WPS6816 Policy Research Working Paper 6816 Estimating the Economic Opportunity Cost of Capital for Public Investment Projects An Empirical Analysis of the Mexican Case Andrea Coppola Fernando Fernholz Graham Glenday The World Bank Latin America and the Caribbean Region Economic Policy Unit March 2014 Policy Research Working Paper 6816 Abstract This paper offers an assessment of the methodologies economic opportunity cost of capital is estimated using employed to estimate the economic opportunity cost of the supply price plus externalities method. For the case capital for public sector projects, relying on the Mexican of Mexico, this paper recommends using a combination case for an applied empirical exercise. The traditional of estimation models (both the top-down and bottom- weighted cost of capital (top-down) approach used in up approaches) to check the consistency of results and the estimation of Mexico’s economic opportunity cost re-estimating the economic opportunity cost of capital of capital is reviewed and compared to the supply price every five years to accommodate for macroeconomic and (bottom-up) approach. With respect to previous studies fiscal changes. More broadly, the paper acknowledges the using the top-down approach, this paper explores the complexities involved in the estimation of the economic contribution of domestic savings and expands the analysis opportunity cost of capital for public investment to include a more detailed examination of the available projects and underlines the relevance of additional macroeconomic, labor, financial, and tax information. considerations, such as changes in global economic The re-estimated top-down economic opportunity trends and country risk ratings, tax distortions, financial cost of capital for Mexico comes to 10.4 percent. To sector improvements, the impact of reforms, and data confirm these results and provide additional insights availability. regarding the alternative bottom-up approach, the This paper is a product of the Economic Policy Unit, Latin America and the Caribbean Region. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. he authors may be contacted at acoppola@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Estimating the Economic Opportunity Cost of Capital for Public Investment Projects: An Empirical Analysis of the Mexican Case Andrea Coppola, Fernando Fernholz, and Graham Glenday JEL Classification: H43 Key Words: Economic Opportunity Cost of Capital (EOCK), social discount rate, public investment projects, weighted cost of capital, supply price approach, tax externalities, Mexico. *The authors would like to thank Gloria Grandolini, Humberto Lopez, Auguste Kouame, and Samuel Freije- Rodriguez for their guidance and support. They are also grateful for the comments provided by Ursula Carreño Colorado (SHCP), Anne-Laure Mascle-Allemand (SHCP), and Luis Fernandez Intriago (SHCP); the inputs of World Bank peer reviewers Nataliya Biletska (PRMPS), Jay-Hyung Kim (PRMPS), and David Rosenblatt (DECOS); as well as for the excellent cooperation and courtesy extended by the Mexican authorities during the research phase. The authors may be contacted at acoppola@worldbank.org. 1 Introduction: Background and significance of the economic opportunity cost of capital (EOCK) 1 The official discount rate that informs decision making on investment options—known as the social discount rate or the economic opportunity cost of capital (EOCK)—is critically important to policy makers. The EOCK is an economic price for the valuation of savings and investments and their contribution to economic production in the coming years. Investment projects use a combination of inputs or real resources (such as land, labor, and capital) to produce outputs that society is willing to pay for, whether directly or indirectly. The decision to fund a public project will displace private investments and consumption. Taking this into consideration and using market information, the EOCK is used as a hurdle rate to determine the desirability of implementing projects. During the last decade, the Mexican finance ministry (Secretaría de Hacienda y Crédito Público or SHCP) used a benchmark real discount rate of 12 percent to estimate the net economic benefits of alternative investment proposals (DOF, 2012). This official rate, which was underpinned by the findings of a study conducted by Sergio L. Rodriguez-Medrano (2009), was correspondingly utilized by all public sector agencies whose project proposals are approved by the investment unit—Unidad de Inversiones—of the SHCP. In view of significant macroeconomic developments in recent years (note, for example, that Mexico’s credit rating has improved considerably in recent years, resulting in a lower cost of equity and debt), the Mexican government took the initiative to reassess the estimation of the country’s EOCK. In this paper we analyze Mexico’s EOCK taking into account the recent improvements in the country’s economic and financial structure as well as in the international capital markets. The findings of this study provide the basis for a new recommended discount rate that could be used in the selection of public sector investment projects in Mexico, and suggest an approach with potentially wide applicability beyond this particular case study. In addition to updating and expanding the Rodriguez methodological approach—a version of the weighted cost of capital (top-down) approach (Harberger, 1972b)—through a detailed analysis of the 2003–2011 period, we propose an enhanced methodology to estimate Mexico’s EOCK going forward. With the proposed approach we arrive at two principal recommendations for the Mexican government: (1) a significant downward adjustment of the EOCK—specifically, a single value EOCK that could be reduced to 10 percent, allowing for some flexibility with higher risk investments; and (2) that the EOCK be regularly updated. In high-risk cases such as market-oriented public projects or large commercial public- private partnerships (PPPs), it may be advisable to use a higher discount rate, which could be assessed using the supply price (bottom-up) approach discussed in this study. In the Mexican case, the high-risk distinction could be particularly important for companies like PEMEX, as well as utilities and infrastructure projects that depend on user fees to cover 1 The World Bank report “Mexico: Estimation of the Economic Opportunity Cost of Capital for Public Investment Projects” includes the analysis and the findings of this paper and a more extensive set of annexes providing further background information. 2 investment costs. In the Mexican context, the data shows that the presence of PEMEX—a large public sector corporation that produces output and is also heavily taxed—significantly affects the SHCP’s determination of the appropriate discount rates. Regularly updating the EOCK is important for several reasons, as illustrated in the Mexican case. Over the medium term, Mexico’s EOCK is expected to experience upward pressure, as monetary supply restraints are restored in the international capital markets in the coming years and as the country seeks enhanced revenues to stabilize its fiscal position. However, the EOCK could also experience downward pressure if improved governance and economic reforms further reduce the country risk premiums in the capital markets. In this paper, we offer a preliminary assessment of the impact Mexico’s recently approved fiscal reform is likely to have on the EOCK. Nevertheless, considering the hitherto uncertain effect of broader reform efforts currently underway in Mexico, we recommend reassessing the EOCK in five years. 2 Assessing the EOCK This section describes the weighted cost of capital method used for the assessment of public sector projects and also discusses alternative methods that have been considered in the literature. 2.1 The weighted cost of capital method The weighted cost of capital method is based on the assumption that the discount rate for capital investments should be the economic opportunity cost of funds obtained from the capital markets. This rate—initially proposed by Arnold C. Harberger (1972, 1980, and 1997) and subsequently expanded and improved by other authors—is a weighted average of the marginal productivity of capital in the private sector and the rate of time preference for consumption or the interest rate on savings. The decision to fund a public project will displace private investments and consumption. The weights for the estimation of the EOCK, as explained in more detail below, will come from the expected displacement of investments in one case or the postponed consumption in the other, evaluated at their respective economic prices. A key advantage of the weighted cost method is the use of market information to estimate prices for the marginal gross-of-tax returns for investors as well as net-of-tax savings rates for consumers or suppliers of capital, both domestic and foreign. The basic equations underlying this method are summarized in Box 1. The weighted cost of capital approach offers certain advantages over alternative methods, particularly with regard to clarity and robustness of the results. It is a comprehensive approach that takes into account the impact of financing public projects on both domestic private investment and consumption, while other approaches tend to focus either on one variable or the other. Unlike alternative methodologies, the weighted cost of capital estimation is based on observed market evidence. For instance, this approach relies on domestic prices such that 3 economic values are compatible with the values in financial flows and budget analysis. Furthermore, it favors the comparability of results across sectors in the economy, hence promoting transparency. Based on these characteristics, the weighted cost of capital methodology has been widely favored by governments and multilateral agencies (including the World Bank). As demonstrated by Burgess (2008), the weighted cost of capital approach tends to offer a higher probability of optimal investment choices than other methodologies. For further discussion on the merits of this approach, see Rajaram et al. (2010). Box 1: The weighted cost of capital The economic opportunity cost of capital (EOCK) is the weighted average of the share of displaced investments (ωd ) priced at the gross of tax return (π) and share of induced savings (ωs = 1 − ωd) at the rate of time preference for consumption (r). EOCK = ω dπ + ω s r For several types of investors and savers, the weighted returns can be expressed as follows: n n − ∑η d j ( I j / S )π j ω dπ = ∑ω d j πj = m j =1 n and j =1 ∑ε i =1 i s ( S i / S ) − ∑η ( I j / S ) j =1 d j m m ∑ε i s ( Si / S )ri ω r= s ∑ω r = s i i m i =1 n and ∑ =1 + ∑ = 1 i =1 ∑ε i =1 i s ( Si / S ) − ∑η ( I j / S ) j =1 d j The weight of demand for displaced investments of type j (ωd j ) is the price elasticity of demand d for investments (ηj ) times the investment share over total savings (Ij ⁄S) for each type of investment. This weight is applied to the gross of tax return (πj) of the type j investment. Similarly, for the weight of supply of savings out of postponed consumption from type i (ωs i ) is the price elasticity of supply of savings (εs i ) times the savings share over total savings for each group (i), times the net of tax returns (ri). This method will be shown in more detail in the next sections. 