FMP 17: Corporate Bonds

Privately placed bonds frequently never trade, the original buyers keeping them to maturity. Publicly offered bonds change hands over the counter, unlike shares, which trade on exchanges. What stands in for an exchange is a network of dealers, trading for their own portfolios or for clients and earning the gap between what they pay and what they achieve. An excess of buyers lifts prices and an excess of sellers lowers them.

What the yield is made of

The yield on a bond is the return earned over its life assuming every payment arrives as promised. One piece is a risk-free return, available on a comparable instrument with no default risk; the other is a credit spread, paid for the possibility of a default. Debt a government issues in its own currency is often called risk-free, since the currency can be printed, though developing countries do sometimes default even then and members of the euro area cannot print euros at all.

Stretch the maturity of a well-rated bond and its credit spread tends to widen, since every extra year is another chance for the issuer to strike trouble. Prices and yields move against each other. If the market expects rates to rise, investors hold out for higher yields and prices fall across the market; an expectation of falling rates works in reverse. News about one company works instead on the spread demanded from it alone.

Liquidity and the dealer inventory

Bond dealers are market makers, quoting bid and ask prices on request and holding inventory. That inventory is an exposure: rising risk-free rates or widening spreads cut bond prices and the dealer loses on what it holds.

Liquidity, meaning how readily an asset converts into cash within a reasonable period at a reasonable price, varies enormously: some bonds trade several times a day, others a handful of times a year. Heavy volume comes with narrow bid-ask spreads and thin volume with wide ones, and part of every yield pays for liquidity risk. The Volcker rule, a component of the Dodd-Frank legislation in the United States, limits how far banks may trade securities for their own account. American banks argue that it drains bond liquidity, since holding inventory for market-making purposes is what it constrains.

Every coupon-paying bond carries two prices. What a dealer quotes is the clean price. What changes hands on settlement is the cash price, called the dirty price, which is the clean price with accrued interest added. Accrued interest is what the seller has earned since the last coupon date without being paid, and the buyer reimburses it before collecting the whole next coupon.

Counting the days

Corporate bonds in the United States accrue interest on a 30/360 basis: every month is assumed to hold 30 days and every year 360, so only the dates at each end matter and a leap year makes no difference.

A bond indenture is the legal contract between issuer and bondholders, and it pins down what matters: the maturity date, the size and timing of interest payments, any callable and convertible features, and the rights holders acquire if the issuer violates it.

Three families of covenants

Most indentures also carry covenants. A negative covenant, also called a restrictive covenant, fences the issuer in, limiting further debt financing, sales of assets, dividend payouts and buybacks of shares. A positive covenant obliges the issuer to act: produce regular financial statements, maintain its properties, carry insurance, spend the money raised as the offering document promised. A financial covenant holds certain financial ratios at acceptable levels, interest coverage and leverage being the usual pair. Interest coverage sets earnings before interest and taxes against interest expense; leverage ratios describe how much of the business is financed with debt. An indenture may also set out what follows a rating downgrade, or a change of control, abbreviated to CoC. Highly creditworthy firms generally issue with few covenants, while riskier firms accept long lists, because lenders want protection where the balance sheet alone does not persuade.

The corporate bond trustee

The issuer appoints and pays a corporate bond trustee, normally a bank or a trust company, and a federal statute in the United States requires one for bond issues above USD 5 million involving interstate commerce.

The trustee looks after the interests of the bondholders and confirms compliance with the indentures. It reports to holders periodically, acts for them when the need arises, and may serve as paying agent and registrar, handling record keeping and passing interest and principal through. Its duties are itemised in the indentures and it need not exceed them. Some indentures let it take the issuer’s own word, or that of the issuer’s attorneys, on whether covenants are being met, in which case no independent investigation is required.

Moody’s, S&P and Fitch publish opinions on how creditworthy bond issuers are. The top of the Moody’s scale is Aaa, given to bonds seen as having virtually no chance of default, and the ladder runs down through Aa, then A, Baa, Ba, B, Caa, and finally Ca and C. The other two agencies use symbols mapping onto the same rungs, writing AAA where Moody’s writes Aaa, AA for Aa, BBB for Baa. Every broad category except Aaa and Ca is cut into three notches, so Moody’s writes Aa1, Aa2 and Aa3 inside Aa, and A1, A2 and A3 inside A, while S&P and Fitch use plus and minus signs.

