in eachcountry has been chosen to be representative of the relative importance of privatization in each country. Thesample includes companies in four sectors: telecom-munications, water, electricity (generation and distri-bution), and transport. In the latter case, the sample is restricted to four companies in the road and port subsectors. They cannot, therefore, be considered repre-sentative of the entire transport sector. Airport conces-sions in particular are not included in the sample. Onaverage the sample concessions have been in operationfor seven years.

4 The Sample

THE SAMPLE

2.

Table 1: The sample of concessions used

Number of concessions Telecom Water Energy Transport Total

Argentina 2 3 4 2 11Bolivia 1 1 1 - 3Brazil 1 1 4 - 6Chile 1 3 - 1 5Colombia - 2 - - 2Mexico 1 - - - 1Panama - - - 1 1Peru 1 - 3 - 4Venezuela 1 - - - 1Total 8 10 12 4 34

The study uses a two-pronged methodology. First, itmeasures the overall return which shareholders in eachselected project earned on the capital they invested inthat project. Second, it determines whether thosereturns are adequate given the risk taken by comparingthem to appropriate hurdle rates.

Absolute measures of returns are meaningless sincethey fail to recognize that private investors are not will-ing to invest in all potential projects for the samereturns. This is because risk-averse investors perceivethat the risks associated with various investments differ.The higher their perception of the riskiness of a specificinvestment, the higher the return they will require inorder to make that investment.

Intuitively, this is because financial managers realizethat, all else being equal, risky projects are less desirablethan safe ones. Therefore, they demand a higher expectedrate of return from risky projects. The observation thatexpected returns are related to risk has been formalizedin the Capital Adequacy Pricing Model developed in the1960s (Sharpe 1964; Lintner 1965).

The Capital Asset Pricing ModelThe CAPM model is based on the idea that investorsdemand higher expected returns if asked to take onadditional risk. More precisely, it tells us that investorsin a project will require earning an expected returnwhich compensates adequately for the risk embedded inthe project. This required rate of return, called the hur-dle rate, is the expected return above which an invest-ment makes sense and below which it does not. For acompany, this hurdle rate is equivalent to its opportu-nity cost of capital; that is, the rate of return the com-pany can otherwise earn at the same level of risk as theinvestment it is considering (see box 1).

The CAPM model shows that this return should be atleast equal to the return the company can earn on a risk-

free investment plus a risk premium that compensatesfor the nondiversifiable risk embedded in the project.

Some risks can be eliminated by appropriate diversi-fication. These risks are called unique risks, becausethey measure the perils that are peculiar to one particu-lar company or project, but can be eliminated by anappropriate portfolio diversification. Investors are notremunerated for these risks, since it is their job to ade-quately diversify their portfolio to eliminate them.However, there are some risks which cannot be avoidedhowever much you diversify. These are called marketrisks. They stem from the fact that there are economy-wide perils which threaten all businesses in an economy.Since investors cannot diversify these risks away, theywill only accept to invest in a risky asset (that is, an assetsensitive to market risks) rather than in safe ones (thatis, an asset nonsensitive to market risks) if they are ade-quately compensated for the extra risk taken.

How Profitable Are Infrastructure Concessions in Latin America? 5

THE METHODOLOGY

3.

ra = rf + ß * (rm – rf)

Where: ra = required return on the asset, i.e. the hurdle rate to use when deciding whether to invest in the asset or not

rf = risk-free rate, i.e. the return of a risk-free investment

ß = beta of the asset

rm = market return, i.e. the return on the market as a whole, that is on a fully diversified portfolio

ß * (rm – rf) is the asset risk premium

The required rate of return on an asset

BOX 1

Therefore, it is futile thinking about how risky aninvestment is in isolation. One needs to measure howsensitive that investment is to market movements. Thissensitivity is called beta (ß). The CAPM theory showsthat the premium investors require in exchange forholding a riskier (more volatile) asset (called the assetrisk premium) varies in direct proportion to its beta.

As the formula suggests, hurdle rates are asset- orproject-specific. Since they represent the rate at which aspecific project makes sense for an investor, given thatproject’s own degree of risk, it is logical that their valuesdiffer from project to project.

Effective returns are the ex-post returns thatinvestors effectively earned on the asset or project theyinvested in. They may be very different from thereturns investors expected to earn ex-ante and on thebasis of which they made their original investmentdecision. Excess returns are returns an investor hasgained in excess of those required originally accordingto the CAPM.

This study intends precisely to investigate whetherthe effective returns earned by private investors in infra-structure projects in Latin America have been commen-surate to their expectations, given the risks taken. Usingthe two-pronged methodology described previously, onefirst needs to define adequate measures of the effectivereturns earned by private concessionaires in LatinAmerica, and then one needs to compare them withappropriate hurdle rates. Each of these issues will nowbe looked at in turn.

Measures of returns Several measures of the effective returns earned by con-cessionaires can be used. In this study, the ShareholderInternal Rate of Return (Shareholder IRR), the Returnon Equity (RoE), the Project Internal Rate of Return(Project IRR), and the Return on Capital Employed(RoCE) is used. Appendix 1 provides detailed defini-tions of each of these measures.

The first two (the Shareholder IRR and the RoE) aremeasures of the returns earned from dividends by equi-ty investors in the project company (the shareholders),while the last two (the Project IRR and the RoCE) aremeasures of the concession’s overall profitability, inde-pendent of their financing structure. These last two aremeasures of the average return earned by equity anddebt investors into the project company, while the firsttwo indicators are measures of the returns earned byequity holders only.

