COMPARATIVE ANALYSIS OF PROJECT
FINANCE IN ELECTRIC POWER SECTOR:
DEVELOPED COUNTIRES AND DEVELOPING COUNTIRES
By
INSUN JUN
MASTER THESIS
SUPERVISED BY Dr. RAFAEL MATTA
AMSTERDAM BUSINESS SCHOOL
MASTER IN INTERNATIONAL FINANCE
UNIVERSITY OF AMSTERDAM
2 Abstract
Project finance is to raise capitals to finance an economically feasible project that is intended to be repaid exclusively out of the project cash flow. Thus, project finance requires careful financial deal structure to allocate the risk and to benefit the involved parties in a manner that is acceptable to all parties. In this study, empirically, I found that electric power sector is leading the project finance in terms of deal frequency and deal amount (loan size). Dividing the data group into developing countries and developed ones did not change the finding in each group except for deal amount in developed countries where electric power sector placed in the second after transportation. Furthermore, I ran multivariate regression test with data exclusively from electric power sector to determine factors to influence loan pricing (spread). The analysis suggests that loan for the project located in developing countries with high country risk had significantly higher spread. Also, the guarantee from sponsors or host government, the involvement of multilateral bank/export credit agencies, and the number of loan arrangers were the most important determinants of the spread with negative relationship. Interestingly, loan maturity was also negatively related to the spread. However, the size of loan was not much significantly related to the spread.
3 Acknowledgement
I would like to express my sincere gratitude, first, to Professor Rafael Matta, who was my thesis advisor. His encouragement and suggestion were invaluable to finish my study.
I also wish to thank my wife, Hyeyun Kim for her endless support. Her help and concerns cheered me to finish this long journey. I also appreciate very much for my father and my mother’s valuable guidance with tenderness. Thanks a lot for everything, and I dedicate this thesis to all of them.
4
Table of Contents
Introduction ... 5 1. Literature Review ... 9 2. Methodology ... 12 2.1. Practical Approach ... 12 2.2. Empirical Approach ... 123. Project Finance Overview in Electric Power Sector ... 14
3.1. Typical Deal Structure in Electric Power Sector (IPP)... 15
3.2. Use of Power Purchase Agreement (PPA) ... 17
3.3. Attractiveness of Project Finance in Electric Power Sector ... 18
3.3.1. Smaller Equity Injection ... 18
3.3.2. Lower Chance of Default ... 18
4. Data ... 20
4.1. Sector/Country Group Distribution of All Project Finance Deals (Years 1990~2014) 21 4.2. Features of the Full Info Data in Power Sector (Years 1990~2014) ... 27
5. Regression Analysis ... 30
5.1. Regression Results ... 31
5.2. Interpretation of Regression Test ... 33
6. Summary and Conclusion ... 36
5 Introduction
It has been a while since private participation and financing increased its investment to infrastructure development project, which is one of the fastest growing sectors in financial industry. Especially, in parallel with economy booming in developing countries, it was inevitable to see fast growth in infrastructure financing because of the scale of funding in need. Interestingly, project finance scheme is often being used for this area.
Project finance normally follows the structure of setting up a Special Purpose Vehicle (SPV) for construction, operation and the administration of cash flow generated exclusively from its project assets. When it comes to raise funds to finance certain projects, lenders monitor both the cash flows of the project and assets of the project company (SPV) to determine whether or not the debt service through SPV can be made stably on time and SPV can provide proper security through its project asset for contingency.
Basically, SPV does not have any business history because it is newly established. Thus, it is impossible for lenders to make decision based on the credit check of the SPV on which the traditional corporate financing credit committee is based. In result, lenders, for their loan recollection, cannot but mainly count on the future profit to be generated from the project itself.
However, it is practically very unrealistic to arrange a huge volume of capital project without any guarantees (non-recourse type) from sponsors (equity investors) or the government because of traditional conservativeness of the lenders. Consequently, the sponsors or project host government commits to provide lenders additional safety net, such as equity as an upfront for risk or sovereign guarantee. We normally call this method as a limited-recourse financing and this is more typical way of project finance.
According to literatures, project finance is considered one of the most important financing vehicles for investment in infrastructure sectors such as electric power plants,
6 telecommunication, transportation, etc. (Esty, 2003). In addition, Kleimeir and Megginson (2001) argued that project finance has evolved into a very efficient, specialized vehicle to fund risky, capital intensive projects that generate predictable cash flows. Also, Gatti (2008) argued that project finance is now recognized as the most common method used in the private infrastructure project financing. According to global Project Finance Review Full Year 2012, the total volume of worldwide project finance capital market in 2012 has reached US$195.4 billion with 541 deal cases, 12.6% down from the US$223.4 billion in 2011. All of the deals are related to the infrastructure industry in one way or another. Now, we cannot imagine the infrastructure projects without thinking about the project finance any more. Why is the project finance so popular in infrastructure financing?
Furthermore, if we zoom in the infrastructure industry, it is categorized by sectors mainly as follow: Mining, Oil & Gas, Power, Telecommunication, Transportation, Water & Sewage etc. Among them, according to Project Finance Review Full Year 2012 again, electric power sector was the main area, U$64 billion (33%) of annual volume of project finance deal followed by oil & gas sector, U$60 billion (31%). Why is electric power project sector so distinctive in terms of utilizing project finance? With regard to financial terms, what is project finance in electric power sector different from between developing countries and developed countries?
On the other hand, Morrison (2012) mentions that “It is a general belief that MLA1/ECA2’s involvement in a project financing can produce a ‘halo’ effect over the commercial bank tranches that are not directly covered. With strong cross-default provisions, common in multi-source project financings, the default of debt service for a commercial lender will typically trigger a default with all lenders, thus bringing the weight of the MLAs and ECAs (and often their home country governments) down upon the defaulting project.”
1 Multi-Lateral Agency: IBRD, IFC, EIB, EBRD, AfDB, ADB etc.
7 How is this effect associated with loan pricing from the perspective of commercial banks when MLAs/ECAs partially participate in the deal together?