Source: the rating scales set out in the chapter, with the grade column added.

Who pays, and what the rating tells you

Public bond issuers normally pay the agencies to rate them, although nothing obliges them to seek a rating. On its face that invites a conflict of interest, since the fee comes from the borrower and not from the investors reading the result. Reputation holds the line: an agency seen to bend toward paying clients loses the standing its opinions rest on.

A rating is a statement about default risk and nothing wider. The agencies publish it in two forms: the probability that a bond with a given rating defaults within n years, for a range of values of n, and the probability that it migrates between rating categories over a stated horizon such as one year or five years. Losses arriving without a default, from widening spreads, thin liquidity or a general rise in interest rates, fall outside what a rating measures.

Everything above the threshold in the ladder is investment grade. Everything below collects interchangeable labels: high-yield, non-investment grade, speculative grade, or plain junk. The line falls at Baa3 for Moody’s and BBB- for the other two agencies, so one notch of downgrade moves an issue from one world to the other.

Three routes into the high-yield market

Young and growing companies form the first group: prospects may be excellent, but they lack the operating record and strong financial statements of an established firm. Fallen angels form the second, issuers once rated investment grade whose finances deteriorated until the rating followed. The third arrives deliberately, since a company with stable cash flows can pile on debt to benefit shareholders, and a leveraged buyout run by a private equity firm is the classic route.

Payment features you rarely see on investment grade paper

These issuers frequently need structures that ease the early cash burden. A deferred-coupon bond pays nothing for a stated period and then begins paying a specified coupon in the ordinary way. A step-up bond raises the coupon as time passes. A payment-in-kind bond settles interest with additional bonds instead of cash, which matters when covenants on other issues forbid a cash payment. Some issues also let the borrower call the bond out of the proceeds of an equity issue.

An extendable reset bond resets its coupon annually or more often, the reset chosen to hold the price at some target such as USD 101. That structure has a flaw. Let the company run into financial difficulty and the coupon needed to hold the target becomes extremely large, draining cash from a company that already lacks it and forcing a higher coupon still at the next reset. A mechanism built to protect the holder can accelerate the failure of the issuer.

Two credit exposures live inside one corporate bond, and only one requires anybody to actually fail. Credit default risk is the risk that the issuer does not make the payments it promised. Credit spread risk is the risk that the market changes the price it charges for bearing credit risk, so the spread widens while the issuer keeps paying on time. The first destroys cash flow; the second destroys market value, with no change in the rating or in the ability to pay. Agencies address the first directly, through default probability and rating transition tables, while the second is market-wide.

What spread widening looks like in numbers

In calm conditions an A-rated bond with seven years to run pays a spread of approximately 100 basis points. Put the seven-year risk-free rate at 3% and the bond yields 4%. Stressed conditions differ sharply. When investors turn averse to risk they shift money toward the safest instruments available, a movement known as a flight to quality, and the same spread can reach 2% or even 3%. No default is needed anywhere in that description; the whole move is a repricing of risk.

Sensitivity to such a repricing is captured by spread duration, which approximates how far the price of a bond moves, in percentage terms, when its credit spread rises by 100 basis points and the risk-free rate stays put. A spread duration of four means that widening costs 4% of the price.

Event risk covers discrete happenings that damage a bond with no gradual deterioration in front of them: a natural disaster wrecking a company’s main facility, or the death of a chief executive officer the business depends on. The variety that matters most to bondholders is a large increase in the leverage of the issuer.

Leverage is the dangerous case precisely because it is chosen, and chosen by people acting for shareholders. In a leveraged buyout a private group of shareholders borrows to buy out the existing shareholders, and the debt lands on the company. Share buybacks, where shares are repurchased from investors, do a milder version, as do certain mergers and acquisitions and various restructurings. The existing bondholder is left holding a claim on a business that turned riskier overnight, at a coupon fixed when it was safer.

