Chapter 2 examines government credit support as a type of support to fossil fuels. It introduces a method elaborated by Professor Deborah Lucas of the Massachusetts Institute of Technology to quantify the support element of government credit assistance. Section 2.1 makes the case for why it is important to measure the support element of loan guarantees and concessional loans. Section 2.2 provides examples of government credit assistance granted by different types of financial institutions. Section 2.3 discusses how and why governments incur a cost when providing credit assistance to fossil fuel energy-related projects. Section 2.4 examines how credit assistance is reported in government budget reports and the implications of different accounting practices. Section 2.5 explains the method used to quantify the subsidy element of government credit support, both in theory and in practice. Section 2.6 provides real world examples of loan guarantees and derives the support element of the specific credit support program. Section 2.7 examines the value of credit assistance to the beneficiaries. The chapter concludes with remarks on how the method could be applied to allow the Inventory to incorporate information on government credit support.
OECD Companion to the Inventory of Support Measures for Fossil Fuels 2018
Chapter 2. The government support component of loan guarantees and concessional loans linked to fossil fuels
Abstract
2.1. Why measure the support component of preferential loans and loan guarantees?1
Governments play an important role in allocating financial resources and risk in the energy markets. Around the world, they provide support to investment in the production of fossil fuels or in the energy sector either through loan guarantees or direct concessional loans. In doing so, they increase access to credit or lower the cost of borrowing for the firm that would have otherwise been excluded from the credit market or penalised by higher interest rates.
When a government provides a loan guarantee for a project, it pledges to repay some or all of the outstanding amount to the lender should the borrower default. As guarantors governments pass the risk underlying such investment to taxpayers, who inevitably become de facto equity-holders in the project. In the event of a default, the government would have to pay back the loan by cutting spending or levying additional taxes to finance this expenditure; debt financing is another way to pay back losses in the short run, but issuing additional debt simply means pushing repayment into the future.
Direct government lending for energy projects is an alternative and widely used form of credit support with costs and benefits comparable to loan guarantees. As for loan guarantees, direct government loans provide support to investors through better contractual terms than those that would have been obtained on private markets, including through favourable interest rates, or repayment conditions. Recipients may, for instance, be granted the right to spread payments over a longer period of time or defer them until the end of the loan so as to maximise their earnings before covering the debt payments. Delaying the start of the repayment period can lower the likelihood of default as it pushes repayment further into the future. However, longer loan maturity can lead to a severer default because it increases total indebtedness and exposes the project to a longer period of uncertainty during which adverse events could occur (CBO, 2004[15]).
Government credit support, therefore, can result in a cost to the government that should enter into the evaluation of government support policies to fossil fuels. The potential subsidy cost of the loan guarantee or direct loans should be expressed in a way that can be compared with other support measures that have been included in the OECD Inventory of government support to fossil fuels.
Subsidised government financing support may take the form of a loan provided at non-market terms and conditions (e.g. with a below-market interest rates or with a tenor which is not available in the private market), a loan guarantee with a below-market credit risk premium, or simply even provision of a loan or loan guarantee that would not otherwise have been offered by a private entity. Such loans and guarantees may also be provided by official sources at costs which are commensurate with purely private financing (i.e. with no measurable support element).
In order to quantify the cost to the government of extending a credit or credit guarantee, one needs to determine the terms and conditions of financing that could have been provided by the private sector for the same transaction had the government not stepped in. The difference between what would have been paid by the debtor and what is actually paid is the subsidy cost of the loan or loan guarantee to the government. The challenge of finding an appropriate market pricing counterfactual, even in the absence of private funding, can be addressed using valuation approaches common in the private sector.
2.2. Examples of government credit assistance
Government credit programmes are generally provided through domestic, bilateral or multilateral financial institutions (mainly development banks), export credit agencies or majority state-owned banks. Examining some of the loan guarantees taken on by governments can provide a glimpse into the size and scope of the projects benefitting from this support measure. The examples below are grouped by the type of financial institution through which the credit assistance is granted. Since an estimate of the support value of the loan guarantee is generally unavailable, only the principal loan amount is provided here.