My research has started from these questions above at the beginning, and the interest has finally been narrowed down as follows after reviewing relevant literatures and retrieving my own experience in practice arranging project finance.
Firstly, I would like to find practical motivations why project finance is comparatively more favored in electric power sector among others.
Second and my main topic is to conduct the comparative analysis of project finance in electric power sector dividing the data group into developing countries and developed countries. I found a relevant empirical literature (Kleimeir and Megginson, 1998), which covered the entire sector of project finance in a broader scale with limited and old data collected in 1990s. I think that it would be interesting to update and compare the study by changing the comparison country boundary with more current and abundant data, specifically focusing on electric power sector. Relevant sub-topics are:
To run multivariate regression analysis (1) to judge which specific terms significantly influence project finance loan pricing [spread above LIBOR (London Inter-Bank Offer Rate) or other base lending rate], (2) to see how much the involvement of MLAs/ECAs affects the loan pricing and (3) to determine whether developing countries loans are priced differently from developed countries loans.
The thesis will be organized as follows. Section 1 will give short review of research literature on the relevant topics stipulated as above. Section 2 will describe the methodology of answering each topic. Section 3 will start with an overview of project finance in electric power sector and derive the possible reasons why project finance is used more broadly in this power sector. Although there are typical characteristics over general project finance deal such as setting up special purpose vehicle (SPV), it is crucial to understand couple of details,
8 specifically the deal structure to figure out why lenders and borrowers feel more comfortable and satisfied with project finance arrangement when it comes to electric power sector. Section 4 will cover descriptive explanation of data used for main topic. Section 5 will present its results based on multivariate regression analysis. Section 6 will summarize the study with conclusion.
9 1. Literature Review
Considering the practical understanding that project finance takes longer time to arrange and is more costly (mainly legal and financial advisory fees) to construct a financially and legally comfortable structure of independent project company than typical corporate financing, it can be a mystery why companies favors using project finance. While Nevitt and Fabozzi (2000) claim that “Project financing can sometimes be used to improve the return on the capital invested in a project by leveraging the investment to a greater extent than would be possible in a straight commercial financing of the project”, Esty (2003) points out that it cannot explain the reason perfectly, especially because this approach is not enough to explain the fact that higher leverages go in parallel with the increases of equity risk and expected distress costs, say, the tradeoff relationship. Esty (2003) argues that the project finance is pursued because it gives the solution of two important financing problems: “1) the reduction of agency cost of debt among project companies and 2) the decrease of opportunity cost of underinvestment due to leverage and incremental distress costs from the perspective of project sponsoring companies”. John and John (1991) also analyzed the idealistic features of project finance structure which reduces the agency costs of debt with similar result. Smith (1980) argues that the legal safety net such as the pledge to escrow account and other restrictive covenants are efficient mechanisms to control the incentive conflict, and the benefit goes to the borrower by lowering the loan pricing. Moreover, although the fundamental approach of project finance is the same in developed and developing markets, it might be also possible that considerable variation exists in terms of average deal size, funding costs, number of corporate sponsors, credit ratings, reliance on banks versus bond markets, cost of insurance, etc. (Morrison, 2012).
10 partial derivative for the expected rate with respect to the size of the loan. Scott and Smith (1986) find that the loan size increases default risk. However, it seems that Time to maturity is arguable because other loan pricing studies have found both significantly positive and negative correlation for loan size. Scott and Smith (1986) find a significant positive impact while Booth (1992) finds the opposite. Interestingly, according to Kleimeier and Megginson (2001), two variables emerge as potentially important measures (positive relationship) in case of general loan financing: loan size and time to maturity. However, it does not seem to apply in the same way when it comes to project finance. From their empirical study, they find that the longer are the maturity of project finance loans, the lower is the risk of financial distress, which ultimately leads lower loan spread. On the other hand, Bonetti et al. (2010) points out a different perspective and approach arguing that the funding cost can be saved through efficient contract structure, appropriately allocating most of the inherent risk factors to external parties. Therefore, “The pricing of project loans is dictated by the sole residual component of risk, i.e. counterparty risk, such as the risk of off-taker default”. Altunbas & Gadanecz (2004) analyzed the decisive factors of the pricing of 5,000-plus syndicated loans mobilized to developing country borrowers between 1993 and 2001. The study shows that riskier borrowers make syndicated loans more expensive than others which is very natural, but the effect of purely microeconomic factors is sometimes weaker when macroeconomic conditions are controlled. Gadanecz & Sorge (2008) argue that “The term structure of credit spreads in project finance is hump-shaped. This contrasts with other types of debt, where credit risk is shown instead to increase monotonically with maturity ceteris paribus.”
On the other hand, Kleimeir and Megginson (2001) argue that project finance loan prices are very strongly correlated with the existence of currency risk, the presence of a loan guarantee by the project sponsor, host country, or MLAs. According to Morrison (2012), “One of the main financial products of MLAs is the A loan/B loan structure, in which MLA
11 funds the A loan and syndicates the B loan to commercial banks. The MLA is the ‘lender of record’ for the B loan and all loan payments are shared pro rata between the MLA and the B loan lenders. A borrower cannot default on a B loan without also defaulting on the A loan, thereby jeopardizing the host country’s relationship with the MLA.”
Hainz & Kleimeier (2006) comments that “The political risk can be mitigated through a loan contract, i.e., through the inclusion of multilateral or national development banks in the syndicate.” The effect can be explained through the fact that the government running out of the development budget counts on the capital from foreign sources, and if the government defaults the loan from MLAs/ECAs, then, it means the government won’t be able to receive any further international loan due to the worst reputation in international capital market. Therefore, MLAs/ECAs are also known as political umbrellas (Buljevich and Park, 1999).
12 2. Methodology
2.1. Practical Approach
Starting from the understanding of specific characteristics of project finance in electric power sector and differences from others in different sector, I will try to derive a rationale in practical viewpoint why several factors make the project finance of electric power sector distinctive from others. The main goal is to share the ideas in practice regarding the popularity of power sector in project finance, so this topic can be studied further in detail and be empirically proven later.