The case that made the market pay attention

Kohlberg, Kravis, Roberts & Co bought RJR Nabisco in a leveraged buyout in 1988, and the consequences travelled well beyond one company. The transaction was worth USD 25 billion, and the leverage it created drove the credit spread on the existing bonds from 100 basis points to 350 basis points. Holders of solid investment grade paper watched it reprice as high-yield debt, and the whole corporate bond market was marked down once investors saw that other issuers could be treated the same way.

The maintenance of net worth clause

Indentures for lower-rated issues sometimes anticipate exactly this and add a maintenance of net worth clause, obliging the firm to keep the value of its equity above a prescribed level. Should equity fall below it, the firm must start retiring debt at par until equity climbs back above the threshold. In some versions the company need only offer to retire the debt at par, and each holder decides. Holders normally accept: a breached covenant means the company is in poor shape, so its bonds are unlikely to trade above par, and taking par beats selling into the market.

Corporate bonds are sorted along several dimensions at once. The first is who is borrowing.

Source: the issuer categories and examples in the chapter, with the third column added. Bonds sold in the United States by issuers in the last category are called Yankee Bonds.

Maturity and interest rate

Original maturity on a corporate bond is at least one year, and anything shorter is commercial paper. Issues maturing within five years are usually called short-term notes, those running between five and 12 years are medium-term notes, and anything beyond 12 years belongs among long-term bonds. Several mechanisms can repay all or part of the principal early.

Fixed-rate bonds pay one rate for life, occasionally in a foreign currency. Floating-rate bonds, also called floating-rate notes or variable rate bonds, pay a floating reference rate plus a spread: a quarterly payer observes the reference rate for each period and adds a fixed 20 basis points. Such a bond can equally be built from a fixed-rate bond and an interest rate swap.

Zero-coupon bonds pay nothing until maturity and sell at a discount to principal instead. In a bankruptcy, a coupon-bearing holder in the United States usually claims the principal, whereas a zero-coupon holder claims what was originally paid plus the interest accrued on it. A zero also carries no reinvestment risk, since nothing arrives to be reinvested, whereas a coupon bond redeploys every payment at prevailing rates, so falling rates leave it behind a zero.

A default leads either to a reorganization or to a liquidation of assets, and a secured bondholder does better in both: paid first from the proceeds of selling the collateral in a liquidation, and negotiating from strength in a reorganization, since the asset can credibly be seized.

Secured structures

A mortgage bond pledges specific assets, homes and commercial property being typical. Default lets the holders sell those assets to satisfy what they are owed, although the courts usually have to give permission first. The terms may also restrict future issues and how far assets bought later can be pledged. An after-acquired clause goes further, requiring property acquired after issue to serve as collateral for the existing bonds.

A collateral trust bond pledges securities instead of property: shares, bonds or other instruments issued by another company, usually a subsidiary of the issuer. Voting rights follow the credit. Absent a default the issuer normally votes the pledged shares, which matters at shareholder meetings; after a default the corporate bond trustee votes them for the bondholders, whose interests may diverge from those of the issuer’s own shareholders. Terms often demand additional collateral if the appraised value of the pledge slips below a stated level.

An equipment trust certificate finances the purchase of a specific asset, aircraft being the common application. Title vests in the trustee, who leases the asset to the borrower for payments large enough to deliver the lenders their promised return, and the borrower takes title once the debt is repaid. The advantage is procedural: the trustee already owns the asset, so no proceedings are needed after a default and it simply leases the aircraft to somebody else.

Unsecured structures and where they rank

Debentures are unsecured, meaning no collateral has been posted. They rank below mortgage bonds and collateral trust bonds and consequently pay more. A debenture indenture commonly caps how many further debentures may be sold, because each new one dilutes the claim of existing holders on the same pool of assets. Suppose holders would recover 30 cents on the dollar, that is 30% of principal, in a default or liquidation. Had twice as many been permitted, with no other significant general creditors, the same assets would spread across twice the claims and recovery would fall to 15 cents on the dollar. A negative pledge clause protects from the other side, stopping the issuer from securing new bond issues on its assets where that weakens the debenture holder’s position.

