1. Introduction
The electric power supply chain is essentially composed of generation, transmission, distribution, and commercialization activities. In order to remunerate the investments made by energy transmission system operators (TSOs), each country’s regulatory commissions must guarantee investment recovery and return on investment. Return on investment is calculated by multiplying the regulatory asset base (RAB) by the discount rate, which is fundamental for the valuation of investments in the energy sector because it represents the minimum expected rate of return on an investment with a specified risk [
1]. To determine the discount rate, many commissions around the world use the weighted average cost of capital (WACC) methodology, which seeks to recognize both the cost of financing and the cost of equity for RAB investments. According to each country’s regulator and conditions, the WACC can be used in its basic form (plain vanilla), where the effect of taxes is not taken into account, as in the case of England; it can also be used before taxes (pre-tax), as in the case of Colombia, or after taxes (post-tax), as in the case of Brazil. Additionally, the WACC can be expressed in nominal or real terms, where the effect of inflation is discounted.
Table 1 shows some countries where regulators have adopted the real pre-tax WACC methodology to calculate the discount rate of TSOs, taking 2019 rates into account. We have chosen this year in particular because we found comparable information that exemplifies the problem.
Table 1 shows that in Colombia and other emerging countries that use the real pre-tax WACC methodology, the discount rates are very high compared to the discount rates of developed countries. As Dobbs [
2] mentions, very high discount rates can lead to financial inefficiency and increased user fees. However, very low discount rates discourage investment, which has negative effects on service quality, which is why the discount rate calculation should be as accurate as possible.
One of the causes for these high rates in emerging countries is the recursive WACC application in countries with underdeveloped stock markets, such as Colombia. The general WACC methodology applied in Colombia can be found in CREG Resolution 95 of 2015 [
7], while the detail on the variables and how they are calculated can be seen in CREG Document 011 of 2018 [
3]. Equation (1) shows how to calculate the pre-tax WACC, and Equation (2) shows how to calculate the real pre-tax WACC:
where
is the nominal weighted average cost of capital in Colombian pesos (cop) before taxes for transmission and distribution activities;
is the weighted cost of debt, equal to 40%;
is the weighted cost of equity and is equal to 60%;
is the cost of debt in pesos at time
t and is equal to 9.62%;
is the cost of equity in cop, equal to the cost of equity in dollars at time
t (the value estimated by CREG was 12.83%); and
is the tax rate in force in Colombia on the calculation date, which is 33% for the purposes of this article and traceability of the provisions of CREG Document 011 of 2018 [
3]. The resulting
in nominal pesos is 15.33%. However, as the applicable value in Colombia is determined in real pesos, an adjustment is made with the inflation expectation
determined at 3.17%, leading to a
of 11.79% using Equation (2).
The weighted cost of debt and cost of equity used by CREG are very close to those used by most regulators, who generally assign a higher weight to the cost of equity. The cost of debt is calculated as the weighted average of commercial loan placement rates; therefore, the value assigned to the cost of debt is readily observable in the market and does not need to be discussed. On the contrary, the cost of equity cannot be observed in the market and is estimated based on the CAPM model, which has been widely discussed for its assumptions and the way it is applied to the calculation of regulated tariffs in emerging Latin American countries, since regulators do not use local variables, but rather U.S. market variables, which results in a less accurate measurement of the cost of capital since the market risk measured by the CAPM is different for each country and industry [
8] and depends on the market index taken as a reference.
Based on the above, the objective of this article is to present the estimation of the cost of equity for TSOs in Colombia according to firms’ returns and the country’s local variables, defining an accurate discount rate that balances final user and TSO interests. To achieve this objective, we estimate the cost of equity through the CAPM and our proposed four-factor model, TD-4. As a target variable in both models, we use the returns of companies in the Colombian stock exchange (BVC) that operate as TSOs, and as determinants of these returns, we use the COLCAP index and Colombian economic variables. Finally, we sensitized the WACC calculation by changing the cost of equity value calculated by CREG with the values found through the two evaluated models.
For this analysis, in
Section 2 of this article, we explain how the cost of equity is currently estimated in Colombia and why it should be changed. In
Section 3, we present a state-of-the-art of asset pricing model and then present the models proposed for the cost of equity estimation. In
Section 4, we present the proposed methods, and in
Section 5, we show the results of the models used to estimate the cost of equity and its effect on the WACC calculation.
2. Cost of Equity for TSOs in Colombia
Like most regulators in the world, CREG uses the CAPM model to estimate the cost of equity by adding a variable called country risk, as shown in Equation (3):
where
is the average risk-free rate at time
t, estimated as the average 10-year U.S. bond yield rate;
is the leveraged beta at time
t; is the average market premium at time
t and corresponds to the arithmetic average of the premium between the Standard and Poor’s 500 (S&P 500) index yield and the 10-year U.S. bond yield; and
is the average country risk premium at time
t and is determined as the difference between the average of the Colombian 10-year CDS and the average of the U.S. 10-year CDS.