2.2. Empirical Approach
Firstly, I will check the position of power sector in overall project finance deals and the comparative features of project finance terms in power sector between developing and developed countries using the empirical data collected for this study.
Secondly, in order to find the determinants of loan pricing for electric power project finance and to check whether the risk factors of developing countries project versus developed one are reflected into the loan pricing, I will run the multivariate regression test. In the previous study, Kleimeir and Megginson (1998) meant “Guarantee” as a guarantee from the host government, the sponsors, or a MLA/ECAs. According to the study, the result of regression test showed significantly high negative coefficient for the “Guarantee” variable compared with other independent variables. I am going to split this term into two (2), one with any guarantee from sponsors and/or from host government and the other with “Involvement” of MLA/ECAs. The involvement means that either MLA/ECAs participated in a direct partial loan provider or they offered a partial guarantee, so that I can see the effect
13 of MLA/ECAs separately this time. Also, I am going to take the number of loan “arranger banks” into consideration as an additional regression independent variable. I expect the number of arranger banks will affect the loan pricing significantly because more participation of loan arranger banks can result in total risk being shared one another so that the risk taken by the individual banks will be ultimately reduced.
Table 1: Expected Regression Variables Used in the Loan Pricing Test
This table presents expected regression variables used in analysis. Dependent variable will be Spread, and others will be independent variables. “Guarantee”, “Involvement” and “Developing” terms are dummy variable (Yes:1, No:0). “Developing” term is additionally included at the second test to check whether country classification is reflected into the loan pricing as a risk factor.
Variable Name Description Intuitive Expected Sign Spread Loan contract rate minus base rate N/A
Size Loan amount in millions of USD Positive Maturity Years to maturity Positive Guarantee
Guarantee from Sponsors and/or Host Government
(Dummy variable, Yes:1, No:0)
Negative
Involvement
Involvement from
MLAs/ECAs/Central/Development Bank
(Dummy variable, Yes:1, No:0)
Negative
Country Risk Country risk grade from OECD
If risk is high, then index is high Positive
Year Financing closure year Not sure Number of Arranger (Banks) The Number of Participation Banks
for Syndication Negative
Developing
Distinction of Sample Group: either developing or developed
(Developing:1, Developed:0)*
14 3. Project Finance Overview in Electric Power Sector
With the basic understanding of project finance from introduction and literature review here above, I would like to go further over general idea of project finance in electric power sector. The fast economic growth in many developing countries in recent era, has naturally incurred the increase of demand at home and in industry for electric power to run the economy efficiently. However, the rapid growth in demand has placed a considerable burden on the power industry, especially due to the shortage of power plants. In recognition of this, governments have started restructuring their power industry through privatization and liberalization and considered the Independent Power Producers (IPPs) as a way of attracting new domestic and foreign investment into power generation. The governments of these developing countries started looking for alternative sources of funding , and the concept of IPPs was very fit to this need where the general project finance scheme can be well utilized with more favorable structure for lenders and borrowers. More details about IPP structure is followed in the following sub-section.
3.1. Typ Compan the Pow the Loc structur where a SPV is shareho 3 Referre 4 O&M: 5 EPC: E pical Deal S The princi ny (Borrow wer Purchas cal Governm re does, the
all the relev set up in th olders in the ed to Peppiatt Operation an Engineering, P Structure in Figure 1. [S ipal parties wer), the Sh ser, the Len ment as we e Project C vant parties he country i e Project C (1995) nd Maintenanc Procurement a Electric Po . Deal Struc Source: Milb of electric hareholder ( nders, the C e can see fr ompany (S s form cont n which the Company m ce and Constructi ower Sector cture (Proje bank Intern power proj (Owners/Sp Contractor (E
from the Fig SPV) is the tractual and e power stat may include ion (IPP)3 ct Finance i nal Documen ect finance ponsors), the EPC5 Cont gure 1 abov essence of d financial r tion will be the EPC C in Power) nt] broadly co e Operator ractor), the ve. As typi f the projec relations wi e constructed Contractor ( onsist of the (O&M4 Pr Fuel Supp ical project ct finance s ith SPV. N d. Interestin (who will b 15 e Project rovider), lier, and finance structure ormally, ngly, the build the
16 Power Station), the Operator (who will operate it once completed), a local partner or partners, and the Power Purchaser (either the local public utility company or a large local business owner who agrees to buy the generating capacity of and electricity from the completed Power Station) which can be shown to have a conflict of interest issue. However, this approach can be also viewed as a commitment of each participant for the success of project and ideal risk allocation. With this typical project scheme in power sector, the project finance, as commented before, takes a crucial role in developing a power plant. In order to obtain the proper financing, the Project Company will enter agreements with various Lenders. These Lenders may include bilateral and multilateral lending agencies (MLAs), export credit agencies(ECAs), development bank and commercial banks, both international and domestic.
On the other hand, the approval from host government can be another important factor for the successful implementation of power plant projects, especially in developing countries, which in most cases will be asked to offer a certain support (at the local or national level, or both) to the project company and possibly to the Shareholders and Lenders. Although the type of governmental support will depend on the particular project, governmental support may take the form of stand-by Letter of Credit, currency exchange guarantees, exemptions from valued added tax or withholding taxes, etc.
The SPV itself also must take care of the operation of the Power Station including the securement of adequate fuel supply for the Station. For this purpose, the SPV may make various types of arrangements with the Power Purchaser. For instance, unless the SPV is not comfortable about sufficient fuel supply for the Power Station on a “spot” basis (i.e. buying fuel in the market when required) to satisfy the obligation (electricity supply) under Power Purchase Agreement (PPA), it will contract with a Fuel Supplier for a guaranteed supply in advance using futures market. An alternative is a “tolling” arrangement which oblige the Power Purchaser to take the responsibility for fuel procurement by obtaining and delivering
17 the fuel to the Power Station and for settling payment. Because the concept of Power Purchase Agreement (PPA) is very important to understand the popularity of power sector in project finance, more details follow in the next sub-section.