A subordinated debenture sits below other debentures and other general creditors in a bankruptcy, so those claims are settled first, and the compensation is a higher interest rate. A guarantee works differently: one company, often the parent, guarantees a bond issued by another, and the holder should receive the promised interest and principal unless issuer and guarantor both fail. The higher the correlation between the financial performance of the two, the more likely they fail together, and the less the guarantee is worth.

Plenty of bonds simply run to maturity and are retired with the proceeds of a fresh issue. Two things tempt an issuer to act sooner. Falling interest rates make replacing an expensive issue with a cheap one attractive, and the saving to the issuer is exactly the loss to the holder, who redeploys the money at the lower rate now available. The second temptation is contractual: a change in the business, or better financial health, can leave covenants that were reasonable at issue looking burdensome, and paying the holders off early is the way to be rid of them.

The call schedule

An indenture may permit the issuer to call the bond, meaning buy it back from the holder, with a schedule setting out when calls are allowed and at what price. That call price generally starts above par at issue and steps down toward par as maturity approaches. A bond is typically not callable during its first few years, and this call protection shields holders from an early fall in interest rates.

Calls and convertible bonds working together

Some bonds convert into equity on terms agreed in advance, and the conversion option is usually paired with a call feature. Left alone, a holder postpones conversion as long as possible, because keeping the bond preserves protection against a fall in the share price. An issuer who wants the bonds turned into equity, so that new debt can be raised, forces the matter by calling once the share price has risen far enough that converting beats handing the bond back at the call price.

Make-whole calls

A make-whole call provision replaces the fixed schedule with a calculation. The call price is worked out as the present value of the interest and principal the holder has not yet received, discounted at the risk-free rate with a stated spread added. For a United States bond with five years left, that discount rate might be the five-year Treasury rate with 10 basis points on top. Such structures cost the holder nothing, so holders do not demand a higher return at issue for granting the option and the issuer gains flexibility without a premium coupon.

A sinking fund retires bonds periodically before maturity, so the debt shrinks in stages rather than falling due in one lump. Two routes satisfy it. The issuer can pass money to the bond trustee, who retires bonds at par value. Or it can buy its own paper in the open market and hand those bonds to the trustee, which is cheaper whenever the bonds sell below par. An accelerated sinking fund provision allows more than the schedule specifies.

Why the debt is made to shrink

The reason is usually the collateral. Assets pledged as security, plant and equipment being the obvious case, depreciate as they age, so holding the principal steady lets coverage deteriorate. A sinking fund can be calibrated so that the amount borrowed falls in step with the collateral as its value drains away. Requirements can sometimes be met by adding to the property held as collateral rather than retiring bonds, and the addition normally exceeds what is required, so the security improves.

Maintenance and replacement funds, asset sales and tender offers

A related arrangement, the maintenance and replacement fund, requires the issuer to keep the value of the collateral intact through property additions, failing which cash must retire debt instead. The difference is one of intent: meeting a sinking fund obligation with property usually increases the collateral, while a maintenance and replacement fund maintains its value rather than building it up.

An indenture normally permits a company to sell pledged assets on one condition, that the proceeds retire the bonds those assets backed, so selling property is a further route to early retirement.

The last route is a tender offer, an offer to buy the bonds from whoever will sell. The price named can be fixed, or worked out as a present value of the cash flows still to come, discounted at the risk-free rate with a pre-specified spread added, exactly as for a make-whole call. What separates it from a call is consent: a call is an option the issuer holds and holders must comply, while a tender offer is a proposal any holder may refuse.

A default happens once an issuer stops making the payments it agreed to make. Everyone owed money then holds a claim against its assets, and two paths open. The company can reorganize in negotiation with its creditors, or its assets can be sold so that the claims, bondholders included, are met from the proceeds.

What Chapter 11 buys the company

Bankruptcy law in the United States is written to make reorganizations workable, and a Chapter 11 filing gives a company time to negotiate with bondholders and other creditors, and throughout it the firm’s own executives stay in control, barred only from actions such as selling fixed assets, arranging new loans, and stopping or expanding operations.