4 2.2 Alternative methods to assess the EOCK The literature offers several alternatives to the weighted cost of capital approach to estimate the EOCK. Below we offer a brief review of the following: (1) the social rate of time preference (SRTP) method; (2) the marginal productivity of capital in the private sector method; and (3) a method that uses an accounting or “sliding” discount rate. The social rate of time preference (SRTP) measures the preference for giving up consumption in favor of savings. A method originally proposed by Marglin (1963), Feldstein (1964), and Dasgupta et al. (1972), the SRTP-based discount rate has become the preferred approach by some European countries (e.g. the United Kingdom) and European multilateral agencies. Often centrally prescribed rather than measured in the capital markets, the SRTP method still generates debate when compared to the weighted cost of capital approach. The latter was the chosen methodology in this paper given its obvious advantages in the Mexican context, where the current discount rate is market-based and estimated using a weighted cost of capital method. The SRTP method could be comparable with the EOCK by valuing investment costs at the shadow price of investment measured in units of consumption. Conceptually, the SRTP must be used in discounting consumption benefits and costs measured in units of foregone consumption. Hence, to make the SRTP comparable to the weighted cost of capital method, investment costs should be valued at the shadow price of investment, which measures the foregone consumption arising from the investment. For a long-lived investment, if investment is financed out of a weighted average of foregone investment and consumption, then the shadow price of investment (S) is the present value of the foregone stream of consumption, or a weighted average of the consumption foregone on investment valued at consumption units (⁄ ) and the share from consumption valued at unity (S = ⁄ + = EOCK/r). Clearly, the shadow price of investment rises with the investment share and the gap between π (cost of foregone investments) and r (SRTP or cost of foregone consumption). 2 Relying on the shadow price of investment when using the SRTP method will yield similar conclusions to those reached when using the EOCK. The SRTP tends to be significantly lower than the rate obtained when using the weighted cost of capital method. In capital market terms, the SRTP can be equated to the rate of time preference for consumption (r), and hence, if the cost of foregone investments (π) exceeds r, then the EOCK exceeds the SRTP. According to Lopez (2008), when taking into consideration the long-term horizons and environmental sustainability of public sector projects, the social discount rate for Mexico should be 3.3 percent. Compared to the weighted average approach, which uses the domestic currency as the numeraire or unit of account, the use of the SRTP implies a loss of transparency and 2 In countries with substantial taxes on capital (corporate, property, and personal income taxes), a large distortion can exist between π and r, such that the shadow price of investment (EOCK/r) can be around 1.5 to 2, making the adjustment to the appraisal significant. 5 simplicity in accounting. The defenders of the SRTP method for discounting justify its use on the basis of normative prescriptions, which, as observed earlier, are not based on market transactions. Another possible approach to assess the discount rate considers the marginal productivity of capital in the private sector. This approach is based on the principle that the government will always seek to maximize the returns to the economy. Under this rationale, all public sector projects would use the rate equal to the marginal productivity of capital in the private sector (Hirshleifer et al., 1960). If private sector returns are higher than the ones generated by the public sector, more funds should then be made available to the private sector to maximize the returns on economic resources. Little and Mirrlees (1969, 1974) developed a cost-benefit method that values costs and benefits in terms of border or world prices in foreign exchange units. 3 The marginal productivity of capital in the private sector method advocates the use of a discount rate based on the cost of foregone investments (or π). This approach to setting the discount rate is appropriate in closed economies with unresponsive private savings. Under these conditions, the weighted average EOCK could be approximated by the cost of foregone investments π as ωd approaches unity. In the context of more open economies with increasingly integrated capital markets, this extreme assumption is no longer appropriate. One possible advantage of this method is the comparability of projects across borders, as they are measured using international monetary units. Some international organizations and UN agencies have used this method in the past. The discount rate based on the marginal productivity of capital in the private sector approach is higher than the weighted EOCK. An unnecessarily higher discount rate for public sector projects, as π is by definition higher than the weighted EOCK, can result in the elimination of worthwhile investments in the public sector, hence leading to real economic losses. Finally, the accounting or sliding discount rate employs a rationing approach that allows public sector projects to be funded, in descending order, as long as there are available resources in the public sector budget. Originally proposed by Little and Mirrlees (1969, 1974) and by Squire and van der Tak (1975), this method recommends the use of an accounting discount rate, which is compared to the marginal returns from public sector projects, within the available budget constraints for the public sector. The accounting discount rate employs a rationing device to fund public sector projects in descending order, on the condition that the marginal project (e.g. the one with the lowest economic internal rate of return or lowest net present value) is accepted subject to available resources in the public sector budget. The accounting discount rate is adjusted upwards or downwards depending on the proposed projects, their returns using this discount rate, and the available budget. 3 An attractive approach to dealing with the value of traded goods, the Little and Mirrlees methodology requires all non-traded goods and labor to be converted to their foreign exchange equivalent and all distortions are expressed in foreign exchange units. 6 From an economic standpoint, the accounting discount rate approach is not optimal. An accepted economic discount rate (as a true opportunity cost to the economy) ensures that only projects deemed to contribute to the economy at that rate should be adopted. Excess funds, if any, should then be used to lower the public sector debt and/or fund private sector projects through the private capital markets. By contrast, an accounting discount rate only selects projects appropriately if the selected discount rate is equal to or higher than the EOCK. When enough funds are available, the sliding rate advocates the approval of public sector projects that yield lower returns than those approved with the weighted EOCK, ultimately generating economic losses. 2.3 Suggested improvements to the weighted cost of capital method Seeking to strengthen the traditional Harberger approach, this study takes into account recent advancements in the weighted cost method and applies them to the Mexican case. First, we employ in each step of the methodology used by the Mexican government the latest innovations in the estimation of the EOCK components for the period 2003–2011. Second, our analysis provides further refinements to the weighted cost of capital approach, as well as examples to demonstrate the advantages that these improvements offer for decision-making purposes. Empirical evidence showed that the traditional weighted average approach might present some challenges. Two significant problems in the traditional approach to the weighted average EOCK stand out: (1) differentials in the mobilization or intermediation costs of financing for certain types of projects that raise the real costs of supplying capital funds (for example, for micro financed projects 4); and (2) the differential costs of systematic risk (both country risk and market risk) that need to be recognized in different types of investment. To illustrate, consider that a public infrastructure project that depends on government revenue for financing ultimately faces lower risk-related costs than a project that is financed by market-determined revenues and, accordingly, bears the full or partial costs of market risk. Government decision makers face the challenge of appraising publicly financed projects as well as public sector market-oriented and market-financed projects, such as those undertaken by state-owned commercial enterprises, regulated utilities, and user fee–financed public private partnerships. Nonetheless, the EOCK can be adjusted to address these challenges. For example, the EOCK can be adjusted down to a “risk-free” and “intermediation-cost-free” estimate, which may be appropriate for purely publicly financed projects. The EOCK, however, requires further adjustment in cases where significant mobilization of funding and systematic market risk costs arise. The alternative and more direct approach employed in this study begins by estimating the minimum financial supply price of capital to the investment project, then factoring in the economic externalities resulting from the use of capital funds. The externalities are a consequence of tax and other distortions that arise as the capital markets adjust to the use of capital funds through reduced investment and expanded domestic and foreign savings. This 4 The cost of supply and administration of micro financed projects is much higher, given that the costs per unit of capital tend to be much higher. 7 approach can offer greater precision and practicality in defining country, market, and project risk considerations, as discussed in greater detail below. The analysis also demonstrates that the enhanced methodology is consistent with the traditional EOCK approach (Glenday, 2010). The supply price approach to the EOCK is based on the estimation of the minimum supply price of capital and on the estimation of the externalities arising in the economy from the use of capital. The minimum supply price in capital markets is the minimum cost of capital required by the financiers of the investment project in a competitive market. This minimum required rate of return would adjust for all the differential costs of risk and the capital mobilization costs that would either make the financiers indifferent between funding the project and allocating the funds to alternative investments, or withdrawing their savings from the capital market and channeling the resources elsewhere. The minimum supply price of capital in the market implicitly captures all the compensating differentials for the specific project (due to location, project, and country risks) without an explicit need to analyze them and account for them. Given the general fungibility of funds over the long run, the externality 5 associated with the long-term investment of capital in an economy can be taken to be a function of the characteristics of the economy (such as the structure of taxation) rather than of the project itself, and hence, it is taken as a single estimate characteristic of the economy. The externality per unit of capital invested over the long run can be thought of as a national parameter. This approach to estimating the EOCK captures the specific characteristics of the project being financed, as well as the general externalities arising from the long-run use of capital funds by the project in a particular country. Following the standard model of economic pricing, this method takes the financial price of a commodity and adjusts for the associated economic externality. For example, when estimating the economic opportunity cost of labor, the minimum market supply price plus externalities approach is applicable, as labor markets experience significant compensating differentials to adjust supply prices for job and location conditions. The minimum supply price based EOCK, which is consistent with the traditional approach (see Box 2), allows to differentiate between pure public sector–financed projects and self- financing projects. Self-financing projects span different sectors of the economy and are characterized by added market risk costs to the financiers (market risk effects). The differentiation between publicly-financed and self-financed projects allows for explicit recognition of the externalities arising from the use of capital funds. As mentioned above, the EOCK consists of the sum of the financial prices paid by the project plus externalities per unit of capital. This is comparable to the economic price of foreign exchange in the economy, which is the market exchange rate plus externalities that result from taxes and subsidies in the relevant markets arising from the use of foreign exchange. 5 This refers to the additional costs to society of investing in a project that are external to the calculations of the investors. 