Valuation process and market efficiency for US Treasury bonds.
This study investigates the valuation process for fixed income securities, using investors' expectation changes associated with the call announcement of Treasury bonds. Because the call was unanticipated and the potential existed for calling an additional $78 billion of bonds, the Treasury's action provides a unique opportunity to analyze the bond market's reaction to the arrival of new information and hence, bond market efficiency. The study extends current empirical research by focusing on US government issues, which are not subjected to many of the confounding effects associated with corporate issues.
According to basic financial theory, investors value assets based on the present value of expected future cash flows. Therefore, prices change when investors revise either cash flow expectations, assessment of risk, or both. The efficient-market hypothesis posits that prices adjust quickly to relevant new information. The information hypothesis and market efficiency have been analyzed extensively for equity issues,(1) but the literature provides limited research on bond market reactions to new information. The existing bond research analyzes corporate securities, acknowledging that considerations such as taxes, default risk, agency problems, restrictive covenants, and leverage complicate the bond valuation process.(2) In addition, corporate bonds experience problems associated with private placements and infrequent trading.
Although finding a negative market reaction to the Treasury's call announcement is not surprising, we use it to enhance our understanding of fixed income markets by providing empirical evidence supporting efficiency and consistency in a default-free setting. Our methodology is motivated by Jordan, Jordan, and Jorgensen's (1995) arguments concerning the implied put options. Our paper provides empirical evidence on the valuation of put options implicit in US Treasury bonds.
The paper is organized as follows. Sections I and II contain background information on past research and on the callable Treasury bond market. Sections III and IV describe the sample selection process and methodology. Section V contains an analysis of the results, and Section VI concludes the paper.
I. Prior Literature
Most callable bonds research focuses on explaining the call's existence, developing theoretical models based on option pricing theory, explaining corporate decisions surrounding the call, or analyzing stock market reactions to bond-related announcements.(3) Daves and Ehrhardt (1993) and Longstaff (1992) break from the corporate bond focus. Daves and Ehrhardt find that zero-coupon Treasury strips created from principal payments typically trade at higher prices than do otherwise identical zero-coupon strips created from coupon payments. They attribute this to liquidity and reconstruction differences. Longstaff analyzes implicit call options on five US Treasury bonds. He finds overpricing for bonds with various maturities and coupons, and this overpricing persists over time. However, Longstaff cannot explain the implied negative option price through liquidity differences or tax arbitrage. He feels that costs and difficulty in establishing the short component may make arbitrage difficult to implement.
Jordan, Jordan, and Jorgensen (1995) reexamine option values on US Treasuries and note that premium callable bonds behave more like bonds with maturities equal to first call. They use a theoretical, tax-timed, bond-pricing model and show that a comparison of callable bonds with non-callable bonds that mature on the first call date provides a valuation for an implied put option. Examples of nonsynthesized case studies demonstrate that the call features on Treasury bonds have positive value. In addition, a triplets methodology, using a large number of observations, confirms overwhelmingly positive implied option values. Jordan et al show that Longstaff's negative findings result from the method used to measure implicit option values, rather than from mispricing. Both articles conclude that the Treasury follows an optimal call strategy.
II. The US Treasury Market and Callable Bonds
Callable Treasury bonds allow the government to refund the bonds at par. Deferred call provisions restrict calls to the last five years of maturity. The bonds require a call notice four months prior to the bond's interest payment date. During the high interest-rate period of the late 1970s and early 1980s, most long-term Treasury bonds included call provisions, and all issues with maturities between 2007 and 2014 are callable. These bonds allow the government the option of replacing high-coupon bonds with lower-coupon issues if interest rates drop. In 1984, after determining that investors are willing to accept lower yields on bonds that are not expected to experience an early call, the government stopped issuing callable bonds.
Six months prior to the call, the government announced that it was considering redeeming $1.8 billion of 7.5%, 20-year bonds issued in 1973. The bonds had two years remaining until maturity and could be refinanced by issuing two-year notes with a 6.875% coupon. However, the call did not take place. The government then reversed its decision, and it called the issue on October 9. Investors were caught by surprise. Mitchell and Murray (1991) report that the 7.5% T-bond price fell from 101.40625 to 100.75 later that day.(4)
III. Sample Selection
To investigate the valuation effects of the call, we identified all callable Treasury bonds that existed in October 1991. To control for term effects and market movements not associated with the call,(5) we selected a matched-control sample of independent noncallable bonds with a maturity date on or immediately following the call date. The final sample contains 15 callable and 15 noncallable bonds.(6)
Descriptive statistics indicate that although the two samples are not perfectly matched, they do exhibit similar characteristics. The mean, range, and standard deviation of the number of months to first call/maturity are similar for both samples. The mean coupon rate for the noncallable sample is greater than that for the callable sample (10% and 8.7%, respectively). The coupon range for noncallable bonds is slightly larger than that of callable bonds, although the standard deviation is substantially smaller.