Multilateral financial institutions
Multilateral financial institutions account for the largest share of government credit support. Membership shares vary across institutions but tend to concentrate in large and high-income countries. Projects receiving the funds, however, are generally located in developing regions outside OECD countries. Table 2.1 lists the major multilateral institutions and the average amount of credit provided for fossil fuel-related projects in 2013-14 by them, as reported by the organisation Oil Change International (OCI). These figures represent the principal amount of the government-backed loan and not its subsidy component.
Table 2.1. Multilateral development bank finance for fossil fuels
USD million
Note: OCI report the total face value of the loan or the loan guarantee to fossil fuels as a subsidy and not the “support element” of the credit assistance.
1. Strictly speaking, the European Investment Bank (EIB) is not classified as a multilateral development bank. See: www.eib.org/about/partners/development_banks/index.htm.
Source: (OCI, 2017[16]).
Export credit agencies
Export Credit Agencies (ECA), present in most OECD countries, are agencies that provide support (in the form or loans, loan guarantees and insurance) for or on behalf of the government for the export of goods and/or services.2 ECAs can be private companies operating on behalf of the government or actually part of the government. ECA operations comprise direct loans and loan guarantees provided for the purchase of exported goods and services. Here, the guarantee issued by the ECA covers the repayment risks of the foreign buyer’s debt obligation.
Some ECAs already quantify the cost of their credit assistance. The United States’ Export-Import Bank (EXIM), for example, is obliged to define the subsidy cost of a credit based on the terms of credit and the estimated probability of default in line with the Credit Reform Act of 1990. The subsidy element of the credit support represents the share of the credit paid by the agency itself, while the remainder is borrowed from the US Treasury at interest rates based on Treasury securities of comparable maturity (EXIM, 2016[17]).
In many cases, the subsidy cost of the individual loan guarantees provided by EXIM is calculated to be negative or zero under this system; this reflects the fact that the fees collected for certain transactions more than offsets the estimated loss that is determined by the US budget scoring model, which does not incorporate a risk premium as a cost. With respect to other ECAs, recent years have witnessed steady positive returns with ECAs more than covering their costs and losses on a cash basis of accounting. Nonetheless, the terms and conditions of most ECA-financed transactions would likely be considered as “below-market” if judged according to a market pricing counterfactual approach.
State-owned enterprises
State-owned enterprises (SOEs) also often benefit from implicit or informal credit guarantees. A leading example is the Tennessee Valley Authority (TVA), wholly owned by the United States government and the largest wholesale supplier of electricity in the United States. Although the TVA relies on debt financing, its debt has consistently been rated AAA. This reflects the implicit guarantee from the US government on its debt obligations. Credit support through this channel may be classified separately from the traditional credit assistance programmes, but omitting this source of support from a comprehensive inventory would under-report the total costs borne by the government. Information on the financial support granted to SOEs is, however, mostly undisclosed. Out of necessity, the analysis in this report will focus on channels for providing government support other than SOEs.
2.3. The cost of government credit assistance
Since money today is more valuable than money tomorrow, the subsidy cost of extending a loan guarantee or a loan arises when the discounted value of the sum of future expected repayments and fees is less than the loan amount disbursed. To compare the value of money today with its value in the future, one would have to use the rate at which money loses its value over time and with risk; the value of money to be received in the future would have to be discounted using this rate (Scott, 2017[18]). Applying this concept to quantify the support element of an assistance direct loan, the sum of the expected cash flows from the repayment of the loan would must be discounted to represent their present value. Whenever the present discounted total value of the loan cash flows is lower than the principal value of the loan disbursed, the government that is backing or granting the loan incurs a cost equal to that difference in value.
Measuring the cost of a loan guarantee can be less straightforward than for a preferential loan because it is harder to infer the appropriate discount rate. However, with regards to the cost incurred by the government, a loan guarantee is conceptually not very different from a direct preferential loan because both expose the government to uncompensated default losses. A loan guarantee can either grant the firm access to credit or lower the borrowing costs in the same way that a direct concessional loan does. When a firm borrows at a lower cost due to government backing, the value of the loan guarantee amounts to the value of the savings from paying a reduced interest rate that does not fully compensate for the default risk. But to translate the future value of cash inflows into its present day, it is important to use the appropriate rate at which future “money loses its value”, i.e. the discount rate. The guarantee cost, and the implied discount rate in guarantee cash flows, can be inferred by comparing the value of an equivalent preferential loan and a risk-free loan with the same promised cash flows.