8 Box 2: Consistency between the minimum supply price of capital approach and the traditional Harberger approach Following Glenday (2010), the consistency between the standard EOCK model (Box 1) and the supply price approach can be readily shown by recognizing that the economic value of the marginal product of a unit of capital () is also equal to the weighted average cost of capital ( ) plus the tax externalities (and any other externalities such as monopoly rents) per unit of capital ( ), or = + . Similarly, on the supply or savings side (r), the cost of added savings (including risk and intermediation costs) is the weighted average financial return received by savers—( ) reduced by any taxes on the return to savings per unit of capital ( ) plus any other external costs incurred by the economy. Hence, the EOCK can be expressed as: = + = ( + ) + ( − ) = + { ( ) + (− )} = supply price of capital + economic externality per unit of capital 3 Empirical estimations: Application to Mexico This section presents and discusses the application of the weighted cost of capital method to the case of Mexico. Section 3.1 considers the traditional approach, starting with an assessment of the applied methodology (Rodriguez, 2009), which covers the 1970–2006 period. The following section covers the recalculation of the 2009 study results, updating them to 2012. Section 3.3 offers a more detailed analysis of this top-down methodology to make explicit the necessary adjustments, based on market and economic information, to obtain rates of return and ultimately the EOCK. 3.1 Estimation of the traditional weighted average EOCK 3.1.1 The Rodriguez methodology and estimates The Rodriguez (2009) estimation of the EOCK for Mexico focuses on the cost associated with the displacement of private investments and the role of foreign savings, while the cost of displacing domestic consumption is considered negligible on the basis of an assumed low or zero price elasticity of supply of savings. Relying on the weighted cost of capital approach, the Rodriguez study focuses solely on the cost associated with the displacement of private investment (gross of taxes) and the expansion of the foreign supply of savings. Displaced consumption is excluded on the basis of the low price elasticity of savings, which, according to Rodriguez, reduces its influence on the EOCK. 9 The 2009 Rodriguez estimation of the return to capital in the private sector is based on a strong assumption on the treatment of labor income that affects the final estimation of the EOCK. The return to capital in the private sector is estimated using national account statistics and adjusted on the basis of a strong single assumption that labor income should be treated as part of the gross operational surplus. This adjustment to the capital income reflects the share of labor income earned in unincorporated activities that are reported as capital income in the national accounts. The study assumes that out of the gross operating capital profits for the private sector recorded in the national accounts, 50 percent can be attributed to labor. To reflect on this, consider that if the initial allocation of gross value added for labor is 33 percent and for capital it is 67 percent, by assuming that 50 percent of capital income is in fact labor income, the result flips to about 67 percent for labor and 33 percent for capital. It is this assumption that leads to the 2009 study results and the recommended EOCK of 12 percent for Mexico. 3.1.2 Updated analysis and commentary With the information provided in the Rodriguez (2009) study, we proceeded to recalculate a series of estimated values for the capital stocks, income to private capital, and rates of return to producible capital. We checked for consistency of the Rodriguez data and key results with more recent national accounts data, although the full data set on investment flows and capital stocks, as well as the yearly values for the interest rates used for the original study, were not available. The 2009 study provides data on capital stocks for the initial (1970) and final (2006) years covered. Based on the initial ratios of value added (VA) for the public and private sectors and rates of return to each sector, we estimated the ratios of reproducible 6 private sector capital stocks, public sector capital stocks, and capital to GDP (K/Y), which are similar to the ratios estimated in the Penn World Table, version 8 (July 2013) 7 and other studies (Hofman, 2000; Loría and de Jesús, 2007). These estimations allowed us to recreate the estimated rates of return to private capital using the simplified weighted cost of capital methodology chosen by Rodriguez (2009). To arrive at the estimated EOCK rate for the entire 1970–2006 period and extend it to 2012 using more recent national accounts data, these results have to be combined with the marginal cost of foreign savings or borrowing (MCFS). For this initial assessment, the MCFS is taken from the Rodriguez (2009) study as a constant variable. Because the price elasticity of domestic savings is assumed to be zero in the original document, there is no data for rates of return on domestic savings. The EOCK estimations for Mexico show a downward trend, mostly driven by the evolution of the rates of return in the private sector. Confirming the results of the Rodriguez (2009) study, the recalculation shows a downward trend for Mexico’s EOCK (the upper trend line in 6 Reproducible here means subject to increase by investment less depreciation. 7 According to the website description, the Penn World Table “provides purchasing power parity and national income accounts converted to international prices for 189 countries/territories for some or all of the years 1950– 2010.” For more information, see https://pwt.sas.upenn.edu/php_site/pwt_index.php. 10 blue shown in Figure 1), which is driven by the downward trending estimated rates of return in the private sector, with a constant marginal impact by the MCFS. This responds to the relatively high elasticity of domestic investment to the interest rate (assumed in the 2009 study as −1.8), which is unusually high, and a lower price elasticity for the supply of foreign savings (1.0) as well as the initial share of foreign savings in total savings for Mexico. As mentioned above, the Rodriguez (2009) study assumes zero price elasticity for domestic savings. Figure 1 EOCK trends for Mexico: An update of the 2009 Rodriguez calculations to 2012 16.0% 14.0% 12.0% EOCK 10.0% 8.0% EOCK (Eds 0.40) 6.0% 4.0% 2.0% 0.0% Source: Authors’ calculations using the methodology by Rodriguez (2009), updated to 2012. After replicating the methodology proposed by Rodriguez (2009), we took into account the estimated contribution of domestic savings, considering sensitivity to alternative price elasticities of supply of savings based on empirical evidence from different sources. Jenkins, Kuo, and Harberger (2010) use a price elasticity of newly stimulated domestic savings of 0.4 and a price elasticity of foreign savings of 3.0 for the re-estimation of the EOCK for Canada. Jenkins and El-Hifnawi (1993) use a price elasticity of savings 0.50 for the estimation of the EOCK for Indonesia. The rates of return on savings were adopted as a constant variable at 5.4 percent in real terms, which Rodriguez (2009) uses as the starting interest rate to estimate the MCFS. The second trend line in Figure 1 (in red) reflects the impact of domestic savings with a price elasticity of supply of domestic savings (Eds) of 0.40. As we can observe from the results, when extending the estimation from 2006 to 2012, we arrive at a final EOCK of about 11 percent without contribution from domestic savings, while taking into consideration the impact 11 of domestic savings we reach an estimated value of 10 percent. To reduce the impact of yearly fluctuations in the data, it would be preferable to use an average for the 2003–2011 period, which comes to 11.8 percent for the series without domestic savings and 10.5 percent when taking displaced domestic consumption—and thus domestic savings—into consideration. We use these initial results, which confirm the EOCK results obtained in the Rodriguez (2009) study, as a starting point for a more detailed analysis in the next section, including reasonable assumptions about the impact of domestic savings. 3.1.3 Methodology and data sources In this section we employ the traditional weighted cost of capital approach, but with some differences in the calculation of net income to capital and in the considerations on the applicable capital stocks. The starting point is the national accounts data from a top-down perspective. From GDP, we deduct indirect taxes and depreciation to arrive at net domestic product at basic (or factor) prices. This, in turn, is divided between income to labor and income to capital (net operating surplus). The data sources employed in the estimation of the EOCK are critically important. In this analysis we rely on several key sources. For the national accounts data, we use figures from the SHCP, the national statistics institute or Instituto Nacional de Estadística y Geografía (INEGI), the Central Bank or Banco de México, the Organization for Economic Cooperation and Development (OECD), the International Monetary Fund (IMF), and the World Bank. For data on labor employment, compensation, and mixed income, we rely on figures from the SHCP, United Nations (UN) Statistics, the International Labor Organization (ILO), and INEGI. For capital market information, we use data from the SHCP, the Banco de México, the UN, and the IMF. Finally, for information on taxes and different sectors, we use OECD and SHCP data. In the case of Mexico, in addition to consistent sets of data on the national accounts, a selected set of statistics on labor, wages, mixed income, as well as the different forms of taxation and the corresponding sector of origin are critical to further strengthen the estimation of the EOCK. In this analysis, the initial data is adjusted for the estimates of labor income contained in the capital income accounts. According to several authors such as Caselli and Feyrer (2007), Karabarbounis and Neiman (2012), and Guerriero (2012), the labor share in national domestic product has to be adjusted for the labor share of mixed income (self-employed, non-paid family labor, unincorporated businesses) reported as capital income. This adjustment increases the labor income share and thus reduces the income accruing to capital. In addition, we deduct from the income accruing to capital the portion that can be considered income accruing to land. We then further adjust the income to capital by adding the proportion of indirect taxes that can be attributed to capital income. The data reported in the national accounts requires an adjustment to ensure proper representation of the share of net national income accruing to labor. In the computation of the labor share, which is a ratio to national income, the labor income component needs to be derived and added to the compensation of employees as reported in the national income accounts. For the computation of its denominator—the income aggregate—a few adjustments are 12 required. First, taxes on production and imports (minus subsidies) are removed from the gross value added at basic prices. The calculations are based on information from the UN System of National Accounts using the main methodologies proposed in the literature on the labor share of income (Bernanke & Gürkaynak, 2001; Diwan, 2001; Harrison, 2002; Jayadev, 2007). We adjust the labor share following the methodology proposed by Guerriero (2012). Based on the six alternative adjustments to the labor share proposed in the Guerriero (2012) study, this paper focuses in particular on the definitions coded by the author as LS4 and LS5: • LS4 = [compensation of employees]/[value added (– indirect taxes – depreciation of fixed capital) – mixed income]. This ratio assumes that the share of labor income is the same in the corporate and non-corporate sectors. Typically, the non-corporate sector is expected to be more labor-intensive than the corporate sector, so that this adjustment will tend to understate the labor share adjustment. Note that in Appendix E of the Guerriero study, the labor share for Mexico under LS4 is 0.48, while under definition LS5 it is 0.60 (Guerriero 2012, 32–6). • LS5 = [compensation