The Shareholder IRR and the Project IRR measurereturns earned over several years, while the RoE and theRoCE are annual measures of returns (see Table 2).

Note that the Shareholder IRR used in the analysis isbased on dividends (and other direct financial flows toshareholders) only. It does not incorporate the value cre-ated by re-investing part of the generated earnings intothe concessions. This value is captured in the overallProject IRR. Shareholders capture it through increasesin the share price or value of their company.1 This valuehas been incorporated in the calculation of ShareholderIRRs by way of a terminal value only. This is because,since most concessions are not listed or sellable, accu-mulated capital gains cannot be cashed-in by sharehold-ers. The only way shareholders can really cash-in theaccumulated value is at the end of the concession, whenit is re-bid, and this, provided they get a fair compensa-tion for the value created. The value added created byretaining earnings into the concessions by way of a ter-minal value has been included instead.

The measures of shareholder returns used would,therefore, underestimate the returns earned by conces-sion shareholders if the latter could freely sell theirshares in the concession companies to cash-in accumu-lated capital gains, or if the value accumulated in theconcessions was fully reflected in the value of their owncompanies.

To conclude, the four measures of return used can beinterpreted as summarized in Table 3.

The Shareholder IRR and the Project IRR have beencalculated over three distinct horizons. The firstincludes historical dividends/free cash flows only. Theythen measure the effective return earned by sharehold-ers/the concession to date. Second, they have been com-puted to include historical dividends/free cash flows andthe future value (FV) of dividends/free cash flows to bereceived annually until the concession’s last year ofoperation. They then measure the potential returnwhich shareholders/the concession can hope to earnuntil the end of the concession from annual flows.Finally, they have been computed including historicaland future dividends/free cash flows and a terminal

6 The Methodology

1 It is, however, difficult to isolate share price movements resultingfrom earnings announcements in concession subsidiaries from otherevents influencing share prices. In addition, recognizing that thevalue accumulated in concessions is subject to political risk, asexamples of expropriations have shown, the markets tend to valueit only partly.

How Profitable Are Infrastructure Concessions in Latin America? 7

Tabl

e 2:

Mea

sure

s of

ret

urn

used

Mea

sure

s of

sha

reho

lder

ret

urns

Mea

sure

s of

con

cess

ion

retu

rns

Char

acte

rist

ics:

Rem

uner

ate:

Eq

uity

hol

ders

Equi

ty a

nd d

ebt

hold

ers

Indi

cate

:Re

turn

ear

ned

by s

hare

hold

ers

on e

quity

inve

stm

ent

Prof

itabi

lity

of c

once

ssio

n as

bus

ines

s en

tity

and

inve

stm

ent

prop

osal

Base

d on

:Fl

ows

avai

labl

e fo

r di

vide

nd d

istr

ibut

ion

or a

ctua

l div

iden

dsFr

ee c

ash

flow

s av

aila

ble

afte

r in

vest

men

t an

d w

orki

ng c

apita

l fin

anci

ng

Inte

rpre

tati

on:

Indi

cato

r:Ba

sed

on:

Indi

cato

r:Ba

sed

on:

Ann

ual r

etur

n Ro

EN

et in

com

eRo

CE

Earn

ings

bef

ore

inte

rest

cha

rges

Tota

l ret

urn

earn

ed u

p to

tod

ay

Shar

ehol

der

IRR

His

toric

al a

nnua

l div

iden

dsPr

ojec

t IR

RH

isto

rical

ann

ual f

ree

cash

flo

w (F

CF)

Tota

l ret

urn

earn

ed o

ver

entir

e Sh

areh

olde

r IR

R w

ith F

VH

isto

rical

ann

ual d

ivid

ends

and

Pr

ojec

t IR

R w

ith F

VH

isto

rical

ann

ual F

CF

and

futu

re

life

of t

he c

once

ssio

n fu

ture

ann

ual d

ivid

ends

unt

il la

st

annu

al F

CF

until

last

yea

r of

ye

ar o

f co

nces

sion

conc

essi

onTo

tal r

etur

n ea

rned

ove

r en

tire

Shar

ehol

der

IRR

His

toric

al a

nd f

utur

e an

nual

Pr

ojec

t IR

R w

ith T

VH

isto

rical

and

fut

ure

annu

al F

CF,

lif

e of

the

con

cess

ion,

incl

udin

g w

ith T

Vdi

vide

nds,

plu

s sh

areh

olde

r pl

us f

air

valu

e of

con

cess

ion

at t

he

for

resi

dual

val

ue a

dded

com

pens

atio

n fo

r fa

ir va

lue

of

end

conc

essi

on a

t th

e en

d

8 The Methodology

value (TV) measuring the fair compensation sharehold-ers should receive for the value added they created in theconcession. They then measure the potential returnwhich shareholders/the concession can hope to earnuntil the end of the concession from annual flows andfrom the compensation they should receive at the end ofthe concession for the value added created.2

Hurdle rates–Cost of capital Once the effective returns earned by project sharehold-ers have been calculated they need to be compared withappropriate hurdle rates. The appropriate hurdle ratedepends on the measure of returns used. In addition, theappropriate benchmark value for each hurdle rate willvary for each project, depending on the country and sec-tor of investment (since market risks vary across coun-tries and sectors).

The measure of returns chosen dictates the nature ofhurdle rates one needs to use. In particular, theShareholder IRR and the RoE, both of which measurereturns earned over equity capital, must be compared tothe appropriate cost of equity, while the Project IRR andthe RoCE, which measure returns earned on the con-cession’s overall capital structure, must be compared tothe weighted average cost of capital.

