By Ted Howard
Societe Generale has a new model that offers treasurers a new way to think about financing.
In a world of historically low interest rates, what is the cost of issuing long-term debt vs. short-term debt? Or more specifically, how do you put a number to the price of extending duration?
This is a question that Societe Generale debt capital markets executives were presented with this year when a client was looking at how it could save money issuing debt. The client’s position was that it was a good credit, with little or no rollover or refinancing risk. With no refinancing risk, issuing shorter repeatedly is usually—in the current environment—cheaper than issuing long-dated bonds. But it is not a risk many treasurers are prepared to assume, however. Rates can go up and short-term markets can experience hiccups that can put a short-term based financing strategy in jeopardy. And finally, issuing repeatedly can up end generally accepted asset-liability management principles.
But still: why should it pay more for a traditional debt mix? In short, what are the limits for monetizing a high credit rating?
Perhaps it’s not such bad idea? To determine if this was so, SocGen created a comparison model that shows various permutations of what kind of tenors to issue and the cost differences between them. The goal of the duration extension calculation model is to allow users to compare the associated costs of issuing long bonds vs. multiple short bonds (which combined, would match the duration of the long bond).
By comparing the discounted cash flows of various bonds, the model identifies a breakeven point which can help users understand how risky the multiple short-bonds strategy was vs. one long bond. In other words, when it’s time to roll over and the company wants to extend, say, another 10 years, should it just issue another 10-year bond or issue two 5-years over 10 years? The answer, as with all multinationals with different cash and debt structures and varying credit ratings is, it depends. But SocGen’s model can now add clarity and help treasurers make a more informed decision.
According to model developer Aris Corti, debt capital markets banker for the Americas at Societe Generale Corporate & Investment Banking, the genesis of the model came from a routine market update meeting with a client. “Because we are in an environment where we have very low and very tight rates, along with very tight spreads, you have the argument from the debt capital markets folks saying that companies should lock in at these [low] rates and at these spreads and issue as long as they can,” Mr. Corti says. “The only problem with that is that when you have a very steep curve from the rate and credit-spread side, the coupon of the long bond actually looks higher than issuing a 5-year because of where rates are.”
And where rates are and have been seemingly forever, is very low. For years the market has been telling investors that rates will eventually go higher, but in reality they haven’t budged. “So that’s where treasurers start asking themselves, ‘Is it really worth it for me to continue issuing in something so long and then locking these rates?’ ” Mr. Corti says. “Then the question becomes, ‘how do you put a number to issuing something longer instead of issuing something shorter multiple times?’ “
That’s why Mr. Corti created the model. “What I wanted to do is basically compare two outcomes that have exactly the same tenor,” he says.
The Model
SocGen’s model is a closed universe where it is assumed a company’s corporate/treasury spread profile will not evolve over time. He starts by taking a snapshot of the credit curve spread-wise (for example a 5-year will be issued at 20 basis points (bps) over Treasuries, a 10-year to be issued at 50bps over Treasuries and a 30-year at 90bps over Treasuries). He then takes that and applies it throughout the life of the bond. At this point in the process, Mr. Corti does not include any refinancing or movement of the credit spread.
On the rate side, he uses Bloomberg to take a snapshot of the forward curve. “What Bloomberg does is calculate the forwards for specific tenors,” he says. “It takes the current treasury curve you have and then allows you to have forward rates.” For illustrative purposes, Mr. Corti points to an easy and simple example: comparing the issuance of a 10-year bond versus issuing a 5-year bond twice (see Figure 1). In the illustration, he shows two different scenarios of issuing the 5-year twice; one using the discount rate and the other using the forward plus the credit spread.
The single 10-year column represents a very simple formula of rates plus credit spreads, which gives users the cash flow that they will generate. “And then the same thing happens with the 5-year twice,” Mr. Corti says. “It’s basically the 5-year today, plus the credit spreads, and then the 5-year in five years, plus the credits spreads; so that’s how you get the two different cash flows.” In Figure 1, those cash flows in the total CF line show some sizeable disparities.
One caveat Mr. Corti points out is that the calculations are done “on an absolute cash-flow point of view.” Therefore, users need to include the concept of time in this specific analysis. This is why the cash flows are also discounted, as seen in the first 5-year twice column in Figure 1. “Those cash flows are going to be discounted by the same rate at which I am issuing the single 10-year, he says. “So in this case if the discount rate for a single 10-year is 2.859, you use the same rate discount in the second set of cash flows.”
Once Mr. Corti determined the cash flows, as noted above, he needed to compare them using time value. He points out that there are perhaps too many assumptions to make when choosing a discount rate. In the first 5-year scenario, it was the same discount used in the single 10-year issuance. But then on the second 5-year scenario, in column three of Figure 1, Mr. Corti uses forward risk plus credit to discount the cash flows. “I am trying to link the discount rate to the time of the bonds that are being issued,” he says. “So in a sense what I’m trying to do is stick to—as much as I can—the life of the specific bond that I’m issuing.”