of employees * total workforce/number of employees]/[value added (– indirect taxes – depreciation of fixed capital)]. This ratio assumes that the wage rates in the unincorporated sector are the same as in the corporate sector, when typically they are expected to be lower in the unincorporated sector. Therefore, this ratio will tend to marginally overstate the labor share adjustment. The actual adjustment is expected to be between LS4 and LS5. Hence, we use an average of the two methods to adjust for the labor share in the unincorporated sector and the self-employed sector for the estimations of labor income in Mexico. Relying on various national statistics sources for Mexico (ILO, UN, SHCP) on labor markets and mixed income, an average estimation of ratios LS4 and LS5 of the labor share of value added was selected from the different proposed and possible ratios. The fact that they reflect a middle range of values allows us to re-estimate the income accruing to capital (on the basis of this methodology and data). The standard definition of capital income implies measuring it net of capital consumption. Therefore, where data are available, consumption of fixed capital has been subtracted from the measure of value added, obtaining a net income measure. After the adjustments described above, the income accruing to capital is divided by the stock of private capital, including the value of inventory stocks and land. The values for capital stocks are calculated using the perpetual inventory method (results that are calculated by dividing investment flows by the sum of depreciation and growth of capital stocks in percentage terms). The resulting series is then compared to GDP and the ratios (K/Y) compared to other Mexico-specific and broader international studies to assess its acceptability. This provides the first set of empirical results under different assumptions for labor’s share and income accruing to capital and the corresponding returns to private capital in Mexico. Because PEMEX affects the estimation of the rates of return in the Mexican economy, the EOCK calculation is likewise affected. Further analysis on the rates of return to capital in 13 Mexico should also consider the impact that large companies can have on the results. If we separate the income accruing to PEMEX and to the rest of the private sector, then divide these incomes by the respective stocks of capital plus inventories, we obtain separate rates of return for each subsector. This process allows us to assess the contribution of PEMEX—a large state- owned enterprise that has very high returns and is also heavily taxed by the state—to the rates of return in the economy. These results can have an impact on the calculations of the EOCK when using the traditional method (top-down approach using national accounts data under different scenarios). In addition, the analysis in separate accounts is useful for the discussion of the bottom-up approach to the EOCK calculation for public sector projects in Mexico. 3.2 Estimation of the gross rates of return for invested private sector capital, 2003– 2011 Based on the discussion in section 3.1, this section focuses on the estimation of the EOCK and the necessary intermediate calculations for the 2003–2011 period, using SHCP data for those years. The base year for the nominal and real values used in the calculations is 2003. 3.2.1 Rates of return to private sector capital The national accounts data allow us to differentiate our assessment of income accruing to capital versus income accruing to labor. We obtain GDP at factor costs from the national accounts data by taking GDP at market prices and deducting taxes on output. We then obtain the amounts accruing to labor, capital, and indirect taxes. The data shows that the income accruing to capital in the national accounts is, on average, more than double the income accruing to labor (see Table 1). Depreciation is reported as part of the gross operating surplus. Table 1 National accounts: Shares of income to factors of production, 2003–2011 (million pesos) Gross Less: Gross value Indirect operating GDP Indirect added (GVA) Capital Year Labor income taxes paid Depreciation surplus (nominal) taxes on at factor income out of GVA (= GVA less output costs labor income) 2003 7,555,803 393,030 7,162,773 2,370,474 304,878 4,487,421 683,354 4,792,299 2004 8,574,823 390,210 8,171,095 2,540,339 399,461 5,231,295 759,183 5,630,756 2005 9,251,737 395,564 8,825,085 2,729,235 519,324 5,576,526 804,314 6,095,849 2006 10,379,091 400,971 9,943,093 2,955,305 579,641 6,408,148 883,697 6,987,788 2007 11,320,836 436,368 10,854,384 3,172,663 614,088 7,067,633 972,737 7,681,721 2008 12,181,256 315,664 11,837,772 3,409,255 971,440 7,457,077 1,095,587 8,428,517 2009 11,937,250 499,027 11,394,220 3,474,162 580,608 7,339,451 1,264,601 7,920,058 2010 13,071,597 543,592 12,485,511 3,678,700 722,627 8,084,184 1,306,076 8,806,811 2011 14,395,833 507,736 13,843,758 3,955,335 932,639 8,955,784 1,382,400 9,888,423 Source: National accounts for Mexico, SHCP (2013) and INEGI. 14 From the national accounts data we see that the share of labor out of GDP at factor costs is 30 percent, while out of net domestic product at factor costs it is 33 percent on average for the period. Correspondingly, the income accruing to capital, as per the national accounts, comes out to 67 percent of net domestic product as an average for the period. In this case, labor income only includes labor employed by corporate entities, not the self-employed or those working for unincorporated employers (see Table 2.1). Table 2.1 Mexico: Labor income shares, 2003–2011 Net domestic Net domestic Gross value product at Labor income product at Labor income Mixed added at Labor income factor costs over net factor costs over [NDP at income Year basic prices over GDP at (GVA – domestic less mixed factor prices [million [million factor costs depreciation) product at income less mixed pesos] pesos] [million factor costs [million income] pesos] pesos] 2003 7,162,773 33% 6,479,420 37% 1,359,937 5,119,483 46% 2004 8,171,095 31% 7,411,912 34% 1,497,569 5,914,343 43% 2005 8,825,085 31% 8,020,770 34% 1,875,614 6,145,156 44% 2006 9,943,093 30% 9,059,396 33% 2,049,152 7,010,244 42% 2007 10,854,384 29% 9,881,647 32% 2,260,332 7,621,315 42% 2008 11,837,772 29% 10,742,185 32% 2,505,541 8,236,644 41% 2009 11,394,220 30% 10,129,619 34% 2,384,723 7,744,896 45% 2010 12,485,511 29% 11,179,435 33% 2,674,661 8,504,774 43% 2011 13,843,758 29% 12,461,358 32% 2,945,621 9,515,737 42% Source: SHCP; for mixed income: UN and SHCP. Once we exclude mixed income earned in self-employed and unincorporated activities, the labor income out of net value added at factor prices increases by about 10 percent. If the labor income contained in mixed income is assumed to be the same share as in the corporate sector, then the labor income in the corporate sector can be adjusted to include this same share of mixed income. This is shown in Table 2.2 for method LS4, such that the adjusted labor share averaged 43 percent of the net domestic product (NDP) at factor costs over the period. This is expected to underestimate the labor share of the unincorporated sector, so the alternative estimate under method LS5 assumes that the wage rates in the unincorporated sector are the same as in the corporate sector. Table 2.2 shows that this raises the adjusted labor income share to an average of 51 percent over the period. Given that wages in the unincorporated sector are expected to be lower on average than in the corporate sector, using the average of these two estimates results in an average labor share of 47 percent (last column in Table 2.2). This adjusted labor income is used in the subsequent estimates. 15 Table 2.2 Labor income adjusted for labor income share of mixed income, 2003–2011 Average Labor income Adjusted labor Labor income Adjusted labor Average labor adjusted labor adjusted for income (LS4) adjusted for income (LS5) income adjusted Year income over mixed income over NDP at mixed income over NDP at for mixed NDP at factor (LS4 method) factor costs (LS5 method) factor costs income costs 2003 3,000,166 46.3% 3,751,545 57.9% 3,375,856 52.1% 2004 3,183,578 43.0% 4,014,771 54.2% 3,599,175 48.6% 2005 3,562,248 44.4% 4,244,344 52.9% 3,903,296 48.7% 2006 3,819,165 42.2% 4,519,693 49.9% 4,169,429 46.0% 2007 4,113,612 41.6% 4,843,730 49.0% 4,478,671 45.3% 2008 4,446,331 41.4% 5,173,867 48.2% 4,810,099 44.8% 2009 4,543,887 44.9% 5,274,945 52.1% 4,909,416 48.5% 2010 4,835,612 43.3% 5,299,665 47.4% 5,067,639 45.3% 2011 5,179,720 41.6% 5,992,660 48.1% 5,586,190 44.8% Source: SHCP; for mixed income: UN and SHCP. 3.2.2 Capital stocks The perpetual inventory method is used to estimate the stocks of capital for the public and private sectors. We also use disaggregated series of investments into private and public stocks of capital divided into construction and equipment. Using standard procedures, we adopted medium levels of depreciation, 2.5 percent per year for construction and 6 percent for equipment. As is the case in many studies, the sensitivity of the ratios of stocks of capital to GDP or the reported rates of return to private capital to different plausible depreciation rates did not have a significant impact, since the depreciation rates enter both the numerator and the denominator of the ratios. The total capital to GDP (or K/Y), which averaged 2.9 over the 2003–2011 period, is in line with many other studies in Mexico (Cervini, 2004). The ratio of private capital over GDP (Kpri/Y) averaged 2.23 over the same period. 3.2.3 Land and inventory stocks The estimates of capital income and returns are further adjusted for the value of land in capital stocks. The value of land calculations are drawn from studies that estimate the composition of assets in a range of countries, including Mexico (see Caselli and Feyrer, 2007; World Bank, 2006). Based on these studies, an estimate of 18 percent is used here as the share of the value of land out of the value of reproducible capital plus land. The 18 percent represents the share of urban land over the total capital stocks. This is low compared to the estimate for the U.S. land share of total capital value, which comes to 20–26 percent over the 1960–2003 period (OMB, 2005). According to Caselli and Feyrer (2007), for world average parameters, the estimates of the share of reproducible capital on the total stock is about 75 percent—in other words, the proportion of land and land-related resources to reproducible capital stocks is close to 25 percent. In this study we adopt a conservative estimate. 16 The land share value in capital stocks gives us a ratio of private capital to GDP, including land, of 2.72 for the 2003–2011 period, from a value of 2.23 without land. Adding the ratio of stock of inventories to GDP, which was estimated with long-term national accounts series at an average of 1.47 over the 2003–2011 period, the private stock of capital, land, and inventories in relation to GDP (the ratio) comes to 4.19 for that same period. The stocks of inventories were constructed using long-term series data for changes in inventory for Mexico. The rates of investment on inventories reported for Mexico are comparatively high in the international context, averaging 4.2 percent of GDP over the 2003–2011 period. Further research is advisable to confirm and understand why Mexico has such a relatively high level of investment in increased inventories. As can be observed in the last column of Table 3, the ratio of capital, plus land and inventories, is quite stable for the period from 2003 to 2011. Table 3 Ratios of reproducible capital, land, and inventories to GDP at market prices, 2003–2011 Private capital + Total capital over Private capital over Private capital + Inventories over Year land + inventories GDP GDP land over GDP GDP over GDP 2003 2.75 2.14 2.61 1.59 4.20 2004 2.75 2.13 2.60 1.56 4.16 2005 2.77 2.15 2.62 1.53 4.15 2006 2.76 2.14 2.61 1.47 4.07 2007 2.79 2.16 2.64 1.42 4.06 2008 2.89 2.22 2.71 1.41 4.12 2009 3.18 2.43 2.97 1.49 4.46 2010 3.11 2.37 2.89 1.41 4.30 2011 3.10 2.34 2.86 1.34 4.20 Source: SHCP; World Bank; UN. To calculate the rate of return on private capital stock, the first step is to adjust the net income earned by capital in the economy for the labor income adjusted for the share of self-employed labor income in mixed income. Net income earned by private capital is the NDP at factor costs (GVA – depreciation) less the average adjusted labor income. Over the 2003–2011 period, this averaged 45.8 percent of GDP (see Table 4), or 52.9 percent of NDP. 8 Second, given the estimated total private capital (reproducible capital, land, and inventories), the rate of return on private capital averaged 10.9 percent over the period. Third, this rate of return includes all corporate income taxes, property taxes, and other indirect taxes paid out of value added, but it excludes any net indirect taxes collected in the product markets on the output of capital. When the share of net indirect taxes in the product markets is allocated to the value added earned by capital, the gross of indirect tax return to capital rises by an average of 0.6 percent, to a gross 8 NDP is about 86.5 percent of GDP for the 2003–2011 period. 