The cost of equityThe cost of equity is a measure of the return investorsrequire on equity investments, given the level of risk ofsuch investments. It is the appropriate hurdle rate formeasures of returns on equity investments. It is usuallyestimated using the CAPM.

In our calculations, the risk premium was brokeninto two components: (a) the stock market risk premi-um, measured by ß * (rm – rf), corresponding to theextra return investors require to invest in stocks ratherthan in a risk-free asset; and (b) the country risk premi-um (CRP), corresponding to the extra return investorsrequire to invest in stocks of companies in a countrydeemed riskier than a less risky country used as bench-mark (see box 2). More details on these two compo-nents are provided in appendix 2.

The weighted average cost of capitalThe weighted average cost of capital (WACC) repre-sents the expected return on all of a company’s securi-ties. It is measured as the average of the returns requiredon each source of capital, such as stocks, bonds, and

2 Such compensation is usually calculated on the basis of the non-amortized value of the concession’s assets, or of the new biddingprice if the concession is re-bid to new private investors. The secondmethod was preferred in this study since it is dynamic and forward-looking. Therefore, the terminal value as a perpetuity of average div-idends/free cash flows over the last three years of operation of eachconcession was calculated, adjusting for exceptional items.

Table 3: Interpretations of various measures of return

Indicator Interpretation

Shareholder Internal Rate of Return Net financial return earned from dividends by shareholders on their (Shareholder IRR) equity investment in the project companyReturn on Equity (RoE) Measure of the annual after tax return the concession is earning on

its equity capitalProject Internal Rate of Return Net financial return generated by the concession in the form of free (Project IRR) cash flows available to remunerate its various financing sources

(including debt and equity)Return on Capital Employed (RoCE) Measure of the annual net operating profitability of the concession,

measuring its ability to service its overall long-term financing structure

CE = rF + ß * (rm –rf) + Crp

Where: rf = risk-free rate

ß = beta of the project

rm = expected stock market return

Crp = country risk premium

Definition of the cost of equity

BOX 2

other debts, weighted by the shares of each source ofcapital in the company’s financing structure. The calcu-lation is often simplified by grouping the varioussources of financing into two categories only, equity andfixed income instruments. It is the appropriate hurdlerate to use for measures of returns on a concession’soverall liabilities (see Box 3).

Interpretations of resultsIn the analysis, the Shareholder IRR and RoE of eachconcession is compared to the appropriate CE and theProject IRR, and RoCE is compared to the appropriateWACC. The results will be interpreted as summarized inTable 4.

How Profitable Are Infrastructure Concessions in Latin America? 9

WACC = E / (D + E) * CE + D / (D + E) * (1-T)* CD

Where: E = book value of equity

D = long-term debt

CE = cost of equity (as measured above)

CD = cost of debt

T = nominal corporate income tax rate

Definition of the weighted average cost of capital

BOX 3

Table 4: Guide to interpretation of results

Result: Interpretation:

Shareholder IRR > appropriate CE The shareholders in the project have earned excess returns compared with those commensurate to the risk taken

Shareholder IRR < appropriate CE The shareholders have not earned returns commensurate to the risk takenRoE > appropriate CE The concession has returned a post-tax profitability on its equity capital superior

to that of alternative investments of similar riskRoE < appropriate CE The concession has returned a post-tax profitability on its equity capital inferior

to that of alternative investments of similar riskProject IRR > appropriate WACC The concession has generated positive net financial flowsProject IRR < appropriate WACC The concession has generated negative net financial flowsRoCE > appropriate WACC The concession’s net operating profitability exceeds the level necessary to

adequately service its debt and equityRoCE < appropriate WACC The concession’s net operating profitability is insufficient to adequately service

its debt and equity

Data related issues

Data consistency, quality, and availabilityTo ensure sufficient quality of information, only audit-ed financial statements and official company pressreleases were used. Concessions have to follow eachcountry’s particular accounting standards. Therefore,the data provided by each concession in the sample maynot be fully consistent since the concessions may use dif-ferent accounting rules to prepare their financial state-ments. Although accounting standards in all the coun-tries under consideration are broadly based on interna-tional accounting standards (IAS), there remain somesignificant discrepancies which may generate differencesin earnings. No attempt has been made to adjust finan-cial statements for differences in accounting standards.

In addition, some data that would have been impor-tant for the analysis are generally not published by com-panies, whatever the country in which they operate.This applies for instance to the fair value of some assets,depreciation/amortization rules, and the detailed classi-fication of costs. It also applies to the market value ofassets and liabilities, so that the analysis is based ontheir book value.

Finally, some argue that regulation by return some-times creates incentives for concessionaires to dress uptheir accounts in order to present the lowest profitabil-ity or return possible. This would happen when regu-lated tariffs are set so as to ensure a minimum return toconcessionaires, who, therefore, have the incentive tominimize their historical return in order to maximizefuture tariff increases. However, many of our sampleconcessions are listed companies (56 percent)3 or partof listed groups. In this case their managers might have

some level of the opposite incentive to maximize theconcession’s profitability to create as much shareholdervalue as possible (through a share price increase). Inany case, financial statements are audited; so with allthe appropriate caveats, the leeway to window dressbalance sheets remains limited. Balancing both effects,the former is bound to dominate the latter. Thus, overall the imputed estimated profitability or returnestimate here is at least a lower bound of the true prof-itability.

Hurdle rates’ time sensitivitySome of the data used to estimate the appropriate hur-dle rates vary constantly over time. This is, for instance,the case of country risk premiums and betas. Both tendto vary as the market incorporates new information onthe country, sector, or company.