3.2. Use of Power Purchase Agreement (PPA)
A Power Purchase Agreement (PPA) is often at the core of bankability for a power project finance. Similar contractual tool is a “take or pay” or “tolling” contract. Under PPA, SPV typically becomes the seller for electricity, and a utility company takes the role of the buyer, which will often be either owned or guaranteed by the government. The most important consideration factor when reviews the PPA is the “tariff,” paid for supplied energy under the agreement which must be more than enough to cover both fixed and variable costs including debt service, fuel costs, operation and maintenance expenses. In most cases, PPA puts the responsibility to procure fuel under the Seller, but sometimes in several countries where only a single, national utility company exists, the Purchaser takes responsibility for fuel supply. When the Power Station is implemented successfully with the start of revenue generation, the Lenders must make sure that the PPA remains in force during the entire term of the project loan and that the risk such as force majeure or other relevant adverse events is stably shared. Typically, PPAs are long-term, usually ranging from an average of 20 to 25 or 30 years. Due to the well-proven PPA contract structure having the government owned utility company as a Purchaser or with a certain back-up guarantee from government, lenders feel very comfortable about the repayment of loan from borrower and consequently let lenders consider increasing the ceiling of leverage rate and extending the maturity, more favorable conditions from borrowers’ viewpoint. Thanks to this relationship, we can expect that power sector is taking the main position among other sectors in terms of utilizing project finance scheme. More details about these benefits of using project finance in power sector are
18 covered in the following sub-section. Also, the empirical data collected for this study also support this expectation. More detailed analysis will be explained later in empirical section.
3.3. Attractiveness of Project Finance in Electric Power Sector
Electric Power Project Financing has the common attractiveness with the one from other sector such as None or limited recourse financing, Off-balance sheet debt financing etc. However, the most distinctive features are the followings.
3.3.1. Smaller Equity Injection
When it comes to power sector in project financing, SPV with support from sponsors generally can achieve higher loan to equity ratio compared to the one for other sector project finance still maintaining the full control of the SPV without an equity dilution. In the power sector, the typical equity contribution goes only between 20% and 30% and sometimes even 10% is allowed. Other sectors normally go above 30%. The main reasons for this phenomenon can be found as discussed above from the use of Power Purchase Agreement or Fuel Feedstock Agreement signed with host government or its equivalent party in long-term basis which goes around more than 15 years. This contract based long-term stable revenue generation structure let lenders feel the certainty on borrower’s debt service although force majeure events such as expropriation or natural disasters sometimes stop the general function of the contract umbrella. For this contingent situation, other safety measures such as insurance coverage are packaged together with the contracts.
3.3.2. Lower Chance of Default
service embrac energy credit r undevel lenders governm more ea transpor The rev fee coll revenue during d on time. F ed with PPA is state-ow risk goes sig
loped coun have mor ment throug asily compa rtation sect venue sourc lection when e after launc due diligenc Figure From this p A will show wned utility gnificantly ntries. Even re room to gh settling d aring to the tor. The fig e is the fee n the loan a ching the to ce. 2. Typical D erspective, w the low vo company o lower altho n in worst o recoup th down proce structure u gure below that road us approval wa oll road, ant
Deal Structu Sour power proj olatility of c or the state ough there scenario w heir repaym ess. We can used in othe shows the sers pay. We as under rev t its uncerta ure in Trans rce: World B ject under t cash flow. E itself as po are certain when the Pu ment direct n find the di er sectors an typical dea We can easily view would inty leads th sportation ( Bank the scheme Especially, i ower projec exceptions urchaser ca ly negotiat ifference of nd I brough al scheme o y imagine th be very diff he lenders t Toll Road S e of project in case the b ct is such c when it go annot pay o ting with t f financial s ht an examp of toll road hat the proje fferent from to be more System) 19 finance buyer of case, the oes with on time, the host structure ple from project. ection of m the real cautious
20 4. Data
Project level data used in this study have been obtained from the Thomson Reuters database, provided by Thomson One. Thomson Reuters database provides details on projects that have successfully achieved financial closure on project financing basis. Project information available in the Thomson Reuters database included sector, funding cost, project cost, loan value amount, domicile country, maturity period, MLAs/ECAs involvement, date of financial closure, and the sponsors of the project etc. The sample period considered to choose for this study is from 1990 to April 2014.
Basic data mainly consists of two parts. The first is the whole available project finance data from database without any specific search condition including power sector during the certain period as mentioned above to check descriptive statistical information for general idea. Some of data do not have all financial information due to its early stage such as “planning” or “financial closing not completed”. The second is the project finance power sector data which has full information to run the comparative and regression analysis for the loan pricing study in power sector. The second data will be grouped into two by country, one is developing countries and the other is developed countries based on the national GDP (nominal) per capita with US$30,000 as a boundary.
Table 2: Analyzed Sample Data
This table presents two groups of samples which were analyzed in this study. Sample 1 includes all available data from database from years 1990 ~ April 2014 while Sample 2 include among Sample 1 only the ones having full information in need for regression analysis such as loan maturity, spread etc. focusing on power sector. * These data excluded the canceled, inactive and unapproved case.
Data Description Available Data (1990~2014) Sample 1 All Project Finance Deals 20,571
21 All Project Finance Deals Sample: This is a collection of [20,571] project finance loans from the period [1990 – 2014]. Among all collected data, only [6,666] had actually been financially closed at the time they were selected, while others were in the planning stage when they were gathered. This sample is used only to show the sector and geographic distribution comparison of project finance loans since 1990.
Full Info Data in Power Sector: This is a set of [982] project loans out of all available data [2,221] in power sector – [304] from developing countries and [678] from developed countries – from the sample described above for which complete information about the loans’ pricing and other features could be collected.