17 return to capital (π in the weighted average EOCK discussed above) of an average of 11.6 percent over the 2003–2011 period. 9 Table 4 Net capital income and gross rate of return to private capital (reproducible capital, land, and inventories), 2003–2011 Average labor Net capital Private capital Indirect taxes share adjusted income (NDP Net capital Return to Gross return + land + over private Year for mixed − adjusted income over private to private inventories capital + land income labor income) GDP capital capital (π) over GDP + inventories [million pesos] [million pesos] 2003 3,375,856 3,103,564 41.1% 4.20 9.8% 0.7% 10.5% 2004 3,599,175 3,812,738 44.5% 4.16 10.7% 0.6% 11.3% 2005 3,903,296 4,117,474 44.5% 4.15 10.7% 0.6% 11.3% 2006 4,169,429 4,889,967 47.1% 4.07 11.6% 0.6% 12.2% 2007 4,478,671 5,402,976 47.7% 4.06 11.8% 0.6% 12.4% 2008 4,810,099 5,932,086 48.7% 4.12 11.8% 0.7% 12.5% 2009 4,909,416 5,220,203 43.7% 4.46 9.8% 0.6% 10.4% 2010 5,067,639 6,111,796 46.8% 4.30 10.9% 0.7% 11.5% 2011 5,586,190 6,875,168 47.8% 4.20 11.4% 0.7% 12.1% Average 10.9% 11.6% Source: SHCP; for mixed income: UN and SHCP. 3.2.4 Estimation of the net rates of return on domestic and foreign savings: Analysis of returns in stock and security markets, pension funds, and banking sector reports To assess the rates of return on savings, we look at real rates of return on government securities, private sector instruments, and pension funds (see Table 5). There are an increasing number of instruments and institutions in the capital markets in Mexico. To construct a representative series of returns for domestic savers, we estimate the real returns obtained by pension funds in Mexico, as well as the equity returns from the stock market index (MXXD) reported by the stock exchange. We also look at returns on interest rates on long-term government bonds and debt in real terms per year. We use yearly average values for equity returns. We then assess the information from the stock exchange using the Erb, Harvey, and Viskanta (1995, 1996) methodology (see Annex II). Using different reports from Banamex and the SHCP, we used equity to debt ratios of 40/60 to weigh 10 the returns, reflecting the market 9 The estimate of the gross return to private capital could be further refined by: (1) adding in the imputed rent on private owner–occupied housing and (2) adjusting the income and capital stocks for the investments in intangible assets. These two adjustments would increase the return on capital, but would be offset by including the full value of private working capital, namely, the cash balances held for transaction purposes and net receivables. 10 The share allocated to debt is taken to include savers’ investments in housing. 18 conditions on average. 11 For consistency, we use the gross returns to capital information (Table 4) in the last column of Table 5 to report the gross returns to saving, by deducting from gross of tax returns, taxes per unit of capital, as well as the value added of financial services divided by the capital stocks plus inventories and land. We can observe that, although there is some small difference in the trends, the weighted average real returns to different instruments and the top- down net returns (gross returns less taxes and capital intermediation) give very similar averages. The top-down calculation is used for the rates of return on savings and the average estimated return on different instruments as the basis to estimate the marginal cost of foreign savings. Table 5 Rates of return to pension funds, private sector, and public sector bonds, 2003–2011 IFS Banxico Top-down gross Siefores Traded World Bank Weighted government corporate returns to pension funds equity external debt average* Year securities bonds (real capital less taxes (real rates of (real rates of (real interest (real rates of (real interest interest and financial return) return) rates) return) rates) rates) intermediation 2003 13.5% 15.8% 5.0% 3.4% 5.9% 9.2% 7.9% 2004 11.6% 14.3% 5.1% 4.9% 6.1% 8.9% 8.5% 2005 10.8% 13.5% 6.4% 8.0% 6.5% 9.6% 8.1% 2006 11.4% 13.0% 4.6% 6.9% 7.0% 8.9% 8.7% 2007 10.4% 11.7% 4.2% 6.1% 6.9% 8.1% 8.9% 2008 7.6% 11.2% 2.3% 4.0% 5.3% 6.8% 8.3% 2009 6.6% 12.6% 5.1% 3.4% 4.8% 7.7% 7.2% 2010 8.8% 12.3% 3.5% 4.6% 4.4% 7.4% 7.9% 2011 8.7% 12.2% 4.3% 4.6% 4.7% 7.6% 9.4% Average 9.9% 13.0% 4.5% 5.1% 5.7% 8.2% 8.3% Source: INEGI; Banco de México; IFS (IMF); Global Finance Statistics; World Bank External Debt Data. * Weighted average return based on 40 percent equity and 60 percent debt at average government, corporate, and external interest rate. 3.3 Results of the EOCK estimation for Mexico for the 2003–2011 period This section shows estimations obtained for Mexico’s EOCK. The shares of investment for the private sector, as well as domestic and foreign savings data, have been incorporated in percentage terms from the national accounts. Based on several reference studies, we also adopt mid-range values for the price elasticities of demand for investments (η = −1.0), for supply of domestic savings (Eds = 0.30), and for supply of foreign savings 12 (Efs = 3.0). 13 In the 11 Note that a large share of private savings is in residential housing that yields a real financial rate of return similar to debt interest rates. 12 Foreign savings are defined as the amount of net inflows of foreign capital funds (debt, portfolio investments, direct investment, etc.) into Mexico reduced by the amount of net outflow of investment by Mexicans outside of Mexico and by the net additions to foreign reserves. 19 calculation we include the gross of tax returns to private capital and the rates of return to domestic savings (rs) and foreign savings (rf) (see Table 6.1). Despite foreign savings forming a relatively small share of total savings, they have a high price elasticity, and hence, form a relatively larger share of capital funds supporting marginal investments. The shares of investments and domestic and foreign savings over GDP in percentage terms have been calculated from the national accounts 14. Table 6.1 Investment, savings, price elasticities, and weights for the EOCK estimation, 2003–2011 Weights for the computation of the EOCK Weights (f1–f3) Private Elasticities Inv. DS FS GFKF + inventory Domestic Foreign Domestic Foreign investment/ savings/ savings/ η Eds Efs Investment savings savings GDP GDP GDP Investment Dom. sav. For. sav. f1 f2 f3 2003 19.1% 22.9% 1.1% −1 0.3 3 64.9% 23.4% 11.7% 2004 20.5% 24.9% 0.9% −1 0.3 3 66.9% 24.3% 8.7% 2005 19.8% 24.4% 1.0% −1 0.3 3 65.6% 24.2% 10.2% 2006 21.8% 26.2% 0.8% −1 0.3 3 68.2% 24.5% 7.2% 2007 21.9% 26.5% 1.4% −1 0.3 3 64.4% 23.4% 12.2% 2008 21.3% 26.9% 1.8% −1 0.3 3 61.4% 23.3% 15.3% 2009 17.7% 23.8% 0.8% −1 0.3 3 64.6% 26.1% 9.3% 2010 18.0% 23.9% 0.2% −1 0.3 3 69.5% 27.6% 2.8% 2011 19.8% 25.1% 0.9% −1 0.3 3 66.3% 25.1% 8.6% Average 20.0% 24.9% 1.0% 65.8% 24.7% 9.6% Source: Authors’ calculations using SHCP data. In addition, for the calculation of the EOCK below, we report the estimated rates on investment, the net of tax real rates on domestic savings, as well as the marginal opportunity cost of foreign savings. For domestic savings, the economic opportunity cost of savings (rs) is taken as the return to saving reduced by the taxes on the income earned on saving at an effective tax rate of 10 percent. For the marginal opportunity cost of foreign savings in Table 6.2, we use the formula: MCFS = rf * (1−ts) * (1+k/Efs), where rf is the weighted average interest rate on savings in real terms (from Table 5); ts is the tax rate on savings; k is the share of foreign capital that is responsive to changes in interest rates; and Efs is the price elasticity of supply of foreign savings. We adopt the following values for our estimation: ts = 10 percent; k = 0.90; and Efs = 3.0. 13 The authors reviewed the literature on estimates of the price elasticities of investments and savings used in the analysis of the EOCK for other countries. These are documented in the World Bank report “Mexico: Estimation of the Economic Opportunity Cost of Capital for Public Investment Projects.” 14 Data on Mexican national accounts were kindly provided by SHCP. 20 Table 6.2 Rates of return on investments and domestic and foreign savings, 2003–2011 EOCK input values Domestic investment Domestic savings rate Foreign savings marginal rates of return (net) cost π rs MCFS 2003 10.5% 7.1% 10.7% 2004 11.3% 7.7% 10.5% 2005 11.3% 7.3% 11.2% 2006 12.2% 7.9% 10.4% 2007 12.4% 8.0% 9.5% 2008 12.5% 7.5% 8.0% 2009 10.4% 6.5% 9.0% 2010 11.5% 7.1% 8.7% 2011 12.1% 8.4% 8.9% Average 11.6% 7.5% 9.7% Source: Authors’ calculations using SHCP, INEGI, BAMEX, IFS (IMF), Global Finance Statistics (IMF), and WDI data. Note: For domestic and foreign savings, effective tax rates of 10 percent have been assumed. 3.3.1 Results of the estimation The estimation of the EOCK for Mexico, which comes out to 10.4 percent (Table 7), is obtained by applying the weights to the relevant returns. By applying the average weights to domestic investment (f1), domestic savings (f2), and foreign savings (f3) that result from the calculation presented in the table, as well as the rates of return reported earlier for gross of tax returns to domestic investment (π), net of tax returns to domestic savings (rs), and marginal cost of foreign savings (MCFS), we can estimate the EOCK for the 2003–2011 period. Table 7 Top-down approach for the EOCK, 2003–2011 Domestic investment Domestic savings Foreign savings f1 f2 f3 Average weights 65.8% 24.7% 9.6% π rs MCFS Rates of return 11.6% 7.5% 9.7% EOCK 10.4% 3.3.2 Discussion and sensitivity analysis of the results The sensitivity analysis confirms that taking into account the price elasticity of domestic savings is critical. As capital markets develop in Mexico, different financial savings instruments are becoming available to savers (households, corporations, pension funds). This will necessarily 21 result in higher elasticities of savings (or the response from these sectors). Assuming a higher price elasticity of savings supply, from 0.30 to 0.60, would tend to lower the EOCK by 0.6 percent, as the last row and middle columns of Table 8.1 and Table 8.2 show. The possible changes with plausible ranges for the price elasticity of domestic investment do not have a significant effect on the EOCK. Table 8.1 Sensitivity analysis of the EOCK for investment and domestic savings price elasticities Investment elasticity −0.8 −1 −1.2 0.10 10.8% 11.0% 11.0% Domestic savings 0.30 10.2% 10.4% 10.5% elasticity 0.60 9.6% 9.8% 10.0% Table 8.2 Sensitivity analysis of the EOCK for foreign savings and domestic savings price elasticities Foreign savings supply elasticity 1 3 5 0.10 11.2% 11.0% 10.7% Domestic savings 0.30 10.6% 10.4% 10.2% elasticity 0.60 9.9% 9.8% 9.7% An estimation of the impacts of foreign savings price elasticities shows that these are not as critical as the price elasticities of domestic savings. In part, this is embedded in the formula for the marginal economic cost of foreign saving (MCFS), as the elasticity in the second term is also in the denominator. The other important reason in the context of Mexico is the relatively small share of net foreign savings in the economy in recent years. 