Elements needed to measure the cost of government credit assistance
To estimate the support component of a direct loan or a loan guarantee, the following loan contractual information is needed:3
principal amount issuance
interest rate charged on the loan
maturity
repayment terms: coupon payments if relevant, or projected cash flow from loan repayments
any other additional fees and costs
Non-contractual information is also necessary to estimate the support component. That includes default probabilities, expected losses in the event of a default, and the appropriate discount rate including a risk premium. That information is generally inferred from data on loans with similar attributes.
In order to arrive at a grant-equivalent of the loan guarantee or a direct loan, one needs to calculate: the discounted value of expected future cash flows of the loan and compare it with the principal amount borrowed. The exercise is to express the sum of future cash flows from the loan in terms of the present value as discussed in the previous section. Given the time value of money and the risk embedded in the loan, the future value of cash flows must be discounted accordingly. The discount rate represents the opportunity cost or the foregone earnings of loaning the money today as opposed to investing it; the greater the risk and thus the opportunity cost of the loan, the lower the present value of future cash flows, and eventually the higher the cost of the credit assistance.
Box 2.1. Calculating the cost of credit assistance to the government: An illustration
To illustrate how the cost of a direct loan and loan guarantee can be equivalent, one can start with a simple example.1 Assume that the government issues a direct loan of USD 100 at its own borrowing cost of 4%. The firm is expected to pay back the loan in one year's time with certainty. The government will receive USD 104 a year from now, which implies that today, the value of the repayment discounted by the government's risk-free rate would be:
where Vrf is the present value of a one-year maturity loan if the firm’s riskiness was equivalent to that of the government’s, in this case, it is the principal amount disbursed.
Let’s assume instead that the firm has a 30% possibility of defaulting on the loan, and in the event of default that the government could recover only 40% of the loan. Because of the default risk, the discount rate on the loan includes a risk premium of, for example, 0.005%, which increases the discount rate to 4.5%. The expected value of the loan repayment would be a weighted average of the amount repaid under a default scenario and the full repayment amount. This would be equal to:
Introducing repayment uncertainty reduces the expected amount that the government gets back; i.e. the value of the risky loan, Vr. The cost of the direct loan, Cdl, is the difference between the present value of the total repayment when the loan is expected to be paid with certainty and when the loan is expected to default with a positive probability. The difference between these two magnitudes is the cost of the direct loan to the government.
Similarly, should the government guarantee the same loan instead of disbursing it directly, the cost of the guarantee would be a weighted average of what the government would pay in the case of a default and when the firm fully pays the loan. In case of a default, the government pays the face value of the loan at maturity less the recovered amount and in case of a full repayment, the government pays nothing. The expected present value of the loan guarantee is thus,
This example introduces the concepts that are crucial to the subsequent discussion on the measurement of government credit assistance. First, it shows that risk-adjusted present value of a loan is an essential element for calculating the cost of credit assistance. Second, it illustrates that the cost to the government of a direct loan or loan guarantee, for the same underlying investment project, should be the same. In reality, loans are granted on much longer maturities than one year and the default and recovery rates are not always available. Also, the same discount rate is used in all three scenarios, implying that the only risk measures used here are the default rate and the rate of recovery. Other factors contribute to the riskiness of the firm and that information is usually captured by the firm-specific discount rate. Thus, the cost of credit assistance to the government hinges on the choice of the discount rate that best describes the firm's credit worthiness.
1. This is based on an example provided in (CBO, 2004[15]).
2.4. Practices in reporting loan guarantees
Governments and multilateral institutions implement different methods to measure and report costs related to credit assistance. These approaches can be categorised under three different methods using: 1) a cash basis approach; 2) an accrual approach using government borrowing cost; and 3) an accrual approach using a market interest rate as a discount rate. The first two approaches, while simpler to implement, suffer from serious limitations as they fail to capture the full extent of the subsidisation. The last approach is the most suited for deriving the cost of credit assistance, but it relies on firm-specific data that is not always available.
