Treasury Revisited

The main focus of our discussion, credit and debt, is seen through the activity of a treasury, its role, and the way certain risks are managed. We have seen how a financial institution and in particular a development institution work; we have seen how discounting plays a crucial role in assessing the value of any financial instruments; more importantly we have seen how the concept of credit and in turn the concept of funding cost is closely linked to it. On the issue of credit we have seen how this is treated under a modeling point of view and what it entails when applied to the lending side of a financial institution (with a focus on development). We have seen how financial instruments behave differently in less developed markets and what this entails in terms of borrowing and lending. We have seen how bonds are priced and, crucially, the role played in this by credit.

The above are some of the most important tools needed to understand the role and activities carried out by a treasury desk and in particular to understand the role played by funding when it comes to treating financial instruments. In this chapter we shall try to show how the goal of a treasury desk is to raise funds by issuing debt in the most attractive way to investors. How to get to this price, where and how to issue it, and the impact of the price on the credit standing (or vice versa) of the issuers are all points that will leverage the knowledge we have acquired in the previous chapters. We shall conclude this chapter with a discussion on borrowing and investment benchmarks with a special focus on these in the domain of development banking.

FUNDING AS AIM ASSET SWAP STRUCTURE

Asset Swaps Revisited

In Section 5.3.2 we defined what an asset swap is. At first Figures 5.2 and 1.3 do not seem identical, but if we focus on what in essence an asset swap means, we shall see how they are.

Probably the most general and essential definition of an asset swap one could give is the one where we define it as a way of capturing the credit risk of a bond through a fixed spread over a LIBOR. This can be written (we generalize Equation 5.11) as

(6.1)

where, in our preferred fashion, we have kept the discount factor expressed in a general form Di and where is a correction to the discount factor driven by the credit standing of the bond issuer. In Equation 6.1 Fj is the coupon of the bond where, in order to generalize, we assume that it can be anything from a fixed rate to a floating rate to a complex structure. As usual Li is the LIBOR and is the asset swap spread. The symbolmeans that each side of the equation is as seen by each counterparty. The right-hand side is the side of the bond issuer, which sees (positive) 1 as the money received from the investors (bond buyers) and the bond payment to the investors as negative, that is, a liability. The left-hand side is the side doing the swap, which sees the LIBOR payment made by the bond issuer as positive and the coupon leg payments as negative. (Had we used a simple equal sign, Equation 6.1 would not have satisfied the simple rules of algebra. In Equation 5.11 we did use an equal sign, but the signs in front of the terms are different: the rules of algebra were satisfied and, at the time, we did not need to be too precise about the directions of the trade.)

With the asset swap framework in mind, a treasury desk would gauge the market appetite for a certain type of bond (e.g., fixed rate or with a more complicated payoff) after which it would issue the bond and, as a consequence of the market response, the bond acquires a price. At this stage, the right-hand side of Equation 6.1, that is,

becomes known (with the bond price driving the value of ). Not only the right-hand side is known but also the term

on the left-hand side is now known. This means that the only remaining element needed for a fair price is the asset swap spread sA. After the trading desk (or any unit acting as the other party in the asset swap) and the treasury desk agree on this, the swap part of the asset swap structure is completed.

In practice the operation does not really follow this order; the issuing and the swapping happens simultaneously. More important the bond and the swap are often legally linked. For example, if something were to happen to the bond in terms of interruption, such as recall or prepayment, the swap would often reflect the consequences. This is slightly different from a normal asset swap situation where the bond holder after having purchased the bond swaps it in order to monetize the bond's risk. This helps to explain the difference between Figure 5.2 and Figure 1.3; however, if we look at an asset swap spread as an easily tradable proxy for a bond's credit, the two situations immediately become very similar.

The bond shown in Equation 6.1 matures at time T, which of course means that the swap also matures at time T. From this we say that the asset swap spread represents the cost of funding or funding level up to year T and we shall rewrite it as s. Similarly if the treasury desk issues a swap with maturity S > T, the asset swap spread will represent the cost of funding up to year S. It would be unlikely for the two costs of funding to be identical. If we repeat the same process (which is of course what happens in practice) of issuing bonds with different maturities we build a term structure of funding levels. For the better issuers in the past this term structure would be almost flat, that is, asset swap spreads were very similar at different maturities. After the financial crisis of 2007 to 2009, however, the overall reconsideration of credit (which we have seen affecting discounting in Chapter 2) has not left any corner undisturbed, and now almost all term structures of funding costs present some (usually upward) movement.

In Table 6.1 we show the asset swap levels of selected bonds issued by the Republic of Italy as a display of a term structure of funding costs. Next to it we show the approximate CDS rate, quoted in USD (see Section 3.2.3), corresponding to the maturity of the bond: this is only for illustration purposes since we know, after discussing it in Sections 5.3.2 and 5.3.3, that asset

TABLE 6.1 Example of USD funding level term structure for the Republic of Italy as of October 17, 2011.

swap spreads and CDS rates, while often close in value, are quite different in principle.

To be more precise we would need to say that the funding levels shown in Table 6.1 are USD funding levels: why is the specification important? A borrowing entity, corporate or sovereign, can decide to issue a bond either in its local market (even multinationals have a principal or local market) or in a foreign one. The decision for an institution to issue debt abroad is driven by the notion that there will be interest in purchasing its own debt in that specific country and in that country's currency.^[1] The question however is, how much interest?

