TRENDS AND POLICY OPTIONS No. 2 · January 2005 37568 HELPING TO ELIMINATE POVERTY THROUGH PRIVATE INVOLVEMENT IN INFRASTRUCTURE How Profitable Are Infrastructure Concessions in Latin America? Empirical Evidence and Regulatory Implications Sophie Sirtaine Maria Elena Pinglo J. Luis Guasch Vivien Foster © 2005 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, NW Washington, DC 20433 Telephone 202-473-1000 Internet www.worldbank.org E-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 necessarily reflect 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 such boundaries. Rights and Permissions The 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 to the 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. TABLE OF CONTENTS 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 rates­Cost of capital 7 The cost of equity 7 The weighted average cost of capital 7 Interpretations of results 9 4. Issues in Measuring Returns and Hurdle Rates: Cost of Capital 10 Data related issues 10 Data consistency, quality, and availability 10 Hurdle rates' time sensitivity 10 Concession data's time sensitivity 10 Concessions versus industry-specific data 11 Estimating reasonable terminal values 11 Adjustments to financial returns 12 Management fees 12 Investments (transfer pricing) 12 Other possible adjustments 13 Acquisition value 13 Pre-acquisition asset value 13 Transfer pricing (other than for investments) 13 Depreciation 13 How Profitable Are Infrastructure Concessions in Latin America? iii 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 18 Overall concession returns 18 Concession returns by sector 19 Concession returns by country 21 Concession returns by concession maturity 21 Individual concession returns 22 Concession return, a conclusion 22 Shareholders' returns 23 Overall shareholder returns 23 Shareholder returns by sectors 25 Shareholder returns by country 26 Shareholder returns by concession maturity 27 Individual shareholder returns 28 Volatility of returns across concessions 28 Volatility of returns from year to year 29 Shareholder returns, a conclusion 29 Concession and shareholder returns, a conclusion 30 Sensitivity analysis 31 Future concession growth 31 Future concession dividend policy 33 Future concession growth and dividend policy 34 Acquisition 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 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 Tables Table 1: The sample of concessions used 4 Table 3: Interpretations of various measures of return 7 Table 2: Measures of return used 8 Table 4: Guide to interpretation of results 9 Table 5: Variation in the cost of equity and WACC over the concessions' lifetime 17 Table 6: Volatility of profitability by sector 21 Table 7: Investment levels by sector 21 Table 8: Investment levels by country 22 Table 9: Historical concession returns by concession maturity 23 Table 10: Dispersion of returns across concessions 25 Table 11: Pay out ratios by sectors 26 Table 12: Management fees by sector 27 Table 13: Pay out ratios by country 28 Table 14: Relation between return and concession maturity 29 Table 15: Dispersion of adjusted RoE and Shareholder IRR across concessions 30 Table 16: Construction of regulatory quality indices 38 Table 17: Summary of regression results 39 Table 18: Summary of regression results 40 Table 19: Summary of second regression results 41 Table 20: Unleveraged betas by sector 51 Table 21: Typical leverage by sector 52 Table 22: Nominal corporate income tax rates by country 52 Table 23: Leveraged betas by sector and by country 53 Table 24: Returns on stocks compared to government bonds 53 Table 25: Country risk premiums 54 Table 26: Cost of debt by country 55 List of Boxes Box 1: The required rate of return on an asset 5 Box 2: Definition of the cost of equity 7 Box 3: Definition of the weighted average cost of capital 9 Box 4: Definition of the shareholder internal rate of return 44 Box 5: Definition of the shareholders' internal rate of return with a terminal value 45 Box 6: Definition of the return on equity 46 Box 7: Definition of the project internal rate of return 46 Box 8: Definition of the project internal rate of return with a terminal value 47 Box 9: Definition of the return on capital employed 47 Box 10: Definition of the cost of equity 49 Box 11: Leveraged and unleveraged betas 51 Box 12: Definition of the weighted average cost of capital 55 How Profitable Are Infrastructure Concessions in Latin America? v List of Figures Figure 1: Private investment in infrastructure, 1990­2002 2 Figure 2: Estimated cost of equity by country, May 2004 14 Figure 3: Estimated cost of equity by sector; United States and Argentina, May 2004 15 Figure 4: Estimated weighted average cost of capital by country, May 2004 15 Figure 5: Estimated WACC by sector: the example of the USA and Argentina, May 2004 16 Figure 6: Comparison of the estimated cost of equity and the estimated WACC 16 Figure 7: Overall concession return 19 Figure 8: Average annual profitability of sample concessions 19 Figure 9: Long-term concession returns by sectors 20 Figure 10: Average operational profitability by sector 20 Figure 11: Average concession returns by country 22 Figure 12: Average annual concession profitability by country 23 Figure 13: Financial RoCE and Project IRR by concession 24 Figure 14: Overall long-term shareholder returns 25 Figure 15: Average annual return on equity of sample concessions 26 Figure 16: Overall shareholder returns by sector 27 Figure 17: Shareholder returns by country 28 Figure 18: Financial RoE and shareholder IRR by concession 29 Figure 19: Evolution of annual RoE by sector 30 Figure 20: Evolution of the annual RoE by country 30 Figure 21: Project IRR sensitivity to concession growth rates 31 Figure 22: Shareholder IRR sensitivity to concession growth rates 32 Figure 23: Shareholder IRR sensitivity to dividend payout 33 Figure 24: Shareholder IRR sensitivity to concession growth rates and dividend payout 34 Figure 25: Project IRR sensitivity to concession acquisition price 35 Figure 26: Shareholder IRR sensitivity to concession acquisition price 36 Figure 27: Evolution of the risk-free rate, 1960­2004 50 Figure 28: Evolution of country risk premiums over time 54 vi Table of Contents ACKNOWLEDGMENTS The authors are all at the Finance, Private Sector and The findings, interpretations, and conclusions Infrastructure Department, Latin America and expressed in this paper are entirely those of the authors Caribbean Region (LACFPSI), the World Bank. J. Luis and should not be attributed in any manner to the Public- Guasch is, in addition, professor of Economics, Private Infrastructure Advisory Facility (PPIAF) or to the University of California, San Diego. The authors are World Bank, to its affiliated organizations, or to mem- most grateful for comments and suggestions from Ian bers of its Board of Executive Directors or the countries Alexander, Soumya Chattopadhyay, Antonio Estache, they represent. Neither PPIAF nor the World Bank guar- Danny Leipziger, Isabel Sanchez Garcia, and Ilias antees the accuracy of the data included in this publica- Skamnelos. Partial funding from the Public Private tion or accepts responsibility for any consequence of their Infrastructure Advisory Facility (PPIAF) is gratefully use. The boundaries, colors, denominations, and other acknowledged. Contact e-mail address jguasch@world- information shown on any map in this report do not bank.org. imply on the part of PPIAF or the World Bank Group any judgment on the legal status of any territory or the endorsement or acceptance of such boundaries. How Profitable Are Infrastructure Concessions in Latin America? vii ABSTRACT This report estimates the returns that private investors ment between rates of return and cost of capital, or did in infrastructure projects in Latin America really made allow for the capture of excessive rents by the investors on their investments, and assesses the adequacy of these or of excessive benefits by the users at the expense of the returns relative to the risks taken--the cost of capital-- investors. The findings of this report are that contrary to and the impact that the quality of regulation had on the general public perceptions, the financial returns of pri- closeness of alignment between returns and the cost of vate infrastructure concessions have been modest and capital. This is done by estimating both historical and that in fact for a number of concessions the returns have projected future returns earned by a sample of private been below the cost of capital. On average telecom and infrastructure concessions, across a variety of Latin energy concessions have fared better than transport and American countries and infrastructure sectors, and com- water. It also shows that the variance of returns across paring them against expected returns given the level of concessions and countries is considerable; that the vari- risk taken--the cost of capital. In this way, it is possible ance of returns across concessions can be partially to evaluate whether private investors did indeed earn explained by the quality of regulation; and that the bet- abnormally high returns on their investments. The ter the quality of regulation the closer the alignment report develops a quality of regulation index and exam- between financial returns and costs of capital, as is desir- ines the extent to which the quality of the regulatory able. Thus this report shows and validates the claim that framework contributed to maintaining a closer align- regulation indeed matters. viii Abstract EXECUTIVE SUMMARY Background and objective Sample and methodology During the 1990s many countries in Latin America The study is based on a sample of 34 concessions that implemented broad privatization and concession pro- are representative of global privatization trends in Latin grams for infrastructure services, with the aim of raising America from close to 1,000 infrastructure concessions fiscal revenues and improving sector performance. A in the region. It includes companies from nine countries decade later many are questioning whether private sec- with widescale privatization programs: Argentina, tor participation yielded the anticipated benefits, and Bolivia, Brazil, Chile, Colombia, El Salvador, Mexico, whether those benefits were equitably distributed Peru, and Venezuela. The number of concessions in each among the different stakeholders to the privatization country has been chosen to be representative of the rela- process. A frequent complaint is that investors may have tive importance of privatizations in that country. The captured a disproportionate share of the benefits in the sample includes companies in four sectors: telecommuni- form of excess profits over and above what was neces- cations, water, electricity (generation and distribution), sary to attract private capital into these sectors. and transport. On average these concessions have been However, to date there has been very little empirical in operation for seven years. It must be noted the data evidence against which to assess this claim. used run to 2001. They are thus largely exempted from The objective of this study is to estimate the returns the impact of the recent crisis in Latin America, and it is that private investors in infrastructure projects in Latin likely that returns would have looked significantly worse America really made on their investments, to assess the had 2002 and 2003 been included in the analysis. adequacy of these returns relative to the risks taken, To ensure sufficient quality of information, only and the impact that the quality of regulation had on audited financial statements and official company press those returns relative to the cost of capital. The study releases were used. The study does not attempt to adjust does not attempt to evaluate the overall impact of pri- financial statements for differences in accounting stan- vatization and concession programs, but simply focus- dards. It is recognized that regulation by return may cre- es on the narrow aspect of profitability. This is done by ate incentives for concessionaires to dress up their estimating both historical and projected future returns accounts to present the lowest profitability or return earned by a sample of private infrastructure conces- possible. As a consequence the profitability results sions across a variety of Latin American countries and imputed here ought to be construed as lower bound esti- infrastructure sectors and comparing them to expected mates of the true profitability of those regulated firms. returns, given the level of risk taken. In this way it is However, the scope for such accounting distortions are possible to evaluate whether or not private investors limited because 56 percent of the sample concessions are earn abnormally high returns on their investments. In listed companies or part of listed groups and financial addition, the study examines the extent to which the statements are audited. quality of the regulatory framework put in place at the Recognizing that private investors can be remuner- time of privatization contributed to maintaining a clos- ated in various ways, two sets of returns are comput- er alignment between rates of return and hurdle rates, ed. First, the financial returns resulting from the or the cost of capital. distribution of dividends from the concession to the How Profitable Are Infrastructure Concessions in Latin America? ix concessionaires' parent companies (mostly abroad) Conclusions and implications are computed. Second, the adjusted returns are com- The analysis shows that concessions are capable of gen- puted by attempting to include indirect forms of divi- erating adequate returns in the long term, and are dends. The most common of these are management potentially interesting business proposals. Concessions fees, based on the assumption that all the explicit in the water sector appear relatively the least attractive, management fees paid by concessions were in fact div- while concessions in the telecommunications sector idends to their strategic shareholders. Another adjust- appear to be the most profitable overall. On average, ment is made for the possibility of investment cost concessions seem to become profitable after about 10 markups, which arise when intragroup purchases are years of operation. However, about 40 percent of the priced above cost, thereby implicitly transferring sample concessions do not seem to have the potential to dividends out of the concession toward the parent generate attractive returns, with this number climbing company. to 50 percent in the energy and transport sectors. The study is built on the Capital Adequacy Pricing Concessions are thus risky businesses. Model (CAPM), which formalizes the observation that Low dividend distribution ratios have, however, not expected returns are related to risk. A two-pronged translated this overall profitability into adequate returns approach is used. The first step is to measure the over- for shareholders to date. In fact, on average, concession all return which shareholders in each selected project shareholders have so far earned negative returns on earned on the capital they invested in that project. The their investments, even including management fees, esti- second step is to determine whether those returns were mated accumulated capital gains, and potential invest- commensurate with the risk taken. Thus, the expost ment markups. returns that investors effectively earned on the asset or With historical growth maintained into the future, project they invested in (effective returns) are compared only telecom concessions would seem to have an inher- with the threshold minimum return given the risk pro- ent profitability high enough to generate adequate file of the project-cost of capital (hurdle rates). returns to their shareholders in the long term, this, pro- The study uses four measures of the effective returns: vided they can capture annually the capital gains accu- the shareholders' internal rate of return (Shareholder mulated in their concessions over all years of operation IRR), the return on equity (RoE), the project internal and that the full value of their management fees corre- rate of return (Project IRR) and the return on capital spond to dividends. In all other sectors, shareholders employed (RoCE). The first two are measures of the can hope to earn long-term returns commensurate to returns earned by equity investors; the last two are the risk taken only if the sectors consistently and signif- measures of the profitability of the concessions overall, icantly outperform historical market growth. This con- independent of their financing structure. clusion would not change if the concessionaires had The measure of returns chosen dictates the nature paid up to 20 percent less for their concessions. The of hurdle rates one needs to use. The Shareholder IRR implication is that to build an adequate return, share- and the RoE, both of which measure returns earned holders must rely both on various sources of remunera- over equity capital, must be compared to the appro- tion (including dividends, management fees, and capital priate cost of equity (CoE), which is a measure of the gains), and on outperforming historical market growth return investors require on equity investments, given consistently, over the entire length of their concession. the level of risk of such investments. The Project IRR These results suggest that concessionaires operate and the RoCE, which measures returns earned on the with long-term perspectives, giving priority to growth- concession's overall capital structure, must be com- enhancing investments in the early years (at the cost of pared to the weighted average cost of capital depressing returns in the short term), and relying on the (WACC), which represents the expected return on all entire concession period to build an adequate return. of a company's securities. Importantly, the appropri- This may be driven by their contractual obligations, ate benchmark value for each hurdle rate varies for which usually require high investments in the early each project depending on the country and sector of years. It implies that early breaks of concession con- investment, reflecting that market risks also vary tracts may have a highly negative impact on expected across countries and sectors. returns. x Executive Summary The results also highlight that management fees may not only a consequence of market conditions and man- be needed to build adequate returns, but that their treat- agerial skills, but also partly a reflection of regulatory ment--from an accounting stand point they ought to be decisions on service tariffs. A good regulator should aim treated more like dividends than costs--ought to be to maintain alignment between a company's rate of more transparent. In addition, allowing concession return and its cost of capital in the medium term. This shareholders to be fairly compensated at the end of the is because a rate of return in excess of the cost of capi- period for the capital gains accumulated during the life tal inappropriately penalizes consumers, while a rate of of the concessions is also an important component of return beneath the cost of capital inappropriately dis- their return. courages further investment. The relatively low returns earned so far by conces- An evaluation of the quality of the regulatory sion shareholders also suggest that either regulators regimes faced by concessionaires in the study sample have been tough at setting tariffs or that concession bid- finds that these do not score very high on average, and ding processes have been successful in creating strong that there is a high variance in the quality of regulato- competition (aggressive bidding) among bidders, bring- ry frameworks across concessions. Furthermore, the ing their offered price to the limits of what made con- quality of regulation is found to be a significant deter- cessions interesting investments for them. That old con- minant of the divergence between the overall prof- cessions are on average more profitable than young itability of the concession and its corresponding hurdle ones suggests that returns may be depressed in early rate, explaining around 20 percent of the variation. concession years by inadequate prices, corrected after Thus regulation does matter. However, regulatory the first price control period (high investments in the efforts seem to be more closely associated with mini- first years of operation may also have a toll on young mizing the simple IRR-WACC differential (and thereby concession returns). This squares with the arguments keeping tariffs as low as possible for current con- and data presented in Guasch (2004), where it appears sumers), than with minimizing the absolute IRR- that a significant number of concessions were won by WACC differential (and thereby keeping profitability aggressive bidding, perhaps too aggressive, and that well aligned with hurdle rates of return). A striking fea- shortly afterward the contracts were renegotiated, often ture of the results is that regulatory quality variables granting better terms to the operators. That would at seem to have overall significance, more than individual least partially explain why old concessions tend to be significance, in determining IRR-WACC