4.1. Sector/Country Group Distribution of All Project Finance Deals (Years 1990~2014)
Table 3 and Figure 3 details the industrial sector distribution of all project finance deals sample (Sample 1). Table 3 shows that project finance loans are highly concentrated in three key industries by deal amount: Power, Transportation and Oil & Gas. This result is very analogous to the result of study by Kleimeier & Megginson (2001) which pointed out the five major industries in project finance: communications, mining & natural resources, oil & gas, electricity and energy utility, and transportation. However, we can see the difference from the portion of power sector. In this study, power sector leads the league taking 28% portion by deal frequency and amount respectively while in Kleimeier & Megginson (2001) study, power sector accounts for only 21% by deal amount. This finding can be understood to imply that power sector is (1) getting more popular to use project finance scheme due to its robust financial structure and is (2) growing very fast to satisfy the demand caused by rapid speed of economic development in developing countries.
22 Table 3: Project Finance Deal (sorted by deal amount)
This table presents the number of deal frequency and deal amount by sector by analyzing the all available data from database from years 1990 ~ April 2014.
Sector Deal Frequency Deal Amount (Mil.US$) Power 7,726 4,767,537
Transportation 3,757 4,247,162 Oil & Gas 2,144 3,106,314
Industry 1,403 1,442,895 Leisure & Property 2,377 910,462
Petrochemical 398 865,667 Mining 1,127 851,489
Telecommunication 588 543,726 Water & Sewage 793 396,366
Waste & Recycling 239 44,593 Agriculture & Forestry 19 19,658
On the other hand, Table 4 and Figure 4 suggests that [Developing Countries] (with [11,223] loan frequency) is attracting more attention from loan arrangers than is the [Developed Countries] (with [9,348] loan frequency). Considering the deal amount, the difference is more distinctive. In developing countries, the total deal amount is 10.4 Trillion US$ while it is about 6.8 Trillion US$ in developed countries. This finding suggests that developing countries market has increased its share of project finance markets which is explainable through the fact that more infrastructure projects are announced in developing countries in parallel with their economic growth rate. Also, taking a look at power sector separately, the finding shows that power sector is taking 1st major portion of project finance deals in both developing (35%) and developed countries (41%) by deal frequency. Interestingly, however, deal amount presents that transportation is slightly higher than power (27% and 26% respectively) in developed countries while power is again ahead of transportation in developing countries (29% and 23% respectively) which might imply that the demand for power in developed countries are getting saturated due to slow economic growth rate or the understanding to have to save energy is more spreading among the people in developed countries to protect the environment.
25 Table 4: Project Finance Deal (Developing/Developed Countries)
This table presents the number of deal frequency and deal amount by sector and also by project country group (developed/developing) analyzing all available data from database from years 1990 ~ April 2014.
Sector\Geography Developed Developing Deal Frequency Deal Amount (Mil.US$) Deal Frequency Deal Amount (Mil.US$) Power 3,837 1,762,194 3,889 3,005,343 Transportation 1,422 1,852,322 2,335 2,394,840
Oil & Gas 755 1,160,153 1,389 1,946,161 Industry 424 487,898 979 954,997
Leisure & Property 1,791 550,740 586 359,721 Petrochemical 87 102,401 311 763,266
Mining 359 314,285 768 537,204 Telecommunication 235 311,843 353 231,882
Water & Sewage 279 182,140 514 214,225 Waste & Recycling 151 28,561 88 16,032
Agriculture & Forestry 8 15,044 11 4,614 Total 9,348 6,767,584 11,223 10,428,290
4.2. Features of the Full Info Data in Power Sector (Years 1990~2014)
Table 5 with sample size of 982 having full information (306 and 678 in developing and developed countries respectively) out of 7,726 all available project data in power sector presents mean values – as well as minimums, maximums and standard deviations – of the following loan characteristics only focusing on power sector for the Developing and Developed countries sub-samples: (1) loan spreads (above LIBOR or another base lending rate); (2) financial closing year; (3) loan maturity in years ; (4) the number of participating banks (arrangers) for the project being financed; (5) the loan size; (6) whether or not loan repayment had a full (or partial) guarantee from the host country government; (7) whether or not MLA/ECAs or development bank are participating as a loan arranger or partial guarantor, any direct involvement from them and (8) country risk defined as an index from OECD of which the lowest risk country is described “1” and the highest one “7” (In case of the official members of OECD, the index is not defined because they are technically considered with least risk, so I categorized them with index “1” for study purpose).
These values suggest that loan spreads and maturities are slightly different in one group while developing countries loans have a loan maturity of 11.09 years versus 9.91 years for developed ones.
It appears to be substantial differences between developing and developed countries loans with respect to mean spreads, financial closing year, and the number of arrangers, loan size, the frequency of a guarantee being offered, and the frequency of the involvement from MLA/ECAs etc. Mean spreads are 241.11 basis points above reference rate in developing countries and 193.49 basis points in developed ones. We can infer from this finding that lenders take the developing countries risk into account when they decide an appropriate spread. On average, developed loans were booked more than three calendar years after developing countries loans (July 2006 versus August 2003). This earlier average booking date
28 for developing countries loans is mostly the result of developing countries’ recent preference on project finance scheme. It is also true that developing countries projects average 3.3 bank arrangers each, while developed countries projects on average have about 4.7 arrangers, and those developing countries loans tend to be much smaller than those booked in developed countries (means of USD 187.75 million versus USD 218.25 million) . We can understand this fact that lenders would like to share total risks allocating them to more participants when the loan size increases, so the individual lender’s risk taken can be decreased one another. Also, guarantees are used in 5 percent of the developing countries loans versus in 18 percent of developed countries projects. This finding seems to be surprising because the guarantees normally used to protect the default risk, and we can expect that the project in developing countries might have more risks than the one in developed countries. However, we might be able to find a rationale for this fact from different status of utility company in each country groups. In case of developed countries, they have started its privatization from early eras and we can see that more utility companies in developed countries have been privatized so it is necessary to have a certain guarantee while most of developing countries still run their utility companies by themselves, which might have the equivalent effect to the guarantee. Finally, Involvement was made in 13 percent of the developing countries loans versus in 7 percent of developed countries projects. We can imagine this result from the mission of MLA/ECAs who wants to support the economic development of poor countries. Country risk is average 3.26 in developing countries versus 1.91 in developed countries.