The cost of equity and weighted average cost of cap-ital have been computed at three different points intime: (a) at the start of each concession—this representsthe hurdle rate which investors would have used whenthey assessed whether to invest in a concession or not;(b) on average over the concession’s life to date—thisrepresents the opportunity cost that investors have facedon average since they invested into the projects; and (c)today—this represents the current opportunity cost ofhaving invested money in a specific project.

It must be noted the data used is up to 2001. It is thusmostly exempted from the impact of the recent crisis inLatin America. It is highly probable that returns wouldhave looked worse were 2002 and 2003 included in theanalysis.

Concession data’s time sensitivityThe concessions’ financial results are usually sensitive totheir life cycle. It is not uncommon to make losses in thefirst years as operational processes are optimized andheavy investments are often made. By contrast, prof-itability usually increases in later years as the system

10 Issues in Measuring Returns and Hurdle Rates-Cost of Capital

ISSUES IN MEASURING RETURNS ANDHURDLE RATES: COST OF CAPITAL

4.

3 The percentages vary by sector, however: Transport (0 percent),Water (30 percent), Energy (75 percent), and Telecommunications(88 percent).

matures and reaches an appropriate level of efficiency.Therefore, mixing concessions at different stages oftheir life cycles is not ideal. As a result, the returnsearned by individual concessions will not be compared,as they may not be comparable. The problem is not asacute, however, when computing averages for the wholesample since our sample includes several concessions ateach stage of their life cycle.

Concessions versus industry-specific dataWhen estimating the cost of equity and the weightedaverage cost of capital, the salient choice is to use indus-try averages for the value of betas, the structure of cap-ital, and the cost of debt, rather than those of the spe-cific company under consideration. This is becauseindustry averages are deemed representative of bestpractices in the sector. This ensures that risks that can beeliminated with prudent financial management are notremunerated. Using an industry average penalizes proj-ects which are highly leveraged, since the hurdle rateresulting from industry averages would include a lowerproportion of debt and, therefore, be higher (assumingthe cost of debt is lower than the cost of equity) than theproject’s own hurdle rate. This means investors consid-ering investing in a highly leveraged project may need toconsider lowering its leverage—and, consequently, itsfinancial risk—to make it attractive.

Another advantage of using industry averages is toensure a higher comparability of results. Otherwise, itmay be difficult to assess whether the extra return gen-erated by a given project is not largely due to its higherleverage.

Finally, industry betas are often more reliable thanindividual betas. This is because individual betas mustoften be estimated from a limited-time series of data,which exposes the results to potentially large estimateerrors. Fortunately, these errors tend to cancel out whenone estimates betas of diversified portfolios.

Some, by contrast, argue that industry values are notrelevant because the objective is precisely to assess theconcession’s specific cost of capital, given its existingfinancial structure.

This study favors arguments in favor of using indus-try averages, and they have been used in this study’s cal-culations. In addition, for many of the concessions ana-lyzed, specific betas were simply not available.

Estimating reasonable terminal valuesIn a net present value computation, the terminal value(TV) is often the largest amount. This is especially truefor young concessions, for which the TV includes a largenumber of future years. Estimating the TV reasonably is

important given the impact it has on results. Thisrequires properly estimating the flows to be projectedinto the future and their growth rate.

As with any projection into a distant future, it isimpossible to forecast it with certainty. In the case ofinfrastructure concessions, projecting future flows ismade even more arduous by their high sensitivity to eco-nomic ups and downs, political events, and so forth. Allone can do is be as reasonable as possible, which is oftenachieved by making simple assumptions.

The flows to be projected into the future were esti-mated on the basis of the average of those of the lastthree available historical years (1999, 2000, 2001).This means that for the Shareholder IRR the averagedividends distributed by each concession over the lastthree years was used, while for the Project IRR theaverage net financial flows generated by each conces-sion over the last three years was used. Averages overthe last three years were taken to compute recurringflows independent of yearly variations and to be repre-sentative of the future.4

It was then assumed the resulting flows would growat a constant rate. The same growth rate for dividendsand concession cash flows were used (which may beinterpreted as meaning that dividend payout ratios willremain constant). The latter was taken as equal to thelong-term historical GDP growth rate of the country ofeach concession. This assumption means that each con-cession will grow at the same rate as the economyaround it and that the future growth rate of the econo-my will be equal on average to its historical level. Eachcountry’s historical growth rate over a 39-year horizon,from 1961 to 1999, was used. This ensures that severaleconomic cycles are included and that the data is notbiased by particular economic circumstances. Theresulting growth rates are: Argentina (2.7 percent),Bolivia (2.8 percent), Brazil (4.8 percent), Chile (4.4percent), Colombia (4.2 percent), Mexico (4.7 percent),Panama (4.6 percent), Peru (3.2 percent), and Venezuela(2.7 percent).5

The resulting flows were projected into perpetuity toobtain the TV.

In practice, if the concessions were to be re-bid at theend of their current contract, a growing perpetuity offuture flows would be only one of the methods used to

How Profitable Are Infrastructure Concessions in Latin America? 11

4 For a very young concession, this can underestimate returns, sincethe last flows may still be of a small amount. Note also that anyexceptional amount from the calculation was removed.

5 Source: IMF, 2002.

estimate their value. Prospective bidders would alsolook at the net asset value of the concession and at trad-ing and transaction multiples (that is, multiples of earn-ings and assets at which similar businesses trade or havebeen transacted). The perpetuity method was preferredbecause it is dynamic and forward looking, while theother methods are mostly static and based on historicalnumbers.