differentials. more profitable than young ones. This is in fact consistent with the fact that performance The analysis also highlights that returns (in particu- along different dimensions of regulatory quality is not lar shareholder returns) are highly volatile across sec- highly correlated, and that the benefits of high regula- tors, concessions, and from year to year. Thus infra- tory quality along one dimension can be completely structure concessions, in Latin America are a high-risk offset by low regulatory quality along another dimen- investment proposal, which explains why the required sion. Thus, for regulation to be effective, one needs the rates of return on such investments are high. whole package of regulatory characteristics. If some of Given that virtually all the concessions included in the key ingredients are missing the effectiveness of reg- this study are regulated monopolies, their profitability is ulation is highly diminished. How Profitable Are Infrastructure Concessions in Latin America? xi 1. OBJECTIVES OF THE STUDY During the 1990s many countries in Latin America reducing ongoing subsidies. After a decade of reform, implemented broad privatization and concession pro- popular support for privatization around the region has grams for infrastructure services. In aggregate, private dwindled, and public debate increasingly questions the participation in infrastructure in less developed and extent to which these reforms delivered the anticipated emerging countries amounted to US$690 billion during benefits, and (if so) whether these benefits were equi- the 1990s (World Bank 2003). The Latin America and tably distributed among different stakeholder groups. the Caribbean Region (LAC) proved to be the investors' With any privatization process, there are a number of preferred destination, receiving 50 percent (US$345 bil- distinct stakeholder groups whose interests are likely to lion) of worldwide private capital flows to the infra- be affected. First, the state has major fiscal interests in structure sectors during the same period (figure 1a). privatization transactions, standing to gain from priva- Within LAC, these flows were predominantly channeled tization proceeds, as well as from any reductions in sub- to the telecommunications and electricity sectors (Figure sidy or increases in tax revenues, often made possible as 1b), and, moreover, heavily concentrated in a handful of a result of privatization. Second, the interests of current the larger economies: Brazil, Argentina, Mexico, and consumers will be affected by the resulting changes in Chile (figure 1c). the price and quality of the services provided, while new LAC's ability to attract such an inordinate share of consumers may be incorporated as service areas expand. infrastructure investment flows can be explained by the Third, the interests of employees will be affected as a region's early opening of its infrastructure sector to pri- result of potential layoffs and changes in the pattern and vate sector participation, the existence of substantial conditions of employment. Fourth, the extent to which levels of unmet demands in practically all infrastructure transactions are designed to generate benefits for the sectors, and perspectives of macroeconomic stability other stakeholder groups, as well as the quality of sub- and reasonably high growth. In addition, to a much sequent regulatory decisions, will affect the residual greater extent than in other regions, LAC went ahead profitability of the enterprise to the private investors. with major divestitures of public enterprises. Thus, it The huge complexity of privatization transactions, as is estimated around 60 percent of these capital flows well as their major ramifications for the economy's gen- were captured by the state as fiscal revenues associated eral equilibrium, make it difficult to generalize as to with asset sales, while the remaining 40 percent were how the costs and benefits of privatization will play out invested directly within the infrastructure sectors across the different stakeholder groups in any particular Latin America's private sector participation in infra- case. However, a relatively new, but growing, literature structure programs was generally part of a broader set of aims to document the economic and distributional policy reforms. The reforms were expected to improve impact of privatization (Andres, Foster, and Guasch much needed sector performance, increase levels of serv- 2004; Birdsall and Nellis 2002; Nellis 2003; McKenzie ice coverage, and attract private sector financing for and Mookherjee 2004; Ugaz and Waddams-Price 2003; long-delayed investments in infrastructure expansion and Chong and Lopez-de-Silanes 2005). The emerging and upgrading, thereby enabling scarce public funds to conclusions of this literature are that the efficiency gains be used for investment in the social sectors and for the and increases in quality in the provision of infrastruc- creation of fiscal benefits by creating sale revenues and ture services and fiscal payoffs of privatization have How Profitable Are Infrastructure Concessions in Latin America? 1 Figure 1: Private investment in infrastructure, 1990­02 (a) Across regions Private investments in infrastructure in less developed countries Latin America & Carribean East Asia & Pacific Other 140 120 100 bn 80 US$ 60 2002 40 20 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 (b) For Latin America, by sector (c) For Latin America, by country 5% 17% 35% 21% Telecom Brazil 46% Energy 7% Argentina Transport Mexico Water Chile Others 16% 28% 25% Source: World Bank 2003. been substantial; that while the layoffs of workers have So far this literature has had relatively little to say been large relative to the size of the industry but small about the extent to which private investors have bene- relative to the workforce as a whole, overall sector fited from the privatization process. There are popular employment levels have increased on average after a few perceptions that investors have profited excessively years from the transactions; that existing customers from privatization transactions, repatriating dividends have generally seen quality improve but have sometimes to their countries of origin instead of reinvesting them had to pay higher prices in return; and that service in the host country. However, there had been no sys- expansion has accelerated bringing major benefits to tematic empirical evidence from which to evaluate those previously unserved. such a claim. 2 Objectives of the Study The objective of this study is, therefore, to estimate private infrastructure concessions, across a variety of the returns that private investors in infrastructure proj- Latin American countries and infrastructure sectors, ects in Latin America really made on their investments and compares them with expected returns given the and to assess the adequacy of these returns relative to level of risk taken--the cost of capital. In this way, it the risks taken. The study does not attempt to evaluate evaluates whether private investors earned abnormally the overall impact of privatization and concession pro- high or low returns on their investments. In addition, grams, since this, as mentioned, has already been under- the study examines the extent to which the quality of taken elsewhere, but simply focuses on the narrow the regulatory framework put in place at the time of pri- aspect of concession profitability. The study estimates vatization was a factor in aligning rates of return and the historical and projected future returns of a sample of cost of capital. How Profitable Are Infrastructure Concessions in Latin America? 3 2. THE SAMPLE As of 2003, there were more than 1,200 infrastructure privatization programs in the region: Argentina, Bolivia, concessions in Latin America with private sector partic- Brazil, Chile, Colombia, El Salvador, Mexico, Peru, ipation. From that universe of private contracts, a sam- and Venezuela. The number of concessions in each ple of 34 concessions was selected, using the following country has been chosen to be representative of the criteria: (a) to include most Latin American countries relative importance of privatization in each country. The with meaningful privatization programs; (b) to include sample includes companies in four sectors: telecom- companies from all main infrastructure sectors; (c) to munications, water, electricity (generation and distri- focus on companies with at least five years of operation bution), and transport. In the latter case, the sample (to have a time series of data of adequate duration for is restricted to four companies in the road and port the analysis); and (d) to focus on companies publishing subsectors. They cannot, therefore, be considered repre- good quality financial statements. sentative of the entire transport sector. Airport conces- The resulting sample of 34 companies are represen- sions in particular are not included in the sample. On tative of global privatization trends in Latin America. It average the sample concessions have been in operation includes companies from nine countries with wide-scale for seven years. Table 1: The sample of concessions used Number of concessions Telecom Water Energy Transport Total Argentina 2 3 4 2 11 Bolivia 1 1 1 - 3 Brazil 1 1 4 - 6 Chile 1 3 - 1 5 Colombia - 2 - - 2 Mexico 1 - - - 1 Panama - - - 1 1 Peru 1 - 3 - 4 Venezuela 1 - - - 1 Total 8 10 12 4 34 4 The Sample 3. THE METHODOLOGY The study uses a two-pronged methodology. First, it free investment plus a risk premium that compensates measures the overall return which shareholders in each for the nondiversifiable risk embedded in the project. selected project earned on the capital they invested in Some risks can be eliminated by appropriate diversi- that project. Second, it determines whether those fication. These risks are called unique risks, because returns are adequate given the risk taken by comparing they measure the perils that are peculiar to one particu- them to appropriate hurdle rates. lar company or project, but can be eliminated by an Absolute measures of returns are meaningless since appropriate portfolio diversification. Investors are not they fail to recognize that private investors are not will- remunerated for these risks, since it is their job to ade- ing to invest in all potential projects for the same quately diversify their portfolio to eliminate them. returns. This is because risk-averse investors perceive However, there are some risks which cannot be avoided that the risks associated with various investments differ. however much you diversify. These are called market The higher their perception of the riskiness of a specific risks. They stem from the fact that there are economy- investment, the higher the return they will require in wide perils which threaten all businesses in an economy. order to make that investment. Since investors cannot diversify these risks away, they Intuitively, this is because financial managers realize will only accept to invest in a risky asset (that is, an asset that, all else being equal, risky projects are less desirable sensitive to market risks) rather than in safe ones (that than safe ones. Therefore, they demand a higher expected is, an asset nonsensitive to market risks) if they are ade- rate of return from risky projects. The observation that quately compensated for the extra risk taken. expected returns are related to risk has been formalized in the Capital Adequacy Pricing Model developed in the 1960s (Sharpe 1964; Lintner 1965). BOX 1 The Capital Asset Pricing Model The required rate of return on an asset The CAPM model is based on the idea that investors demand higher expected returns if asked to take on ra = rf + ß * (rm ­ rf) additional risk. More precisely, it tells us that investors Where: ra = required return on the asset, i.e. the hurdle in a project will require earning an expected return rate to use when deciding whether to invest which compensates adequately for the risk embedded in in the asset or not the project. This required rate of return, called the hur- dle rate, is the expected return above which an invest- rf = risk-free rate, i.e. the return of a risk-free ment makes sense and below which it does not. For a investment company, this hurdle rate is equivalent to its opportu- ß = beta of the asset nity cost of capital; that is, the rate of return the com- rm = market return, i.e. the return on the market pany can otherwise earn at the same level of risk as the as a whole, that is on a fully diversified investment it is considering (see box 1). portfolio The CAPM model shows that this return should be at least equal to the return the company can earn on a risk- ß * (rm ­ rf) is the asset risk premium How Profitable Are Infrastructure Concessions in Latin America? 5 Therefore, it is futile thinking about how risky an The Shareholder IRR and the Project IRR measure investment is in isolation. One needs to measure how returns earned over several years, while the RoE and the sensitive that investment is to market movements. This RoCE are annual measures of returns (see Table 2). sensitivity is called beta (ß). The CAPM theory shows Note that the Shareholder IRR used in the analysis is that the premium investors require in exchange for based on dividends (and other direct financial flows to holding a riskier (more volatile) asset (called the asset shareholders) only. It does not incorporate the value cre- risk premium) varies in direct proportion to its beta. ated by re-investing part of the generated earnings into As the formula suggests, hurdle rates are asset- or the concessions. This value is captured in the overall project-specific. Since they represent the rate at which a Project IRR. Shareholders capture it through increases specific project makes sense for an investor, given that in the share price or value of their company.1 This value project's own degree of risk, it is logical that their values has been incorporated in the calculation of Shareholder differ from project to project. IRRs by way of a terminal value only. This is because, Effective returns are the ex-post returns that since most concessions are not listed or sellable, accu- investors effectively earned on the asset or project they mulated capital gains cannot be cashed-in by sharehold- invested in. They may be very different from the ers. The only way shareholders can really cash-in the returns investors expected to earn ex-ante and on the accumulated value is at the end of the concession, when basis of which they made their original investment it is re-bid, and this, provided they get a fair compensa- decision. Excess returns are returns an investor has tion for the value created. The value added created by gained in excess of those required originally according retaining earnings into the concessions by way of a ter- to the CAPM. minal value has been included instead. This study intends precisely to investigate whether The measures of shareholder returns used would, the effective returns earned by private investors in infra- therefore, underestimate the returns earned by conces- structure projects in Latin America have been commen- sion shareholders if the latter could freely sell their surate to their expectations, given the risks taken. Using shares in the concession companies to cash-in accumu- the two-pronged methodology described previously, one lated capital gains, or if the value accumulated in the first needs to define adequate measures of the effective concessions was fully reflected in the value of their own returns earned by private concessionaires in Latin companies. America, and then one needs to compare them with To conclude, the four measures of return used can be appropriate hurdle rates. Each of these issues will now interpreted as summarized in Table 3. be looked at in turn. The Shareholder IRR and the Project IRR have been calculated over three distinct horizons. The first Measures of returns includes historical dividends/free cash flows only. They Several measures of the effective returns earned by con- then measure the effective return earned by sharehold- cessionaires can be used. In this study, the Shareholder ers/the concession to date. Second, they have been com- Internal Rate of Return (Shareholder IRR), the Return puted to include historical dividends/free cash flows and on Equity (RoE), the Project Internal Rate of Return the future value (FV) of dividends/free cash flows to be (Project IRR), and the Return on Capital Employed received annually until the concession's last year of (RoCE) is used. Appendix 1 provides detailed defini- operation. They then measure the potential return tions of each of these measures. which shareholders/the concession can hope to earn The first two (the Shareholder IRR and the RoE) are until the end of the concession from annual flows. measures of the returns earned from dividends by equi- Finally, they have been computed including historical ty investors in the project company (the shareholders), and future dividends/free cash flows and a terminal while the last two (the Project IRR and the RoCE) are measures of the concession's overall profitability, inde- pendent of their financing structure. These last two are 1It is, however, difficult to isolate share price movements resulting measures of the average return earned by equity and from earnings announcements in concession subsidiaries from other debt investors into the project company, while the first events influencing share prices. In addition, recognizing that the value accumulated in concessions is subject to political risk, as two indicators are measures of the returns earned by examples of expropriations have shown, the markets tend to value equity holders only. it only partly. 6 The Methodology al (FCF) opos financing e , the pr FCF at flow pital futur of ca charges est cash and year annual investment inter eefr e FCF last concession working e futur of and until and befor annual annual and value entity on: FCF fair nings Based Ear Historical Historical annual concession Historical plus end investment business eturnsr as after holders available FV TV concession concession debt of with with of flows es and IRR IRR IRR cash ofitability ee oject oject oject Measur Equity Pr Fr Indicator: RoCE Pr Pr Pr last and of until dividends value annual e eholder actual dividends dividends fair end investment shar or dividends futur for the annual annual and plus at equity on: annual concession on income e of distribution eturnsr Based Net Historical Historical futur year Historical dividends, compensation concession eholders FV eholder shar dividend by for with shar IRR IRR IRR of ned es holders ear n available eholder eholder eholder TV Measur Equity Retur Flows Indicator: RoE Shar Shar Shar with e e used today entir entir to including eturnr up over over of added ned ned ned es n ear ear ear value n n concession n concession, eturr Measur on: etation: 2: acteristics: eturr eturr the eturr the of of esidualr ableT Char Remunerate: Indicate: Based Interpr Annual otalT otalT life otalT life for How Profitable Are Infrastructure Concessions in Latin America? 7 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 company Return on Equity (RoE) Measure of the annual after tax return the concession is earning on its equity capital Project 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 value (TV) measuring the fair compensation sharehold- The cost of equity ers should receive for the value added they created in the The cost of equity is a measure of the return investors concession. They then measure the potential return require on equity investments, given the level of risk of which shareholders/the concession can hope to earn such investments. It is the appropriate hurdle rate for until the end of the concession from annual flows and measures of returns on equity investments. It is usually from the compensation they should receive at the end of estimated using the CAPM. the concession for the value added created.2 In our calculations, the risk premium was broken into two components: (a) the stock market risk premi- Hurdle rates­Cost of capital um, measured by ß * (rm ­ rf), corresponding to the Once the effective returns earned by project sharehold- extra return investors require to invest in stocks rather ers have been calculated they need to be compared with than in a risk-free asset; and (b) the country risk premi- appropriate hurdle rates. The appropriate hurdle rate um (CRP), corresponding to the extra return investors depends on the measure of returns used. In addition, the require to invest in stocks of companies in a country appropriate benchmark value for each hurdle rate will deemed riskier than a less risky country used as bench- vary for each project, depending on the country and sec- mark (see box 2). More details on these two compo- tor of investment (since market risks vary across coun- nents are provided in appendix 2. tries and sectors). The weighted average cost of capital The measure of returns chosen dictates the nature of The weighted average cost of capital (WACC) repre- hurdle rates one needs to use. In particular, the sents the expected return on all of a company's securi- Shareholder IRR and the RoE, both of which measure ties. It is measured as the average of the returns required returns earned over equity capital, must be compared to on each source of capital, such as stocks, bonds, and the appropriate cost of equity, while the Project IRR and the RoCE, which measure returns earned on the con- cession's overall capital structure, must be compared to the weighted average cost of capital. BOX 2 Definition of the cost of equity CE = rF + ß * (rm ­rf) + Crp Where: rf = risk-free rate 2Such compensation is usually calculated on the basis of the non- amortized value of the concession's assets, or of the new bidding ß = beta of the project price if the concession is re-bid to new private investors. The second method was preferred in this study since it is dynamic and forward- rm = expected stock market return looking. Therefore, the terminal value as a perpetuity of average div- Crp = country risk premium idends/free cash flows over the last three years of operation of each concession was calculated, adjusting for exceptional items. 