Menu
Home
Log in / Register
 
Home arrow Business & Finance arrow Treasury Finance and Development Banking
< Prev   CONTENTS   Next >

5.5 NUMERICAL EXAMPLE: ESTIMATING THE COUPON OF AN EMERGING MARKET DEBT INSTRUMENT

Let us apply the combined knowledge we have obtained from the previous two chapters in a numerical exercise in which we imagine we are helping a developing world entity to issue a bond.

Let us imagine we are working closely with Electricity of Vietnam to issue a fixed-rate two-year bond and we need to decide what the coupon of the bond is going to be either in local currency, Vietnamese Dong (VND), or in USD.

Vietnam is a developing country with a rudimentary financial market. It is dynamic and with a growth rate averaging around 7% over the past decade. Electricity of Vietnam is a public utility company backed by the government of Vietnam. In establishing what the coupon should be in a two- year fixed-rate bond, we should, as we have seen throughout this chapter, take into account considerations of credit and the general interest rate landscape of the denominating currency. These two considerations need to be combined so that, irrespective of the currency in which we issue the bond (USD or VND), given the chosen coupon, at inception the bond price is equal to par.

We have chosen the example of a fixed-rate bond because this is what happens in practice at the developing stage of a debt market. Investors want to purchase debt that is easy to hedge with the instruments available in the market. In order to hedge a floating-rate bond one needs floating-rate instruments, which are not easily available in developing markets. Although interest rate swaps are traded in Vietnam, they are not very liquid and/or the

TABLE 5.2 Data, on interest rate instruments as of March 1, 2012, relevant in the assessment of the coupon of a two-year Electricity of Vietnam bond.

Instrument

type

Currency

Maturity

Rate

VNIBOR

VND

1Y

13.50 %

VNIBOR

USD (loc.)

1Y

4.10 %

VNIBOR

USD (fgn.)

1Y

2.36 %

Swap rate

VND

2Y

13.00 %

Prime rate

VND

1M

9.00 %

Swap rate

INR

2Y

5.85 %

JIBOR

IDR

1Y

4.87 %

JIBOR

USD

1Y

1.49 %

Swap rate

IDR

2Y

6.05 %

sizes of contracts would put a serious cap on the principal of the bond we are trying to issue.[1] As we have mentioned many times before, the first instruments to appear in the market are FX forwards, which are ideal instruments to hedge fixed-rate bonds. As the coupon is fixed, we are only concerned about the combination of interest rate and credit, which will result in that discount factor leading to an initial par level.

Vietnam, like many other emerging economies as outlined in Section 4.2.4, exercises capital control: VND cannot be exchanged freely and, as a consequence, the dealing of USD within the country is also controlled. From Table 5.2 we see that there is one interbank rate, VNIBOR, fixing in VND and representing the rate at which banks lend VND to each other. We also see that there is another VNIBOR rate fixing in USD and representing the rate at which banks lend, within Vietnam, USD to each other.

It is not surprising that the first rate is higher than the second (even allowing for the further detail that the VNIBOR rate in USD depends also on the local versus foreign nature of the bank that does the lending). We offer the additional example of Indonesia where we also see a different fixing in the interbank rate (JIBOR) depending on whether the fixing is in USD or in Indonesian Rupees (IDR). Again, the USD fixing is lower. For Vietnam in Table 5.2 we have also shown the prime rate, which is the rate, set by the government, at which commercial banks can lend on a monthly basis. Intuitively we would expect it, being shorter maturity, being set by the government and applying only to commercial banks, to be lower than the interbank rate. We see that this is indeed the case.

In Section 5.4.3 we have shown how, when trying to price an illiquid or an unusual debt instrument, we proceed with a strategy of looking for proxies. Since in the case of Electricity of Vietnam we do not have a measure of the credit standing of the issuer (the only information we have on its credit is a USD loan expiring in three years issued at 290 bps over LIBOR) we need to look at similar entities and see what we can achieve by imitation. The process consists of modifying little by little what we mean by similar until we have built a set of proxies we are confident with.

In this particular example we decide to focus on three criteria of similarity: geographical proximity, type of industry, and role of the company within the economy and/or its relation to the government. Using these criteria we decide to focus on India, Indonesia, and Thailand for similar geographical location and type of country (i.e., fast growing and relatively wealthy by developing world standards). Within these countries we then focus on Electricity Generating Authority of Thailand (Thai EGA in Table 5.3) and Perusahaan Listrik Negara of Indonesia (PLN in Table 5.3) as two examples of electricity companies. Finally we add Indian Oil Corporation (Indian Oil in Table 5.3) as an example of a government-owned power company. The bonds were chosen such that those that needed to be compared pairwise were issued close to each other in time and therefore in a similar credit environment.

