A Long‐Term Perspective on the Determinants of Treasury Bond Stripping Levels

• DOI: http://doi.org/10.1111/fmii.12018
• Published date: 01 November 2014
• Date: 01 November 2014

A Long-Term Perspective on the Determinants

of Treasury Bond Stripping Levels

BYMARCK BULTER,MILES LIVINGSTON,AND LEI ZHOU

We examine the proportion of individual Treasury bonds held as strips over the entire

history of the STRIPS program. First, we document a secular decline in the Treasury bond

stripping levels from 1985 to 2010, coincidental with the long-term decline in the interest

rates. This pattern suggests that investors purchase strips to avoid reinvestmentrisk and to

lock in the high interest rates in the 1980s and 1990s. Second, higher coupon and longer

maturity bonds are shown to be more heavily stripped. Third, the suspension of new issues

of 30-year bonds from 2001 through 2006 created a gap in the maturity structure of Treasury

bonds and induced heavy stripping of 30-year bonds issued post 2006. Our findingssuggest

that stripping is motivated by several factors, including interest rate risk management, tax

concerns and market completion.

I. INTRODUCTION

In January 1985, the U.S. Treasury introduced the STRIPS (Separate Trading of

Registered Interest and Principal of Securities) program, which allows the coupon

and principal payments of eligible Treasury notes and bonds to be traded sepa-

rately as zero-coupon bonds, called Treasury strips. Two years later, the Treasury

allowed reconstitution of the original Treasury notes and bonds by rebundling

corresponding Treasury coupon strips and principal strips. The STRIPS program

has been a huge success since its introduction in 1985. As of December 2009,

$152 billion, or 21% of the par value of all outstanding Treasury bonds, were

held in stripped form. In addition, the amount of stripping and reconstituting is

very large. For example, in the single month of December 2009, $25.4 billion of

Treasury bonds were stripped and $23.7 billion were reconstituted.

Over time, strips haveplayed an increasingly important role in government bond

markets. Following the U.S. Treasury, 15 other countries introduced similar pro-

grams for their Treasury debts as well, including Germany, France, UK, Canada,

Japan, Spain, Italy, Sweden, and the Netherlands. Because underlying bonds can

be both stripped and reconstituted, arbitrage between strips and underlying bonds

means that strips are the basic building blocks of the government bond markets,

resembling Arrow-Debreu state securities. In addition, the yields on strips are

increasingly used by the market to measure the term structure of interest rates, as

evidenced by Merrill Lynch maintaining indexes of strips with maturities from 1

to 30-year.

Despite its growing importance in government bond markets of the United

States and other countries, there is relatively little analysis on this important

topic. Furthermore, the few studies on this active market, using data in the 1990s,

Corresponding author: Miles Livingston, University of Florida, miles.livingston@Warrington.ufl.edu

C2014 New YorkUniversity Salomon Center and Wiley Periodicals, Inc.

180 Bulter et al.

focus on potential arbitrage opportunities and monthly stripping and reconstitu-

tion activities. Conspicuously missing is an investigation of the determinants of

the relative level of stripping of individual Treasury bonds. While a significant

amount of Treasury bonds are held in stripped form, there is very large cross sec-

tional variation in the stripping of individual Treasury bonds. For example, as of

December 2009, less than 8% of the February 2029, 5.25% coupon rate Treasury

bond was held in stripped form, while more than 60% of the May 2030, 6.25%

coupon rate Treasury bond was held in stripped form. In addition, there is signif-

icant intertemporal variation in stripping of Treasury bonds. In the early 1990s,

about 45% of the par value of all eligible Treasury bonds was held in stripped

form. The level of stripping has dropped to about 20% in the late 2000s.

This paper attempts to explain the cross-sectional and intertemporal variations

in the Treasury bond stripping levels.1Our empirical findings shed some light

on the three major theories on the Treasury bond stripping: the Interest Rate

Risk Hypothesis (Sundaresan, 2002), the Tax Hypothesis (Livingston & Gregory,

1989), and the Market Completion Hypothesis (Grinblatt & Longstaff, 2000).

According to the Interest Rate Risk Hypothesis, some investors want to avoid

the reinvestment risk of bond coupons by holding strips. In addition, if investors

expect future interest rates to decrease, they will hold strips to lock in the high

prevailing interest rates. The Tax Hypothesis indicates that a portfolio of strips

has a greater after-tax value than an underlying Treasury bond for upward sloping

term structures. The Market Completion Hypothesis argues that strips allow some

investors to create cash flow sequences that are unavailableusing coupon bearing

bonds.

Our empirical findings are consistent with all three hypotheses, suggesting

that Treasury stripping is not driven by one single consideration but by several

factors. First, we find a secular decline in the percentage of Treasury bonds held as

strips from 1985 to 2010. This secular decline is coincidental with the long-term

decline in the level of interest rates. This positive correlation between the level

of Treasury bond stripping and the level of general interest rates is illustrated in

Figure 1, which plots the monthly percentage of par value of Treasury bonds held

in stripped form and the monthly average daily yields on 10-year Treasuries. In

the 1980s and 1990s interest rates were high by historical standards. As a result,

many investorspreferred to hold strips to lock in the historically high interest rates.

Indeed, we estimate that a change of 1% in the 10-year Treasury yield translates

to an approximately 2% to 2.5% change of the Treasury bond stripping level in

the same direction. This empirical pattern is consistent with the Interest Rate Risk

hypothesis.

Second, we find a positive correlation between Treasury bond stripping and

coupon rate. Figure 2 plots the average life-to-date percent of par value held in

1We concentrate on Treasurybonds instead of Treasury notes for two reasons. First, the proportion of

Treasury notes held in stripped form is relatively small. Second, the maturities of Treasury notes are

tightly bunched making it very difficult to distinguish between different motivesfor stripping of the

underlying securities.

Determinants of Treasury Bond Stripping Levels 181

2.00%

3.00%

4.00%

5.00%

6.00%

7.00%

8.00%

9.00%

10.00%

11.00%

12.00%

20.00%

25.00%

30.00%

35.00%

40.00%

45.00%

50.00%

55.00%

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

Yield on 10-Year Treasury

Percentage Held in Stripped Form

Average Percent Held in Stripped Form T-10 Yield

Note: This figure depicts the monthly average of the U.S Treasury bonds held in stripped

form as a percent of the total amount outstanding and the yields on 10-year Treasury notes

from May 1985 to February 2010.

Figure 1: Treasury Stripping and Interest Rates.

stripped form and coupon rate for each Treasury bond. As shown in Figure 2, the

more recently issued Treasury bonds, with much lower coupon rates, are not as

heavily stripped as the high coupon bonds issued in the 1980s and early 1990s.

We estimate that a difference of 1% in coupon rates leads to a 10% difference in

the Treasury bond stripping level. This finding is consistent with both the Interest

Rate Risk and the Tax Hypotheses.

In addition, as shown in Figure 2, almost all bond issues with local maximum

coupon rates are heavily stripped compared to adjacent maturities, suggesting that

patterns of coupon rates of successive issues have an impact on stripping as well.

Third, the non-issuance of 30-year Treasury bonds between 2001 and 2006

created a gap in the maturity spectrum of Treasury securities as indicated by the

gap in the x-axis of Figure 2. The first three Treasurybond issues after the Treasury

re-initiated the 30-year program in February 2006 are actively stripped, despite

their historically low coupon rates. The 30-year Treasury bonds issued beginning

in 2006 play the unique role of allowing the creation of zero-coupon Treasuries

with maturities after 2031, which cannot be otherwise synthesized by alternative

means. This finding supports the Market Completion Hypothesis.