8 The Methodology 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 taken RoE > appropriate CE The concession has returned a post-tax profitability on its equity capital superior to that of alternative investments of similar risk RoE < appropriate CE The concession has returned a post-tax profitability on its equity capital inferior to that of alternative investments of similar risk Project IRR > appropriate WACC The concession has generated positive net financial flows Project IRR < appropriate WACC The concession has generated negative net financial flows RoCE > appropriate WACC The concession's net operating profitability exceeds the level necessary to adequately service its debt and equity RoCE < appropriate WACC The concession's net operating profitability is insufficient to adequately service its debt and equity other debts, weighted by the shares of each source of BOX 3 capital in the company's financing structure. The calcu- lation is often simplified by grouping the various Definition of the weighted average cost of capital sources of financing into two categories only, equity and fixed income instruments. It is the appropriate hurdle WACC = E / (D + ) * CE + D / (D + E) * (1-T)* CD E rate to use for measures of returns on a concession's Where: E = book value of equity overall liabilities (see Box 3). D = long-term debt Interpretations of results CE = cost of equity (as measured above) In the analysis, the Shareholder IRR and RoE of each CD = cost of debt concession is compared to the appropriate CE and the Project IRR, and RoCE is compared to the appropriate T = nominal corporate income tax rate WACC. The results will be interpreted as summarized in Table 4. How Profitable Are Infrastructure Concessions in Latin America? 9 4. ISSUES IN MEASURING RETURNS AND HURDLE RATES: COST OF CAPITAL Data related issues some level of the opposite incentive to maximize the concession's profitability to create as much shareholder Data consistency, quality, and availability value as possible (through a share price increase). In To ensure sufficient quality of information, only audit- any case, financial statements are audited; so with all ed financial statements and official company press the appropriate caveats, the leeway to window dress releases were used. Concessions have to follow each balance sheets remains limited. Balancing both effects, country's particular accounting standards. Therefore, the former is bound to dominate the latter. Thus, the data provided by each concession in the sample may overall the imputed estimated profitability or return not be fully consistent since the concessions may use dif- estimate here is at least a lower bound of the true prof- ferent accounting rules to prepare their financial state- itability. ments. Although accounting standards in all the coun- tries under consideration are broadly based on interna- Hurdle rates' time sensitivity tional accounting standards (IAS), there remain some Some of the data used to estimate the appropriate hur- significant discrepancies which may generate differences dle rates vary constantly over time. This is, for instance, in earnings. No attempt has been made to adjust finan- the case of country risk premiums and betas. Both tend cial statements for differences in accounting standards. to vary as the market incorporates new information on In addition, some data that would have been impor- the country, sector, or company. tant for the analysis are generally not published by com- The cost of equity and weighted average cost of cap- panies, whatever the country in which they operate. ital have been computed at three different points in This applies for instance to the fair value of some assets, time: (a) at the start of each concession--this represents depreciation/amortization rules, and the detailed classi- the hurdle rate which investors would have used when fication of costs. It also applies to the market value of they assessed whether to invest in a concession or not; assets and liabilities, so that the analysis is based on (b) on average over the concession's life to date--this their book value. represents the opportunity cost that investors have faced Finally, some argue that regulation by return some- on average since they invested into the projects; and (c) times creates incentives for concessionaires to dress up today--this represents the current opportunity cost of their accounts in order to present the lowest profitabil- having invested money in a specific project. ity or return possible. This would happen when regu- It must be noted the data used is up to 2001. It is thus lated tariffs are set so as to ensure a minimum return to mostly exempted from the impact of the recent crisis in concessionaires, who, therefore, have the incentive to Latin America. It is highly probable that returns would minimize their historical return in order to maximize have looked worse were 2002 and 2003 included in the future tariff increases. However, many of our sample analysis. concessions are listed companies (56 percent)3 or part Concession data's time sensitivity of listed groups. In this case their managers might have The concessions' financial results are usually sensitive to their life cycle. It is not uncommon to make losses in the first years as operational processes are optimized and 3The percentages vary by sector, however: Transport (0 percent), Water (30 percent), Energy (75 percent), and Telecommunications heavy investments are often made. By contrast, prof- (88 percent). itability usually increases in later years as the system 10 Issues in Measuring Returns and Hurdle Rates-Cost of Capital matures and reaches an appropriate level of efficiency. important given the impact it has on results. This Therefore, mixing concessions at different stages of requires properly estimating the flows to be projected their life cycles is not ideal. As a result, the returns into the future and their growth rate. earned by individual concessions will not be compared, As with any projection into a distant future, it is as they may not be comparable. The problem is not as impossible to forecast it with certainty. In the case of acute, however, when computing averages for the whole infrastructure concessions, projecting future flows is sample since our sample includes several concessions at made even more arduous by their high sensitivity to eco- each stage of their life cycle. nomic ups and downs, political events, and so forth. All one can do is be as reasonable as possible, which is often Concessions versus industry-specific data achieved by making simple assumptions. When estimating the cost of equity and the weighted The flows to be projected into the future were esti- average cost of capital, the salient choice is to use indus- mated on the basis of the average of those of the last try averages for the value of betas, the structure of cap- three available historical years (1999, 2000, 2001). ital, and the cost of debt, rather than those of the spe- This means that for the Shareholder IRR the average cific company under consideration. This is because dividends distributed by each concession over the last industry averages are deemed representative of best three years was used, while for the Project IRR the practices in the sector. This ensures that risks that can be average net financial flows generated by each conces- eliminated with prudent financial management are not sion over the last three years was used. Averages over remunerated. Using an industry average penalizes proj- the last three years were taken to compute recurring ects which are highly leveraged, since the hurdle rate flows independent of yearly variations and to be repre- resulting from industry averages would include a lower sentative of the future.4 proportion of debt and, therefore, be higher (assuming It was then assumed the resulting flows would grow the cost of debt is lower than the cost of equity) than the at a constant rate. The same growth rate for dividends project's own hurdle rate. This means investors consid- and concession cash flows were used (which may be ering investing in a highly leveraged project may need to interpreted as meaning that dividend payout ratios will consider lowering its leverage--and, consequently, its remain constant). The latter was taken as equal to the financial risk--to make it attractive. long-term historical GDP growth rate of the country of Another advantage of using industry averages is to each concession. This assumption means that each con- ensure a higher comparability of results. Otherwise, it cession will grow at the same rate as the economy may be difficult to assess whether the extra return gen- around it and that the future growth rate of the econo- erated by a given project is not largely due to its higher my will be equal on average to its historical level. Each leverage. country's historical growth rate over a 39-year horizon, Finally, industry betas are often more reliable than from 1961 to 1999, was used. This ensures that several individual betas. This is because individual betas must economic cycles are included and that the data is not often be estimated from a limited-time series of data, biased by particular economic circumstances. The which exposes the results to potentially large estimate resulting growth rates are: Argentina (2.7 percent), errors. Fortunately, these errors tend to cancel out when Bolivia (2.8 percent), Brazil (4.8 percent), Chile (4.4 one estimates betas of diversified portfolios. percent), Colombia (4.2 percent), Mexico (4.7 percent), Some, by contrast, argue that industry values are not Panama (4.6 percent), Peru (3.2 percent), and Venezuela relevant because the objective is precisely to assess the (2.7 percent).5 concession's specific cost of capital, given its existing The resulting flows were projected into perpetuity to financial structure. obtain the TV. This study favors arguments in favor of using indus- In practice, if the concessions were to be re-bid at the try averages, and they have been used in this study's cal- end of their current contract, a growing perpetuity of culations. In addition, for many of the concessions ana- future flows would be only one of the methods used to lyzed, specific betas were simply not available. Estimating reasonable terminal values In a net present value computation, the terminal value 4For a very young concession, this can underestimate returns, since (TV) is often the largest amount. This is especially true the last flows may still be of a small amount. Note also that any for young concessions, for which the TV includes a large exceptional amount from the calculation was removed. number of future years. Estimating the TV reasonably is 5Source: IMF, 2002. How Profitable Are Infrastructure Concessions in Latin America? 11 estimate their value. Prospective bidders would also services or that their value is frequently inflated, so that look at the net asset value of the concession and at trad- all or part of them should be considered as dividends. In ing and transaction multiples (that is, multiples of earn- this case, the concession's operating and net income ings and assets at which similar businesses trade or have should be corrected (that is, increased), and the fees paid been transacted). The perpetuity method was preferred should be added to dividends in the computation of the because it is dynamic and forward looking, while the internal rate of return. All returns would thus be higher. other methods are mostly static and based on historical To be as conservative as possible, returns have been numbers. corrected, assuming that all the explicit management Adding a TV may, however, overestimate the com- fees paid by concessions were in fact dividends to their pensation concessionaires can reasonably expect. First, strategic shareholders. It is important to understand not all concession contracts include the payment of that this assumption implies that none of the services or compensation to shareholders for the value created at know-how the management fees were supposed to the end of the concession's operation. Second, it cannot remunerate are considered real. If the concession really be excluded that a future government negotiates a needed to use those services or to buy that know-how, change to or is not in the capacity to pay such com- and had to acquire them from an outside party, their pensation. Third, compensations are often based on the costs would be real costs to the concession and no cor- nonamortized value of assets, which may be very dif- rection would be made to its RoE and RoCE, while the ferent from the value of estimated future flows. In addi- Shareholder IRR should only be increased by the mar- tion, concessionaires generally expect to earn a suffi- gin earned by shareholders on those services. Hence, by cient return over a reasonable period of time and do including them entirely as dividends, it is assumed the not rely on the TV to get a sufficient return. They will concession could have reached the same operating and tend to rely on returns including the future value (FV) net incomes without any of these services or know-how of flows until the last year of the concession, but transfer. This is clearly a strong assumption. excluding the value of flows beyond the concession It must also be noted that some concessions may not operation period. recognize the management fees paid to their concession- aires as such. The latter might be disguised under other Adjustments to financial returns cost categories, such as technical assistance costs. As explained earlier, financial returns were adjusted rec- Investments (transfer pricing) ognizing that the financial accounts of a company may When concessions invest in new machinery and equip- not always be fully representative of its economic situa- ment or build new facilities, they tend to use related par- tion. Two adjustments were made, related to manage- ties to execute the works. Some argue that related par- ment fees and investments. Other possible adjustments ties are likely to charge higher prices for these services were run as sensitivities. and works than if competitive bidding was organized, Management fees due to the monopolistic nature of the transaction. As a For many of the concessions, a management contract result, the transfer price is inflated and the concession's was signed between the concession company and its income reduced. shareholders (see table 12, page 27). These contracts It is obviously very difficult to assess the fairness of typically called for various services to be provided to the such transactions since there is precisely no competitive concession by its shareholders, including the transfer of bid with which to compare them. In fact, most countries technology, development of general policies, prepara- include transfer pricing audits in the audits required tion of detailed strategic plans, design of formal organi- from concessions. These require full disclosure of intra- zational structures, hiring of qualified staff, and formu- group transactions, and compare their prices with those lation of annual operating budgets. These services are of similar transactions internationally. Therefore, signif- generally remunerated with a fee payable to the conces- icant inflation of transfer prices should not go unnoticed sionaire by the concession, normally defined as a per- by auditors. Also, not all concessions' investments are centage of sales or operating profits. made by related parties. In the concession's accounts, these fees are usually Therefore, the value of the total investments made treated as tax-deductible operating costs. Therefore, by each concession were reduced by only 10 percent, they reduce the concession's operating and net incomes. and, likewise, the annual depreciation charges, assum- However, it is often argued they do not remunerate real 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 by the goodwill amortization charges (spread over a longer 33 percent. period of time), so that the net effect on returns is neg- With such adjustment, the operating and net incomes ative. One could adjust returns for the difference are higher (as depreciation charges are lower), which between the goodwill amortization charges and the leads to higher return on equity and on capital depreciation charges on the overvalued assets. employed. Concession returns are also higher because From the point of view of shareholders, however, investments are smaller. The shareholder IRR is also what matters is the amount they effectively paid for higher since it was assumed these investment cost the assets and on the basis of which they hope to gain markups were direct benefits for the strategic share- a sufficient return. From that angle, no adjustment is holders' companies, which could then distribute them to needed. themselves as dividends (after tax). Pre-acquisition asset value Other possible adjustments Companies often revalue their assets shortly after their In addition to the two adjustments made, several other privatization. A revaluation increases the concession's adjustments could have been made. However, these fur- asset base but also depreciation charges, and reduces ther adjustments would have been very difficult to esti- rates of return. Frequently, little information is provid- mate and quite uncertain. They have, therefore, not ed on the basis of these revaluations, and doubts arise been considered in this analysis. on their economic foundations. However, the lack of Acquisition value information makes any reasonable adjustment impossi- When bidding for a concession, concessionaires often ble to make. pay an acquisition premium over the fair value of the Transfer pricing (other than for investments) target firm, known as goodwill. This acquisition premi- In the same way investments are often carried out by um is triggered by the competition created by the bid- related parties to a concession, many other services can ding process. The buyer is willing to pay such premiums be provided to the concession by group companies. The if the asset provides a competitive advantage, such as an most common suspicious accounts are repairs and entry point in a new market, or a stronger brand name. maintenance, consulting services, technical assistance, Normally, the goodwill appears on the balance sheet rent, and general services. Again, the prices of the relat- of the acquirer in the amount by which the purchase ed services are likely to be inflated as a result of the price exceeds the net tangible assets of the acquired monopolistic nature of the transactions. However, com- company, and the goodwill is amortized over a few panies tend to disclose little information on transfer years, creating a recurrent cost in the company's prices other than for large transactions, so that it is not accounts. For all the concessions analyzed, however, possible to make any reasonable assumption on their the goodwill was not recognized. Rather, the value of true extent. In addition, all intragroup transactions are the assets of the acquired companies was increased so supposed to be audited for transfer prices so that there the net asset value of the acquired company was made should be limited room for overpricing. In any event, equal to the purchase price. This implies the entire pur- regulators are encouraged to require concessions to doc- chase price was incorporated in the capital base of the ument and disclose their transfer pricing policies as concession, which reduces returns (in particular the much as possible. Project IRR). However, the acquisition premium should be left in the concession's capital base only if it repre- Depreciation sents real market value. If it represents a premium the Depreciation periods for infrastructure assets can some- acquirer was willing to pay for strategic reasons, for times be shorter than the actual life of the assets (accel- instance, then it represents additional value for the cur- erated depreciation). This increases depreciation rent buyer, but probably not for a future purchaser. In charges in the short term and reduces returns. Even this case, it needs to be progressively eliminated by though accelerated depreciation is normally allowed by amortizing the goodwill. most countries' tax and accounting rules, from an eco- In addition, overvalued assets lead to higher depreci- nomic point of view depreciation should be made over ation charges, which lower concessions' returns on equi- the actual life of the assets. However, adjusting for this ty or capital employed further (by reducing operating would only be possible if a detailed depreciation sched- and net incomes). These charges are usually larger than ule was available for each concession's assets. How Profitable Are Infrastructure Concessions in Latin America? 