29 Table 5: Characteristics of The Full Info Data in Electric Power Sector
This table presents Mean, Standard Deviation, Minimum and Maximum value for several features of data having full information for regression analysis specifically in power sector. Country Risk index follows the table used in OECD.
Characteristics Value Loans to Developing Country Loans to Developed Country
Sample # 304 678
Spread (basis point) Mean 241.11 193.49
St. dv. 231.88 111.52 Minimum 1.5 12.5 Maximum 1375 887.50 Financial Closing Year (1990 ~ 2014) Mean 2003.70 2006.59 St. dv. 5.66 4.56 Minimum 1993 1991 Maximum 2014 2014
Maturity (year) Mean 11.09 9.91
St. dv. 5.06 6.45
Minimum 0.438 0.192
Maximum 21.86 27.44
Bank Arrangers # Mean 3.26 4.69
St. dv. 2.79 4.33
Minimum 1 1
Maximum 16 32
Loan Size (Mil. US$) Mean 187.75 218.25
St. dv. 205.86 294.65 Minimum 1.31 0.234 Maximum 1080 1,586.85 Guarantee (Yes:1, No:0) Mean 0.05 0.18 St. dv. 0.21 0.39 Minimum 0 0 Maximum 1 1 Involvement (Yes:1, No:0) Mean 0.13 0.07 St. dv. 0.33 0.26 Minimum 0 0 Maximum 1 1 Country Risk (Low:1~High:7) Mean 3.26 1.91 Minimum 1 1 Maximum 7 3
30 5. Regression Analysis
Having reviewed that developing and developed project finance loan in power sector are funded with different characteristics, we now conclude our empirical analysis by examining the factors that impact the pricing of the loan spreads. For this purpose, I have assumed a model of loan pricing that specifies the loan’s pricing (interest rate spread) as a following equation (1) based on the literature review and practical intuition.
The dependent variable is the loan spread above LIBOR, in basis points, and the independent variables are those presented and discussed in Table 5. We employ standard OLS regression estimation techniques and adjust for heteroskedasticity using the methodology proposed by White (1980). The model estimated is:
Spread = α + β1 Maturity+ β2 Size + β3 Close_Year + β4 Country_Risk +
β5 Arranger + β6 Developing + β7 Guarantee + β8 Involvement --- (1)
Where:
Maturity = Loan maturity, in years; Size = Loan size, in US$ millions;
Close Year = Financial Close Year; Arranger = Number of Arranger Bank for funding
Country Risk = OECD Country Risk Index6 (Risk Low ~ High = 1 ~ 7, OECD Members
classified as lowest risk index 1 for this study7)
Developing = Dummy variable taking the value of 1 if a loan was funded to developing
countries (GNP per capita US$30,000 below) and 0 otherwise;
Guarantee = Dummy variable taking the value of 1 if a loan has a guarantee and 0 otherwise; Involvement = Dummy variable taking the value of 1 if MLA/ECAs and development banks
are participating to the loan as a lender or partial guarantor and otherwise 0
6 http://www.oecd.org/tad/xcred/cre-crc-current-english.pdf
31 5.1. Regression Results
The following discussion of independent variables and their expected impact on the spread is summarized in Table 6. The dependent variable, the loan spread, is defined as the contract rate minus the base rate. The currency in project finance in power sector is the US dollar and the data collected from Thomson database has been already converted into dollar amounts if necessary for comparison purpose.
Since one key objective of this loan pricing analysis is to determine whether developing countries have higher loan rates than developed countries, after all other risk factors are controlled, I included a relevant specific dummy variable named “Developing”. If the estimated coefficient of this dummy variable is positive, we can conclude that developing countries loans include a developing countries risk premium in their spreads which is priced. The first regression includes all of the pricing variables but not the “Developing” dummy variable, the second regression includes all of the pricing variables to compare whether the country group (developed/developing) the loan is booked in impacts its loan price. The third column includes all of the pricing variables but not the “Involvement” variable to check how much “Guarantee” variable affects the loan pricing without “Involvement” variable and the fourth regression includes all of the pricing variables except for “Close Year” variable to check how much the “Close Year” means to the pricing. Finally, the fifth column excludes only “Size” variable.
32 Table 6: Regression analyses of the determinants of project finance loan spreads (loan pricing)
This table presents the results of an ordinary least squares regression analysis of the determinants of loan pricing spreads for project finance in power sector based upon data availability. The first regression includes all of the pricing variables but not the “Developing” dummy variable, the second regression includes all of the pricing variables to compare whether the country group (developed/developing) the loan is booked in impacts its loan price. The third column includes all of the pricing variables but not the “Involvement” variable to check how much “Guarantee” variable affects the loan pricing without “Involvement” variable and the fourth regression includes all of the pricing variables except for “Close Year” variable to check how much the “Close Year” means to the pricing. Finally, the fifth column excludes only “Size” variable. T-statistics in parentheses are based on heteroskedasticity robust errors according to White (1980) and reported in parentheses. *, **, *** indicates that the reported coefficient is significant at the 100%, 5%, 1% level, respectively.