Will a bond issued by, say, Coca Cola, an American company, be considered more valuable in the United States, Coca Cola's local market, issued in USD or in Japan issued in JPY? This is a very important question that, once answered (through subscribed debt issuances) by the market over multiple occasions, enables an issuer to build term structures of funding levels in different currencies. These funding levels, one in USD and one in JPY, will almost certainly be different.

Remaining with Coca Cola's example, let us try to answer the question on the view of the credit of the company as held by American versus Japanese investors. Probably a Japanese investor would think that, should a default of Coca Cola occur, an American investor would have easier access to the spoils of the firm, so to speak.^[2] In this case, a Coca Cola bond in Japan would be worth less than a similar Coca Cola bond in the United States. We should know by now, after reading Chapter 5, that to be worth less means a higher yield, which in turn could mean a higher coupon paid by the bond (in order to attract Japanese investors). Using Equation 6.1 we can easily see that the higher the coupon Ej, the higher the asset swap spread sa must be. This means that the asset swap level, that is, the cost of funding, for Coca Cola in Japan would be greater. Unless something changes in our simple reasoning (which could happen since our reasoning was very simple indeed and ignored many possible market scenarios), Coca Cola has little interest in issuing debt in Japan. Since the same principle we have used for Japan would apply to any non-American market, it should concentrate in issuing debt only in the United States.

TABLE 6.2 Example of EUR funding level term structure for the Republic of Italy as of October 17, 2011. Indicative levels are shown for the currency basis swap as spread to be paid over EURIBOR versus USD LIBOR flat.

Not many institutions issue debt in multiple currencies, that is, in foreign markets: it is interesting to see what happens to the few that do. Table 6.2 shows bonds issued by the Republic of Italy in EUR: the maturities and issue dates of the bonds have been chosen so as to be as similar as possible to those of the bonds shown in Table 6.1. The values are considerably different. Let us express each situation as an asset swap. The Republic of Italy could issue a EUR-denominated bond and enter into an asset swap

(6.2)

where is the principal in EUR, are the EUR discount factors, and is the EURIBOR floating rate. Alternatively it could issue a USD- denominated bond and enter into an asset swap

(6.3)

where, similarly, is the bond's principal in USD, are the USD discount factors, and is the USD LIBOR. These two bonds would have two different asset swap spreadsandindicating two different costs of funding, one in each currency. From Table 6.2 it would appear that the government

of Italy could issue bonds in EUR at a cheaper (i.e., more advantageous) level than if it did the same in USD. This is true, but the extra column in Table 6.2 explains part of the difference. The government of Italy could either issue a bond in USD and pay or issue a bond in EUR and pay . Isn't there something we have encountered before that enables us to compare floating rates from different currencies? In Section 2.2.5.2 we have seen that a EUR versus USD cross currency basis swap would be structured as

where bC is the cross currency basis. From Table 6.2 we see, for example, that the cross currency basis quote with tenor similar to the first bond is (negative) -65 bps: this means that, in the floating rate exchange, the EURIBOR level is higher than the USD-LIBOR level by 65 bps. This also means that, because the rate itself is higher, an asset swap spread/funding level quoted in EUR must be intuitively lower.^[3] As a consequence, in order to compare two funding cost levels in a meaningful way, one should at least add the currency basis. This means that we need to compare

with

in order to compare with . The basis bC can be either added to the floating rate or subtracted from the asset swap spread. This, confusingly since it is negative, means that if we want to compare the asset swap spread given in EUR in Table 6.2 to the one given in USD in Table 6.1 we need to subtract the –65 bps from the 207.96 bps of the asset swap spread leading to a total of 272.96 bps. This value is still very much lower than the 385.40 bps we see quoted for a USD bond with similar maturity, leading us to conclude that for Italy it is cheaper to issue debt in EUR than in USD. The fact that a quick glance at Italy's outstanding debt shows that EUR debt principal is roughly 100 times larger than USD debt principal would confirm this.

The relationship between funding levels and cross currency basis cannot be stressed enough. In the example above we have used cross currency levels to manipulate asset swap spreads, but the real relationship starts at the other end. In Section 2.3 we said how the cross currency basis is the level that represents the ease with which investment banks borrow in a certain currency, this meaning that it represents the view that a certain country's market has of the credit standing of the average foreign bank. This means that cross currency basis swaps exist exactly, because when a bank tries to issue a bond outside its native market, the cost of funding is different. Bond prices, or more generally borrowing instruments, drive the basis. To conduct an exercise similar to the one we have done for Italy on several investment banks would be more difficult since banks issue a greater number of smaller denominated bonds with varying liquidity and varying complexity of payoff (we shall describe in Section 6.2 why this is an issue). Such a study however would lead to interesting results when compared to the traded cross currency basis levels.

• [1] We assume that a bond issued in a certain country will be denominated in the currency of that country. While this is not always the case it is the most common situation.
• [2] Which is not necessarily true but market sentiment, particularly when it comes to credit, can be fairly irrational or at least overcautious.
• [3] ShouId this seem hard to grasp one can think that floating rate plus asset swap spread must be very roughly equal to the coupon value: if the floating rate is lower, the asset swap spread must be higher and conversely if the floating rate is higher, the asset swap spread must be lower.