13 5. COMPUTATION OF THE HURDLE RATES: COST OF CAPITAL Computation of the cost of equity would only need to earn an internal rate of return of 6 As explained above, the cost of equity has been calcu- percent to invest in an American concession and 7 per- lated on the basis of CAPM. The required parameters cent in a Chilean concession. have been estimated as described in Appendix 2. Figure Since the main discriminating factor is the country 2 shows the resulting current costs of equity for each risk premium (see Appendix 2), the cost of equity tends country. to vary significantly across countries, but less so across The figure indicates, for instance, that an investor sectors within a country (see Figure 3). looking to invest today in a concession in Argentina would need to earn an internal rate of return of at least Computation of the weighted average cost 19 percent on the investment. If the concession's finan- of capital cial projections show the internal rate of return on the The weighted average cost of capital has also been investment is likely to be less than 19 percent, the invest- derived from CAPM. The required parameters have ment would not be worth it, as the investor should be been estimated as described in Appendix 2. Note the able to find alternative investments with a similar level cost of debt used in the calculation is nominal and does of risk but a higher expected return. The same investor not include the cost of potential debt renegotiations. It Figure 2: Estimated cost of equity by country, May 2004 25% 19% 20% 18% 19% 17% 18% 15% 13% 14% 10% 7% 7% 6% 5% 0% ru Bolivia Brazil Chile nama Pe Mexico States Pa nezuela Argentina Colombia Ve United Source: Authors' calculations. 14 Computation of the Hurdle Rates-Cost of Capital may understate the effective cost of debt and the result- worth it, as the investor should be able to find alterna- ing WACC. Figure 4 shows the resulting estimates of tive investments with a similar level of risk but a higher the weighted average cost of capital per country. expected return. The same investor would only need to Figure 4 indicates, for instance, that an investor look- earn a project internal rate of return of 3 percent to ing to invest today in a concession in Argentina would invest in an American or Chilean concession. need to earn a project internal rate of return of at least As for the cost of equity, the WACC varies more 14 percent on the investment. If the concession's finan- across countries than across sectors (see Figure 5). cial projections show the project internal rate of return Since the estimated cost of debt is lower than the cost is likely to be less than 14 percent, the investment is not of equity, the resulting WACCs are lower than the costs Figure 3: Estimated cost of equity by sector; United States and Argentina, May 2004 United States Argentina 10% 25% 9% 9% 22% 20% 19% 19% 8% 20% 7% 18% 17% 7% 6% 6% 6% 15% 5% 5% 4% 4% 10% 3% 2% 5% 1% 0% 0% . . ter . . Roads Ports ter Wa Telecom distr gener Roads Ports Wa distr Telecom gener Energy Energy Energy Energy Source: Authors' calculations. Figure 4: Estimated weighted average cost of capital by country, May 2004 16% 14% 14% 14% 13% 13% 12% 12% 10% 9% 8% 8% 6% 3% 4% 4% 3% 2% 0% Peru Bolivia Brazil Chile Mexico States nezuela Argentina Colombia Panama Ve United Source: Authors' calculations. How Profitable Are Infrastructure Concessions in Latin America? 15 Figure 5: Estimated WACC by sector; United States and Argentina, May 2004 United States Argentina 5% 5% 16% 15% 15% 4% 3% 15% 3% 3% 3% 14% 14% 14% 2% 2% 14% 13% 13% 2% 13% 13% 13% 1% 12% 12% 0% 11% . . . . Roads Ports Water Telecom distr gener Roads Ports Water distr Telecom gener Energy Energy Energy Energy Source: Authors' calculations. Figure 6: Comparison of the estimated cost of equity and the estimated WACC CoE WACC 25% 20% 19% 19% 18% 17% 18% 15% 14% 14% 14% 13% 13% 12% 13% 10% 9% 8% 7% 7% 6% 5% 3% 4% 3% 0% Bolivia Brazil Chile Peru Mexico States Argentina Colombia Panama Venezuela United Source: Authors' calculations. of equity (see Figure 6). This means equity investors try, the regulatory environment, and so forth), and as expect higher returns than debt holders, a logical conse- the risk-free rate moves. Therefore, three hurdle rates quence of their taking on more risk. for each concession have been computed: at their start, on average over their operation, and at the end of 2001. Variability during the concessions' lifetime It can be seen from Table 5 that, despite small varia- As explained in Appendix 2, both the cost of equity and tions, on average the cost of equity was relatively stable the weighted average cost of capital vary constantly, as until 2001. The average weighted cost of capital of the new information is incorporated into betas and country sample concessions, by contrast, has been falling over ratings (on the company's risk level, the sector, the coun- the life of the concessions (by up to 2 percentage points 16 Computation of the Hurdle Rates-Cost of Capital on average). This fall results from concessions' higher hurdle rates over each concession's historical years of overall indebtedness level in the telecommunications operation are compared. One has to keep in mind that and energy sectors. In the analysis, returns to average hurdle rates are higher today. 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% WACC Water 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. How Profitable Are Infrastructure Concessions in Latin America? 17 6. CONCESSION AND SHAREHOLDER RETURNS As explained in the methodology, two sets of returns Figure 7 shows that without including the future have been computed. First, concession returns, those value to be created by the concessions (future and ter- measuring the overall attractiveness of the concessions minal values)--that is, including historical years only-- as business entities, were computed. Second, share- our concessions reach a financial return of negative 24 holders returns, those effectively earned by project percent, well below their average WACC of 13 percent. shareholders from the distribution of dividends or This results in part from our sample concessions' other sources of funds generated by the concession, low operating profitability compared to their average were computed. WACC. Figure 8 shows the average annual return on For each sets of returns, financial returns derived capital employed generated by our sample conces- directly from the financial statements of the concessions sions so far was 7 percent (oscillating between 4 and were first computed. Then, to account for some poten- 9 percent from year to year), well below the average tial economic distortions, with the objective of estimat- WACC of 13 percent. (The impact of adding up man- ing adjusted returns representative of the economic sit- agement fees and excess depreciation is minimal, the uation of each concession, these returns were adjusted. overall average annual ROCE rising to 9 percent). Finally, three values were provided when returns Investments, which averaged 27 percent of our con- based on an internal rate of return calculation cessions' annual revenues, have also had their toll on (Shareholder IRR and Project IRR) were measured. First net profitability. these returns were computed only over historical years Despite this low historical profitability, Figure 7 to account for the returns already generated by the con- shows the sample concessions should on average be able cessions or earned by their concessionaires to date. to generate an internal rate of return (Project IRR with Second, it was estimated what they might generate or TV) above their average WACC (14 percent) if their earn over the concessions' remaining years of operation future growth is at least equal to each country's average by projecting flows until each concession's last year of historical economic growth and the residual value operation. Third, to estimate the returns they would added is taken into account. In this sense, infrastructure generate or earn if they were adequately compensated concessions are interesting business proposals for poten- for the value created at the end of each concession, a ter- tial investors, providing them with an adequate long- minal value was added. In this section, the results of the term return compared to the risk taken. analysis, focusing on each of these measures of return, In fact, based on this study's adjusted measures of are presented. returns, they even seem able to generate some excess returns. If no management fees were paid to the con- Concessions' returns cessions' operators, the concessions' average long-term Overall concession returns Project IRR would reach 15 percent; and if, in addition, Starting with the most aggregated level first, the average the cost of investments was not possibly inflated by 10 returns earned by our sample of 34 concessions were percent, their internal rate of return would reach 19 computed. Obviously, these overall returns must be percent. interpreted with caution given the variety of countries It must be noted, however, that these adjustments and sectors included in the sample and the wide dis- may not always be feasible or desirable for the con- crepancy of results across them. cessions' shareholders. Management fees may be 18 Concession and Shareholder Returns Figure 7: Overall concession return Project IRR, Overall average 25% 19% 20% 14% 15% 13% 15% 10% 7% 5% 0% IRR, with FV IRR, with TV IRR with fees IRR with fees WACC -5% IRR, no TV adjusted and investment -10% markup adjusted -15% -20% -25% -24% -30% Source: Authors' calculations, based on concessions' historical financial statements and the authors' growth assumptions. Figure 8: Average annual profitability of sample concessions ROCE Vs WACC 16% 14% 12% 10% 8% 6% 4% 2% 0% 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. unavoidable if the services they pay for are necessary Concession returns by sector for the concession company, and the cost of invest- Returns seem to vary somewhat across sectors, although ment may not be reducible. In addition, reducing the general conclusions presented above apply to all sec- management fees and possible investment markups tors. As Figure 9 shows, historical returns are negative would reduce shareholders' returns, while this analy- in all sectors. The water sector stands out, with an aver- sis will show later that these are below the required age historical return significantly lower than those in cost of equity. other sectors. How Profitable Are Infrastructure Concessions in Latin America? 19 Figure 9: Long-term concession returns by sectors Project IRR vc WACC 25% 5% -15% -35% -55% Water Transport Telecom Energy IRR, no TV IRR, with FV IRR, with TV IRR with fees adjusted IRR with fees and investment WACC markup adjusted Source: Authors' calculations, based on concessions' historical financial statements and the authors' growth assumptions. Figure 10: Average operational profitability by sector ROCE Vs WACC 20% 15% 10% 5% 0% 1993 1994 1995 1996 1997 1998 1999 2000 2001 -5% -10% Water Transport Telecom Energy Avg WACC Source: Authors' calculations, based on concessions' historical financial statements. This results because in all sectors the net average ating profitability of concessions in the water sector has annual operating profitability of concessions has not been the lowest. It has also been the most volatile. been high enough compared to their respective levels of Telecommunications is the sector with the highest aver- risk (see Figure 10). Table 6 confirms the average oper- age operational profitability. 20 Concession and Shareholder Returns Large investments have also reduced the net prof- Table 6: Volatility of profitability by sector itability of concessions, as illustrated in Table 7. Standard deviation As Figure 9 shows, water is the only sector where the Average RoCE of RoCE long-term financial return of concessions is expected to Water 4.3% 4.3% remain below the sector's corresponding WACC (9 per- Transport 5.2% 2.9% cent Project IRR with TV, versus 11 percent WACC). Telecom 8.2% 3.8% However, if management fees are added back to the Energy 7.2% 3.5% concessions' net income, the average return of water Global average 6.3% 3.7% concessions equals their average WACC. This means Source: Authors' calculations, based on concessions' historical financial that if water concessions could reach our projected prof- statements. itability without paying any management fees, they would be interesting business proposals.6 If they could also reduce the cost of their investment by at least 10 Table 7: Investment levels by sector percent, their average long-term return would exceed the corresponding WACC. Average annual level In all the other sectors, the estimated long-term of investments in returns all equal or exceed the corresponding sector % of revenues WACC, even without adjustment. When return adjust- Water 32 ments are introduced, the long-term returns of conces- Transport 35 sions in all three sectors exceed their corresponding Telecom 26 WACC. Energy 21 Concession returns by country Overall 27 Source: Authors' calculations, based on concessions' historical financial Mexico, Panama, and Venezuela were excluded from statements. the 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- Colombia stands out with above average expected ly across countries. Historical returns (IRR with no TV) returns partly because the average historical profitabili- have been negative in all countries, except in Colombia, ty of Colombian concessions has been the highest in the where they are marginally positive. sample (see Figure 12). The lower level of investments in Looking at long-term returns, Colombian conces- percentage of revenues also contributes to these conces- sions really stand out as the only ones capable of gener- sions' higher profitability (see Table 8). ating a long-term return well above their corresponding In Bolivia, the country where concessions have WACC, even without any adjustment. Adjustments recorded the lowest return, investments have been the would bring their expected return above 40 percent. highest compared to revenues, while the operational Brazil follows, with concessions estimated to gener- profitability of concessions has been the lowest (except- ate long-term financial returns equal to their WACC. ing Argentina). In Argentina and Chile, long-term returns become equal to the corresponding country WACC only with Concession returns by concession maturity adjustments. Returns vary also as a function of each concession's Peru and even more so Bolivia stand out as the only maturity, that is, its years of operation. As the concession countries where our sample concessions are expected to matures, profitability usually increases (heavy invest- earn a long-term return below the country WACC, even ments and operating restructuring often penalize the when adjustments are taken into account. In Bolivia, the early years). Table 9 shows that concessions with more long-term adjusted return of concessions is projected to than 10 years of operation have returned an adjusted be negative. Project IRR higher than the overall average WACC (16 percent compared to 14 percent). This confirms that concessions are economically profitable businesses in the long term, possibly after as few as 10 years of operation. 6Because management fees are supposed to remunerate some transfer The concessions' Project IRR increases systematically of knowledge and other services, suppressing them could lower the concessions' profitability. as the sample concessions' number of years of operation How Profitable Are Infrastructure Concessions in Latin America? 21 Figure 11: Average concession returns by country Project IRR vs WACC by Country 60% 40% 20% 0% -20% -40% -60% -80% Chile Brazil Argentina Colombia Bolivia Peru IRR, no TV IRR, with FV IRR, with TV IRR with fees IRR with fees and investment Avg WACC markup adjusted Source: Authors' calculations, based on concessions' historical financial statements and the authors' growth assumptions. rises. The average RoCE increases, too, with conces- have generated a positive net operating profitability, the sion maturity, although less so. This seems to suggest majority of them have so far generated a return below that concession returns are driven up in later years by their WACC or even negative. lower investments. It may also be that for many con- Table 10 confirms that, while the majority of our sam- cessions returns are depressed in early years by inade- ple concessions have so far generated a negative return, quate prices that are then corrected after the first price about 60 percent of them should generate an adjusted control period. return above their WACC in the long term. This means that about two-thirds of our sample concessions have the Individual concession returns potential to become interesting business proposals. The intention here is not to provide an analysis of However, it also means that about 40 percent of our each concession's returns, but to draw some general sample concessions do not seem to have the potential to conclusions. generate adequate long-term returns under our growth Individual concession returns are highly volatile. assumptions. These unattractive concessions are spread Figure 13 shows that while all our concessions but one in all sectors, but with a lower concentration in the telecommunications sector (where 25 percent of our concessions fall into that category). Table 8: Investment levels by country Concession return, a conclusion Average annual level The analysis above shows that if concessions continue of investments in to grow at least as fast as the economies around them, % of revenues their average long-term financial profitability will be Chile 30 equal to their average WACC. Under these assumptions, Brazil 27 they are on average interesting business proposals. Argentina 22 Returns are highly volatile, however. They vary across Colombia 15 sectors, countries, and concessions. Concessions in the Bolivia 32 water sector appear less attractive than in others, as do Peru 28 concessions in Peru and Bolivia. Concessions in the Overall 26 telecommunications sector and concessions in Colombia Source: Authors' calculations, based on concessions' historical financial state- appear overall more profitable than in other sectors and ments; the calculation includes all years of operation of each concession. countries in the sample. The largest differences are 22 Concession and Shareholder Returns Figure 12: Average annual concession profitability by country ROCE 14% 12% 10% 8% 6% 4% 2% 0% -2% 1995 1996 1997 1998 1999 2000 2001 Chile Brazil Argentina Colombia Bolivia Peru Chile Brazil Argentina Colombia Bolivia Peru Global Avg RoCE 7.5% 9.7% 3.0% 10.7% 7.1% 7.3% 6.3% St dev 7.6% 3.6% 3.9% 4.4% 3.5% 2.6% 1.5% Source: Authors' calculations, based on concessions' historical financial statements. observed by concession maturity: Concessions with over Therefore, concessions seem, in general, to be an attrac- 10 years of operation appear clearly more profitable tive but highly risky business. This is especially true in than younger ones. This may result from the way invest- the energy and transport sectors, where as many as half ments were forecast (future investment flows were based of our sample concessions do not seem able to generate on historical investment amounts, penalizing young con- adequate returns in the long term (under our base case cessions that usually suffer from heavy early invest- assumptions). ments). As mentioned earlier, it may also be that returns are often depressed in early years by inadequate prices Shareholders' returns that are then corrected after the first price control period or by favorable renegotiation. Whatever the reason, it Overall shareholder returns means that returns are built over many years. Figure 14 shows that with a growth rate equal to his- Returns also vary widely across concessions. torical economic growth, the shareholders of our sam- Therefore, even if, overall, about 60 percent of our sam- ple concessions would on average earn a long-term ple concessions have the potential to generate attractive financial return well below the average required cost of returns in the long term, 40 percent do not seem to have equity (negative 27 percent compared to 18 percent). the potential to ever generate attractive returns. When management fees and investment cost markups Table 9: Historical concession returns by concession maturity Number of years of operation Avg ratio of Economic Project IRR (as of Dec. 2001) investment to revenue Adjusted RoCE (no TV) < 2 46% 7% -91% 2 to 4 18% 5% -45% 5 to 7 29% 8% -30% 8 to 10 29% 9% 2% > 10 24% 11% 16% Overall 27% 8% -19% Source: Authors' calculations, based on concessions' historical financial statements. How Profitable Are Infrastructure Concessions in Latin America? 