A reorganization can end in several ways: part or all of the business may be sold, the amount owed on loans cut, the interest rate reduced, and debtholders may end up as equity holders. An immediate liquidation instead forces asset sales at distressed prices and ends the business, so reorganization preserves value that liquidation destroys. Many large American companies have filed under Chapter 11 and come through it, General Motors, Kmart and United Airlines among them.

Who gets paid first

Once a default has happened, what decides how much each investor recovers is the ranking of claimants, meaning which claims are satisfied first from the funds available. One rule is absolute: bondholders always rank above equity holders. Below that the picture is set by contract. Holders of one bond issue may rank ahead of holders of another, and some bonds rank ahead of trade creditors such as suppliers owed money for goods delivered. Working out where a bond sits means reading its indenture, checking what collateral has been pledged, and confirming whether the issue is subordinated.

Rating agencies publish two statistics used constantly: the default rate and the recovery rate. Annual default experience can be expressed in two ways, and the answers can diverge sharply. An issuer default rate counts how many bonds defaulted and divides by the number of issues outstanding. A dollar default rate takes the par value that defaulted and divides by the par value outstanding, so the first treats a small issue and a giant one as equals while the second weights every default by size.

Recovery rates

A defaulted bond rarely leaves its holder with nothing. Tracking what claimants eventually collect is slow, so the recovery rate is read off where the bond trades a few days after the default, against par value. A bond changing hands at USD 40 per USD 100 of face value straight after defaulting therefore has a recovery rate of 40%, and loss given default, one minus the recovery rate, is 60%.

Published research reports several findings. Average recovery across defaults comes to 38%, and the distribution of outcomes is bimodal. One finding records no link between average recovery in a year and that year’s default rate, while another records a negative link, so the pair sit awkwardly together. Downturns and distressed industries both depress recovery, and industries built on tangible assets recover more, which supports the negative reading, since heavy default years are the years assets sell badly.

Why the distribution has two peaks

Bimodality comes from seniority.

Combining the two statistics gives the expected loss rate on a bond over a year.

Set the annual default probability at 0.5% alongside a recovery rate of 40%. Multiplying 0.5% by a loss given default of 0.6 produces an expected loss rate of 0.3%. Assume nothing is recovered and the same probability yields 0.5% instead.

Expected return on a corporate bond has three components: begin at the risk-free rate, add the credit spread, and take away the loss expected from defaults.

The spread is larger than the loss rate at every rating

It would be natural to assume that a bond paying 100 basis points, that is 1%, expects to lose 1% a year to defaults, leaving an expected return equal to the risk-free rate. The data say otherwise. Results reported by Hull (2018) for bonds maturing in about seven years put expected loss well below the spread in every rating category, and the excess broadly widens as credit quality falls, with the Ba category standing out because its excess exceeds that of the B category.

Source: results from Hull (2018) as presented in the chapter, ordered from the weakest rating upward, the final column being spread divided by loss rate.

That final column, absent from the published table, sharpens the excess column. A holder of Aaa paper is paid 39.00 times the loss expected, while a holder of Caa paper is paid 1.53 times. One entry does not reconcile with the two figures beside it, since 3.22 and 1.30 differ by 1.92 rather than the 2.92 recorded; the published values stand as they are, and the point that the Ba excess beats the B figure of 1.83 survives either reading.

Where the extra return comes from

One explanation is that Treasuries are the wrong yardstick. Traders use the Treasury rate as their risk-free reference, but research suggests a higher benchmark, the inter-bank borrowing rate for instance, suits corporate bond yields better. Swapping it shrinks the numbers without touching the conclusion, since expected return still exceeds the risk-free rate and the spread still exceeds expected default loss. Poor liquidity is a second candidate, yet research indicates the liquidity component is relatively small and cannot account for the gap.

The main explanation is that bonds do not default independently. Default rates run low while the economy is doing well and climb through a recession, so defaults arrive in clusters exactly when investors can least afford them. That is systematic risk, tied to the performance of the market as a whole, and diversification does not remove it, so traders have to be paid to carry it. Non-systematic risk can in theory be diversified away, but doing so is far harder in bond markets, because each bond contributes only a small probability of default: around 30 well-chosen stocks eliminate non-systematic risk in equities, while thousands of bonds are needed for the same effect. Traders may therefore be paid for both kinds of risk at once.