Adding a TV may, however, overestimate the com-pensation concessionaires can reasonably expect. First,not all concession contracts include the payment ofcompensation to shareholders for the value created atthe end of the concession’s operation. Second, it cannotbe excluded that a future government negotiates achange to or is not in the capacity to pay such com-pensation. Third, compensations are often based on thenonamortized value of assets, which may be very dif-ferent from the value of estimated future flows. In addi-tion, concessionaires generally expect to earn a suffi-cient return over a reasonable period of time and donot rely on the TV to get a sufficient return. They willtend to rely on returns including the future value (FV)of flows until the last year of the concession, butexcluding the value of flows beyond the concessionoperation period.

Adjustments to financial returnsAs explained earlier, financial returns were adjusted rec-ognizing that the financial accounts of a company maynot always be fully representative of its economic situa-tion. Two adjustments were made, related to manage-ment fees and investments. Other possible adjustmentswere run as sensitivities.

Management feesFor many of the concessions, a management contractwas signed between the concession company and itsshareholders (see table 12, page 27). These contractstypically called for various services to be provided to theconcession by its shareholders, including the transfer oftechnology, development of general policies, prepara-tion of detailed strategic plans, design of formal organi-zational structures, hiring of qualified staff, and formu-lation of annual operating budgets. These services aregenerally remunerated with a fee payable to the conces-sionaire by the concession, normally defined as a per-centage of sales or operating profits.

In the concession’s accounts, these fees are usuallytreated as tax-deductible operating costs. Therefore,they reduce the concession’s operating and net incomes.However, it is often argued they do not remunerate real

services or that their value is frequently inflated, so thatall or part of them should be considered as dividends. Inthis case, the concession’s operating and net incomeshould be corrected (that is, increased), and the fees paidshould be added to dividends in the computation of theinternal rate of return. All returns would thus be higher.

To be as conservative as possible, returns have beencorrected, assuming that all the explicit managementfees paid by concessions were in fact dividends to theirstrategic shareholders. It is important to understandthat this assumption implies that none of the services orknow-how the management fees were supposed toremunerate are considered real. If the concession reallyneeded to use those services or to buy that know-how,and had to acquire them from an outside party, theircosts would be real costs to the concession and no cor-rection would be made to its RoE and RoCE, while theShareholder IRR should only be increased by the mar-gin earned by shareholders on those services. Hence, byincluding them entirely as dividends, it is assumed theconcession could have reached the same operating andnet incomes without any of these services or know-howtransfer. This is clearly a strong assumption.

It must also be noted that some concessions may notrecognize the management fees paid to their concession-aires as such. The latter might be disguised under othercost categories, such as technical assistance costs.

Investments (transfer pricing)When concessions invest in new machinery and equip-ment or build new facilities, they tend to use related par-ties to execute the works. Some argue that related par-ties are likely to charge higher prices for these servicesand works than if competitive bidding was organized,due to the monopolistic nature of the transaction. As aresult, the transfer price is inflated and the concession’sincome reduced.

It is obviously very difficult to assess the fairness ofsuch transactions since there is precisely no competitivebid with which to compare them. In fact, most countriesinclude transfer pricing audits in the audits requiredfrom concessions. These require full disclosure of intra-group transactions, and compare their prices with thoseof similar transactions internationally. Therefore, signif-icant inflation of transfer prices should not go unnoticedby auditors. Also, not all concessions’ investments aremade by related parties.

Therefore, the value of the total investments madeby each concession were reduced by only 10 percent,and, likewise, the annual depreciation charges, assum-ing that about 30 percent of all investments were made

12 Issues in Measuring Returns and Hurdle Rates-Cost of Capital

by group companies with a price inflated on average by33 percent.

With such adjustment, the operating and net incomesare higher (as depreciation charges are lower), whichleads to higher return on equity and on capitalemployed. Concession returns are also higher becauseinvestments are smaller. The shareholder IRR is alsohigher since it was assumed these investment costmarkups were direct benefits for the strategic share-holders’ companies, which could then distribute them tothemselves as dividends (after tax).

Other possible adjustmentsIn addition to the two adjustments made, several otheradjustments could have been made. However, these fur-ther adjustments would have been very difficult to esti-mate and quite uncertain. They have, therefore, notbeen considered in this analysis.

Acquisition valueWhen bidding for a concession, concessionaires oftenpay an acquisition premium over the fair value of thetarget firm, known as goodwill. This acquisition premi-um is triggered by the competition created by the bid-ding process. The buyer is willing to pay such premiumsif the asset provides a competitive advantage, such as anentry point in a new market, or a stronger brand name.

Normally, the goodwill appears on the balance sheetof the acquirer in the amount by which the purchaseprice exceeds the net tangible assets of the acquiredcompany, and the goodwill is amortized over a fewyears, creating a recurrent cost in the company’saccounts. For all the concessions analyzed, however,the goodwill was not recognized. Rather, the value ofthe assets of the acquired companies was increased sothe net asset value of the acquired company was madeequal to the purchase price. This implies the entire pur-chase price was incorporated in the capital base of theconcession, which reduces returns (in particular theProject IRR). However, the acquisition premium shouldbe left in the concession’s capital base only if it repre-sents real market value. If it represents a premium theacquirer was willing to pay for strategic reasons, forinstance, then it represents additional value for the cur-rent buyer, but probably not for a future purchaser. Inthis case, it needs to be progressively eliminated byamortizing the goodwill.