23 Figure 13: Financial RoCE and Project IRR by concession 40% Project IRR 20% RoCE 0% -20% -15% -10% -5% 0% 5% 10% 15% 20% 25% -20% -40% -60% -80% -100% -120% Source: Authors' calculations, based on concessions' historical financial statements. are added to dividends, the average Shareholder IRR It must also be noted that the prospective average increases substantially but remains below the average return of 14 percent with management fees and invest- cost of equity (at 14 percent). Figure 14 also shows that ment markups includes a terminal value. This means historical returns earned so far by concession share- that it would only be earned by concession shareholders holders from dividends only have been highly negative, if they are compensated fairly at the end of their con- at ­49 percent on average.7 cession's operation period for the value added they cre- With our assumptions, investment markups make ated by reinvesting part of the concession's earnings into up a large share of the prospective returns concession the concession. Most concession contracts include some shareholders might hope to earn. This is the result of form of compensation to their shareholders for the the large investments made so far by all our concessions value created. Such compensation often takes the form (see Table 7), on the basis of which future investments of a payment equal to the nondepreciated value of assets were projected. In reality, future investments are likely or to the market value of the concession company at to be high for most of our concessions, but sharehold- that time. ers may not earn any markup on these (for instance, if Such payments are not free of risk, however. Since they are implemented by companies outside the group they are made by the government, they are exposed to or at market prices). Therefore, in such cases share- the same credit risks as any other government liabilities. holders will have to rely on dividends and management The history of concessions is, unfortunately, too short to fees only. assess how fairly concessionaires tend to be treated at This suggests that to earn a sufficient return share- the end of their concession. Without such payment, holders will have to achieve concession growth rates however, the returns shareholders can expect are well superior to those of the economies around them. below the required cost of equity. Again, this does not include the potential impact that retained earnings 7Note that this does not include the potential increase in their own might have had on their own share price during the life share price as a result of the earnings accumulated in their conces- of their concession, although this value added is subject sion subsidiaries. 24 Concession and Shareholder Returns Table 10: Dispersion of returns across concessions (number) With Adjusted Project IRR (no TV) With Adjusted Project IRR (with TV) Below Above Below Above % unattractive Sector Negative WACC WACC Negative WACC WACC concessions Water 7 1 2 2 2 6 40 Transport 2 2 0 0 2 2 50 Telecom 2 3 3 1 1 6 25 Energy 8 2 2 2 4 6 50 Globally 19 8 7 5 9 20 41 Source: Authors' calculations. to the same political risk as the overall concession (as it firmed in Table 11, which shows the average dividend could be captured by an expropriating government). payout ratio of our sample concessions is 30 percent. As Figure 15 shows, the low return earned so far by This means that on average 70 percent of net income concessionaires results largely from the low average has been reinvested every year in the concessions. annual return on equity earned by our sample conces- The large differences between Project and Share- sions. Over the last 10 years, the latter has been signifi- holder IRRs confirm that shareholders have so far not cantly below the average cost of equity. (Management extracted much of the value they created in their con- fees and investment markups have a marginal impact on cessions by way of dividend distributions. the average RoE.) In fact, their average return on equi- Shareholder returns by sector ty (5.8 percent) has been just below the cost of equity in As Figure 16 shows, shareholders' returns vary widely the United States (6.1 percent). across sectors. As for our overall results, without adjust- The large difference between the average RoE and ments, the long-term returns concession shareholders the average historical Shareholder IRR suggests that on can expect to earn are below the required cost of equity average a significant portion of net income has been in all sectors. In fact, such return is negative in all reinvested in each concession every year. This is con- sectors, except in telecommunications, where it is Figure 14: Overall long-term shareholder returns Shareholder IRR, Overall average 30% 18% 20% 14% 10% -3% 0% IRR, no TV IRR, with FV IRR, with TV IRR with fees IRR with CoE -10% adjusted fees and -20% investment markup adjusted -30% -28% -27% -40% -50% -49% -60% Source: Authors' calculations, based on concessions' historical financial statements and the authors' growth assumptions. How Profitable Are Infrastructure Concessions in Latin America? 25 Figure 15: Average annual return on equity of sample concessions RoE versus CoE 20% 18% 16% 14% 12% 10% 8% 6% 4% 2% 0% 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Avg RoE Avg RoE with economic adjustments Avg CoE Source: Authors' calculations, based on concessions' historical financial statements. marginally positive. Management fees bring sharehold- Shareholder returns by country er returns above the required cost of equity in the Figure 17 shows that shareholders' financial returns telecommunications sector, while in water, management vary quite widely across countries. In financial terms fees and potential investment markups are needed to (with dividends only) they are positive only in Colombia bring shareholder returns to the required level. In the (with TV). Colombia also stands out as the only coun- other sectors, energy and transport, even with such try where, under our assumptions, concession share- adjustments, shareholder returns remain below the holders would earn a long-term return above their required cost of equity. In the transport sector, they even required cost of equity just from dividends. If manage- remain negative. This is the result, to a large extent, of ment fees and potential investment markups are added the lower dividend payout concessionaires have benefit- to dividends, their returns would be above 40 percent. ed from in the transport sector (see Table 11). As Table In Bolivia, some concessions have paid large man- 12 shows, the relatively high management fees that agement fees to their shareholders, so that if they were transport concessions have paid their strategic share- to be maintained in the future, shareholders would earn holders on average have not been sufficient to compen- returns in line with their required cost of equity. In sate for the low dividends. Brazil, potential investment markups could possibly bring returns very close to the required cost of equity (as investments have been high in the past and are project- ed to remain so), but again, there may not be any invest- Table 11: Payout ratios by sectors ment markups in these concessions. % net income Chile stands out as a country where shareholder distributed in dividends returns would remain negative if things were to contin- Water 18 ue on a steady state path, even including management Transport 2 fees and potential investment markups. This is one Telecom 44 of the countries with the lowest payout ratios (after Energy 41 Argentina). Shareholders in Chile, Argentina, and Peru Overall 30 would only earn an adequate return in the long term Source: Authors' calculations, based on concessions' historical financial if the future rate of growth of income is higher than statements. 26 Concession and Shareholder Returns Figure 16: Overall shareholder returns by sector Shareholder IRR versus CoE 25% 5% -15% -35% -55% -75% -95% Water Transport Telecom Energy IRR, no TV IRR, with FV IRR, with TV IRR with fees adjusted IRR with fees and investment CoE markup adjusted Source: Authors' calculations, based on concessions' historical financial statements and the authors' growth assumptions. in our projections. It is interesting to note that share- made in general, if their shareholders have been remu- holders in the Argentinean concessions in the sample, nerated through possible investment markups, they who are most frequently accused of having extracted might in fact have already earned a return higher than resources from Argentina via large dividend flows, their cost of equity. In all the other countries, concession have paradoxically the lowest average historical payout shareholders have earned negative returns so far, even if (see Table 13). management fees and potential investment markups are Figure 17 shows also that historical returns have included (but excluding the potential increase in their been much lower. Looking at dividends only, sharehold- own companies' share price or value). ers in all countries have earned negative returns so far. Shareholder returns by concession maturity If management fees are included, only Colombian con- Shareholder returns vary also as a function of each con- cessions have paid their shareholders a positive return, cession's maturity, or the number of years in operation. just one percent below the required cost of equity. Given This is because shareholders simply benefit from more the large investments Colombian concessions have years of dividend distribution. Table 14 shows that it is only for concessions with more than 10 years of opera- tion that average shareholder returns from dividends Table 12: Management fees by sector become positive (but they remain much lower than the % Avg average cost of equity). The average annual return concessions with management on equity of our concessions increases systematically management fee in % of with their maturity, becoming positive after five years contracts total sales and close to (but below) the average cost of equity after Water 60 2 10 years. Transport 100 7 Given that, on average, concessions with more than Telecom 75 2 10 years of operation returned a Project IRR above the Energy 42 1 required WACC, the discrepancy with shareholder Overall 62 2 returns (which are below the required cost of equity) Source: Authors' own calculations. confirms that concessionaires have reinvested most Note:The low level of management fees (as a percent of total sales) is confirmed earnings in their concessions. It also suggests that the by an analysis of electric utilities in Latin America by Deutsche Bank Securities overall economic profitability of concessions has not Inc, which shows that only 14 percent of all electric utilities in Latin America have established management fees in their contracts. How Profitable Are Infrastructure Concessions in Latin America? 27 Figure 17: Shareholder returns by country Shareholder IRR versus CoE 60% 40% 20% 0% -20% -40% -60% -80% -100% Chile Brazil Argentina Colombia Bolivia Peru IRR, with no TV IRR, with FV IRR, with TV IRR, with TV, with fees IRR with TV, with fees and Avg CoE investment markup Source: own calculations, based on concessions' historical financial statements and the authors' growth assumptions. been shared yet between equity and debt holders pro- the concession or its contractual terms could be renego- portionally to their investments, even for those conces- tiated shortly after. The evidence supports that hypoth- sions that have been in operation for more than 10 years esis, since on average the incidence of renegotiation of (unless shareholders have been able to cash in the value infrastructure concessions in Latin America was 42 per- added accumulated in the concessions by way of share cent, (but much higher in the water and sanitation sec- price increases in their own companies). tor, 75 percent, and in the transport sector, 55 percent), There is another argument that suggests old conces- the average time interval between the granting of the sions should be more profitable than young ones. As concession and renegotiation was less than three years, reported in Guasch (2004), there has been a tendency of and the outcome of the renegotiations usually favored aggressive bidding by potential operators to secure the the operator (Guasch 2004). This would provide a pro- rights to the concession. That aggression, or overbidding, file consistent with low or negative returns early on and led to contractual terms that were not financially sus- larger returns later on (after renegotiation). tainable from the start, with rates of return not covering Individual shareholder returns the cost of capital. The rational for that overbidding has The main characteristic of shareholder returns by con- been the expectation--quite often well founded--that cession is their high volatility, not only across conces- sions, but also from year to year. Table 13: Payout ratios by country, average over concession life Volatility of returns across concessions Payout ratio Looking at the volatility of returns across concessions Chile 24% first, Figure 18 shows no two concessions are alike in Brazil 37% terms of their RoE and Shareholder IRR. Nevertheless, Argentina 10% some concentration can be found, in particular in the Colombia 27% corner, with highly negative financial Shareholder IRR Bolivia 68% (without FV or TV) and a small positive RoE. Peru 54% Table 15 confirms that with adjusted returns the Overall 30% majority of our sample concessions have a Shareholder IRR below the required cost of equity, even with ter- Source: Authors' calculations. minal value, and that for more than half our sample, 28 Concession and Shareholder Returns Figure 18: Financial RoE and Shareholder IRR by concession 40% IRR 20% RoE 0% -50% -40% -30% -20% -10% 0% 10% 20% 30% -20% -40% -60% -80% -100% -120% Source: Authors' calculations. concession shareholders would only be adequately sions to date and from low dividend distribution policies remunerated with higher growth rates. (for which limited management fees do not compensate). Transport and energy stand out as sectors where 75 If concessions were to grow in the future at the same percent of all concessions do not have the potential to rate as the economies around them (all other things generate adequate returns in the long term to their being equal), in the long term shareholders could hope shareholders without higher growth. to earn an overall positive return only if they continued to earn management fees and potential investment Volatility of returns from year to year markups. Their return would, however, remain below Looking at the volatility of returns from year to year, the required cost of equity. This means that on average Figure 19 shows that RoEs have varied substantially shareholders can only hope to earn returns commensu- over the period covered, especially in the water sector. rate with the risks taken if there is superior concession Surprisingly, telecommunications, where sales are sup- growth in the future. posed to be more volatile than for truly basic services, has had one of the most stable RoEs. The standard devi- ation of RoE over the last 10 years was 15 percent in Table 14: Relation between return and concession maturity water, 6 percent in transport, 7 percent in telecommuni- Number of years Adjusted caitons, and 8 percent in energy. of operation Adjusted Shareholder Figure 20 shows that returns have been volatile in (as of Dec. 2001) RoE IRR (no TV) each country. < 2 10% -78% Shareholder returns, a conclusion 2 to 4 -1% -70% The returns earned so far by our sample concession 5 to 7 6% -34% shareholders from dividends and other financial flows 8 to 10 13% -6% (excluding potential appreciations in their own share > 10 15% 1% price) have been negative in all sectors and countries. Overall 8% -30% This results from the low average profitability of conces- Source: Authors' calculations. How Profitable Are Infrastructure Concessions in Latin America? 29 Table 15: Dispersion of adjusted RoE and Shareholder IRR across concessions Number of concessions with: Adjusted Shareholder IRR Adjusted Shareholder IRR (with TV) % unattractive Sector Negative Below CoE Above CoE Negative Below CoE Above CoE concessions Water 6 0 4 2 2 6 40 Transport 3 0 1 2 1 1 75 Telecom 2 1 5 1 1 6 25 Energy 10 2 0 4 5 3 75 Globally 21 3 10 9 9 16 53 Source: Authors' calculations. Figure 19: Evolution of annual RoE by sector 30% 20% 10% 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 -10% -20% -30% -40% Water sector Average Transport sector Average Telecom sector Average Energy sector Average Source: Authors' calculations. Figure 20: Evolution of the annual RoE by country 30% 20% 10% 0% 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 -10% -20% -30% Chile Brazil Argentina Colombia Bolivia Peru 30 Concession and Shareholder Returns Figure 21: Project IRR sensitivity to concession growth rates Project IRR versus WACC 40% 20% 0% -20% -40% -60% Water Transport Telecom Energy IRR, no TV IRR, with FV IRR, with TV IRR IRR with fees and investment WACC markup adjustments Source: Authors' calculations. Returns vary widely, however, across sectors, coun- their average WACC, which means they do constitute tries, and concessions. In the telecommunications sec- potentially interesting business proposals (with wide tor, shareholders can hope to earn adequate returns if variation). By contrast, their shareholders should not be the present conditions are perpetuated, provided man- able to earn long-term returns in line with their average agement fees and investment markups are included. In cost of equity, even when management fees and invest- Colombia, shareholders might hope to earn adequate ment markups are included. The discrepancy between returns even without management fees and investment shareholder and concession returns in the long term markups. As the overall profitability of concessions (including a terminal value)9 suggests that equity hold- increases with their maturity, so do shareholder ers will not receive a share of profits proportional to returns. However, even for concessions with more than their participation and risk taking in the concessions' 10 years of operation, shareholder returns remain funding without more generous dividend policies. below the average cost of equity. The conservative dividend distribution policies of concessionaires suggest that they tend to have a long- Concession and shareholder returns, a conclusion term perspective when investing in concessions: They Overall, the returns earned so far by our sample con- tend to reinvest a significant part of earnings in the cession shareholders from dividends have been signifi- concessions in the first years of operation in the hope cantly lower than the overall returns generated by their these investments will increase their overall return, concessions. This results from low distribution policies. even if it is at the cost of reducing their immediate It may be partially compensated by increases in the financial returns. shareholders' company share price, however.8 It also means that to earn a return commensurate to If things continue on a steady path, concessions the risks they took, concessionaires probably intend should overall generate long-term returns in line with either to distribute more generous dividends in the later years of their concessions (to cash in part of the accu- mulated value) or to reach concession growth rates 8However, concessions' retained earnings are usually not fully reflect- superior to those of the economies around them. ed in their mother companies' share price because of the associated political risk. An analysis of the relationship between earnings retained in subsidiary concessions and a mother company's share 9 The latter captures potential increases in shareholder prices--but prices would indicate to what extent retaining earnings creates slightly underestimates it, given that US$1 is worth more today than immediate value for concession shareholders. tomorrow. How Profitable Are Infrastructure Concessions in Latin America? 