And this would then give users of the model the differential in terms of what the cost of the duration is. Once the duration is matched, there are now two sets of cash flows. Mr. Corti says this is where users can start looking at which strategy is or will be more expensive and thus, less viable.
The next question then is, what needs to happen for either strategy to stop being viable? So Mr. Corti’s next goal was finding the breakeven point. “A breakeven for me, the way I define it, is what’s the overall cost for the issuer?” he says. “By how much do [rates] have to change in order to match the cash flows of the first option. By how much do I need spreads or rates in the second strategy to move in order for that strategy not to be viable anymore?”
Since there are two different calculations to discount, Mr. Corti says the model needs to have two different breakeven points—for both the long and multiple-short scenarios. Mr. Corti uses an example of issuing a 5-year and a 10-year today. “And then immediately a minute later, I assume that there is a change in rates; so what I did was to lock in the first coupon of the first 5-year and then after that, things change.” This leaves the question of by how much do things need to change? In Figure 2, Mr. Corti answers this by using changes in the fed funds rate. In the first 5-year column in all examples, rates would need to change much more significantly than in the second 5-year option.
In the very first scenario in Figure 2, Mr. Corti compares doing a single 10-year versus doing twice the 5-year, with a breakeven 59.6 bps. This means that if a user assumes it can discount everything by the same rate as issuing the 10-year, and between today and the time that the user issues its second 5-year bond, rates have moved above 59.6 bps, it becomes clear that they should have issued the 10-year instead of the 5-year twice. But if it’s below 59.6 bps it tells the issuer that the option of issuing the 5-year twice is the most competitive price-wise.
Mr. Corti stresses this isn’t so much a prediction, or a model to put a treasurer in the position of prognosticator (they would never want to be in that position) but it just adds perspective to the conversation of what should be issued. “My objective is to understand why the breakevens are different,” he says. “It is for me to put things in perspective.” But if a treasurer has a point of view that in the next five years, rates are going to increase, but perhaps not by more than 60bps then in that case it’s better to issue the 5-year. But if the point of view is that rates are going to go much higher than those 60bps in the next five years, then in that case the 10-year is a better option.
The Client’s Goal
According to the client SocGen spoke to, the start of the conversation began when the company was looking into the costs of going to the market to fund cash it wanted to add to its pension plan.
As is standard practice, it began telling the banks but was getting a wide variety of answers in terms of tenor. “I didn’t know if it was right,” the client said. As a result the client started to do some funding strategy analysis and in doing so, one question she came up with is, “How do you measure the cost of extending out of commercial paper?” And this is where the conversation with SocGen started. Initially the client felt that since the company had neither rollover nor refinancing risk, she wanted to use commercial paper. But the size of the resulting CP balances would be too large; banks would balk and rating agencies “would scream bloody murder.”
After considering debt service and how tall the company’s “debt towers” should be each year, the client started putting ideas in place around risk tolerance and what issues are creating risk for the business. “And it just kind of evolved into some parameters,” the treasurer said. “That’s when we started thinking about comparing the cost of issuing longer tenors vs. issuing a series of shorter ones.
Specifically, what is the present value of $100mn at every tenor extension? “If you think about it this way: to extend from commercial paper at that time out into the 3-year space, with say $300,000 additional that you were paying to make that move; to extend from that 3-year to that 30-year was an additional $20mn for every $100mn you finance on a present-value basis; so you could then start putting hard dollars around every time you extended, it was costing you that much to make that extension.” However, again, “if you don’t have rollover and you don’t have refinancing risk, then why are you paying for that extension? And that’s how we kind of came to it in the spring. Let’s issue on the short-end of the curve and fill in those gaps where we’ve been open.” In reality, though, any large debt issuance is subject to market conditions and to fill it, sometimes you have to accept the points on the yield curve you can get.
Shorter May Be Better?
Shorter-duration debt is cheaper overall but a lopsided debt portfolio does not fit the risk appetite of most corporate treasurers. A combination of different debt durations is a more prudent approach as it will limit the volatility of debt costs over time. Those who want to apply rigorous analysis to the selection of durations at different times will like a thoroughly backtested model such as the one SocGen has developed, perhaps with some tinkering about what assumptions fit their own thinking.
Looking ahead, in addition to the breakeven concept, Mr. Corti also is exploring another way of gauging refinancing risk. In this approach, he looks at how to capture refinancing risk by analyzing the evolution of rates over time. In doing so, he applied the capital asset pricing model (CAPM) strategy that investors use to find their efficient portfolios (i.e., maximizing returns while minimizing volatility) to the risk concerns of treasurers. “From this point of view,” he says, “I wanted to see if back-testing the evolution of rates over time could provide any guidance in terms of efficient portfolios.” What he wanted to find out was whether or not there is a specific combination of tenors “that would allow me to have an efficient funding portfolio but also minimize the average cost and volatility.” The debt CAPM model for treasurers, then, tries to tackle this challenge and “allows me to understand if history can be of any help when treasurers decide what tenors to issue.”