In addition, overvalued assets lead to higher depreci-ation charges, which lower concessions’ returns on equi-ty or capital employed further (by reducing operatingand net incomes). These charges are usually larger than

the goodwill amortization charges (spread over a longerperiod of time), so that the net effect on returns is neg-ative. One could adjust returns for the differencebetween the goodwill amortization charges and thedepreciation charges on the overvalued assets.

From the point of view of shareholders, however,what matters is the amount they effectively paid forthe assets and on the basis of which they hope to gaina sufficient return. From that angle, no adjustment isneeded.

Pre-acquisition asset valueCompanies often revalue their assets shortly after theirprivatization. A revaluation increases the concession’sasset base but also depreciation charges, and reducesrates of return. Frequently, little information is provid-ed on the basis of these revaluations, and doubts ariseon their economic foundations. However, the lack ofinformation makes any reasonable adjustment impossi-ble to make.

Transfer pricing (other than for investments)In the same way investments are often carried out byrelated parties to a concession, many other services canbe provided to the concession by group companies. Themost common suspicious accounts are repairs andmaintenance, consulting services, technical assistance,rent, and general services. Again, the prices of the relat-ed services are likely to be inflated as a result of themonopolistic nature of the transactions. However, com-panies tend to disclose little information on transferprices other than for large transactions, so that it is notpossible to make any reasonable assumption on theirtrue extent. In addition, all intragroup transactions aresupposed to be audited for transfer prices so that thereshould be limited room for overpricing. In any event,regulators are encouraged to require concessions to doc-ument and disclose their transfer pricing policies asmuch as possible.

DepreciationDepreciation periods for infrastructure assets can some-times be shorter than the actual life of the assets (accel-erated depreciation). This increases depreciationcharges in the short term and reduces returns. Eventhough accelerated depreciation is normally allowed bymost countries’ tax and accounting rules, from an eco-nomic point of view depreciation should be made overthe actual life of the assets. However, adjusting for thiswould only be possible if a detailed depreciation sched-ule was available for each concession’s assets.

How Profitable Are Infrastructure Concessions in Latin America? 13

Computation of the cost of equityAs explained above, the cost of equity has been calcu-lated on the basis of CAPM. The required parametershave been estimated as described in Appendix 2. Figure2 shows the resulting current costs of equity for eachcountry.

The figure indicates, for instance, that an investorlooking to invest today in a concession in Argentinawould need to earn an internal rate of return of at least19 percent on the investment. If the concession’s finan-cial projections show the internal rate of return on theinvestment is likely to be less than 19 percent, the invest-ment would not be worth it, as the investor should beable to find alternative investments with a similar levelof risk but a higher expected return. The same investor

would only need to earn an internal rate of return of 6percent to invest in an American concession and 7 per-cent in a Chilean concession.

Since the main discriminating factor is the countryrisk premium (see Appendix 2), the cost of equity tendsto vary significantly across countries, but less so acrosssectors within a country (see Figure 3).

Computation of the weighted average cost of capitalThe weighted average cost of capital has also beenderived from CAPM. The required parameters havebeen estimated as described in Appendix 2. Note thecost of debt used in the calculation is nominal and doesnot include the cost of potential debt renegotiations. It

14 Computation of the Hurdle Rates-Cost of Capital

COMPUTATION OF THE HURDLE RATES: COST OF CAPITAL

5.

Source: Authors’ calculations.

Figure 2: Estimated cost of equity by country, May 2004

19% 18%17%

7%

18%

7%

13% 14%

6%

19%

0%

5%

10%

15%

20%

25%

Arge

ntin

aBo

livia

Braz

il

Chile

Colo

mbi

aM

exico

Pana

ma

Peru

Unite

d St

ates

Vene

zuela

may understate the effective cost of debt and the result-ing WACC. Figure 4 shows the resulting estimates ofthe weighted average cost of capital per country.

Figure 4 indicates, for instance, that an investor look-ing to invest today in a concession in Argentina wouldneed to earn a project internal rate of return of at least14 percent on the investment. If the concession’s finan-cial projections show the project internal rate of returnis likely to be less than 14 percent, the investment is not

worth it, as the investor should be able to find alterna-tive investments with a similar level of risk but a higherexpected return. The same investor would only need toearn a project internal rate of return of 3 percent toinvest in an American or Chilean concession.

As for the cost of equity, the WACC varies moreacross countries than across sectors (see Figure 5).

Since the estimated cost of debt is lower than the costof equity, the resulting WACCs are lower than the costs

How Profitable Are Infrastructure Concessions in Latin America? 15

14%13% 13%

3%

12%

4%

8%9%

3%

14%

0%

2%

4%

6%

8%

10%

12%

14%

16%

Arge

ntina

Boliv

iaBra

zilCh

ile

Colom

bia

Mexic

o

Pana

ma Peru

Unite

d Stat

es

Vene

zuela

Source: Authors’ calculations.

Figure 4: Estimated weighted average cost of capital by country, May 2004

United States

7%6%

5%

9%

6%

4%

0%1%2%3%4%5%6%7%8%9%

10%

Road

sPo

rtsWa

ter

Teleco

m

Energ

y dist

r.

Energ

y gen

er.

Argentina

20% 19%18%

22%19%

17%

0%

5%

10%

15%

20%

25%

Road

sPo

rtsWa

ter

Teleco

m

Energ

y dist

r.

Energ

y gen

er.

Source: Authors’ calculations.

Figure 3: Estimated cost of equity by sector; United States and Argentina, May 2004

of equity (see Figure 6). This means equity investorsexpect higher returns than debt holders, a logical conse-quence of their taking on more risk.