31 Figure 22: Shareholder IRR sensitivity to concession growth rates Shareholder IRR, Overall average 30% 20% 18% 20% 10% 0% 0% IRR, no TV IRR, with FV IRR, with TV IRR with IRR with CoE -10% fees adjusted fees and investment -20% markup -24% adjusted -30% -27% -40% -50% -49% -60% Shareholder IRR versus CoE 40% 20% 0% -20% -40% -60% -80% -100% Water Transport Telecom Energy IRR, no TV IRR, with FV IRR, with TV IRR with fees adjusted IRR with fees and investment CoE markup adjusted Source: Authors' calculations. The analysis highlights also that returns, in particu- turn explains why investors in such concessions lar shareholder returns, are highly volatile. While the demand high expected returns. standard deviation of the average return on capital Sensitivity analysis is presented in the next section. employed in our sample concessions over each conces- sion's years of operation is relatively low at 1.5 percent, Sensitivity analysis the standard deviation of the return on equity is signif- icantly higher at 10 percent. In addition, the analysis Future concession growth shows that as many as 40 percent of our sample con- The analysis above has shown that most concessions cessions are unlikely to generate adequate returns would not be able to remunerate their shareholders ade- unless they outperform the economies around them. quately compared to the risks they take, unless the con- The combination of a high volatility of returns and a cessions grow faster than the underlying economy. high probability for returns to be lower than required The analysis above assumed that concessions would confirms that infrastructure concessions in Latin grow in the future at the same rate as the underlying America are a high risk investment proposal. This in economy and that the economies would grow at their 32 Concession and Shareholder Returns Figure 23: Shareholder IRR sensitivity to dividend payout Shareholder IRR, Overall average 30% 22% 18% 20% 10% 0% IRR, no TV 8% IRR with CoE -10% IRR, with FV IRR, with TV IRR fees and -12% -4% investment -20% markup -30% adjusted -40% -50% -49% -60% Shareholder IRR versus CoE 40% 20% 0% -20% -40% -60% -80% -100% Water Transport Telecom Energy IRR, no TV IRR, with FV IRR, with TV IRR with fees adjusted IRR with fees and investment CoE markup adjusted Source: Authors' calculations. average historical growth rate, calculated to vary A sensitivity analysis was conducted, increasing the between 2.7 to 4.7 percent per annum, depending on future growth rate of each concession to 7 percent per the country. In reality, the sample concessions have so annum. Maintaining their historical average growth far been growing at a much faster rate (17 percent on rate of 17 percent did not seem reasonable, since this average for their operating income).10 This may be the was achieved when most concessions were just starting result of their young age (seven years on average) and to operate (which usually leads to growth rates much includes wide discrepancies across concessions (the higher than those in the later years of the concession's standard deviation is 41 percent). life, as the new private management reduces costs and increases efficiency mostly in the first years). In addi- 10Based on the sample concessions, energy has been the fastest grow- tion, conditions and growth prospects in the region ing sector, with an average growth rate of operating income of 23 appear to have worsened for most concessions, since percent, followed by water at 18 percent, transport at 12 percent, when they were awarded. Seven percent per annum and telecommunications at 7 percent. The calculation excludes years of exceptionally high or low growth for each concession. seemed a reasonably optimistic proposal. How Profitable Are Infrastructure Concessions in Latin America? 33 Figure 24: Shareholder IRR sensitivity to concession growth rates and dividend payout Shareholder IRR, Overall average 50% 40% 40% 26% 30% 18% 20% 15% 7% 10% 0% IRR, no TV IRR, with FV IRR, with TV IRR with fees IRR with CoE -10% adjusted fees and -20% investment markup -30% adjusted -40% -50% -49% -60% Shareholder IRR versus CoE 80% 60% 40% 20% 0% -20% -40% -60% -80% -100% Water Transport Telecom Energy IRR, with no TV IRR, with FV IRR, with TV IRR, with fees adjusted IRR with fees and CoE investment markup Source: Authors' calculations. Figures 21 and 22 show that concession returns are Future concession dividend policy quite sensitive to the future rate of growth of operating Figure 23 shows shareholder returns using a higher div- cash flows. In fact, shareholder returns remain largely idend payout (85 percent per annum)--but keeping negative and below their corresponding cost of equity, everything else unchanged. This scenario approximates but concession returns all rise to levels above their cor- the returns shareholders would earn if they could cap- responding WACC without adjustments (the average ture every year most of the value added created in their Project IRR without adjustments reaches 21 percent). concessions (while keeping its profitability unchanged), Shareholder returns remain insufficient, however, which rather than having to wait until the concession is re-bid, means that an accelerated growth would not be enough as our previous analysis assumes. to ensure them of adequate remuneration. They would The figure shows the long-term financial return share- also need to capture more of the value accumulated into holders would earn remains largely below the required their concessions, either via a more generous dividend cost of equity in all sectors. It becomes higher than the distribution policy or by cashing in capital gains. required cost of equity in the telecommunications sector 34 Concession and Shareholder Returns Figure 25: Project IRR sensitivity to concession acquisition price Project IRR versus WACC 30% 10% -10% -30% -50% Water Transport Telecom Energy IRR, no TV IRR, with FV IRR, with TV IRR with fees adjusted IRR with fees and investment WACC markup adjusted Source: Authors' calculations. if management fees are included and in the water sec- The figure shows that, under those circumstances, tor if potential investment markups are also included. concessions in all sectors have the potential to generate Overall the return becomes sufficient only when possi- (on average) adjusted shareholder returns superior to ble investment markups are included, but the existence their corresponding cost of equity (including manage- of the latter is clearly uncertain. ment fees only). With investment markups, returns This suggests that, with historical growth assump- would even become quite higher than the required cost tions, capturing annually most of the value added creat- of equity. Average financial returns become positive but ed by concessions would not be enough to ensure share- remain lower than the cost of equity in all sectors. holders of an adequate return, except in telecommuni- This suggests that shareholders in concessions cations concessions (assuming that management fees expect to earn sufficient returns on their investment by can be considered as dividends rather than operating maintaining growth rates superior to those of the costs). This suggests that telecommunications conces- economies in which they operate. It also shows they sions might be the only ones where concessionaires rely on both dividends and capital gains, and on man- should be able to reap adequate returns without signifi- agement fees. Without any one of these sources their cantly outperforming the market, provided they can returns would remain insufficient even with a higher capture annually accumulated capital gains. rate of growth. Future concession growth and dividend policy Acquisition price Figure 24 shows concession and shareholder returns Figure 25 shows the returns that would have been using both a higher growth rate (7 percent per earned had concessionaires paid 20 percent less for their annum) and a higher dividend pay out (85 percent per concession and invested 20 percent less equity in its cap- annum). This scenario measures the returns that ital (assuming debt would not have changed).11 The fig- shareholders would earn if they could capture every ure shows the average Project IRR becomes equal to the year most of the value added created in their conces- average WACC in all sectors, without adjustments. In sions and if the overall growth of their concessions was superior to the historical growth of each conces- 11This would, however, lead to an increase in leverage, which might sion's country of location. not have been acceptable to the project's financiers. How Profitable Are Infrastructure Concessions in Latin America? 35 Figure 26: Shareholder IRR sensitivity to concession acquisition price Shareholder IRR versus CoE 40% 20% 0% -20% -40% -60% -80% -100% Water Transport Telecom Energy IRR, no TV IRR, with FV IRR, with TV IRR with fees adjusted IRR with fees and investment CoE markup adjusted Source: Authors' calculations the energy sector, the average concession return management fees (and in water with uncertain invest- becomes significantly higher than the average WACC. ment markups). Bringing them in line with the average However, it has a marginal impact only on share- cost of equity in all sectors without adjustments would holders' returns. It becomes superior to the required still require increasing the payout ratio or future con- cost of equity in the telecommunications sector with cession growth rate as previously illustrated. 36 Concession and Shareholder Returns 7. THE IMPACT OF REGULATION ON PROFITABILITY The results reported above indicate that concessions the quality of regulation, but also on the chosen regula- have the potential to be profitable businesses over the tory regime, including elements of the concession full life of their contracts. However, they are risky busi- design. Under rate of return regulation, the regulator nesses with only 60 percent of concessions in the sam- has the possibility of making frequent price adjustments ple having the potential to generate attractive returns, to keep realigning the company's rate of return with its and these returns, moreover, are premised on manage- cost of capital. Under price cap regulation, the regulator ment fees and other additional sources of revenue, as sets tariffs so that expected returns match the cost of well as (in some cases) outperforming historic market capital ex ante, but allows these returns to diverge ex trends. post during the periods between regulatory reviews. Unlike normal competitive business sectors, the prof- In practice in Latin America, the distinction between itability of concessions is not simply a reflection of mar- price cap and rate of return regulation is somewhat ket conditions and managerial competence, but is to a blurred. Although most regulatory regimes provide for considerable extent determined--or at least circum- periodic tariff reviews (typically at five-year intervals), scribed--by regulatory decisions. The companies ana- suggesting a multi-annual price cap framework, con- lyzed in this study operate mostly under a monopoly tracts are often renegotiated between reviews (Guasch regime and are subject to regulation of tariffs and other 2004). The short interval between the granting of a con- aspects of enterprise performance. Thus, the observed cession and its renegotiation, about two years, and the profitability of these concessions in part reflects the per- outcome of the renegotiation process, makes the result- formance of the regulators that oversee them. ing regime a hybrid of price caps and rate of return.12 Moreover, review methodologies sometimes take into Conceptual framework account historic divergences between the rate of return Regulation is needed both to protect consumers from and the cost of capital in adjusting future prices, which abuse of monopoly power and to protect investors from goes against the forward-looking principles of price cap opportunistic behavior by the government, given the regulation. Thus, in practice both types of regulatory politically sensitive nature of infrastructure tariffs and regime tend to converge to a hybrid, suggesting that rate the large sunk-cost characteristics of the companies' of return should track the cost of capital more closely investments. In consequence, regulatory decisions have than under a pure price cap, but less closely than under a substantial impact on the profitability of companies. pure rate of return. Ideally, the regulator's objective should be to maintain Accordingly, instead of focusing on the dichotomy alignment between a company's rate of return and its between price cap and rate of return regulation, the cost of capital. This is because a rate of return in excess of the cost of capital inappropriately penalizes con- 12Guasch (2004) shows the incidence of renegotiation is about 42 per- sumers, while a rate of return beneath the cost of capi- cent of all concessions and about 55 percent and 75 percent for con- tal inappropriately discourages further investment. The cessions in the transport and water sectors, respectively. And the closeness of that alignment will depend, among other incidence is even much higher for concessions regulated under a things, on the quality of regulation. price cap regime. Even more striking is how fast those renegotiations take place. The time interval between the granting of the concessions Nevertheless, the closeness with which the rate of and renegotiation is about 2.1 years, and for water concessions it is return tracks the cost of capital will not only depend on even quicker, about 1.6 years. How Profitable Are Infrastructure Concessions in Latin America? 37 Table 16: Construction of regulatory quality indices Weight Scoring Legal solidity 0.33 1 if regulatory framework established by law, 0 otherwise Financial capacity 0.33 Sum of scores on factors below · Financial independence 0.17 · 1 if funded from regulatory levy, 0 if funded from public budget · Financial strength 0.17 · Regulatory budget as percentage of sectoral GDP normalized on [0,1] scale Decision-making autonomy 0.33 Sum of scores on factors below · Independence of appointment 0.11 · 0 if appointed directly by executive, 1 if screened by legislature · Duration of appointment 0.11 · 1 for a single fixed term, 0 for indefinite appointment · Collegiality of decisions 0.11 · 1 if headed by regulatory commission, 0 if by individual regulator Note: Scores between 0 and 1 are given for intermediate cases. approach taken is to develop a measure of the overall quality of regulation. A second element of financial quality of the regulator that oversees each of the com- strength is the stability of the funding for the regulatory panies in the sample. agency. The principle is that stability and predictability The purpose of this section, then, is to empirically of resources--and thus better quality of regulation-- evaluate the impact of the quality of regulation on the comes when the funding is linked to the sales of the reg- profitability of the firms. The hypothesis is that the bet- ulated sector, rather than a yearly appropriation from ter the quality of regulation, the closer the correspon- the general government budget that can be changed at dence between the firm's rate of return and the firm's the discretion of the executive. cost of capital. Decision-making autonomy is key to securing good regulatory quality. This variable tries to measure the Measuring regulatory quality likelihood that the regulatory decisions are based on To test this hypothesis a quantitative measure of regula- technical and professional assessment of the contract, tory quality is needed. Good regulation is defined by law, and existing evidence as opposed to decisions based clear, stable, and predictable rules, a purely profession- on--or influenced by--a government's political agenda al and technical interpretation of the law and contract, or investors' influence or capture. This can be captured and the ability to withstand influences and pressures by three aspects: independence of appointment, which from the stakeholders, such as government and opera- measures the extent to which the appointment process tors. And for it to be effective regulation, it needs to be avoids a purely political appointee without adequate supported by a predictable and sufficient allocation of technical knowledge of the sector; duration of the resources. In consequence, the Regulatory Quality appointment, which indicates whether a regulator can Index developed here considers three key aspects of reg- be reappointed and hence might be less likely to act ulatory quality: legal solidity, financial strength, and independently and issue professionally and technically decision-making autonomy. based decisions; and collegiality of decisions, which Legal solidity refers to the stability, and thus pre- measures the relative difficulty of regulatory capture, dictability, of the regulatory regime. The stronger the thought to be lower when multiple regulators act jointly legal foundation of regulation, the more stable the within a board structure. regime is, and thus the better the quality of regulation. The construction of each of these indices and the The strongest legal foundation is when the regulatory associated scoring method are detailed in Table 16. The framework is embedded into a law, as opposed to a factors can either be considered individually or aggre- decree, contract, or other lesser legal instruments. gated into three broader regulatory quality indices Financial strength refers to the resources the regula- using the indicated weighting scheme, each of which tory agency has to undertake its functions. In principle may be summed together to obtain an overall regulato- and within limits, the more resources the agency has the ry quality indicator. For the sample of companies cov- better it can perform its function and the better the ered in this study, the average score on the index of 38 The Impact of Regulation on Profitability Table 17: Summary of regression results IRR-WACC simple IRR-WACC simple differential with differential with terminal value and IRR-WACC simple IRR-WACC simple terminal value and management fee and differential without differential with adjustment for adjustment for Dependent variable terminal value terminal value management fee transfer pricing Financial independence ­0.340 ­0.174 ­0.151 ­0.135 Financial strength ­0.372 ­0.332** ­0.355** ­0.370** Legal solidity ­0.026 0.077 0.070 0.080 Independence of appointment ­0.109 ­0.068 ­0.101 ­0.109 Duration of appointment ­0.125 ­0.011 ­0.038 ­0.030 Collegiality of decisions 0.455** 0.256** 0.271** 0.267** Constant ­0.341 ­0.047 ­0.022 0.002 P-value 0.156 0.072* 0.052** 0.045** Adjusted R-squared 0.124 0.208 0.237 0.248 No. of observations 32 30 30 30 Notes: Regressions based on 30 observations; *, **, *** indicate significance at 10 percent, 5 percent, and 1 percent levels, respectively. overall regulatory quality is 0.51, suggesting that the tion purely from a short-term, consumer's perspective, quality of regulation is not very high overall. However, since the smaller the IRR-WACC differential (including there is significant variation in quality across countries negative values), the lower the resulting tariffs for con- and sectors, with scores ranging widely between 0.12 sumers. However, this constitutes a myopic view, since a and 0.85. The highest average score is obtained on legal negative IRR-WACC undermines investment incentives solidity, 0.65, as against decision-making autonomy, and ultimately penalizes consumers through declining 0.56, and financial strength, 0.34. service quality, decelerating service expansion, and Pair-wise correlations between each of the regulatory potential flight of investors. Therefore, the absolute quality measures are typically low, at around 0.20, and IRR-WACC differential is taken as a second relevant in no case greater than 0.57. In some cases, pair-wise measure. According to this indicator, what matters is correlations even take negative values, suggesting that minimizing the distance between IRR and WACC, with high regulatory quality along one dimension is correlat- positive and negative differentials regarded as equally ed with low regulatory quality along another dimen- reflective of poor regulatory decisions. sion. This result illustrates that few countries have con- sistently applied all of the design principles needed to Simple differential (myopic ensure good quality regulation. consumer-protection perspective) These indices of regulatory quality are used to try to The results for the first set of regressions are reported in explain differences in the divergence between rate of Table 17, using each of the four measures of IRR- return and cost of capital across the different companies WACC differential developed in the study. Despite small in the sample. This is done by regressing the Project sample sizes, three out of the four models show the reg- IRR-WACC differential against this set of explanatory ulatory quality variables are significant in overall terms, variables. The hypothesis is that the greater the quality and are on their own capable of explaining 20 to 25 per- of regulation, as measured by the described index, the cent of the IRR-WACC differential. Moreover, some of smaller the differential should be, suggesting that the the regulatory quality variables are also individually sig- regulatory quality subindexes would enter the regres- nificant. Thus, the financial strength variable is signifi- sion with negative signs. cant at the 5 percent level in most of the regressions, Two separate measures of the IRR-WACC differen- with the expected negative sign indicating that regula- tial are considered. The first measure is the simple IRR- tors with larger budgets tend to have greater success in WACC differential. This captures the quality of regula- minimizing the IRR-WACC differential. In addition, the How Profitable Are Infrastructure Concessions in Latin America? 