Regression 1 2 3 4 5 Number of Observations 982 982 982 982 982 Explanatory Variables Intercept -22,554.53 (11.42)*** -23,547.93 (11.39)*** -23,548.25 (11.37)*** 228.87 (17.97)*** -23,650.70 (11.25)*** Maturity (years) -2.79 (4.03)*** -2.99 (4.46)*** -3.10 (4.66)*** -4.03 (-5.82)*** -2.92 (4.29)*** Loan Size (US$) 0.02 (1.07) 0.03 (1.17) 0.03 (1.12) 0.04 (1.44) Close Year (1990 ~ 2014) 11.35 (11.54)*** 11.85 (11.49)*** 11.85 (11.48)*** 11.90 (11.35)*** Country Risk (1 ~ 7) 26.59 (6.22)*** 13.23 (3.19)*** 12.94 (3.15)*** 5.98 (1.42) 13.00 (3.14)*** Arranger # -6.75 (4.90)*** -5.75 (4.49)*** -5.99 (4.62)** -3.06 (2.60)*** -5.25 (4.87)*** Developing (Yes:1, No:0) 60.22 (3.82)*** 59.02 (3.78)*** 43.43 (2.71)*** 59.75 (3.81)*** Guarantee (Yes:1, No:0) -17.95 (2.12)** -8.48 (1.04) -8.83 (1.09) 8.73 (0.97) -13.35 (1.73)* Involvement (Yes:1, No:0) -18.89 (1.67)* -24.07 (1.91)* -24.19 (1.86)* -22.81 (1.86)* Adjusted R2 0.15 0.17 0.17 0.05 0.17
33 5.2. Interpretation of Regression Test
First of all, let’s take a look at the effect how “Developing” term affects the loan pricing. If we compare the column 1 and 2 of Table 6, it is clear that “Developing” term is positively related to “Spread” and the developing countries are paying significant risk premium versus developed countries. The coefficient of “Developing” term is 60.22, which means that developing countries’ loan spread is 60.22 basis point higher than developed countries if other variables are controlled in the same way. We can see the effect was consistent in the other regression test shown from test 3 to test 5. This finding can be understood that the developing countries are still risky in terms of law enforcement when things go wrong, like the situation such as expropriation, termination of currency exchange or foreign remittance etc. Another variable “Country Risk” also implies the same understating with the one we could infer from “Developing” term. As in Table 6 again, Country Risk was also positively correlated with loan pricing in all 5 regression tests. Among them, only one case resulted in the coefficient with lower than 10% significance level. However, the degree of coefficient effect compared with “Developing” term was much lower, ranging from 5.98 to 26.59 versus from 40.43 to 60.22. The two terms just follow my intuitive expectation stipulated at Table 1.
One of interesting results is related to “Maturity” term. In my first expectation, “Maturity” should have positive correlation considering that the chance of repayment would be much more uncertain if the maturity is longer just like we can see in the option pricing. However, the result was negative relation and this result is just the same with the previous study by Kleimeir and Megginson (2001). In this study, they justified this finding with this expression, “Without a negative spread/term relationship, long tenor loans would be prohibitively expensive”. However, the rationale does not seem to be enough. From my practical point of view, lenders feel much more comfortable about lending money longer
34 specifically under project finance scheme because the cash flow generation is very transparent and stable. Lenders control and monitor the cash flow through sophisticated deal structure and security safety net such as bank account pledge etc. This fact is supported in second hand by the finding in their same study, which shows that general corporate financing had positive correlation between loan maturity and spread, exactly inverse result compared with project finance case.
On the other hand, my additional interest in this study was the effect of “Involvement” term. As reviewed in the previous literature, the relationship was expected in negative way, and the test result did not disappoint. From Table 6, we can see that the “Involvement” was significantly affecting the loan price from all four tests within 10% level. Once MLA/ECAs get involved into loan funding, it is considered a big burden for borrowers, especially from developing countries to default this because this can cause a significant adverse effect in terms of attracting further economic development aid from international capital market. If we compare the coefficient of “Involvement” term in test 1 and test 2, we can see the absolute amount of coefficient is larger when “Developing” countries were taken into consideration for the second regression test. (24.07 versus 18.89). “Guarantee” term was also significantly negatively related in two cases with 5% and 10% level respectively. The coefficient was all negative except for one case. The exceptional case was, however, not significantly meaningful, so we can conclude that “Guarantee” term is also negatively related to loan pricing as expected intuitively. One of other interesting points is that the effect of “Involvement” is larger than the one of “Guarantee”. This means that lenders might consider the involvement of MLA/ECAs more secured than the guarantee provided from sponsors or host government.
In case of “Arranger” term, the result was as expected. We can think that lenders feel more comfortable when they can share total risks with more participants. Thus, the more
35 participating banks for funding, the less risk premium for spread is possible. The test result was very sustainable in all 5 tests result with 99% significance level.
The size of loan does not seem to be very important to lenders. I guess when lenders feel comfortable with the overall scheme of cash generation and security provided, then they can get the internal approval for larger loan amount more easily. This might be especially because this test is being done with project finance in power sector. As we discussed in Section 3, the power sector in project finance is the hottest area due to its long proven reliable record in terms of repayment, technology etc. The fact that the size of loan is not much critical in the model can be explained from the test 5. When the “Size” term was excluded, the “Adjusted R2” was still around 0.17, not much different from other tests. The only exception was test 3, but in this case, the leaving “Close Year” term out from the test was the main reason for low Adjusted R2. In other words, the “Close Year” term is very important to explain the change of loan pricing in this model. As shown in Table 6, “Close Year” term has positive correlation with loan spread and we can say that the spread of the loan for recent project is increasing. The reason for this could be found that the demand of project finance is increasing due to its popularity or the recent capital market has been severely damaged by the economic crisis in 2008 and not recovered yet. Especially, considering the fact that the main players in project finance were major European banks, this phenomenon can be understood.
36 6. Summary and Conclusion
This study mainly tried to derive several significant determinants for loan pricing of project finance, especially in electric power sector. Using data sample of 20,571 collected from entire industry sector using project finance scheme during the period between 1990 and 2014 through Thomson Reuters database, I managed to extract 982 data samples only from electric power sector which had full information used for my multivariate regression test.