IV. Abnormal Returns and Cross-Sectional Methodologies
We obtained security characteristics and daily bid and ask quotes for each bond from the Federal Reserve Bank of New York Composite 3:30 p.m. quotes for the US government securities market. We computed daily bond returns for the mean of the bid and ask prices using the mean-adjusted, abnormal-returns event-study methodology developed for common stocks by Masulis (1980) and modified for bonds by Handjinicolaou and Kalay (1984) (HK).(7) We further adjust HK's methodology by following Datta and Dhillon (1993), introducing the geometric mean as the normal rate of return in obtaining standardized abnormal returns on bonds.(8) We analyze both the April 10 and October 10 announcements.
A cross-sectional analysis investigates whether the extent of bond price changes at the Treasury's announcement relates to specific bond characteristics. Limited Treasury bond issues rule out exact term and coupon matching between the callable and noncallable bonds. We control for these differences by incorporating coupon and time to call into a cross-sectional regression model. To examine the association of bond returns to relevant bond characteristics, and to analyze the consistency of bond market movements with investors' expectations, we perform modifications of the following cross-sectional regression:
[AR.sub.i] = [b.sub.0] + [b.sub.1][CALL.sub.i] + [b.sub.2][COUPON.sub.i] + [b.sub.3][TERM.sub.i] + [b.sub.4][CPCALL.sub.i] + [b.sub.5][TMCALL.sub.i] + [b.sub.6][JOINTINT.sub.i] + [e.sub.i] (1)
In the equation, the dependent variable (AR) is the two-day abnormal return, which includes the date of the Treasury announcement and the date of the subsequent WSJ article. This period corresponds to the window of time in which the information becomes publicly available and investors could revise expectations.
CALL is a bivariate variable in which one indicates a call provision. It tests the differential impact of a call option after controlling for other bond characteristics. A significant coefficient for CALL indicates that the Treasury announcement is associated with differential expectations for callable and noncallable bonds. Since the announcement reveals a change in the Treasury's willingness to call bonds, we expect investors' expectations of the value of callable bonds to drop at the announcement and [b.sub.1] to be negative.
COUPON is the bond's coupon rate. In general, higher-coupon bonds have a shorter duration and are less sensitive to changes in investors' expectations. Therefore, higher-coupon bonds should be less negatively affected by changing expectations, and we expect [b.sub.2] to be positive.
TERM is the number of months to maturity for noncallable bonds and the number of months until first call for callable bonds. Because longer-term bonds are generally more sensitive to changing expectations, we expect [b.sub.3] to be negative. Unlike the event study, the ordinary-least-squares (OLS) model does not control for interest rate movements. Changes in interest rates will also impact the TERM variable.
CPCALL is the product of COUPON and CALL and examines the differential impact of call provisions on higher-coupon bonds. Higher-coupon bonds have a greater risk of being called. Thus, we expect higher-coupon callable bonds to have greater negative reactions to the announcement and [b.sub.4] to be negative.
TMCALL is the product of TERM and CALL and reveals the differential effect of call provisions on longer-term bonds. Longer-term bonds provide greater protection from calls because of the delayed call provision. Thus, we expect longer-term callable bonds to be less negatively affected by the call and [b.sub.5] to be positive.
JOINTINT is the product of 1/TERM, COUPON, and CALL and tests the interactive effect of the call provision, term to maturity, and the coupon rate. A bond is more likely to be called if market rates drop below the coupon rate as the call period approaches. JOINTINT includes the inverse time to maturity, because shorter-term bonds with higher-coupons are the most likely to be called. Thus, we expect [b.sub.6] to be negative.
This section discusses the results of the empirical tests of the response of callable and noncallable bonds to the call announcement for US Treasuries.
A. Event-Study Results
Abnormal returns for April 10, 1991 do not show a significant market reaction. Thus, investors' expectations are consistent with analysts' beliefs that the bonds would not be called. In contrast, the market experienced a significant negative reaction at the October 9, 1991 call announcement.
Table 1 presents the abnormal returns for the difference between the callable and control-matched noncallable Treasury bonds. The first column identifies the trading day relative to the Treasury announcement of the call. The following columns show portfolio mean abnormal bond returns, portfolio mean standardized abnormal returns, t-statistics, and percent of returns that are positive.
During the two weeks prior to the announcement, daily abnormal return differences between callable and control bonds are not significant. These results indicate that investors did not anticipate the call. On the day of the announcement (Day 0), callable bond returns fall 0.18% more than noncallable bond returns. This difference is statistically significant at the 0.01 level with a t-statistic of - 3.42. The announcement returns indicate that the Treasury bond market quickly incorporates investors' changing expectations concerning the probability of a call.
Following the announcement, the abnormal returns between the two samples are not statistically different. The event-study results demonstrate a concentrated market reaction to the Treasury announcement that is more negative for callable bonds. This indicates that the bond market efficiently incorporates new information into the bond valuation process.