39 Table 18: Summary of regression results IRR-WACC simple IRR-WACC simple differential with differential with terminal value and IRR-WACC simple IRR-WACC simple terminal value and management fee and differential without differential with adjustment for adjustment for Dependent variable terminal value terminal value management fee transfer pricing First principal factor 0.064 0.041 0.029 0.024 Second principal factor -0.114 -0.017 -0.026 -0.019 Third principal factor -0.219** -0.115** -0.130** -0.127** Collegiality of decisions 0.400** 0.202** 0.216** 0.210** Constant -0.649 -0.136 -0.133 -0.095 P-value 0.065* 0.057* 0.040** 0.043** Adjusted R-squared 0.164 0.186 0.212 0.207 No. of observations 32 30 30 30 Notes: Regressions based on 30 observations; *, **, *** indicate significance at 10 percent, 5 percent, and 1 percent levels, respectively. collegiality of the decision variable is also significant at vations in an already small sample, a log-linear specifi- the 5 percent level, but takes a positive sign. This sug- cation is used to ensure that there is adequate variation gests that, arguably contrary to expectations, regulatory for the purposes of the regression. Overall, this second entities headed by a single superintendent do a better job set of regressions does not perform as well as the first. at reducing the IRR-WACC differential than broader Nevertheless, two of the models show overall signifi- based regulatory commissions.13 cance at the 5 to10 percent level and are able to explain To learn if there is a statistically more efficient way about 20 percent of the variation in the IRR-WACC dif- of combining the information embodied in the six indi- ferential. As before, the financial strength variable cators of regulatory quality, factor analysis is per- proves to be significant in some specifications, although formed. Factor analysis helps to ensure adequate not always with the expected sign. On the other hand, orthogonality between explanatory variables and also the collegiality of decisions is no longer statistically allows statistical information to be condensed into a significant. smaller number of variables, thereby economizing on The lower level of significance and explanatory degrees of freedom. Table 18 summarizes the results of power associated with this second set of regressions may the best specification found for factor analysis. This simply be reflecting that regulatory efforts are more preserves the collegiality of decisions variable, but com- strongly motivated by the short-term considerations of bines the other five variables into three principal fac- keeping prices as low as possible for current consumers tors. This leads to a slight improvement in the overall than by the long-term considerations of keeping returns significance of the regressions, while the pattern of sig- as close as possible to hurdle rates for investors. nificance hardly changes, except that the negative and The conclusion of this analysis is that regulation mat- significant coefficient of the financial strength variable ters in aligning cost of capital and rate of return. Overall, is picked up by the third principal factor. the existing regulatory frameworks are not rated very highly by the regulatory quality index, with an average Absolute differential (protecting both score of 0.51. Nevertheless, variations in quality across consumers and investors) regulatory regimes are significant and material in deter- The results of the second set of regressions are reported mining the size of the IRR-WACC differential. However, in Table 19. Given that taking the absolute value of the regulatory efforts seem to be more closely associated IRR-WACC differential reduces the spread across obser- with minimizing the simple IRR-WACC differential (and thereby keeping tariffs as low as possible for current con- sumers), than with minimizing the absolute IRR-WACC 13One weakness of regulatory commissions, perhaps captured here on differential (and thereby keeping profitability well the estimates, is the higher political intervention, since often each rel- aligned with hurdle rates of return). evant political party gets to designate its own commissioner. 40 The Impact of Regulation on Profitability Table 19: Summary of second regression results IRR-WACC simple IRR-WACC simple differential with differential with terminal value and IRR-WACC simple IRR-WACC simple terminal value and management fee and differential without differential with adjustment for adjustment for Dependent variable terminal value terminal value management fee transfer pricing Financial independence 1.071 -0.653 -0.001 0.071 Financial strength 2.619** -2.478 -2.488** -2.140** Legal solidity -0.697 0.928 0.412 0.844** Independence of appointment 1.147 0.974 0.577 -0.050 Duration of appointment -0.478 1.412 1.053 0.767 Collegiality of decisions -1.771 -0.810 -0.456 -0.243 Constant -1.104 -2.618** -2.365** -2.487** P-value 0.094* 0.273 0.125 0.049** R-squared 0.171 0.069 0.156 0.242 No. of observations 32 30 30 30 Notes: Regressions based on 30 observations; *, **, *** indicate significance at 10 percent, 5 percent, and 1 percent levels, respectively. Another striking feature of the results is that regula- another dimension. For example, a regulatory frame- tory quality variables seem to have overall significance, work that rests on a solid legal basis for regulatory deci- more than individual significance, in determining IRR- sions but does not provide the regulator with any finan- WACC differentials. This is consistent with the point cial resources is unlikely to be very effective. So for reg- that performance along different dimensions of regula- ulation to be effective, one needs the whole package of tory quality is not highly correlated and that the bene- regulatory characteristics. If some of the key ingredients fits of high regulatory quality along one dimension can are missing the effectiveness of regulation is highly be completely offset by low regulatory quality along diminished. How Profitable Are Infrastructure Concessions in Latin America? 41 8. CONCLUSIONS AND POLICY IMPLICATIONS The analysis has shown that concessions are profitable less for their concessions. The implication is that to (although risky) businesses overall, capable of generat- build an adequate return, shareholders must rely both ing adequate returns in the long term. Concessions in on various sources of remuneration (including divi- the water sector appear relatively less attractive than dends, management fees, and capital gains), and on others, while concessions in the telecommunications outperforming historical market growth consistently, sector appear to be the most profitable in relative terms. over the entire length of their concession. On average, concessions seem to become profitable The fact that concessionaires have maintained low after about 10 years of operation. However, about 40 dividend payouts so far, despite knowing that such a percent of our sample concessions do not seem to have choice would have a significant impact on short-term the potential to generate attractive returns, with this returns, indicates they have given priority in the alloca- number climbing to 50 percent in the energy and trans- tion of earnings to investments. This suggests that con- port sectors. Concessions are thus risky businesses. cessions had no other economic way to finance their The profitability results imputed here are likely to investment obligations, probably because of their size understate the true profitability of the firms, since, as and because of constraints on availability of funding opposed to firms in nonregulated sectors, the incentives from other sources. of the regulated firms is to underreport true earnings Furthermore, these results suggest that concession- through creative accounting, so as not to trigger reduc- aires operate with long-term perspectives, giving prior- tions in tariffs at the corresponding tariff reviews. ity to growth-enhancing investments in the early years Low dividend distribution ratios, however, have not (at the cost of depressing returns in the short term), and to date translated this overall profitability into adequate relying on the entire concession period to build an ade- returns for shareholders. In fact, on average, concession quate return. This may be driven by their contractual shareholders have so far earned negative returns on obligations, which usually require high investments in their investments, even including management fees, esti- the early years. It implies that early breaks of conces- mated accumulated capital gains, and potential invest- sion contracts may have a highly negative impact on ment markups. expected returns. With historical growth maintained into the future, The results also highlight that management fees, only telecom concessions would seem to have an inher- although not widespread, may be needed to build ade- ent profitability high enough to generate adequate quate returns. In addition, allowing concession share- returns to their shareholders in the long term, provided holders to be fairly compensated at the end of the period they can capture annually the capital gains accumulated for the capital gains accumulated during the life of the in their concessions over all years of operation and that concession is an important component of their return. the full value of their management fees corresponds to That old concessions are on average more profitable dividends. In all other sectors, shareholders can hope to than young ones suggests that returns may be depressed earn long-term returns commensurate to the risk taken in early concession years by inadequate prices or overly only if the sectors consistently and significantly outper- aggressive bidding, corrected after the first price control form historical market growth. This conclusion would period (high investments in the first years of operation not change if concessionaires had paid up to 20 percent may also take a toll on young concession returns) or 42 Conclusions and Policy Implications through renegotiations. The high incidence of renegoti- priately penalizes consumers, while a rate of return ation in infrastructure contracts, about 42 percent, beneath the cost of capital inappropriately discourages within three years of the award date, as reported in further investment. Guasch (2004), renders support to the hypothesis that An evaluation of the quality of the regulatory regimes operators have tended to submit overly aggressive faced by concessionaires in the study sample finds that bids--which are not financially viable--so as to secure the regimes do not score very highly on average and that the concessions, with the belief or expectation of rene- there is a high variance in the quality of regulatory gotiating the contract shortly afterward and securing frameworks across concessions. Yet, the quality of reg- better terms. The evidence reported also in Guasch ulation is found to be a significant determinant of the (2004) shows the outcome of the renegotiations tended divergence between the overall profitability of the con- to benefit the operator with better terms than those con- cession and its corresponding hurdle rate, explaining tracted or bid at the time of the award. about 20 percent of the variation. However, regulatory The analysis also highlights that returns (in particu- efforts seem to be more closely associated with mini- lar shareholder returns) are highly volatile across sec- mizing the simple IRR-WACC differential (and thereby tors, concessions, and from year to year. Thus, infra- keeping tariffs as low as possible for current con- structure concessions in Latin America are a high risk sumers), than with minimizing the absolute IRR-WACC investment proposal, explaining why the required rates differential (and thereby keeping profitability well of return on such investments are high. aligned with hurdle rates of return). Then the policy Finally, this report shows that regulation matters in implications are clear. Securing effective regulation to aligning cost of capital and rate of return. Given that protect both consumers and investors should be a key virtually all of the concessions included in this study are priority. For regulation to be effective the regulatory regulated monopolies, their profitability is not only a framework and institutional structure of a regulatory consequence of market conditions and managerial skills agency should have (a) legal solidity--embedded in a but also partly a reflection of regulatory decisions on law; (b) adequate financial capacity--capacity and service tariffs. A good regulator should aim to maintain strength; and (c) decision-making autonomy. Moreover, alignment between a company's rate of return and its it is imperative that all three regulatory elements be in cost of capital in the medium term. This is because a place. The absence of any of them significantly decreas- rate of return in excess of the cost of capital inappro- es the effectiveness of regulation. How Profitable Are Infrastructure Concessions in Latin America? 43 APPENDIX 1 MEASURES OF SHAREHOLDERS' EFFECTIVE RETURNS The Shareholder Internal Rate of Return The Shareholder Internal Rate of Return (IRR) is a measure of the profits that shareholders have distributed to remunerate their equity investment in the proj- ect. It is the return that makes the net present value (NPV) of the flow of divi- dends distributed (usually annually) by the concession to its shareholders, less the flow of capital injections made by them into the concession, equal to zero (see Box 4). It is the Shareholder IRR of the net flows received by the conces- sion's shareholders from their investment to date.14 For young concessions especially, the Shareholder IRR represents the return earned so far by shareholders, but it underestimates the total return they can expect to earn over the entire life of their investment, since it ignores the flow of future dividends. To capture the value of these future expected dividends, the Shareholder IRR was also computed, adding a terminal value (TV) to the stream of net cash flows. The TV has been estimated assuming that future dividends grow at a constant rate and that there would be no new capital injection. The formula is included in Box 5. The formula assumes concessionaires will not be compensated at the end of the concession for its value at that time. In most concession contracts this is the case. In fact, in most concessions the concessionaires do not own the BOX 4 Definition of the shareholder internal rate of return y Shareholder IRR is the rate which ensures CFt / (1 + IRR)t = 0 t=0 Where: CF = net cash flow to shareholders, that is, dividend ­ capital injection t = 0 is the first year of operations y = last year of historical data available (2001) 14Note the IRR computed represents the average IRR earned by all shareholders in the concession. It does not attempt to differentiate the returns of shareholders who might have purchased their shares at a later stage than when the concession was initially awarded, nor of those who might have sold their shares already. 44 Measures of Shareholders' Effective Returns BOX 5 Definition of the shareholders' internal rate of return with a terminal value y Shareholder IRR is the rate which ensures (CFt) + TV / (1 + IRR)t = 0 t=0 With:15 g ­ r ­ g r ­ g r Where: D = last historical dividend g = growth rate of future dividends r = cost of equity (see below for calculation details) n = concession's last year of operation concession's assets, which remain the government's property. However, some concession contracts state that at the end of the concession the government will pay compensation to the concessionaire, usually based on the concession's asset value at that time, which is intended to capture the value of nonamortized investments accumulated by the concessionaire. It was decided to ignore these compensations because of their uncertainty (their value will likely be subject to intense discussion). This may understate some results, although slightly, since uncertain and distant cash flows have a highly discounted value.16 The Return on Equity The Return on Equity (RoE) is a measure of the profits a company is able to generate given the resources provided by its shareholders. It is the ratio of the concession's net income divided by the shareholders' equity investment in the concession (see Box 6). Although in theory the concession's net income is available for distribution to equity holders, a portion of it is typically reinvested every year in the con- cession. This portion of net income is only really earned by shareholders in future years, at the latest at the concession's end, when it is sold or transferred back to the government, provided the transaction is at market prices and con- cessionaires are compensated for the value they created by reinvesting the money in the concession. To the extent that part of the concession's net income is not earned immediately by the concessionaires and that they might not be compensated at the end of the concession for its extra value, the RoE tends to overestimate their effective returns. The RoE might be seen as a ceiling capping shareholders' potential returns. Compared to the RoE, the Shareholder IRR presents two advantages as a measure of shareholder returns. First, it summarizes in one single number the 15Formula for a growing annuity. 16Doing so better corresponds to the investors' viewpoint, for whom uncertain distant cash flows do not have much value. How Profitable Are Infrastructure Concessions in Latin America? 45 BOX 6 Definition of the return on equity RoE = Net income/Shareholders' equity Where: Shareholders' equity (also called net worth) = total assets minus total liabilities = book value of common and preferred stocks Net income = after-tax profit return earned by shareholders up to now (without terminal value) or over the entire life of the project (with terminal value), while the RoE is an annual meas- ure. Second, it takes into account the exact net financial flows received by shareholders every year. The Shareholder IRR is, therefore, a superior measure of the concessionaires' financial returns, the RoE being more of a general meas- ure of the company's efficiency. However, the information needed to calculate a Shareholder IRR is not always available. In such cases, computing an average RoE over the concession's life is a reasonable alternative measure of returns. The Project Internal Rate of Return The Project Internal Rate of Return (Project IRR) is a measure of how effec- tively a company uses the funds invested in its operation, independent of the nature of these funds (see Box 7). It is the rate that makes the net present value of the net financial flows generated by the concession before financing equal to zero. The net financial flows generated by a concession are calculated as its earnings before interest, taxes, depreciation, and amortization (EBITDA)--a measure of the financial flows generated by the concession's operation, inde- pendent of its financial structure--minus the investments, increases in work- ing capital, and taxes financed by these flows, and minus the price initially paid for the concession. As for the Shareholder Internal Rate of Return (Shareholder IRR) earned by concession shareholders, the project IRR/including a TV equal to the estimated future value of the net flows generated by the concession was computed (see Box 8). It was assumed the net flows would grow at a constant rate. BOX 7 Definition of the project internal rate of return y Project IRR is the rate which ensures CFt/(1 + IRR)t = 0 t=0 Where: CF = net cash flow generated by the concession, that is, EBITDA ­ (the investments ­ working capital ­ bid price) t = 0 is the first year of operation y = last year of historical data available (2001) 46 Measures of Shareholders' Effective Returns BOX 8 Definition of the project internal rate of return with a terminal value y Project IRR is the rate that ensures (CFt) + TV / (1 + IRR)t = 0 t=0 g With: TV = CF r ­ g ­ r ­ g r Where: CF = last historical net cash flow generated by concession (EBITDA ­ Inv. ­ WC g = growth rate of future cash flows r = weighted average cost of capital (see below for definition) n = concession's last year of operation The Return on Capital Employed The Return on Capital Employed (RoCE) is a measure of the returns a compa- ny is getting from its capital at large (see Box 9). It is calculated as the ratio of profits before interest divided by the difference between total assets and current liabilities. The denominator is, in fact, a measure of the long-term financing structure of the company (that is, its capital and long-term liabilities). The numerator is a measure of the income generated by the company independent of its financing structure. The resulting ratio represents the efficiency with which long-term financing is being utilized to generate net operating revenues. Compared to the RoE, the RoCE is not strictly limited to the return earned on shareholders' invested funds. It is based on a broader notion of capital than equity capital, to incorporate all the long-term funds that were invested in the concession. In addition, it is based on the company's net operating income (before financial charges and revenues), rather than its net income to exclude the impact of the choice of financial structure. Therefore, it is a meas- ure of the concession's ability to generate sufficient operating income (after tax) to cover its financial obligations of all kinds. As with the RoE, the RoCE is an annual measure. BOX 9 Definition of the return on capital employed RoCE = EBIT * (1 ­ T) / Capital Employed Where: EBIT = Earnings before interest and taxes T = nominal corporate income tax rate Capital Employed = Fixed assets + current assets ­ current liabilities = Shareholders' funds + long-term liabilities How Profitable Are Infrastructure Concessions in Latin America? 