To find the beta value of Equation (3), CREG uses the comparable beta method where CREG search for a set of U.S.-listed TSOs and calculates their covariance with the S&P 500. The resulting beta is then deleveraged based on Hamada’s equation [
9] and then leveraged again with Colombia’s financial structure. After finding the beta and multiplying it by the market premium, the average of the so-called country risk premium is added as a constant, which causes the rate to increase. Finally, to find an equivalent rate in Colombian pesos, CREG uses the Fisher equation. As shown, this procedure ignores the risk-return relationship of the TSOs listed in the Colombian Stock Exchange (BVC). This is because, for a long time, it has been assumed that there are few instruments and little information available in the Colombian market, which has led to the implementation of alternative procedures for calculating the cost of equity. However, to date, we have information from the last 14 years of the COLCAP index, which is the most representative index in Colombia, as well as information on electric utilities representative of Colombia and other macroeconomic variables that allow other cost of equity estimation models to be explored, which would produce alternative WACC calculations. The use of electric power infrastructure firms and local variables for the cost of equity estimation has recently been incorporated in emerging market papers [
10,
11,
12], where the need to find a new calculation methodology was also identified.
3. Literature Review and the Four-Factor Model
Despite advances in asset pricing models, regulators continue to use traditional models such as the CAPM to estimate the cost of equity [
13]. This model assumes that the idiosyncratic risks of an industry can be canceled out by efficient diversification, meaning only the market risk exists, which is common to all investments and cannot be diversified [
14,
15,
16]. This model has received two relevant criticisms in the financial literature. The first criticism is that of Roll [
17], who argued that it is not possible to efficiently measure market risk because there is no instrument that includes all investment classes. The second criticism is this model’s poor empirical results [
18]. The CAPM model has also been questioned for remunerating investments in electric power systems since the market risk factor is reduced because these investments are used to provide an essential service, where there is a natural monopoly and regulations that protect these investments to varying degrees in the face of changes in market price. This can be summarized as follows: for electric power infrastructure operators, market risk is low and individual risks are high [
19,
20,
21], which is contrary to the fundamental assumptions of the CAPM.
A different approach is presented by Chen et al. [
22], stating that investment returns can be explained by macroeconomic factor movement, as mentioned by Franc-Dabrowska et al. [
8]. A similar approach is popularized by Fama and French [
23,
24,
25,
26], who have shown that investment returns can be explained by multiple factors or characteristics. The most relevant criticism of these multifactor approaches is the difficulty in recognizing the risk factors that impact a given investment. For example, Harvey et al. [
27] found that at least 316 determinants had been identified in the financial literature, which is why this approach has been referred to as the “zoo of factors” [
28]. However, despite the difficulty in finding the determinant risk factors of an industry, in our literature review for the utilities industry (including TSOs), we found that in different articles published in the last 25 years [
29,
30,
31,
32,
33], interest rates have been identified as a determinant risk factor for explaining utility returns, since they are seen as a conservative investment that competes with fixed income instruments—when interest rates on these instruments increase, utility returns are negatively impacted. Thus, in addition to the market risk premium, we determined two macroeconomic factors that involve interest rates to explain the returns on Colombia’s electric utilities. The first is called
INT and corresponds to the risk premium between the rates of the 360-day and 90-day fixed-term certificates of deposit (CDTs) in Colombia. The second, called
DEF, corresponds to the spread between the interest rates of Colombian government treasury securities (TES) and U.S. Treasury bonds.
The other risk factor for the public utilities industry that we identified in the literature is regulatory risk, which has been discussed theoretically and empirically in different studies [
34,
35,
36,
37]. Since public utilities, including electric utilities, are natural monopolies, they are subject to different regulatory schemes such as cost-plus pricing, price cap, or revenue cap regulations that introduce different levels of risk in these companies’ performance. In cost-plus pricing regulation, there are few associated risks since all the firms’ operating costs are remunerated. In the revenue cap regulation, there is an approved revenue cap for operators that encourages them to be more cost-efficient to increase their profit margins. One aspect to take into account in this scheme is that, since there is a guaranteed revenue cap, network operators do not depend on increases in service demand, i.e., there is no demand risk. This is contrary to the price cap scheme, where the price for the service has a limit that cannot be exceeded. Therefore, the profit margins of firms under the revenue cap scheme depend on operating cost efficiency and service demand, i.e., there is a demand risk. Accordingly, a risk factor we propose is the risk premium between revenue cap and price cap regulation schemes.