Cash basis
Governments using cash-based accounting only report the realised cash flows from the subsidised loan or loan guarantee in the year instead of reporting the cost incurred when the guarantee or loan was granted. By reporting the per-period payments made on the loan guarantee, the guarantee often appears to be profitable because fees are received upfront and the cost of the guarantee is deferred into the future until a credit-default materialises. Using this approach, a very risky loan or guarantee can be made to look less costly compared with a less risky loan or guarantee since the cash-based approach abstracts from the timing and uncertainty associated with a loan’s cash flows.
When comparing the cost of a comparable loan guarantee with that of a direct loan, the loan guarantee under this method would appear to result in a lower cost for the government, since the cost of the direct loan would appear at the time of the disbursement as the total amount of the principal extended,, whereas the cost of the loan guarantee would be reported only in the event of a default. Some governments exclude the cost of credit assistance from budgets to avoid these issues, but in doing so, they compromise their budgetary transparency as they understate the costs.
Accrual basis
Accounting practices that report the value of a loan on an accrual basis address the shortcomings of cash-based reporting. An accrual-based approach to estimating the cost of credit assistance uses the difference between the amount of the loan disbursed and the present discounted value (PDV) of the expected repayments and fees from the loan. In order to translate the value of the expected payments into their present value, the time value of money and uncertainty are accounted for. Two methods are used to derive the present value of the loan guarantee or direct loan: one is to use the government's borrowing cost as the interest rate with which to discount future loan-related cash flows; another is to use a discount rate that reflects the risk underlying the loan in addition to the time value of money. The latter reflects a market discount rate that investors would use for loans of similar risk.
An accrual-based method for valuing the cost of government credit support using the interest rate on government debt assumes that a government is fully able to diversify the underlying risk and therefore benefits from relatively low borrowing rates which can be used to measure its cost of capital. The question of whether this assumption holds true has given rise to several papers. While to some extent it is true that a government has greater capacity to eliminate its exposure to idiosyncratic risk, it cannot eliminate completely economy-wide uncertainty. Therefore, the government's cost of capital should reflect the time value of money and the market (non‑diversifiable) risk associated with the investment project (CBO, 2004[15]).
Using a government’s borrowing rate on government debt to discount the value of the loan guarantee lowers the cost of credit assistance relative to a more comprehensive cost measure because such a discount rate embeds only the borrowing cost for the government but not the investment-specific risk passed on to taxpayers. Financial economics provides several approaches to determine an appropriate firm-specific, project-specific or credit instrument-specific discount rate. The common thread among all the existing methods is the pricing of the riskiness of the firm. While the cash flows of a firm can be financed via equity or debt issuance, in the end it is uncertainty about total cash flows that underpins the discount rate for valuing the firm and not its financing structure (Tirole, 2006[19]).4
Similarly for the government, its discount rate for a guaranteed loan should not be tied down by its borrowing cost but by the riskiness of the loan. When a government shoulders the credit risk associated with an energy project and thereby provides the loan at a lower price, it incurs an opportunity cost because of the under‑pricing of the risk; government stakeholders, i.e. taxpayers, subsidise the loan. Following this logic, the market-risk based method proposed by (Lucas, 2017[20]) prevails as the preferred approach for estimating the support element of government credit assistance.5
2.5. Quantifying the support element of government credit assistances
Quantifying the support element of government credit assistance in theory
Deriving the present value of the loan under market pricing, or its fair value, requires three elements: a default rate, a recovery rate, and a market risk premium, in addition to the abovementioned contractual information.6 The default rate is the probability that the firm does not meet its repayment obligations, and the recovery rate is the share of the loan that the lender can get back in the event of default. The default rate and the recovery rate are linked and allow for the calculation of the expected value of future cash flows from the loan.7 The market risk premium is the component of the discount rate that represents the undiversifiable aggregate uncertainty; it captures the risk that is related to economic business cycles and aggregate changes in asset values. Consider a loan with the following features:
maturity T
full promised payment C
a default probability of d
a recovery rate g, and
a market interest rate r.