Variability during the concessions’ lifetimeAs explained in Appendix 2, both the cost of equity andthe weighted average cost of capital vary constantly, asnew information is incorporated into betas and countryratings (on the company’s risk level, the sector, the coun-

try, the regulatory environment, and so forth), and asthe risk-free rate moves. Therefore, three hurdle ratesfor each concession have been computed: at their start,on average over their operation, and at the end of 2001.

It can be seen from Table 5 that, despite small varia-tions, on average the cost of equity was relatively stableuntil 2001. The average weighted cost of capital of thesample concessions, by contrast, has been falling overthe life of the concessions (by up to 2 percentage points

16 Computation of the Hurdle Rates-Cost of Capital

United States

3% 3%2%

5%

3%2%

0%

1%

2%

3%

4%

5%

Road

sPo

rtsWa

ter

Teleco

m

Energ

y dist

r.

Energ

y gen

er.

Argentina

14% 13%13%

15%

14%13%

11%12%12%13%13%14%14%15%15%16%

Road

sPo

rtsWa

ter

Teleco

m

Energ

y dist

r.

Energ

y gen

er.

Source: Authors’ calculations.

Figure 5: Estimated WACC by sector; United States and Argentina, May 2004

19%18% 17%

7%

18%

7%

13% 14%

6%

19%

14%13% 13%

3%

12%

4%

8% 9%

3%

14%

0%

5%

10%

15%

20%

25%

Arge

ntina

Boliv

iaBra

zilCh

ile

Colom

bia

Mexic

o

Pana

ma Peru

Unite

d Stat

es

Vene

zuela

CoE WACC

Source: Authors’ calculations.

Figure 6: Comparison of the estimated cost of equity and the estimated WACC

on average). This fall results from concessions’ higheroverall indebtedness level in the telecommunicationsand energy sectors. In the analysis, returns to average

hurdle rates over each concession’s historical years ofoperation are compared. One has to keep in mind thathurdle rates are higher today.

How Profitable Are Infrastructure Concessions in Latin America? 17

Table 5: Variation in the cost of equity and WACC over the concessions’ lifetime

At the start of On average during each each concession concession’s life time 2001

Cost of equity Water concession average 15% 16% 15%Transport concession average 15% 17% 16%Telecommunications concession average 20% 20% 20%Energy concession average 18% 19% 18%Overall 17% 18% 17%

WACCWater concession average 10% 11% 11%Transport concession average 11% 11% 10%Telecommunications concession average 16% 15% 13%Energy concession average 16% 15% 13%Overall 14% 13% 12%

Source: Authors’calculations.

As explained in the methodology, two sets of returnshave been computed. First, concession returns, thosemeasuring the overall attractiveness of the concessionsas business entities, were computed. Second, share-holders returns, those effectively earned by projectshareholders from the distribution of dividends orother sources of funds generated by the concession,were computed.

For each sets of returns, financial returns deriveddirectly from the financial statements of the concessionswere first computed. Then, to account for some poten-tial economic distortions, with the objective of estimat-ing adjusted returns representative of the economic sit-uation of each concession, these returns were adjusted.

Finally, three values were provided when returnsbased on an internal rate of return calculation(Shareholder IRR and Project IRR) were measured. Firstthese returns were computed only over historical yearsto account for the returns already generated by the con-cessions or earned by their concessionaires to date.Second, it was estimated what they might generate orearn over the concessions’ remaining years of operationby projecting flows until each concession’s last year ofoperation. Third, to estimate the returns they wouldgenerate or earn if they were adequately compensatedfor the value created at the end of each concession, a ter-minal value was added. In this section, the results of theanalysis, focusing on each of these measures of return,are presented.

Concessions’ returnsOverall concession returnsStarting with the most aggregated level first, the averagereturns earned by our sample of 34 concessions werecomputed. Obviously, these overall returns must beinterpreted with caution given the variety of countriesand sectors included in the sample and the wide dis-crepancy of results across them.

Figure 7 shows that without including the futurevalue to be created by the concessions (future and ter-minal values)—that is, including historical years only—our concessions reach a financial return of negative 24percent, well below their average WACC of 13 percent.

This results in part from our sample concessions’low operating profitability compared to their averageWACC. Figure 8 shows the average annual return oncapital employed generated by our sample conces-sions so far was 7 percent (oscillating between 4 and9 percent from year to year), well below the averageWACC of 13 percent. (The impact of adding up man-agement fees and excess depreciation is minimal, theoverall average annual ROCE rising to 9 percent).Investments, which averaged 27 percent of our con-cessions’ annual revenues, have also had their toll onnet profitability.

Despite this low historical profitability, Figure 7shows the sample concessions should on average be ableto generate an internal rate of return (Project IRR withTV) above their average WACC (14 percent) if theirfuture growth is at least equal to each country’s averagehistorical economic growth and the residual valueadded is taken into account. In this sense, infrastructureconcessions are interesting business proposals for poten-tial investors, providing them with an adequate long-term return compared to the risk taken.

In fact, based on this study’s adjusted measures ofreturns, they even seem able to generate some excessreturns. If no management fees were paid to the con-cessions’ operators, the concessions’ average long-termProject IRR would reach 15 percent; and if, in addition,the cost of investments was not possibly inflated by 10percent, their internal rate of return would reach 19percent.