4 Supply price approach to Mexico’s EOCK for government, state-owned enterprises, and regulated investment projects 4.1 Using the supply price approach to estimate the EOCK This section estimates Mexico’s EOCK by adopting an approach that considers the minimum supply price of capital plus the economic externalities of the use of capital. As discussed in section 2.3 above, this “bottom-up” methodology considers the tax externalities per unit of capital and uses the same weights as the traditional “top-down” weights from the national 22 accounts methodology for domestic investments, domestic savings, and foreign savings. In order to obtain the EOCK for the economy, the result of the overall tax externality per unit of capital invested must be combined with the minimum supply price of capital in the market (im in percentage terms per year), which comes from debt and equity sources. This part of the discussion assumes that we are dealing with the opportunity cost of capital supplied to an investment project that represents the financing of the foregone investment and the same capital funds supplied in the top-down case. This discount rate thus serves as the economic reference price for public sector projects in the absence of variations in the cost of capital for different projects. We will show that, in principle, following the top-down or bottom-up methodology for the average investment should yield similar or comparable results. First, the tax collection per unit of capital was estimated based on data available for general government tax revenues by major category. The financial sum of taxes collected from the operations of the private sector (income taxes, property taxes, indirect taxes paid out of capital income, plus the share of indirect taxes in the product markets attributable to the contribution of capital to value added) was divided by the sum of capital stocks of the private sector, including the value of land and inventories, to obtain an average rate of taxation to capital in Mexico for the 2003–2011 period. The result was 2.3 percent of capital stocks, including land and inventories (Table 9). Table 9 Taxes on capital over GDP, 2003–2011 averages Taxes as share Type of tax of capital Taxes on capital income and assets of corporate and capital share of non-corporate businesses as a share of private reproducible capital + land 0.7% as a share of private reproducible capital + land + inventories 0.5% Taxes on property values as a share of private reproducible capital + land 0.1% as a share of private reproducible capital + land + inventories 0.1% Indirect taxes paid out of capital income (taxes on mining and oil) as a share of private reproducible capital + land 1.7% as a share of private reproducible capital + land + inventories 1.2% Share of indirect taxes in product markets (VAT, customs and excise) attributable to capital as a share of private reproducible capital + land 1.0% as a share of private reproducible capital + land + inventories 0.6% Total tax distortion on capital assets as a share of private reproducible capital + land 3.4% as a share of private reproducible capital + land + inventories 2.3% Source: SHCP; OECD tax data. Second, the weighted average externality for invested capital was estimated. We combine the tax externality on capital of 2.3 percent (from Table 9) with the tax gain from domestic 23 savings and the net externality from foreign savings. The tax gain is the result of the multiplication of the savings rate of 8.3 percent times the assumed average tax on savings of 10 percent for a negative tax externality (or a tax gain) of −0.8 percent. The tax externality on foreign savings has a tax gain of 8.2 percent times a 10 percent assumed withholding tax rate (tw) or −0.82 percent and a positive tax externality of im*(1 − tw)*k/Efs. With k = 0.90 (share of foreign savings responsive to changes in the market interest rate), tw = 10 percent (withholding tax), and Efs = 3.0 (price elasticity of supply of foreign savings), this results in a positive tax externality of 2.22 percent. Considering both the positive and the negative tax externalities yields a total of 1.4 percent (2.22 – 0.82 percent). Based on these estimations, the different tax distortions—such as the tax distortion per unit of capital invested, the tax distortions on savings (e.g. the interest rate times the tax on savings income, with a negative value or negative externality), and the tax distortion on foreign savings—have to be compounded using the same weights as the traditional EOCK. Using the same weights as in the previous exercise (f1 to f3 from Table 7), the weighted externality per unit of capital used in Mexico would be a relatively low 1.4 percent per unit of capital (Table 10). Table 10 Externality per unit of capital Domestic investment Domestic savings Foreign savings f1 f2 f3 Average weights 65.8% 24.7% 9.6% Return 8.3% 8.2% Tax rate 10% 10% Distortions/unit of capital 2.3% −0.8% 1.4% EOCK externalities/unit of capital 1.4% The results for the EOCK are robust when compared to those obtained through the alternative approaches considered. If we add a weighted cost of private funds (im) of 8.9 percent (Table 11) and the EOCK externality per unit of capital of 1.4 percent (Table 10), we obtain an EOCK for Mexico of 10.3 percent, similar in magnitude to the value of 10.4 percent that was estimated using the top-down approach from national accounts data. Some differences can be explained by the average values for both the prices of debt and equity, as well as the ratios in the weights, which are an approximation. 4.2 Consistency of the results and interpretation As we can observe, the results for the top-down and bottom-up approaches are very consistent—both methodologies point to a new recommended EOCK for Mexico of about 10 percent. In this study we used the approximations from market information about Mexico (corporate bonds 30 percent, stock exchange 40 percent, and housing investments 30 percent of market shares representing the private sector in Mexico) and the rates of return from Table 5 and Annex III. We obtain 10.4 percent EOCK from the top-down approach and 10.3 percent from the bottom-up approach. These results, in addition to those obtained from reviewing the existing 24 methodology updated to 2012 (with an EOCK of about 10 percent), give us an indication of the recommended EOCK value for Mexico: a rounded 10 percent discount rate. Table 11 Comparing the two approaches to estimate the EOCK, 2003–2011 Bottom-up and top-down estimations Gross return excluding share of indirect taxes (Table 4) 10.9% Taxes paid out of return on capital (Table 9, less indirect taxes) 1.7% Net of tax return, private sector 9.2% [top-down] Weights in Real rates private Market representation: of return investment Corporate bonds (Table 5) 5.1% 30% 1.5% Real return on equity (Table 5) 13.0% 40% 5.2% Housing return (from Annex III) 8.3% 30% 2.5% 9.2% [same] Domestic Foreign Investment savings savings f1 f2 f3 Average weights 65.8% 24.7% 9.6% Net of tax return on investment or 9.2% 8.3% 8.2% gross of tax return on savings Weighted supply cost of capital (private sector) 8.9% WACC [A] Adding back the tax externality: Weighted average cost of capital (WACC for private sector) 8.9% [A] EOCK externality/unit of capital 1.4% [B, from Table 10] EOCK, bottom-up approach 10.3% [A+B] Domestic. Foreign Investment savings savings Top-down f1 f2 f3 Average weights 65.8% 24.7% 9.6% Rates 11.6% 7.5% 9.7% EOCK, top-down approach 10.4% [same magnitude] 4.3 Estimating the tax externality in the case of disaggregation of PEMEX from the rest of the economy In the final step of our analysis we assess the impact of PEMEX on EOCK externalities. The impact of the tax externalities per unit of investment in PEMEX and the rest of the economy is investigated. From the tax revenue for general government information and from national 25 account information, we obtain a PEMEX capital distortion of 50.8 percent per unit of capital plus inventories invested, in addition to a staggering share of all taxes of 30.6 percent (see Table 13). 4.3.1 The private return to capital in PEMEX and the rest of the private sector In the next stage of the analysis we distinguish between two scenarios: one in which we disaggregate PEMEX and another in which we do not. In Table 12, the income to private capital and stocks of capital are now disaggregated into PEMEX and non-PEMEX accounts, based on available PEMEX data. The net capital income (or net operating surplus) in the private sector is split between PEMEX (based on its accounts) and the residual to non-PEMEX entities in the private sector. Table 12 Estimation of the rates of return to capital in PEMEX and the remainder of the private sector, 2003–2011 Share of Net income to Share of net Net Share Private Gross capital (or net Return indirect capital capital of capital + land return on operating on total taxes in income income capital + inventories total surplus) capital product to private over GDP assets over GDP capital [million pesos] markets sector PEMEX 2003 367,567 11.8% 4.9% 2.7% 0.11 43.5% 0.5% 44.0% 2004 455,201 11.9% 5.3% 2.8% 0.11 46.2% 0.6% 46.8% 2005 498,755 12.1% 5.4% 2.9% 0.12 44.6% 0.6% 45.2% 2006 581,348 11.9% 5.6% 3.0% 0.12 45.7% 0.6% 46.3% 2007 590,431 10.9% 5.2% 3.1% 0.13 41.0% 0.5% 41.5% 2008 571,112 9.6% 4.7% 3.3% 0.13 34.8% 0.4% 35.2% 2009 428,277 8.2% 3.6% 3.6% 0.16 22.5% 0.3% 22.8% 2010 546,456 8.9% 4.2% 3.9% 0.17 25.1% 0.4% 25.5% 2011 672,191 9.8% 4.7% 4.0% 0.17 27.9% 0.4% 28.3% Average 36.8% 37.3% Non-PEMEX 2003 2,735,997 88.2% 36.2% 97.3% 4.09 8.9% 0.6% 9.4% 2004 3,357,537 88.1% 39.2% 97.2% 4.05 9.7% 0.6% 10.2% 2005 3,618,719 87.9% 39.1% 97.1% 4.03 9.7% 0.5% 10.2% 2006 4,308,619 88.1% 41.5% 97.0% 3.95 10.5% 0.6% 11.1% 2007 4,812,545 89.1% 42.5% 96.9% 3.93 10.8% 0.6% 11.4% 2008 5,360,974 90.4% 44.0% 96.7% 3.99 11.0% 0.6% 11.7% 2009 4,791,926 91.8% 40.1% 96.4% 4.30 9.3% 0.5% 9.9% 2010 5,565,340 91.1% 42.6% 96.1% 4.13 10.3% 0.6% 10.9% 2011 6,202,977 90.2% 43.1% 96.0% 4.04 10.7% 0.6% 11.3% Average 10.1% 10.7% Source: Authors’ calculations using PEMEX and SHCP data. 26 Over the 2003–2011 period, PEMEX captured 10.6 percent of the net capital income. In terms of assets, based on the book assets and investment of PEMEX, a perpetual inventory method asset series was created, in order to make it comparable with the private sector asset series. This asset series raises the value of PEMEX assets by 17.5 percent on average over the timeframe covered. While PEMEX assets only averaged 3.2 percent of the total private sector capital assets during the 2003–2011 period, PEMEX investment in fixed capital assets averaged 10.9 percent of total private investment. PEMEX net capital income and assets generated a high rate of return of 36.8 percent on average, but, interestingly, the rate of return dropped from 42.6 percent in the 2003–2008 period to 25.2 percent in the 2009–2011 period, while investment levels by PEMEX rose from 7.8 percent of GDP to 11.1 percent over the same two periods. Finally, the share of the indirect taxes in the product markets attributed to capital income (about 0.1 percent) was added in order to assess the average gross rate of return to capital, which came out to 36.9 percent. Table 12 also shows the non-PEMEX entities earning the remaining 89.4 percent of net capital income on the residual 96.8 percent of capital stocks in the private sector. This results in a change of the gross rate of return to the residual private capital from an average of 11.6 percent (see Table 4) to an average of 10.7 percent (Table 12). After adjusting for the income that accrues to land and the stocks of long-term inventories, the returns to private capital net of indirect taxes are now on average 10.1 percent for the 2003–2011 period; and gross of indirect taxes attributable to capital, they go up to 10.7 percent for the same period. The rates of return to capital for PEMEX are higher than the rest of the private sector. As we can observe in Table 12, if we disaggregate both income and capital shares into the private sector and PEMEX, we obtain separate and composite rates of return. From the results, we can see that the rates of return to capital are quite high for PEMEX, as the capital stock itself is small compared to the rest of the private sector. The estimated rates of return to PEMEX have been declining since 2007. Although the PEMEX rates of return on capital are on average 37 percent (and declining), their contribution to the overall returns in the economy is small due to its capital size and participation in gross operating profits in relative terms. Nevertheless, we can see