Using this information, the expected value of the loan guarantee is the difference between the value of the promised loan payments if they were risk-free and the value of expected loan payments taking into account default losses. The expected loan cash flow at a future time , if the loan has not already defaulted, is the weighted average between the recovered amount in the event of a default and the full promised payment with no default dtgtCt + (1 – dt)Ct. This expected cash flow, in , depends on the firm not defaulting up until now, for t – 1 periods, since the first disbursement of the loan. The probability that default did not occur thus far is expressed as the multiplicative term , where k – 1 represents the number of time periods that default did not take place. Lastly, the total expected repayment value must be discounted using the market interest rate. The fair value of the loan when all the information elements are available can be derived using the following expression,
The cost of the loan guarantee is then the difference between the risk-free value of the loan Vrf, i.e. when the default rate is null, and its fair value Vr, .
Quantifying the support element of government credit assistance in practice
Explicit firm or project-specific default probabilities, recovery rates, and market discount rates are not always available. However, if a firm’s debt is publically traded, its market price will reflect the market rate that private financial institutions required along with their beliefs about default and recovery rates. This discount rate is the yield or return the lenders demand for holding a risky asset in their portfolio, applied to promised cash flows. To calculate the value of the direct loan or the loan guarantee that would result from private lending, the appropriate risk-adjusted discount rate would be the interest rate charged by the private sector.
Information about a firm’s creditworthiness can be extracted from its credit-ratings. In practice, one candidate proxy is the credit rating of the energy-project sponsor (Lucas, 2017[20]). Rating agencies assign a grade to an issuer or a security to measure its credit worthiness (or likelihood of default), taking into account the borrower's risk-related factors: capital, cash flow, liquidity, capability, and at the firm's line of business. Since credit ratings and yield spreads are strongly correlated because they can inform on the default risk and recovery rate of a firm, one can use the firm’s credit rating to back-out the corresponding yield spread (the difference between the yield on a firm’s debt and the corresponding government rate) (Figure 2.1). The most precise credit-rating would be the one that is specific to the issued debt, but the firm's credit rating or the credit rating of a similar debt instrument – of equivalent magnitude, issuance date, and maturity – can be used as a proxy. Figure 2.1 illustrates the link between a given credit rating (e.g. from AAA to CCC) and its associated average corporate bond yield as calculated by BofA Merrill Lynch. It is evident from the graph that the relationship is monotonically decreasing, i.e. the lowest credit ratings (CCC) are related to the highest yields.
Figure 2.1. Corporate credit ratings and bond yields (in %)
Note: The chart above is based on quarterly US corporate bond yields from BofA Merrill Lynch for different credit ratings.
Source: (BofA Merrill Lynch, 2017[21]).
Since the yield on government bonds is usually readily available for different bond maturities, the yield on the firm’s debt can either be derived from the risk-free bond yield augmented by the firm- (or issue-) specific yield spread, or using a firm’s credit rating and applying the associated spread. The fair value of the loan can be derived by an equivalent expression of the present value of the loan cash flows using the risk-adjusted discount factor based on the firm-specific yield,8
The credit rating-based discount rate captures the average risk characteristics of firms that are assigned the specific grade. More explicit data on firm or project are preferred, but given the lack of disaggregation or the unavailability of data, this method benefits from its simplicity while approximating firm-specific risk sufficiently enough.
2.6. Real-world examples of loan guarantees and estimation of government support
The following employs three examples used in Lucas (2017[20]) to illustrate how government support is measured for different projects with differing levels of data availability.
US Export-Import Bank loan guarantee to Pemex
In July 2012, Petróles Mexicanos (Pemex), the Mexican state-owned oil company, issued USD 1.2 billion in bonds backed by the US Export-Import Bank to purchase US-made goods and services. The calculations of the support element from the loan guarantee imply that the support value is USD 206 million from the Ex-Im Bank.