It must be noted, however, that these adjustmentsmay not always be feasible or desirable for the con-cessions’ shareholders. Management fees may be

18 Concession and Shareholder Returns

CONCESSION AND SHAREHOLDER RETURNS

6.

unavoidable if the services they pay for are necessaryfor the concession company, and the cost of invest-ment may not be reducible. In addition, reducingmanagement fees and possible investment markupswould reduce shareholders’ returns, while this analy-sis will show later that these are below the requiredcost of equity.

Concession returns by sectorReturns seem to vary somewhat across sectors, althoughthe general conclusions presented above apply to all sec-tors. As Figure 9 shows, historical returns are negativein all sectors. The water sector stands out, with an aver-age historical return significantly lower than those inother sectors.

How Profitable Are Infrastructure Concessions in Latin America? 19

Project IRR, Overall average

-24%

7%

14% 15%19%

13%

-30%

-25%

-20%

-15%

-10%

-5%

0%

5%

10%

15%

20%

25%

IRR, no TV IRR, with FV WACCIRR, with TV IRR with feesadjusted

IRR with feesand investment

markup adjusted

Source: Authors’ calculations, based on concessions’ historical financial statements and the authors’ growth assumptions.

Figure 7: Overall concession return

ROCE Vs WACC

0%

2%

4%

6%

8%

10%

12%

14%

16%

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Average RoCE Average ROCE with eco. adjusted Average WACC

Source: Authors’ calculations, based on concessions’ historical financial statements.

Figure 8: Average annual profitability of sample concessions

This results because in all sectors the net averageannual operating profitability of concessions has notbeen high enough compared to their respective levels ofrisk (see Figure 10). Table 6 confirms the average oper-

ating profitability of concessions in the water sector hasbeen the lowest. It has also been the most volatile.Telecommunications is the sector with the highest aver-age operational profitability.

20 Concession and Shareholder Returns

-55%

-35%

-15%

5%

25%

Water Transport Telecom Energy

Project IRR vc WACC

IRR, no TV IRR, with FV IRR, with TVIRR with fees adjusted IRR with fees and investment

markup adjustedWACC

Source: Authors’ calculations, based on concessions’ historical financial statements and the authors’ growth assumptions.

Figure 9: Long-term concession returns by sectors

ROCE Vs WACC

-10%

-5%

0%

5%

10%

15%

20%

1993 1994 1995 1996 1997 1998 1999 2000 2001

Water Transport Telecom Energy Avg WACC

Source: Authors’ calculations, based on concessions’ historical financial statements.

Figure 10: Average operational profitability by sector

Large investments have also reduced the net prof-itability of concessions, as illustrated in Table 7.

As Figure 9 shows, water is the only sector where thelong-term financial return of concessions is expected toremain below the sector’s corresponding WACC (9 per-cent Project IRR with TV, versus 11 percent WACC).However, if management fees are added back to theconcessions’ net income, the average return of waterconcessions equals their average WACC. This meansthat if water concessions could reach our projected prof-itability without paying any management fees, theywould be interesting business proposals.6 If they couldalso reduce the cost of their investment by at least 10percent, their average long-term return would exceedthe corresponding WACC.

In all the other sectors, the estimated long-termreturns all equal or exceed the corresponding sectorWACC, even without adjustment. When return adjust-ments are introduced, the long-term returns of conces-sions in all three sectors exceed their correspondingWACC.

Concession returns by countryMexico, Panama, and Venezuela were excluded fromthe by-country analysis because there is only one con-cession for each of the countries in the sample.

Figure 11 shows concession returns vary significant-ly across countries. Historical returns (IRR with no TV)have been negative in all countries, except in Colombia,where they are marginally positive.

Looking at long-term returns, Colombian conces-sions really stand out as the only ones capable of gener-ating a long-term return well above their correspondingWACC, even without any adjustment. Adjustmentswould bring their expected return above 40 percent.

Brazil follows, with concessions estimated to gener-ate long-term financial returns equal to their WACC.In Argentina and Chile, long-term returns becomeequal to the corresponding country WACC only withadjustments.

Peru and even more so Bolivia stand out as the onlycountries where our sample concessions are expected toearn a long-term return below the country WACC, evenwhen adjustments are taken into account. In Bolivia, thelong-term adjusted return of concessions is projected tobe negative.

Colombia stands out with above average expectedreturns partly because the average historical profitabili-ty of Colombian concessions has been the highest in thesample (see Figure 12). The lower level of investments inpercentage of revenues also contributes to these conces-sions’ higher profitability (see Table 8).

In Bolivia, the country where concessions haverecorded the lowest return, investments have been thehighest compared to revenues, while the operationalprofitability of concessions has been the lowest (except-ing Argentina).

Concession returns by concession maturityReturns vary also as a function of each concession’smaturity, that is, its years of operation. As the concessionmatures, profitability usually increases (heavy invest-ments and operating restructuring often penalize theearly years). Table 9 shows that concessions with morethan 10 years of operation have returned an adjustedProject IRR higher than the overall average WACC (16percent compared to 14 percent). This confirms thatconcessions are economically profitable businesses in thelong term, possibly after as few as 10 years of operation.

The concessions’ Project IRR increases systematicallyas the sample concessions’ number of years of operation

How Profitable Are Infrastructure Concessions in Latin America? 21

Table 6: Volatility of profitability by sector

Standard deviation Average RoCE of RoCE

Water 4.3% 4.3%Transport 5.2% 2.9%Telecom 8.2% 3.8%Energy 7.2% 3.5%Global average 6.3% 3.7%

Source: Authors’ calculations, based on concessions’ historical financial statements.

Table 7: Investment levels by sector

Average annual level of investments in

% of revenues