47 The Project IRR, compared to the RoCE, has the same two advantages as the Shareholder IRR, compared to the RoE. First, it summarizes in one single number the return earned by a concession over its entire life (while the RoCE is an annual measure of profits). Second, it takes into account the exact net financial flows generated by the concession every year. For instance, it deducts from earnings the entire cost of the investments financed in any given year, while the RoCE is only affected by the smaller related depreciation charges. The Project IRR is a superior measure of the concession's financial return, the RoCE being more of an economic measure of the company's efficiency. However, the information needed to calculate the Project IRR is not always available. In such cases, computing an average RoCE over the concession's life is a reasonable alternative measure of returns. The Return on Investment (ROI) is sometimes used as an alternative to the RoCE. It is the ratio of net income divided by the company's capital employed. Compared to the RoCE, it deducts financial charges from operating income. The pre-financial charges RoCE were preferred to measure the profitability of the concession independent of its financial structure. 48 Measures of Shareholders' Effective Returns APPENDIX 2 COMPUTATION OF THE COST OF EQUITY AND THE WEIGHTED AVERAGE COST OF CAPITAL Computation of the cost of equity As explained earlier, the cost of equity is calculated with the following formu- la in Box 10. BOX 10 Definition of the cost of equity 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 The required parameters have been estimated as follows. Risk-free rate The risk-free rate is a theoretical interest rate that would be returned on an investment completely free of risk. Interest rates on government bonds are often used as proxies for the risk-free rate. This implicitly assumes that governments are default-free entities. However, in some emerging markets, governments have sometimes failed to meet their financial obligations. Many governments are clearly not risk-free. Because of its default-free track record, the United States government is usually used as a proxy for risk-free entity, and the interest rate on the U.S. three-month treasury bill is usually considered the best approximation of a virtually risk-free interest rate. At the end of May 2004, the risk-free rate was 0.98 percent. Figure 27 shows the evolution of the risk-free rate since 1960. Beta Beta is a quantitative measure of the volatility of a given stock relative to the overall market. The S&P 500 index is usually used as reference for the overall stock market. Beta is defined using an index of 1:1, that is, a beta above 1 means that the stock is more volatile than the market, and a beta below 1 means it is less volatile than the market. Volatility is defined here as the rate at How Profitable Are Infrastructure Concessions in Latin America? 49 Figure 27: Evolution of the risk-free rate, 1960­04 In % 16 14 12 10 8 6 4 2 0 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 T bill (3 month) interest rate Source: Federal Reserve Board of the United States. which the price of a security moves up and down and is usually measured as the annualized standard deviation of daily changes in price. The relative volatil- ity of a stock is measured as the ratio of the covariance between the stock's and the market's return divided by the variance of the market's return. Betas are regularly estimated and updated by a number of specialized private companies. Some companies use the simple covariance method described above, based on historical stock prices to get a historical beta. While some stud- ies have shown betas appear reasonably stable,17 historical betas are only imper- fect guides to the future since the market risk of a company can genuinely change. Some other companies, such as Barra,18 therefore, use more forward- looking methodologies, where historical betas are adjusted to take into account some forward-looking quantitative and qualitative information about the com- pany and its environment (including the regulatory framework). The resulting betas are called predicted or fundamental betas. These are deemed superior to historical betas since they incorporate new information that may influence the future volatility of the stock. There are, therefore, better predictors of an asset's future response to market movements. However, companies such as Barra do not calculate betas for nontraded companies nor for small companies with limited liquidity, especially in emerg- ing markets. Therefore, one has to use proxies. The betas of the sample con- cessions were proxied using the average predicted betas estimated by Barra for American companies in the same sectors.19 The resulting betas are summarized in Table 20. 17See for instance, Sharpe and Cooper, 1972. 18Barra, an American company founded in 1975, became famous for its multi-factor model for measuring the risk of stock portfolios. Its estimates of betas are used by many investment banks and stock brokers. 50 Computation of the Cost of Equity and the Weighted Average Cost of Capital Table 20: Unleveraged betas by sector Unleveraged Number of companies fundamental beta in Barra sample Transportation -- roads 0.57 16 Transportation -- ports 0.49 44 Water 0.34 17 Telecom services 0.85 195 Energy distribution 0.51 6 Energy generation 0.34 2 Source: Barra, end 2003 The table shows the average betas of all the infrastructure sectors considered in this analysis are below 1. This means that stocks of companies in those sec- tors are usually less volatile than the market,20 so that investments in those sec- tors are less risky than in other sectors with higher betas. This reflects the fact that these sectors enjoy more stable economics, in particular a more stable demand, than other sectors. One of the lowest betas is that of water companies (0.34). This reflects to a large extent the fairly stable demand for water, which tends to immunize water companies from large market shocks. The highest beta in the sample is that of telecommunications companies (0.85), which reflects the higher variability of the demand for telecommunications services and, there- fore, their higher sensitivity to market shocks. The other sectors fall between these two extremes. Leveraging betas To isolate risks resulting from the financing structure of a company from its fundamental business risk, betas are usually calculated assuming the company has a hypothetical unleveraged financial structure (they are then called BOX 11 Leveraged and unleveraged betas ßL = ßU * (1 + D / E * (1- T)) Where: ßL = Leveraged beta ßU = Unleveraged beta D = Outstanding long-term debt E = Total shareholders' funds T = Corporate income tax rate 19European companies were used when Barra's sample of American companies was not available. For instance, Barra does not separate energy distributors and generators for U.S. companies. In that case, Barra's samples of European energy companies were used. Note that Alexander, Mayer, and Weeds (1996) show that asset beta values are higher the more incentive­based the regulato- ry regime. Therefore, U.S. beta values reflect the regulatory regime in place in the U.S. in each sec- tor. The adjustments to account for other regimes would, however, be small. 20Assuming the market has a beta of 1; some argue that the market itself has a beta below 1, in which case, these assets' volatility could be in line with the market. How Profitable Are Infrastructure Concessions in Latin America? 51 Table 21: Typical leverage by sector Typical leverage (D/E) Corresponding D/Assets Transportation 162% 62% Water 156% 61% Telecom services 119% 54% Energy 127% 56% Source: Authors' calculation based on sample, December 2001. Note: Leverage is defined as the ratio of total liabilities to total shareholders' funds. Differences across countries appeared insignificant in our sample and were ignored. Table 22: Nominal corporate income tax rates by country 2002 Nominal income tax rate Argentina 35% Bolivia 37.5% Brazil 20% Chile 35% Colombia 35% Mexico 25% Panama 30% Peru 30% Venezuela 34% Source: PriceWaterHouse; Latin Business Chronicle. Note: When the country's nominal corporate income tax rate varies depending on business sizes or income levels, an arithmetic average of all nominal levels has been computed. This is the case of Brazil (two slabs at 15 and 25 percent), Bolivia (25 or 50 percent, depending on nature of business), and Mexico (six slabs from 15 to 35 percent). unleveraged or unlevered betas). The betas presented in Table 20 were all unleveraged betas. They then need to be re-leveraged to account for the extra risk embedded in the company's leveraged capital structure (leveraged or lev- ered betas), using the formula in box 11. As explained earlier, unleveraged betas were transformed into leveraged betas using a capital structure typical of each sector (rather than the specific capital structure of each concession). For each sector, the average leverage of sample companies was used. Table 21 shows these sectors are usually highly leveraged. This again is a reflection of their relative stability, but also of the usually high investments to be financed. The tax rates used are each country's nominal corporate income tax rates (Table 22). The resulting leveraged betas are summarized in Table 23. The table shows that once leverage is taken into account the expected volatility of each sector increases as expected, but the impact is larger in the telecommunications and transport sectors. The average betas in these sectors tend to become higher than 1. In the water and energy sectors, by contrast, leveraged betas tend to remain lower than 1. Market risk premium The market risk premium represents the additional return that investors will require to hold a risky investment (shares) rather than a risk-free asset. It is gen- erally measured as the historical average annual excess return on the U.S. stock 52 Computation of the Cost of Equity and the Weighted Average Cost of Capital Table 23: Leveraged betas by sector and by country As of Transportation Energy Oct. 28, 2003 Roads Ports Water Telecom Distribution Generation Argentina 1.17 1.00 0.68 1.51 0.94 0.61 Bolivia 1.14 0.98 0.67 1.48 0.92 0.60 Brazil 1.31 1.12 0.76 1.66 1.03 0.68 Chile 1.17 1.00 0.68 1.51 0.94 0.61 Colombia 1.17 1.00 0.68 1.51 0.94 0.61 Mexico 1.26 1.08 0.74 1.61 1.00 0.66 Panama 1.21 1.04 0.71 1.56 0.97 0.64 Peru 1.21 1.04 0.71 1.56 0.97 0.64 Venezuela 1.18 1.01 0.69 1.52 0.94 0.62 Source: Authors' calculations. Table 24: Returns on stocks compared to government bonds Average total annual return: Average excess Stocks (S&P 500) Treasury bonds (USA) return on stocks 1960­2004 10.7% 5.8% 5.0% Sources: Global Financial Data Inc, Federal Reserve Board, the United States. market (using returns on the S&P 500) above the risk-free rate (Table 24). The geometric average of these excess returns over the period 1960­2004 were used. A market risk premium of 5.0 percent was used. Country risk premium The country risk premium corresponds to the extra return investors require to invest in stocks of companies in a country deemed riskier than a less risky coun- try used as benchmark (often the United States). The country risk premium reflects the potential volatility of investments in a given country due to defaults associated with political or other events in that country. Country risk premiums are usually estimated as the average spread over the U.S. treasury bond (assumed to be risk-free) of U.S. corporate bonds with a credit rating equivalent to that of the country under consideration (called the default spread). To estimate these spreads default spreads estimated by Reuters for a large number of utilities worldwide were used. Some argue that the country risk premium is likely to be higher than the country's default spread. Therefore, they multiply the latter by the ratio of the volatility of the equity market to that of the bond market in the country under consideration (sometimes proxied by the same ratio globally of 1.5). Such adjustment has not been made, to be as conservative as possible. Table 25 shows the country risk premium is the most discriminating factor among countries, varying from less than 1 percent in Chile to 13 percent in Argentina and Venezuela. It is also highly volatile, as Table 26 shows, varying, for example, from 7 percent in 1990 to 13 percent in Argentina in 2003. This is because it is influenced by many factors and subject to frequent shocks and variations, including exchange rate risk, political risk, regulation risk, and so forth. It is for this reason that one needs to compare expected returns at any given time with the cost of equity at the same time. How Profitable Are Infrastructure Concessions in Latin America? 53 Table 25: Country risk premiums Moody's Long-Term Default As of Rating (foreign spread Country Oct 28, 2003 currency bonds) (in basis points) risk premium Argentina Caa1 1420 13.4% Bolivia B3 1320 12.4% Brazil B2 1170 10.9% Chile Baa1 146 0.7% Colombia Ba2 780 7.0% Mexico Baa2 163 0.9% Panama Ba1 720 6.4% Peru Ba3 830 7.5% United States Aaa 71 0.0% Venezuela Caa1 1420 13.4% Sources: Moody's, bondsonline Figure 28: Evolution of country risk premiums over time (percent) 14 Argentina Colombia 12 10 Brazil Bolivia Peru 8 6 Venezuela 4 Mexico Panama 2 Chile 0 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Argentina Bolivia Brazil Chile Colombia Mexico Panama Peru Venezuela Source: Authors' calculation based on data from Moody's and Bondsonline. 54 Computation of the Cost of Equity and the Weighted Average Cost of Capital Table 26: Cost of debt by country Cost of debt Pre tax Post tax Argentina 20.6% 13.4% Bolivia 19.6% 12.2% Brazil 18.1% 14.5% Chile 7.9% 5.1% Colombia 19.2% 12.5% México 8.1% 6.1% Panama 13.6% 9.5% Peru 14.7% 10.3% United States 7.2% 5.2% Venezuela 20.6% 13.6% Source: Authors' calculations. Computation of the weighted average cost of capital As explained above, the weighted average cost of capital is derived from the formula in Box 12. BOX 12 Definition of the weighted average cost of capital 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 in Box 10) CD = cost of debt T = nominal corporate income tax rate The required parameters have been estimated as described below. Book value of equity and long-term debt As explained in chapter 6, a typical capital structure for each sector was used, rather than concession-specific leverages. The typical leverage levels used are presented in Table 21. Cost of equity The estimated cost of equity was used and is presented in Box 2. Corporate income tax rate The nominal corporate tax rates of each country were used, as presented in Table 22. How Profitable Are Infrastructure Concessions in Latin America? 55 Cost of debt A typical cost of debt for each country was used, estimated on the basis of the cost at which a hypothetical corporate issuer could issue local currency bonds in each country, given that country's rating. This cost was estimated to be equal to the sum of the risk-free rate, the country risk premium, and a 20 basis point premium for corporate issues over sovereign issues.21 Note also there was no attempt to try to estimate the cost of potential debt renegotiation or restruc- turing. Table 27 shows the resulting estimates. 21Note that the same country risk premium (a historical country risk premium) to compute the cost of equity was used. The country risk premium relevant to compute the cost of debt may, howev- er, be different because the relevant horizon is usually shorter (it would be higher if the risk of investing in a given country is perceived as higher in the short term than in the long term, and vice versa). 56 Computation of the Cost of Equity and the Weighted Average Cost of Capital REFERENCES Alexander, I., A. Estache and A. Oliveri. 2001. "A few things transport regulators should know about risk and the cost of capital," Utilities Policy, Volume 9, 1­13. Alexander, I., C. Mayer, and H. Weeds. 1996. Regulatory Structure and Risk and Infrastructure Firms: An International Comparison, Policy Research Working Paper, 1698, the World Bank. Andres, Luis, V. Foster and J. Luis Guasch. 2004. The Impact of Privatization in Latin America: A Cross Country Analysis, Washington, DC: World Bank. Basanes, F. and R. Willig. 2002. Second generation reforms in infrastructure services, Inter-American Development Bank. Birdsall, N. and Nellis, J. 2002. Winners and Losers: Assessing the Distributional Impact of Privatization, Working Paper No. 6, Center for Global Development, Washington, D.C. Campos, J, A. Estache and L. Trujillo. 2003. "Processes and accounting matter for regulators: Learning from Argentina's railways privatization," Journal of Network Industries, 4(1), pp. 3­28. Chong, Alberto, and Florencio Lopez-de- Silanes. 2005. Privatization in Latin America: Myths and Reality, Washington, DC: World Bank and Stanford University Press. Correia da Silva, L, A. Estache and S. Jarvela. 2004. "Is Debt Replacing Equity in Regulated Privatized Infrastructure in Developing Countries," Policy Research Working Paper No. 2971, Washington, DC: World Bank. Estache, A. 2004. "Argentina Privatization: A Cure or a Disease?" In Hirschhausen, Christian, Thorsten Beckers and Kay Mitusch, Trends in Infrastructure Regulation and Financing: International Experience and Case Studies from Germany, Edward Elgar, Cheltenham, UK and Northampton, Massachusetts. Estache, A. and M.E. Pinglo. 2004. Are Returns to Public-Private Infrastructure Partnerships in Developing Countries Consistent with Risks since the Asian Crisis, Policy Research Working Paper, No. 2970, Washington, D.C.: World Bank. Estache, A. V. Foster and Q. Wodon. 2002. Accounting for Poverty in Infrastructure Reform--Learning from Latin America's Experience, Studies in Development Series, World Bank Institute, Washington, D.C. Estrada, J. 2002. "Risk in emerging markets: the D-CAPM," Emerging Markets Review, 4, pp. 365­379. Estrada, J. 2001. "The Cost of Equity in Emerging Markets: A Downside Risk Approach" (II), Emerging Markets Quarterly, Spring, pp. 63­72. Fischer, R., and Serra, P. 2000. "Regulating the Electricity Sector in Latin America," Economia, 1(1), pp. 155­198. Galetovic, A. and R. Sanhueza. 2002. "Regulación de Servicios Públicos: Hacia donde Debemos Ir," Estudios Públicos, Verano, pp.101­134. How Profitable Are Infrastructure Concessions in Latin America? 57 Gomez-Ibanez, J.A. 2004. Regulation of Private Infrastructure: Monopoly, Contracts and Discretion, Harvard University Press. Guasch, J.L. 2004. Granting and Renegotiating Infrastructure Concessions: Doing it Right, World Bank Institute Development Studies, Washington, D.C. Guasch, J.L., and P. Spiller. 1999. Managing the Regulatory Process: Design, Concepts, Issues, and the Latin America and Caribbean Story, World Bank Latin American and Caribbean Studies, Washington, DC: World Bank. Godfrey, S., and R. Espinosa. 1996. "A Practical Approach to Calculating Costs of Equity for Investment in Emerging Markets," Journal of Applied Corporate Finance, Fall 80­89. Hail, L., and C. Leuz. 2004. International differences in the cost of equity capital: do legal institutions and securities institutions matter?, Working Paper No. 04­06. Wharton Financial Institutions Center. IMF. 2002, Global Development Finance and World Development Indicators, Washington, DC. Lintner, J. 1965. "The valuation of risk assets and the selection of risky investments in stock portfolios and capital budgets," Review of Economics and Statistics, 47, pp. 13­37. Lopez de Silanes, F. and A. Chong. 2003. The Truth about Privatization in Latin America, Yale School of Management Working Papers No. 436, Yale School of Management, Newhaven. Manzetti, L. ed. 2000. "Regulatory Policy in Latin America: Post-Privatization Realities," North-South Press Center, University of Miami, Miami. McKenzie, D. and D. Mookherjee. 2004. "The Distributive Impact of Privatization in Latin America: Evidence from Four Countries," Economia, Spring 2003, 161­218. Nellis, J. 2003. Privatization in Latin America, Working Paper No. 31, Center for Global Development, Washington, D.C. Nellis, J. and N. Birdsall, N. 2004. "Glass Half Full: Assessing the Distributional Impact of Privatization," The Center for Global Development, Washington, D.C. Pereiro, L. 2001. "The Valuation of Closely-Held Companies in Latin America," Emerging Markets Review, 2, pp. 330­370. Sharpe, W.F. 1964. "Capital asset prices: a theory of market equilibrium under condi- tions of risk," Journal of Finance, 19, pp. 425­442. Sharpe, W.F., and G. Cooper. 1972. "Risk-Return Classes of New York Stock Exchange Common Stocks, 1931­1967," Financial Analysts Journal, March/April 1972: 46­54, 81, 101. Ugaz, C. and C. Waddams Price, ed. 2003. "Utility Privatization and Regulation: A Fair Deal for Consumers?", Edward Elgar, Northampton, MAS, USA (with O. Chisari and C. Waddams Price). World Bank. 2003. "Private Participation in Infrastructure: Trend in Developing Countries in 1990­2001." Washington, D.C. 58 References