Source of information
The subsidy estimate relies on information from public sources, including a press release from the US Ex-Im bank, information releases from rating agencies, Pemex's Form 20-F Report filed with the US SEC for 2012, and other media coverage. The available information is gathered in Table 2.2.
Contract terms: USD 400 million
The guaranteed loans reportedly have a ten-year repayment term that matches terms typically offered by other export credit agencies. The total USD 1.2 billion raised was spread across three separate Pemex offerings:
One note with a fixed interest rate of 2.0%, issued on 6 July 2012.
One note with a fixed interest rate of 1.95%, issued on 6 July 2012.
One note with a fixed interest rate of 1.7%, issued on 26 July 2012.
For the purpose of these subsidy calculations, and in the absence of information about repayment terms, the bonds are assumed to pay an annual coupon at the stated interest rate and to return the principal in a lump-sum at maturity.
Risk-adjusted discount rate
The subsidy value is calculated based on discounting the promised cash flows at a yield on comparable non-guaranteed bonds of similar maturity and risk. In this case, information is available on several other Pemex issues of a similar maturity that same year. The interest charged is a fixed rate ranging from 3.5% to 4.875%.
Another point of reference for the market yield spread comes from Pemex's credit ratings for foreign current offerings that were issued around that time by Fitch and S&P. Both agencies rated Pemex BBB, citing the strong backing from the Mexican government, which was rated AAA at the time. Moody's rated a recent Pemex issue as Baa3, and noted that its stand-alone rating without the implicit support of the Mexican government would fall to b3.
For the subsidy calculations, 4% is taken to be the market discount rate, based on several considerations: the ten-year US Treasury rate at the time of issuance was 1.65%. The BBB credit spread was 2.4%. The sum of the two is 4.05%, which is consistent with, although slightly higher than, the rates paid on the two direct loans issued at about the same time with similar maturities but without a guarantee. Direct loans often have higher priority in bankruptcy than do bonds, which may be a factor in the slightly lower rates charged.
The loan information from the first line of the table will be used in the following way:
The promised cash flows on the guaranteed bonds and the cost of the guarantee to the government are summarised in Table 2.3.
Table 2.3. Calculating the cost of the loan guarantee
Note: Besides the support component stemming from the US Ex-Imp Bank guarantee, there is an additional guarantee provided by the Mexican government to Pemex. As a state owned enterprise, Pemex benefits from a higher credit rating than what its standalone rating. A similar approach would be applied to estimate the Mexican government cost of support for Pemex's investment project.
Source: (Lucas, 2017[20]).
KfW loan to Electroprivreda Srbije for Kolubara project
PE Elektroprivreda Srbije (EPS) and German Development Bank (KfW), on the behalf of and with financial support of the government of Germany, signed a loan agreement in late 2012 for EUR 65 million and a grant of EUR 9 million to be used for the implementation of project “Energy Efficiency through Efficient Coal Quality Management in MB Kolubara”.
Contract terms
The total funds needed for the project were reported to be EUR 181.6 million. The European Bank for Reconstruction and Development (EBRD) supplied EUR 80 million and EPS committed to provide EUR 27 million from its own funds. Few details about the deal are publicly available. Nevertheless, the example is useful in illustrating the principles that would be used to value the support from KfW using information that certainly was known to KfW.
Risk-adjusted discount rate
Because EPS is a wholly state-owned company, its credit risk is at least as high as that of the Serbian government. That is because if Serbia defaults on international debt, that default is likely to include cessation of payments on EPS debt. There is the additional risk that if EPS were to experience large unanticipated losses, it could default on its debt even if Serbia honoured its other credit obligations.
Serbia was rated BB- with a negative outlook by Standard and Poor’s (S&P) in August 2012, and it reaffirmed that rating in March 2013. The rating is mapped to a discount rate by reference to yield spreads and taking the AAA European government bond rate as the base rate. In late 2012, the BB yield spread was 3.8%, and the B yield spread was 5.17%.9 Because Serbia falls into the lower range of BB ratings, and because the risk of EPS is likely to be higher than that of the Serbian government, the relevant yield spread is taken to be 4.5%.