In Section 5.3.1 we have shown how one of the most common benchmarks for a bond is the debt of the most creditworthy government issuing in that currency. As the Vietnamese government is the only issuer of debt in VND, in estimating the coupon of our (VND) bond we shall try to work toward establishing a benchmark value between the Electricity of Vietnam bond we are issuing and a similar government bond. When it will be the moment of estimating the value of the coupon of our USD bond, the benchmark will be against a U.S. government treasury bond. Benchmarks against government bonds are given in the form of yields. Since our problem is not to price a bond but is the easier one of estimating the coupon of a bond about to be issued, we can use the yield as a coupon. We know that at inception the yield is equal to the coupon. In the spirit of showing the additive nature of yields as quoted, we show as an example the yield of the 10-year U.S. Treasury note: at 1.98% is indeed 254 bps (plus or minus a few basis points) lower than the yield of 4.55% of the November 2021 PLN bond.

TABLE 5.3 Data, on debt instruments as of March 1, 2012, relevant in the assessment of the coupon of a two-year Electricity of Vietnam bond.

Instrument

type

Currency

Maturity

Entity

Price/yield/

spread

Benchmark if spread

Loan

USD

27-Oct-2015

Electricity of Vietnam

290 bps

USD LIBOR

Bond

VND

20-Feb-14

Vietnam

10.98%

Bond

USD

29-Jan-20

Vietnam

5.39%

Bond

USD

29-Jan-20

Vietnam

395 bps

US7YT

CDS

USD

20-Mar-19

Vietnam

361 bps

Loan

USD

22-Mar-16

Indian Oil

120 bps

USD LIBOR

Bond

USD

2-Aug-21

Indian Oil

5.28%

Bond

USD

2-Aug-21

Indian Oil

327 bps

US10YT

Bond

INR

21-Dec-16

Indian Oil

9.32%

Bond

INR

21-Dec-16

Indian Oil

79 bps

IN1YT

Bond

INR

12-Dec-41

India

8.61%

Bond

USD

17-Jan-42

Indonesia

4.82%

Bond

USD

17-Jan-42

Indonesia

169 bps

US30YT

Bond

IDR

15-Jul-21

Indonesia

5.73%

CDS

USD

20-Mar-22

Indonesia

213 bps

Bond

USD

22-Nov-21

PLN

4.55%

Bond

USD

22-Nov-21

PLN

254 bps

US10YT

Bond

IDR

8-Jul-22

PLN

7.77%

Bond

IDR

8-Jul-22

PLN

224 bps

ID10YT

Bond

THB

27-Aug-14

Thai EGA

3.50%

Bond

THB

27-Aug-14

Thai EGA

41 bps

TH2YT

CDS

USD

20-Mar-14

Thailand

77 bps

Bond

USD

2-Aug-21

USA

1.98%

In Table 5.3 we show a set of bonds belonging to the set of proxies we have chosen. When relevant we have shown the same bond twice, once with its value in the form of yield and once in the form of spread with respect to some benchmark. We have one bond in USD and one in VND issued by the government of Vietnam; we have one bond in USD and one in Indian Rupees (INR) issued by Indian Oil; we have one bond issued in INR by the government of India; we have one bond in USD and one in IDR issued by the government of Indonesia; we have one bond in USD and one in IDR issued by PLN; finally, we have one bond in Thai Baht (THB) issued by EGA.

In the table we have also included some CDS levels to test and display what we have learned in the previous chapters. For example, we see that the January 2020 Vietnamese bond yields 395 bps over the relevant U.S.

Treasury note. Since U.S. government bonds are as close as one can get to a risk-free instrument, one can (very) roughly consider them as a purely interest rate instrument. We have said many times that a CDS rate is not something one can apply as it is, as tempting as it might be, onto an interest rate to add a credit element. However we have seen in Equation 5.14 that we can manipulate a CDS rate so that we can do something in that direction. Let us take Vietnam's CDS rate shown in the table and scale it, taking into consideration the recovery rate

This number, with the not unreasonable assumption of 15% recovery, is quite close[2] to the spread of 395 bps yielded by the Vietnam bond over the U.S. Treasury note. The excitement about seeing two numbers making sense of each other needs to be dampened by applying the same test to Indonesia. We see that the January 2042 bond yields 169 bps over the 30-year U.S. Treasury note. By taking the CDS spread shown in Table 5.3 (the required 30 years CDS rate is equal to the 10 years shown) we can calculate

which is almost 100 bps higher than the spread between the government bond's yield and the U.S. Treasury note. The fact that we are treating very long dated and illiquid instruments means that in this context 100 bps is not a very large difference; however, it also shows that these calculations should only be used to support a general view and not to demand high precision. We have used 20% recovery simply on the basis that one would think that a country with a higher credit rating than Vietnam would have a higher recovery rate. Using the same recovery rate of 15% would lead to a value of 250 bps, which is close to the spread over the U.S. Treasury.