The result suggests that the guarantee from sponsors/host government, the involvement of multilateral bank/export credit agencies(MLA/ECAs), and the number of loan arrangers were the most influential determinants of the spread with negative relationship. Especially, the impact of the involvement of MLA/ECAs was overwhelming and the reason might be the fact that once MLA/ECAs gets involved into loan funding, it is considered a big burden for borrowers, especially from developing countries to default this because this can cause a significant adverse effect in terms of attracting further economic development aid from international capital market. In case of the number of loan arrangers, it can be understood that lenders feel more comfortable when they can share total risks with more participants
Loan maturity was also negatively related to the spread, but the effect was comparatively lower than other negative factors. In my opinion, lenders feel much more comfortable about lending money longer, specifically under project finance scheme because the cash flow generation is very transparent and stable. The size of loan was not much significantly related to the spread. I figure that it is because when lenders feel comfortable with the overall scheme of cash generation and security provided, larger loan amount is not much a big deal to get their internal approval any more.
On the other hand, project host location also affected the loan spread. The loan for the project in developing countries with higher country risk had significantly higher spread
37 than the one in developed countries with lower country risk. This finding implies that the developing countries are considered more risky and we can find the reason from their poor law enforcement system, unstable political status, etc.
Apart from the empirical study, I also tried to find the reason for higher attractiveness in electric power sector than in other sectors in terms of project finance arrangement. From practical point of view, the finding could be summarized with two factors: (1) smaller equity injection in favor of shareholders, and (2) lower chance of default in favor of lenders. Thanks to well proven stable deal structure for a longer period than other sectors, especially using power purchase agreement, shareholders and lenders are taking more benefits from project finance scheme in electric power sector. The popularity of project finance arrangement in electric power sector was shown through statistical description of 20,571 collected data for this study. Project finance loans were highly concentrated in three key industries in order by deal amount: Power, Transportation and Oil & Gas and by deal frequency: Power, Transportation and Leisure & Property.
However, it is undeniable that the explanatory variables to influence loan pricing were limited in this study, so the study can be more extended including more variables for more accurate analysis. Also, in this study, the approach to find the distinctive advantages in electric power sector in terms of project finance arrangement was mainly based on my practical experience and I hope this practical understanding can be supported by the comparison analysis which assures that the spread and the maturity of project finance in electric power sector have more favorable conditions than the ones for other sector through collecting relevant empirical data.
38 References
Altunbaş, Y., & Gadanecz, B., 2004, Developing country economic structure and the pricing of syndicated credits, Journal of Development Studies, 40(5), 143-173.
Babbar, S., & Schuster, J., 1998, Power project finance: experience in developing countries, The
World Bank.
Bonetti, V., Caselli, S., & Gatti, S., 2010, Off-taking agreements and how they impact the cost of funding for project finance deals: A clinical case study of the Quezon Power Ltd Co., Review of
Financial Economics, 19(2), 60-71.
Booth, J. R., 1992, Contract costs, bank loans, and the cross-monitoring hypothesis, Journal of
Financial Economics, 31(1), 25-41.
Buljevich, E. C., & Park, Y. S., 1999, Project financing and the international financial markets, Boston, Klwuer Academic.
Esty, B. C., 2003, The economic motivations for using project finance, Harvard Business School, 28.
Farrell, L. M., 2003, Principal-agency risk in project finance, International Journal of Project
Management, 21(8), 547-561.
Gadanecz, B., & Sorge, M., 2008, The term structure of credit spreads in project finance,
International Journal of Finance & Economics, 13(1), 68-81.
Gatti, S., 2008, Project finance in theory and practice: designing, structuring, and financing private and public projects, Burlington, Academic Press.
Gemson, J., Gautami, K. V., & Thillai Rajan, A., 2012, Impact of private equity investments in infrastructure projects, Utilities Policy, 21, 59-65.
39
syndicated lending, Governance and the Efficiency of Economic Systems (GESY), SFB/TR, 15.
John, T. A., & John, K., 1991, Optimality of project financing: Theory and empirical implications in finance and accounting, Review of Quantitative Finance and Accounting, 1(1), 51-74.
Klapper, L. F. and Love, I., 2004, Corporate Governance, investor protection, and performance in emerging markets, Journal of Corporate Finance, Vol. 10(5), 703–728.
Kleimeier, S., & Megginson, W. L., 1998, A comparison of project finance in Asia and the West,
Project Financing in Asia: A Redefining of Premises, 57-90.
Kleimeier, S., & Megginson, W., 2001, An empirical analysis of limited recourse project finance,
Working Paper Series, University of Oklahoma, Michel F. Price College of Business.
Morrison, R., 2012, The principles of project finance / edited by Rod Morrison, Burlington, Gower.
Nevitt, P. K., & Fabozzi, F. J., 2000, Project financing, Euromoney books.
Peppiatt, S., 1995, Introduction to power station project financing, 13 INT’L Tax & Bus. Law.
Scott, J. A., & Smith, T. C., 1986, The effect of the bankruptcy reform act of 1978 on small business loan pricing, Journal of Financial Economics, 16(1), 119-140.
Shah, S., & Thakor, A. V., 1987, Optimal capital structure and project financing, Journal of
Economic Theory, 42(2), 209-243.
Sheppard, R., 2011, Effectiveness of multilaterals, Project Finance International, March 9, Issue 452.
Smith Jr, C. W., 1980, On the theory of financial contracting: the personal loan market, Journal of
monetary economics, 6(3), 333-357.
Suazo, Jose A., 2001, A comparative analysis of independent power projects in three developing countries – India, Indonesia, and Peru, Master Thesis, MIT.
40
White, H., 1980, A heteroskedasticity-consistent covariance matrix estimator and a direct test for Heteroskedasticity, Econometrica 48, 817-838.
Milbank Project Finance
http://digital.milbank.com/i/209256/79 Accessed 30 May 2014
OECD Country Risk Classifications
http://www.oecd.org/tad/xcred/cre-crc-current-english.pdf Accessed 2 June 2014
Thomson Reuters, 2012. Project Finance Review Full Year 2012
http://dmi.thomsonreuters.com/Content/Files/Q42012_Project_Finance_Review.pdf Accessed 29 April 2014.