The base yield to which the yield spread is added depends on the maturity of the KfW loan, which is not reported. The yields on AAA-rated 5-year, 10-year and 20-year bonds in late 2012 were 0.9%, 2.0%, and 2.8%, respectively.10
The subsidy also depends on the unknown interest rate charged by KfW. (Lucas, 2017[20]) shows the subsidy cost as a function of the yield spread charged on the loan and a ten-year-maturity of the loan. The reported subsidies are calculated by deriving promised cash flows based on maturity and assumed bond yield (yield spread charged plus base AAA rate), and discounting by the base AAA rate plus the 4.5% assumed market yield spread to find the value of the promised cash flows. The difference between the loan principal and the present value of the promised cash flows is the implied subsidy.
The value of the estimated subsidy ranges from USD 5 million to USD 28 million depending on how concessionary the interest rate charged and the maturity of the loan.11 Longer maturity loans entail higher subsidies because the below-market rate advantage is realized over a longer period. Development banks often provide longer-term financing, suggesting that the subsidies on the ten-year loans may be the most indicative of the true subsidy amount. Note that those credit subsidies significantly exceed the value of the EUR 9 million grant, which would have been the only subsidy accounted for under current practice (Table 2.4).
The loan information in the case that the interest rate charged is derived from the assumption that the yield spread used is 0.5% will be used in the following way,
The promised cash flows on the guaranteed bonds and the cost of the guarantee to the government are summarised in Table 2.5.
Table 2.5. Calculating the cost of a loan guarantee for ten-year maturity bonds
Source: (Lucas, 2017[20]).
2.7. The value of loan guarantees for the firm
To assess the net value of government credit assistance, there are several costs and benefits to be accounted for, such as the environmental and social externalities associated with energy projects (positive or negative), the impact of the credit support on market prices for loans, on investment patterns, as well as on the firm's own financing structure and leverage. A discussion of the cost of government credit assistance cannot be without mention of the benefits conferred onto the borrower that go beyond the access to credit or the reduced borrowing cost.
Investment projects in the energy sector necessitate large-scale long-term financing. Financial institutions cannot always accommodate the needs of such undertakings due to the lack of full information on the viability of the project or of financial constraints they face. The existing informational asymmetry, i.e. the firm disposing more information about its balance sheet and growth prospects than the lending institution, creates a friction in financial markets. Financial institutions end up confounding high-risk borrowers with low-risk borrowers and therefore mispricing their respective risk.12 In doing so, low-risk borrowers are penalised with high borrowing costs and high-risk borrowers benefit from loans terms that do not fully capture their level of risk. This informational asymmetry results in a misallocation of funds that excludes low-risk borrowers from the credit market and brings in higher-risk types. This market failure can be assuaged with government-backed financing through lower borrowing costs or increased access to credit.
The benefit of credit support to the recipient may go well beyond the value of the support element of the loan. The feedback of credit support on a firm’s credit worthiness can affect its future capacity to raise funds, known as the leverage effect. The debt granted via government support changes the capital structure of firm since the firm usually finances its investments through a combination of debt and equity. The private value to firms benefitting from government credit support is even harder to ascertain because it would depend on: a borrower’s particular tax status; its financial situation, including the profitability of the project and the borrower’s access to credit markets; leverage effects; and the competitiveness of the industry. These dynamic effects would not be reflected in an Inventory of support measures which has thus far only captured the revenue loss from granting the support.
2.8. Implications for the OECD Inventory of Support to Fossil Fuels
Government credit support to fossil fuel-related projects is pervasive and can result in inefficient allocation of public resources by locking-in long-lived carbon-intensive capital assets. Over the period between 2013 and 2015, G20 countries and multilateral development banks granted an average of USD 71.8 billion annually to fossil-fuel related projects (OCI, 2017[22]). According to preliminary estimates of the share of government credit support, it contributed to an additional support component ranging from USD 2.2 billion to 14 billion.