In trying to assess what should be the relation (spread) between the bond we are about to issue and the Vietnamese government bond, let us begin with the local currency bond. From the table we see that the spread between the

Indian Oil INR bond and the one-year Indian Government bond is 79 bps; the spread between the EGA bond and the two-year Thai government bond is 41 bps; the spread between the PLN IDR bond and the 10-year Indonesian government bond is 224 bps. The average between these three values is 115 bps and this would be the first guess for a spread over the Vietnamese government bond. We could set the coupon at 12.15% by rounding up the sum of 10.98% (the government bond's yield) and 115 bps.

There are few considerations that could make this value higher or lower.

On one side Vietnam offers a riskier environment, judging from the CDS levels, than Indonesia and Thailand (there are no CDS levels for India) and therefore our value should be higher than the arithmetic average. The immediate counterargument, however, is that we are discussing relative values, that is, we are not concerned by how risky the government of Vietnam or Electricity of Vietnam are, but by how Electricity of Vietnam is riskier than the government.

On the other side the spread for the PLN bond over the Indonesian government bond is for 10 years, a maturity far greater than the one we are considering and one where we would assume that spreads increase. According to this argument our value should be lower than the arithmetic average. Since both these arguments are valid at first order[3] we could then accept the value of 115 bps.

We could round up the value of the coupon, in the name of uncertainty, to, say, 12.50% but we need to be aware, in case we are tempted to increase it even more, that we are soon reaching an important upper bound.

In Table 5.2 we show the value of the one-year VNIBOR rate to be equal to 13.50% and the value of the two-year swap rate to be equal to 13.00%. We know little about the liquidity of these numbers and their market depth (meaning how large the size of the average trade is), however, intuitively we would expect that a government-owned company would be able to raise capital at a lower cost than a financial institution.

We now focus on estimating the coupon of the USD bond Electricity of Vietnam is wishing to issue. A first approach would be to consider the spreads over U.S. Treasury notes of the two other companies for which we have data. PLN's bond trades at 254 bps over the U.S. 10-year note and Indian Oil's bond trades at 327 bps over the same note: the average is 290 bps. We could then argue, although admittedly we have information only for Indonesia's CDS levels, that we should scale this number by the

ratio in CDS levels between the countries from which we have worked out the spread and the country in which we are trying to apply our calculation. We are basically trying to answer the question: this spread is valid for a credit regime given by one certain CDS level, how would this spread change in a different CDS environment?

We are switching again to a relative view. PLN and Indian Oil are to Indonesia and India what Electricity of Vietnam is to Vietnam: if the spread between the two companies and the USD note is of a certain value in a country with a certain CDS level, wouldn't the spread between Electricity of Vietnam and the U.S. note be increased by the fact that that company is in a country with a larger CDS rate? Indonesia's CDS level is 213 bps and Vietnam's is 361. Therefore, the spread we are looking for could be calculated in the following way

The coupon of our bond should be then given by the U.S. Treasury note's yield plus 491 bps.

A second approach could be to take the number we have obtained in the previous calculation, when we were estimating the coupon of the VND bond, as the spread between Electricity of Vietnam and the Vietnamese government bond (115 bps) and add it on top of the spread between the government of Vietnam's bond and the relevant U.S. Treasury note, which, as shown in Table 5.3, is 395 bps. This operation would lead to a spread of 510 bps, meaning that the coupon of our bond should be given by the U.S. Treasury note's yield plus 510 bps. We note that the numbers obtained with these two methods are not far from each other (which is a good sign).

With these calculations we have shown how traders and bond issuers try to make sense of very scattered and illiquid information. We have produced numbers that at first seem reasonable. This was far from a scientific process: the same numbers could be argued and proved meaningless from several points of view. The main role played by the data provided is to offer a view of the situation and a feel for the credit environment, the rest is open to debate. We could finish this section with two statements often heard among physicists. The first is to never run a simulation unless one already knows the result. The second, which could be applied to the type of data we deal with, is that statistics to a scientist is like a lamppost to a drunk, it is more of a support than a source of illumination.

  • [1] Since investors need to hedge the bond, we cannot issue a bond with a principal that exceeds the principal of the hedging instruments. For example, if the average interest rate swap is traded with a size of 100M it would be risky to issue a bond with a principal much greater than, say, 50M. We want the hedging instrument's size to be not only larger than the size of the instrument that needs hedging but much larger in order to avoid any market distortion.
  • [2] When comparing instruments that are not quite the same and particularly in the realm of emerging markets, the definition of close becomes considerably generous. The author remembers that when pricing illiquid EM bonds, after having considered all inputs, spreads of up to 250 bps were added in the name of liquidity. It was more like saying, we are walking blind, in the dark, in a room with a potentially low ceiling, let us put a large pillow in front of our face in case we bump into something.
  • [3] A more thorough analysis would involve a more serious assessment of liquidity and the legal relationship between each of these companies and their respective governments.
 
Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >
 
Subjects
Accounting
Business & Finance
Communication
Computer Science
Economics
Education
Engineering
Environment
Geography
Health
History
Language & Literature
Law
Management
Marketing
Philosophy
Political science
Psychology
Religion
Sociology
Travel