the effect that PEMEX has on the rest of the economy: it makes the EOCK or the discount rate slightly higher because of its relatively high share in investment. Table 13 The impact of PEMEX on the EOCK: Rates of return and tax distortions relating to PEMEX and non-PEMEX entities, 2003–2011 Breakdown by entity type 2003–08 2009–11 2003–11 PEMEX Taxes paid out of profits as a share of total tax revenue of Mexico 31.9% 28.3% 30.6% Taxes as a share of assets 71.4% 43.6% 62.1% After-tax return on assets −1.8% −1.8% −1.8% Required return on assets 9.5% 9.5% 9.5% Capital tax distortion (including share of indirect product market taxes) 60.0% 32.3% 50.8% Non-PEMEX Capital income and property taxes as a share of assets 0.5% 0.6% 0.5% Capital tax distortion (including share of indirect product market taxes) 1.1% 1.0% 1.1% Source: PEMEX accounts; SHCP; OECD Tax Data. 27 To assess the impact of PEMEX on the EOCK, we disaggregate investment flows for PEMEX and other sectors. Assuming the same price elasticity of demand for investment funds as the rest of the private sector (Table 14), we recalculate the weights in this scenario for PEMEX, other investments, domestic savings, and foreign savings. The averages are included in Table 15. With the assumptions above, we generate the new weights and apply these to the tax distortions reported above for the four sectors (PEMEX, other investors, domestic savers, and foreign savers). The results, reported in Table 15, show a higher EOCK externality of 2.4 percent. The size of the PEMEX distortion has fallen in recent years as a result of its declining surpluses—a trend that could potentially continue over the long term—which suggests a more reasonable estimate of the economic externality of close to 2 percent. Table 14 Investment and savings shares and assumed price elasticities for the disaggregated scenario, 2003–2011 Other private + PEMEX investment DS/ FS/ η η inventory investment Eds Efs over GDP GDP GDP PEMEX (rest) over GDP 2003 1.5% 17.6% 22.9% 1.1% −1 −1 0.3 3 2004 1.4% 19.1% 24.9% 0.9% −1 −1 0.3 3 2005 1.4% 18.4% 24.4% 1.0% −1 −1 0.3 3 2006 1.4% 20.4% 26.2% 0.8% −1 −1 0.3 3 2007 1.5% 20.4% 26.5% 1.4% −1 −1 0.3 3 2008 1.7% 19.6% 26.9% 1.8% −1 −1 0.3 3 2009 2.1% 15.5% 23.8% 0.8% −1 −1 0.3 3 2010 2.1% 16.0% 23.9% 0.2% −1 −1 0.3 3 2011 1.9% 18.0% 25.1% 0.9% −1 −1 0.3 3 Average 1.7% 18.3% 24.9% 1.0% −1 −1 0.3 3 Source: Authors’ calculations using PEMEX and SHCP data. Table 15 Disaggregating PEMEX and the estimation of the EOCK tax externalities per unit of capital PEMEX investment Other investment Domestic savings Foreign savings f0 f1 f2 f3 Average weights 5.5% 60.2% 24.7% 9.6% Distortions/unit 32.3% 1.1% −0.8% 1.4% EOCK externalities/unit of capital 2.4% Source: Authors’ calculations on the basis of previous tables. The reason for the increase in the tax externality when PEMEX is separated out is that PEMEX not only has a high tax distortion, but the investment share of PEMEX is higher than its asset share—over the 2003–2011 period it averaged 4.1 times higher. This higher investment share puts a relatively higher weight on the PEMEX distortion when it is 28 disaggregated and assumed to respond to the market in the same way as other investors. If PEMEX investment is, in fact, less responsive to changes in the cost of capital than other business entities in Mexico, then the tax externality will be lower, as can be seen in Table 16. For a lower (in absolute terms) price elasticity of demand for investments of −0.5 for PEMEX and a slightly higher one of −1.5 for non-PEMEX entities, we would still obtain a per unit capital externality of 1.4 percent. Table 16 Sensitivity of the EOCK tax externalities/unit of capital to investment price elasticities by PEMEX and non-PEMEX entities Rest of the economy investment elasticity −0.5 −1 −1.5 PEMEX 0.0 0.4% 0.6% 0.7% investment −0.5 1.7% 1.5% 1.4% elasticity −1.0 2.9% 2.4% 2.1% 5 Conclusions: Lessons and considerations drawn from the Mexican case The main weakness of the traditional top-down single value approach to the EOCK is that it has difficulty in accommodating the different capital mobilization transaction costs and, more importantly, the different systematic risk premiums required on capital funds supplied to different types of project investments. While the average commercial project (essentially financed in the long run by market sales revenues and using average debt leverage) bears the market risk premium on its equity, projects bearing more or less market risk will face higher or lower systematic risk premiums, or, in other words, their EOCK will vary accordingly to reflect the changes in costs incurred on capital financing. By contrast, all routine projects (non-mega projects with modest unsystematic project risks) ultimately financed out of government revenues (largely taxes) will bear no systematic market risk premium, such that the costs of capital supplied in a competitive market will be lower than the commercial projects. Hence, there are arguments for having two discount rate regimes: one for public sector projects funded out of budget resources, and the other for public sector market-oriented projects. For the latter, the EOCK would either be based on a weighted average cost of capital plus the economic externality per unit of capital or on the actual weighted average cost of capital for each type of commercial project. Nevertheless, the institutional arrangements and capacities in handling a high volume of public investment projects favor using a single EOCK to simplify the project selection process and bring greater transparency and control to the process. Based on the empirical results of this paper’s case study, we recommend that a single EOCK rate for Mexico be set at 10 percent. However, it is also recommended that a degree of flexibility be allowed for the upward adjustments to the EOCK in the case of large-scale commercial projects in regulated sectors or PPPs, where above-average systematic market risks increase the costs of finance. 29 In the Mexican case in particular, a number of areas would benefit from further research and data gathering in order to sharpen the EOCK estimates. These include, first, data on the tax revenues raised yearly from domestic and foreign savings, in order to more accurately estimate the externalities involved in using additional domestic and foreign savings. Second, more details about the breakdown of domestic savings between households, corporations, and the government and the portfolio of household savings, which would sharpen the understanding of the importance and returns on domestic savings. Third, a confirmation and understanding of the relatively high investment in added inventories in Mexico, which would be helpful in estimating the stocks of capital. Based on this study’s applied analysis of the Mexican case, we propose the following general and Mexico-specific recommendations: • Choice of models: We recommend that both the traditional weighted average method for the estimation of the single rate and the minimum supply price of capital plus externalities approach be used to check consistency of results. The parameters of these models should be revised and re-estimated at least every five years to take into account the changing macroeconomic and fiscal circumstances. • Single versus differentiated EOCK: For Mexico, we recommend a single EOCK for all public investment projects at 10 percent. However, it is also recommended that a degree of flexibility be allowed for the upward adjustments to the EOCK in the case of large-scale commercial projects in regulated sectors or PPPs, where above-average systematic market risks increase the costs of finance. • Implications of the economic trends for a forward-looking estimate: As the world economy pulls back from the monetary stimulus adopted in response to the recent global recession, careful consideration of upward trends in the international cost of funds is required in the adoption of an EOCK rate. In addition, the effects of underlying improvements in capital markets, the possible greater degree of integration of the Mexican economy into international capital markets, as well as the impacts of macro-economic policies in the United States, need to be carefully included in the consideration of the minimum supply cost of funds and the marginal economic costs of foreign savings or borrowing. • An improving country risk rating: Credit rating changes affect the EOCK. Mexico’s credit rating has significantly improved since its collapse during the financial crisis of the 1980s (see Annex II), which has in turn resulted in lower costs of external as well as domestic funding. Although this positive trend was partially interrupted during the last global financial crisis, after the recovery and the continuation of expansionary monetary policies, especially in the United States, 10-year Treasury bills and corporate rates continued to go down from 2010 to 2012. However, according to the Federal Reserve Bank (capital market rates), there has been a recent upward shift in 10-year Treasury bills, from a low of about 1.5 percent in early 2013 to about 3 percent as of September 2013. These higher rates will affect international markets, including Mexico. While it is expected that international borrowing costs will increase over the medium term, Mexico still has significant upside potential to 30 improve its governance and economic management in order to continue improving its country risk ratings and bring about systematic reductions in its costs of debt and equity financing. • Changes in tax distortions: The impact of a fiscal reform needs to be taken into account when assessing the EOCK. As a result of the fiscal reform recently approved in Mexico, some of the taxes and actual revenue collected will change. These changes need to be considered in any recalculation of the EOCK. More detail is required on the actual revenue measures and expected revenue impacts. However, rough simulations of a net increase in revenues of 3 percent of GDP over the medium term (as announced in 2013) that is widely spread across the various tax types is expected to raise the EOCK by about 0.2 to 0.3 percentage points. This will ultimately depend on the details of the economic reform package and tax measures, as well as its implementation. • Improvements in the financial sector: From the analysis, we conclude that there is a huge potential for the financial sector to grow in Mexico. The size of the financial sector in relation to GDP is relatively small in the international context. Increased competitiveness, availability of more savings and investment instruments, and greater integration to world markets could lower the cost of capital and reduce the estimated EOCK. The countervailing forces (slightly higher tax externality effects and lower capital costs due to financial sector reforms and greater competitiveness in the sector) will need to be evaluated in the revisions of the EOCK. • The impact of reforms: The impact of the multiple reforms approved in Mexico during the last months is still uncertain and could also affect the EOCK. It is recommended that a reassessment of the EOCK take place in about five years in order to capture the impact of the reforms and the evolution of economic trends and country risk rating. • Improvements required in data collection and availability: In addition to national accounts information and data on labor markets, expanded data collection on country, sector risk, and costs of finance by different instruments and for different sectors and forms of funding will be required so as to improve future estimations of the EOCK. In addition, a clear roster data on tax collection by type of instruments and payers, as well as by type of transactions—especially domestic and foreign savings, as they affect different sectors of the economy—should be developed and maintained using different sources and checking for internal consistency with the national accounts and budgets in the public sector. 31 6 References Bernanke, Ben S., and Refet S. Gürkaynak. 2001. “Is Growth Exogenous? Taking Mankiw, Romer, and Weil Seriously.” National Bureau of Economic Research (NBER) Working Paper 8365, Cambridge, MA. Burgess, David. 2008. “Removing Some Dissonance from the Social Discount Rate Debate.” Economic Policy Research Institute Working Paper 2008-2, University of Western Ontario, London, Ontario, Canada. Caselli, Francesco, and James Feyrer. 