According to this literature review, we propose estimating the cost of equity with the traditional CAPM [
14,
15,
16] using local variables for the Colombian market (local CAPM), shown in Equation (4), as well as with a new proposed model with four risk factors (TD-4), shown in Equation (5):
where
is the monthly return on the Colombian electric utilities portfolio at time
t used to estimate the cost of equity;
is the risk-free rate estimated as the monthly equivalent rate of the zero coupon rate of Colombian TES in pesos for a term of 1 year at time
t;
is the Colombian market risk premium at time
t and is calculated as the premium between the yield of the Colombian Stock Exchange’s (BVC) COLCAP index and the risk-free rate;
is the risk premium per regulation scheme estimated as the return of a regulated electric utilities portfolio under the price cap scheme and another regulated electric utilities portfolio under the revenue cap scheme;
is the interest rate risk at time
t and is calculated as the spread between the interest rate of the 360-day and 90-day fixed-term certificates of deposit (CDTs);
is the default risk estimated as the spread between Colombian TES rates and U.S. Treasury bond rates at time
t and is represented by the Emerging Markets Bond Indicator (EMBIG) for Colombia. The beta coefficients
represent the sensitivity of the returns on the Colombian electric utilities portfolio to risk factors. The coefficient α is the regression intercept expected to be zero, and
represents the residuals of the model at time
t. The hypotheses proposed with this model are:
Hypothesis 1 (H1). There is a significant positive relationship between the Colombian market risk premium and the returns on the electric utilities portfolio in Colombia.
Rationale. Only the market risk premium is relevant when explaining the returns on financial assets since the other risks can be diversified. Therefore, asset movements are a function of market risk [
14,
15,
38].
Hypothesis 2 (H2). There is a significant negative relationship between the regulatory risk premium and the returns on the electric utilities portfolio in Colombia.
Rationale. The returns of public utility firms regulated under a revenue cap scheme are lower than those of firms regulated under a price cap scheme because they are not exposed to the risk of service user demand [
35,
37].
Hypothesis 3 (H3). There is a significant negative relationship between the interest rate spread of 90-day and 360-day CDTs and the returns on the electric utilities portfolio in Colombia.
Rationale. Investments in electric utilities are considered conservative by investors and therefore compete with other low-risk investments, such as fixed-income instruments. Therefore, when interest rates on these fixed-income instruments increase, investors’ investment preferences change, and electric utilities’ returns are negatively impacted [
29,
30,
32].
Hypothesis 4 (H4). There is a statistically significant positive relationship between the spread of Colombian TES and U.S. Treasury bond rates represented by the EMBIG and the returns on the electric utilities portfolio in Colombia.
Rationale. An increase in the EMBIG represents a higher risk of default in emerging countries due to economic and political factors [
39,
40]. This risk is offset by higher investment returns.
6. Conclusions
In this article, we showed that in Colombia and other emerging countries that use the real pre-tax WACC methodology to establish the discount rate for TSOs, the discount rates are higher compared to those of developed countries. One of the causes of this situation is the way in which the cost of equity is calculated using the CAPM model with the comparable beta method. This study proposed the calculation of the cost of equity using a local CAPM model and a model with four risk factors called TD-4, where electric utilities operating as TSOs in Colombia were used, and the risk factors were constructed with local variables.
We found that the two proposed models were robust and statistically significant in all the proposed factors. However, the TD-4 model was able to explain 73% of the variations in the electric utilities portfolio returns, while the CAPM model explained 60% of the variations. Finally, we calculated the real pre-tax WACC with the cost of equity calculated using the two proposed models and found that both resulting values were lower than what was estimated using the regulator’s methodology. A higher discount rate to remunerate investments in energy transmission may affect the 14 million census households in Colombia, considering that spending on housing, water, electricity, and other fuels represents 29.4% of total household expenditures, being the highest item of household expenditures, according to the National Administrative Department of Statistics (DANE) [
43]. For example, for the year 2019, 19 energy transmission projects with an approximate investment of USD 291 million entered into operation [
44], with the rate recognized in 2019 by the regulator of 11.79%, the return on investment is USD 34 million, however, with the rates estimated in this study of 5.89% and 5.28% the remuneration should be USD 17 million or USD 15 million.
The results presented in this article can help formulate a new methodology for calculating the discount rate for power transmission activities involving companies and risk factors specific to the Colombian market, thus establishing a better rate for both users and power transmission companies. Future research could use the TD-4 model with local variables from other emerging countries with similar problems to calculate the cost of equity and could also go more in-depth in establishing the weighted cost of equity and cost of debt where there is generally no support for the established values.