The cost of bearing the risk of granting credit to such investments results in a revenue forgone that could be quantified and integrated in the Inventory. In order to provide estimates on credit support akin to the estimates of tax expenditures and direct budgetary transfers that are included in the Inventory, two streams of information would need to be collected: information on the conditions associated with all government loans and loan guarantees, and information on the credit ratings of the firms or projects benefiting from such support and on the yield on government debt.
For information on the conditions associated with government loans and loan guarantees, a starting point would be to harness the information on the fossil fuel projects benefitting from credit support and the principal amount they received that has been collected by different institutions, such as Oil Change International (OCI). Finding loan specific information in a systematic way is no simple task. However, working closely with governments on disclosing information, and resorting to publically available data are a way forward to tackling this task and contributing to greater transparency on the use of public resources.
For information on the credit ratings of the firms or projects benefiting from such support, credit agencies, such as Standard & Poor's (S&P), Moody's, and Fitch Group, could provide such information, as well as the more consolidated databases, such as Datastream, which report credit ratings from the three aforementioned credit agencies. Datastream and other proprietary databases can also provide information on yield spreads that correspond to different credit ratings. The two data elements would be used to obtain the risk-adjusted discount rate and eventually the support component of credit assistance.
Expanding the Inventory in this direction could bring to the fore information on support to fossil fuel-related projects with long lifespans that emerged only because they were granted this type of assistance. Investment of the kind today can widen the existing infrastructure gap between what is needed to achieve climate policy objectives and the present situation (OECD, 2017[23]). Given the concerted efforts to decarbonise economies and move to less-environmentally-harmful energy sources, credit assistance directed to carbon-intensive infrastructure is incongruent with such efforts.
Several institutions have been taking stock of fossil fuels projects that benefit from government credit assistance, and the OECD could become part of this stream of work. However, given the intensity of effort and resources needed to gather the necessary data, the OECD would need to explore the options to conducting such work and assess whether it would be worthwhile pursuing. Data on government credit support is nevertheless an important element that shed light on government contributions to carbon-intensive infrastructure and to the risk of stranded capital assets. Work on gathering and reporting such information could provide a more accurate picture of the grant-equivalent value of government-mediated credit instruments than would information on the principal value of those instruments alone.
Notes
← 1. The OECD commissioned Professor Deborah Lucas from the Massachusetts Institute of Technology to develop a paper on quantifying the support element of government credit assistance, i.e. direct loans and loan guarantees, for fossil-fuel related projects. This chapter is in large part an abridged version of the work by Professor Deborah Lucas.
← 2. A complete list of ECAs can be found at: www.oecd.org/trade/xcred/eca.htms.
← 3. The discount factor is a crucial element to appraise the subsidy component of credit assistance, but it does not appear in the loan contract.
← 4. The assertion refers to the Modigliani-Miller Theorem, which abstracts from any market incompleteness or friction. While these assumptions have proven to be too restrictive, the main message is that capital cost is tied down by uncertainty and not capital structure (Tirole, 2006[19]).
← 5. Risk relevant to the government is non-diversifiable market risk. Governments have the ability to pool financial assets to reduce exposure to idiosyncratic risk, but the capacity to lower aggregate risk is limited. To price the risk of a government loan, it is important to use an appropriate measure to capture the relevant type of uncertainty, which in this case is aggregate economy-wide uncertainty.
← 6. The fair value of a loan is the price received if the firm were to sell or exchange the asset on the market.
← 7. Usually, the higher the default rate, the smaller the recovery rate.
← 8. V ̃ and r ̃ are the counterparts to the value of the loan guarantee and the discount factor using credit rating-based yield spreads that might not be specific to the project or issuance.
← 9. From the index value of option-adjusted spreads as reported by Bank of America Merrill Lynch.
← 10. Yield spreads are based on the ECB Euro area yield curves data.
← 11. Calculating the subsidy cost for each maturity (5-year, 10-year, and 20-year) and yield spread (0.5%, 1.5%, 2.5%) combination is detailed in (Lucas, 2017[20]). For illustrative purposes, support cost in this chapter is calculated for a ten-year maturity loan. The higher the maturity, the higher the cost of the loan guarantee.
← 12. Known as the “lemons problem” (Akerlof, 1970[25]).