2007. “The Marginal Product of Capital.” The Quarterly Journal of Economics 122 (2): 535–68. Cervini Iturre, Héctor Félix. 2004. “El costo de oportunidad social de los fondos públicos y la tasa social de descuento en México, 1970–2001.” Informe Final, Centro de Estudios para la Preparación y Evaluación Socioeconómica de Proyectos (CEPEP), México, D.F. Dasgupta, Partha, Amartya Sen, and Stephen Marglin. 1972. Guidelines for Project Evaluation. New York: United Nations (UN Industrial Development Organization, UNIDO). Diario Oficial de la Federación (DOF). 2012. “Lineamientos para la elaboración y presentación de los análisis costo y beneficio de los programas y proyectos de inversión.” SHCP, 27/04/2012. Diwan, Ishac. 2001. “Debt as Sweat: Labor, financial crises, and the globalization of capital.” World Bank, Washington, DC. Mimeo. Erb, Claude, Campbell R. Harvey, and Tadas E. Viskanta. 1995. “Country Risk and Global Equity Selection.” The Journal of Portfolio Management 21 (2): 74–83. ———. 1996. “Expected Returns and Volatility in 135 Countries.” The Journal of Portfolio Management 22 (3): 46–58. Feldstein, Martin S. 1964. “The Social Time Preference Discount Rate in Cost-Benefit Analysis.” The Economic Journal 74 (294): 360–79. Glenday, Graham. 2010. “The New (but Old) Approach to the Economic Opportunity Cost of Capital.” In Discount Rates for the Evaluation of Public Private Partnerships, by David F. Burgess and Glenn P. Jenkins, eds., pp. 161–90. Kingston, Ontario: McGill-Queens University Press. Guerriero, Marta. 2012. “The Labour Share of Income around the World. Evidence from a Panel Dataset.” Development Economics and Public Policy Working Paper No. 32/2012, Institute for Development Policy and Management, University of Manchester. 32 Harberger, Arnold C. 1972a. Project Evaluation: Collected Papers. Chicago: The University of Chicago Press. ———. 1972b. “Professor Arrow on the Social Discount Rate.” Project Evaluation: Collected Papers. Chicago: The University of Chicago Press, pp. 123–31. ———. 1980. “Vignettes on the World Capital Market.” American Economic Review 70 (2): 331–7. ———. 1997. “New Frontiers in Project Evaluation? A Comment on Devarajan, Squire, and Suthiwart-Narueput.” The World Bank Research Observer 12 (1): 73–9. Harrison, Ann E. 2002. “Has Globalization Eroded Labor’s Share? Some Cross-Country Evidence.” University of California-Berkeley and National Bureau of Economic Research, Inc. (NBER). Hirshleifer, Jack, James C. DeHaven, and Jerome W. Milliman. 1960. Water Supply: Economics, Technology, and Policy. Chicago: University of Chicago Press. Hofman, Andre A. 2000. “Standardized capital stock estimates in Latin America: A 1950–94 update.” Cambridge Journal of Economics 24 (1): 45–86. Jayadev, Arjun. 2007. “Capital Account Openness and the Labour Share of Income.” Cambridge Journal of Economics 31 (3): 423–43. Jenkins, Glenn P., and Mostafa Baher El-Hifnawi. 1993. “Economic Parameters for the Appraisal of Investment Projects: Bangladesh, Indonesia, and The Philippines.” Report prepared for the Asian Development Bank, Manila. Jenkins, Glenn P., Chun-Yan Kuo, and Arnold C. Harberger. 2010. “The Economic Opportunity Cost of Capital for Canada, an Update.” In Discount Rates for the Evaluation of Public Private Partnerships, David F. Burgess and Glenn P. Jenkins, eds. Kingston, Ontario: John Deutsch Institute, Queen’s University. Karabarbounis, Loukas, and Brent Neiman. 2012. “Declining Labor Shares and the Global Rise of Corporate Saving.” NBER Working Paper No. 18154, National Bureau of Economic Research, Inc. Little, I.M.D., and James A. Mirrlees. 1969. Manual of Industrial Project Analysis. Paris: OECD Development Centre. ———. 1974. Project Appraisal and Planning. London: Heinemann. Lopez, Humberto. 2008. “The Social Discount Rate: Estimates for Nine Latin American Countries.” Policy Research Working Paper No. 4639, World Bank, Washington, DC. 33 Loría, Eduardo, and Leobardo de Jesús. 2007. “Los Acervos de Capital de México: Una estimación, 1980.I–2004.IV.” El Trimestre Económico LXXIV-2 (294): 475–85. Marglin, Stephen A. 1963a. “The Social Rate of Discount and the Optimal Rate of Investment.” The Quarterly Journal of Economics 77 (1): 95–111. ———. 1963b. “The Opportunity Costs of Public Investment.” The Quarterly Journal of Economics 77 (2): 274–89. Office of Management and Budget (OMB). 2005. “Analytical Perspectives, Budget of the United States Government, Fiscal Year 2005.” Executive Office of The President of the United States, Washington, DC. Rajaram, Anand, Tuan Minh Le, Nataliya Biletska, and Jim Brumby. 2010. “A Diagnostic Framework for Assessing Public Investment Management.” Policy Research Working Paper No. 5397, Public Sector Reform and Capacity Building Unit and Poverty Reduction and Economic Management Network, Public Sector Unit, World Bank, Washington, DC. Rodriguez-Medrano, Sergio Luis. 2009. “The Economic Opportunity Cost of Capital (EOCK) in Mexico: 1970–2006.” In Essays on International Economics and Economic Development: A Heckscher-Ohlin View of Differences in Income per Worker Across Countries and the Economic Opportunity Cost of Capital in Mexico, pp. 111–256. University of California, Los Angeles, ProQuest UMI No. 3450992. Squire, Lyn, and Herman G. van der Tak. 1975. Economic Analysis of Projects. Baltimore: Johns Hopkins University Press (for the World Bank). World Bank. 2006. Where is the Wealth of Nations? Measuring Capital for the 21st Century. Washington, DC: World Bank. World Bank. 2014. Mexico: Estimation of the Economic Opportunity Cost of Capital for Public Investment Projects. Washington, DC: World Bank. 34 7 Annexes Annex I Potential impacts of the tax/fiscal reform proposals for the EOCK in Mexico Tax reforms can be expected to change the tax distortions in the markets affected by the use of capital funds, and hence the weighted average tax externality will also be affected; however, an estimation of the changes in tax revenues as a result of the fiscal and tax reforms in Mexico indicates that the net impact on the estimated tax externality per unit of capital will be relatively small. The tax reforms announced to date propose to raise revenues through taxes on capital and labor incomes as well as indirect taxes in the product markets. The modest impacts from the proposed tax increases are expected to raise the tax externality by some 0.1 to 0.3 percentage points. As an illustration, we assume that 3 percent of GDP in added tax revenues is raised equally through the three tax types. The combination of 1 percent of GDP through taxes on capital and a further 1 percent through indirect taxes would increase the tax distortion on invested capital by about 0.35 percent per unit of total capital (the basic calculation is presented below). Given a weight of about two-thirds on foregone investment in the EOCK, this would raise the EOCK by about 0.23 percentage points. To the extent that taxes on domestic and foreign savings are also increased (the proposed reforms include raising the top tax rate on personal income, taxing capital gains on traded securities, and establishing a special 10 percent corporate tax on dividends and profit distributions), the marginal economic cost of domestic and foreign savings is expected to drop slightly with a corresponding small reduction (or contribution) in the weighted average tax externality. For example, if tax changes increase the effective taxes on both domestic and foreign savings by 1 percentage point, then the change in the weighted average EOCK would go down by 0.03 to 0.02 percent. This indicates that the range of increase in the weighted average tax externality would be between about 0.1 to 0.3 percentage points, depending on the mix of tax changes. Ultimately, the detailed tax proposals will need to be settled, and detailed estimates of the revenue impacts and changes in the effective tax rates—categorized by the major tax types relative to the forecast base case scenarios for the economy—will need to be carried out in order to reach a more accurate estimate of the impact of the tax reforms on the EOCK. Tax revenue assumptions and impacts To estimate the impact on the EOCK externality per unit of capital, we assume that a tax increase of about 3 percent of GDP as part of the proposed reforms would result in a total of 22 percent, compared to the 19 percent tax/GDP performance in 2011. For an estimation of a range of impacts, we assume that the 3 percent increase in tax revenues will originate from income taxes on capital at 1 percent of GDP and indirect taxes in the product markets adding up to another 1 percent of GDP. We also assume that other measures will result in taxes on labor and savings, effectively contributing the other 1 percent of GDP in new tax revenues, which will not affect our calculations. As our estimation shows, the range of impacts of the tax reform proposals on 35 the tax externality per unit of capital could range from 0.10 to 0.30 percent, with a median value of about 0.20 percent. Table I.1 Estimation of the impact of the proposed tax reforms on the EOCK externality per unit of capital Effects of proposed tax changes as part of the fiscal reform Incremental tax charged on: Capital income Labor and (corporate, domestic Indirect taxes in property, and and Total product market other taxes paid foreign out of profits) savings Incremental tax revenue as a share of GDP 1% 1% 1% 3% Contribution to tax distortion on investment 100% 0% 59% Tax distortion on capital as share of GDP 1.0% 0.0% 0.6% 1.6% Private capital (reproducible + land + inventories) 4.50 over GDP Tax distortion over private capital assets 0.35% Weight on foregone investment in EOCK 66% Contribution to tax externality per unit of capital 0.23% Change in overall tax externality in EOCK Investment 0.23% Domestic savings −0.02% Foreign savings −0.01% Economic externality (change) 0.20% 36 Annex II Risk issues Recent trends in country risk and cost of funding for Mexico According to the methodology developed by Erb, Harvey, and Viskanta (1995), the current differentials in country credit rating (IICR of 89 for the United States and 68 for Mexico), result in a risk spread of about 5 percent in nominal terms between Mexico and the United States. These rates are used to assess the EOCK tax externality and the rates of interest for the calculation of the EOCK for Mexico. Source: Institutional Investor. As can be seen from the upper figure, the credit rating for Mexico has been improving since the mid-1980s (from 30 to 70). Correspondingly, the equity rates of return or cost of equity estimated for investors in US$ in nominal terms, show a significant decline up to 2012. 37 Source: Authors’ calculations using Erb et al. (1995) methodology and Institutional Investor data. The methodology to calculate cost of equity traces the average real returns of the Mexican Stock Exchange (MXXD) on a decade-by-decade basis. Table II.1 Real rates of return to equity Real rates of return Averages Decade 10.0% 1970s 24.2% 1980s MXXD Mexican stocks 17.1% 1990s 14.3% 2000s Source: Authors’ calculations using Mexican Stock Exchange data. 38 Annex III Real return on housing The stock of housing is calculated using the perpetual inventory method, consistent with the capital stock calculations for the rest of the economy. Dividing gross value added for housing over the capital stocks, we obtain the rates of return in the last column, with an average value of 8.3 percent, which is used in Table 11 of the text. Table III.1 Real rate of return on housing, 2003–2011 Private Housing real Housing Housing Housing + capital + rate or return; stock + GVA Year investment land over land + GDP GVA housing/ land over housing over GDP GDP inventories housing stocks GDP over GDP + land 2003 5.3% 6.6% 1.21 4.20 7,555,803 796,264 8.7% 2004 5.5% 6.9% 1.26 4.16 8,574,823 882,985 8.2% 2005 5.5% 6.9% 1.25 4.15 9,251,737 958,469 8.3% 2006 5.8% 7.2% 1.31 4.07 10,379,091 1,051,038 7.7% 2007 5.7% 7.1% 1.30 4.06 11,320,836 1,122,930 7.6% 2008 5.8% 7.3% 1.32 4.12 12,181,256 1,210,219 7.5% 2009 5.0% 6.3% 1.14 4.46 11,937,250 1,218,135 9.0% 2010 4.7% 5.9% 1.07 4.30 13,071,597 1,269,831 9.1% 2011 4.7% 5.9% 1.08 4.20 14,395,833 1,334,523 8.6% Average 8.3% Source: Authors’ calculations, national accounts data, SHCP. 39