Voice from the Past
Chapter 18

At a closing in New York, some time in the late 1970's, I was told of a new device in municipal bond issues: one or more maturities of an issue bear no current interest but are sold at a discount that produces a significant yield on redemption at par when due. My informant, a bond dealer, was excited because the bond was so well received by the market; it sold bearing a modest fraction of the yield on ordinary bonds that paid interest every six months. He predicted, correctly, that as soon as this device caught on, zero coupon bonds would be sold in such numbers that their yield would soon rise to approximately that of conventional bonds.

A few months later I was asked to prepare documents for an issue that included such bonds in the final maturities. I asked why only those maturities, and was told that this was what the numbers people ordered. Why the numbers people chose only those maturities was a mystery, like so many of their demands.

This was in the days after the introduction of personal computers but before spreadsheets. To persuade a computer to be of any help you had to write a program for it. I had learned to write BASIC programs, and decided to see how much difference it made in the amount of money required to pay a bond issue of the same yield to maturity if zero coupon bonds were included in various different maturities. Assuming that the annual payments of principal and interest combined were the same to the nearest $5,000 or so, it turned out that including such bonds in the last maturities did in fact reduce the cost of repaying the issue. On the other hand, if the zero coupon bonds fell due in the earlier years, it would take more money to pay the bond issue. This experience left me wondering about the fickleness of mathematics.

Eventually someone pointed out that because, under normal yield curves the late maturing bonds of an issue bear interest at a higher rate than the early maturities, eliminating the payments of relatively high-rate interest from the early years makes more money available to pay principal in these years. This shortens the average life of the issue and allows a greater proportion of the issue to comprise shorter, lower rate bonds. At first this sounded to me more like a rationale than a reason, but when I ran the numbers on various hypothetical bond issues, it turned out to be correct. This feature might come in useful when necessary to keep the average life of the issue within some limit imposed by law, regulation, or documents governing prior issues. I never had to face the question of how zero-coupon bonds would be treated for State debt limit or election requirement purposes. Since the amount of money actually borrowed at the time of issue of such a bond is relatively small, the initial value of the bond might put the issue within the limit on the date of delivery; yet if no bonds are retired in the next few months, an excess of debt might result from the increase of the principal amount of the bond every six months by the amount of accrued interest. Alternatively, if the amount of the debt is determined only at delivery, and if the issue containing the zero coupon bonds is then within the limit, the increase might use up future debt incurring capacity under a constitutional or statutory debt limit. It might also be that, if the voters authorize only the initially issued amount, the transubstantiation of interest into principal violates the election requirement.

The 1980's became a time of very high interest rates. The psychology of investors was bemusing because some considered that this was an excellent time to buy long-term investments; yet others felt that they should not invest long because rates might go even higher. In the case of zero coupon bonds, the effect of a change in market interest rate makes much more of a difference in price than when current interest bonds are involved; and this effect increases with the length of the maturity. So what now seems a sure way to have gotten rich looked too risky to some. As tax exempt rates exceeded ten per cent per annum, this latter group were guided more and more by fear that rates would keep on going up.

However, there were investors willing to take that chance with some of their money. I know of one who paid about $28,000 for the entire final maturity of a 32-year issue of single family mortgage revenue bonds. It comprised a zero-coupon bond in the face amount of $1,000,000, yielding 11.559%. Needless to say, it was subject to early call for redemption, and that call was exercised, bit by bit over the years, until nothing was left. Yet it was a good ride as long as it lasted. This points out another advantage to the issuer of zero coupon bonds in those cases (such as single family mortgage revenue bonds) where there will probably, but not certainly, be a lot of money (such as from the prepayment of mortgages) available to redeem bonds prior to maturity: the greatest future liabilities can be reduced at particularly great advantage when the compounding effect is terminated early.

The market for zero coupon bonds was not limited to those issued by States and local governments. Salomon Brothers and Merrill Lynch came to market with CATS and TIGRS. Each such obligation comprises a right to payment from the proceeds of a certain future installment of interest on, or principal of, a government bond held in trust. They sold well enough so that the U.S. Treasury decided to get into the business, too, with STRIPS.

Anyone who has tried to determine which SLGS (State and Local Series Government obligations) would best fill an advance refunding escrow, with its fixed demand for payments of principal of and interest on refunded bonds, knows how awkward, and often inefficient, it is. Consider trying to fit hexagonal pegs into octagonal holes. At one time, when Treasury - IRS proposed regulations governing SLGS, I suggested that they ought to be issuable as zero coupon obligations; this would eliminate the problem of having to keep interest received on SLGS lying around unused for months, and would make filling the escrow almost trivial; whoever was assembling the SLGS would simply buy a SLG for each payment out of the escrow at whatever available discount was convenient, so long as the yield on the entire package of SLGS did not exceed the yield limit. The idea got nowhere.

One feature of zero coupon municipal bonds has puzzled me: their treatment for State income tax purposes. In general, States base taxable income for State purposes on Federal taxable income with adjustments, including the addition of interest on municipal bonds not exempt under State law. It is clear that interest on zero yield corporate bonds is Federally taxable as it accrues from year to year, and not entirely when paid at maturity. So, by analogy, one would think that accruals of interest on zero coupon municipal bonds would be taxable by the States from year to year in the amount of each accrual. However, the accrual concept for corporate bonds is specifically established by Section 1272 of the Internal Revenue Code. If State law declares municipal bond interest taxable in the year received, and does not cover accruals, the analogy (even though supported by sound public policy) and the law are at odds. I vaguely recall a common law doctrine to the effect that if a person voluntarily pays a tax under a mistaken interpretation of law, he can't get it back. This situation might make a good subject for someone who chooses to write about the present, rather than the past.

Manly W. Mumford