The event-study results for Day 0 and Day + 1 might be driven by changes in interest rates and maturity effects rather than the call announcement, as long-term interest rates dropped on October 9 and 10. However, we greatly reduce the potential impact of interest-rate movements by taking the difference in returns of callable and corresponding noncallable bonds. We also estimate the event-study results by constructing a duration-matched control sample of noncallable bonds.(9) We obtain similar results. Moreover, a matched control sample based on final maturity does not indicate statistically significant abnormal returns for Day 0 and Day + 1. These results further strengthen the finding that the reported results are an outcome of the Treasury's call announcement, rather than being due to interest-rate movements and/or maturity effects.(10)
The results provide insight into the valuation of put options implicit in US Treasury bonds. The value of the implied put is the price of the callable bond minus the price of a noncallable bond that matures on the first call date. As Jordan et al. (1995) point out, when interest rates rise and the price of the callable bonds falls, the value of the implied put option should increase as the call becomes less likely. Although the implicit option could influence the differences in returns between samples, the effect should be relevant on all days not just the day of the call announcement. The event-study results, together with interest-rate movements surrounding the call announcement, indicate that the results are not driven by changes in interest rates and that the option effect does not dominate the announcement effect.
B. Cross-Sectional Results
We compute cross-sectional regression estimates on the full sample of 30 callable and noncallable bonds. We estimate the full model and four partial models to examine the sensitivity of the estimates to the exclusion of individual variables. The results are in Table 2. The estimated coefficient for the control variable TERM has the expected negative sign and is significant at the 0.01 level in all models. Thus, longer-term bonds have a larger negative reaction to changing investors' expectations.
The control variable COUPON is not significantly different from zero in any of the models. When CALL is added to the control variables, the estimated coefficient has the predicted negative sign and is significant at the 0.05 level. The negative relation indicates that investors expected callable bonds to be more negatively affected by the announcement.
The differential impact on short-term callable bonds [TABULAR DATA FOR TABLE 1 OMITTED] (TMCALL) and on high-coupon callable bonds (CPCALL) have expected signs and are significant at the 0.01 level when they are added to the basic control variables. The positive coefficient on TMCALL indicates that shorter-term callable bonds are more negatively affected at the announcement, indicating greater concern about the probability of a call. The negative coefficient on CPCALL is consistent with the increased probability that higher-coupon bonds will be called. The coefficient of JOINTINT is negative and significant at the 0.01 level, indicating that the joint effect of high coupons and short term-to-maturity further diminishes the value of callable bonds.
This paper investigates the efficiency of the bond valuation process by analyzing the effect of a call [TABULAR DATA FOR TABLE 2 OMITTED] announcement on the US Treasury bond market. Since the government had not called Treasury bonds over the previous 30 years, many investors were surprised by the Treasury's action. Thus, this announcement provides a unique opportunity to examine the impact of new information on investors' changing expectations of bond prices.
The results of our study extend the literature on market efficiency to the bond market and help to further our understanding of the valuation process for fixed income securities that are not subjected to default risk and other confounding factors. Moreover, the study provides evidence on the behavior of implied put options as discussed by Jordan et al. (1995).
Prior to the October 9, 1991 announcement, callable bond returns fluctuate, but the differences between callable and noncallable bonds are not statistically significant. However, on the day of the Treasury's call announcement, the abnormal return difference between callable bonds and term-matched noncallable bonds is a significant 0.18 % (t-value - 3.42). Following the announcement, the abnormal returns between the two samples are not statistically different. The event-study results demonstrate a swift negative market reaction to the Treasury announcement that was more severe for callable bonds. This indicates that the bond market efficiently incorporates new information into the bond valuation process. When we use a cross-sectional regression to control for general bond characteristics that can impact returns, we find that callable bonds are more negatively affected than noncallable bonds. Additionally, the drop in returns is greater for bonds most likely to be called, i.e., shorter-term callable bonds and higher-coupon callable bonds that are closer to maturity.
Overall, the study demonstrates that the bond market's reaction to new information is similar to that of the stock market. Investors revise their expectations of bond values with the availability of new information. The revision is concentrated around the announcement, and the strength of the reaction is related to those bond characteristics most likely to be affected by the new information.
We thank Brent Ambrose, Ben Greene, Susan Jordan, Eileen Malitz, Joseph Vu, and the referees for helpful comments and suggestions.
1 Klein (1989) provides a survey of research relating to stock market adjustments associated with stock splits. Lev (1989) summarizes stock market research relating to earnings announcements. Haugen (1996) and Shanken and Smith (1996) discuss some of the major issues affecting the analysis of stock market efficiency.
2 Kim, Ramaswamy, and Sundaresan (1993) determine that the valuation of corporate bonds is different from that of default-free Treasury bonds. Mauer, Barnea, and Kim (1991) discuss the capital structure and tax implications associated with corporate bonds. Crabbe and Helwege (1994) test for agency effects of callable corporate bonds that do not exist for government issues.
3 See for example Ederington, Caton, and Campbell (1997) for a discussion of theories for calling convertible bonds. For examples of corporate bond valuation studied, see Datta and Dhillon (1993); Cowan, Nayar, and Singh (1993); Warga and Welch (1993); Handjinicolaou and Kalay (1984); and Dann (1981). Kihn (1996) analyzes municipal bond performance but does not consider call features.
4 The timing of the market price change reported by Mitchell and Murray (1991) is inconsistent with the published data, which indicate the closing bid on Wednesday, October 9 at 101.40625 and the closing bid on Monday, October 15 at 100.75.
5 Fama and French (1993) identify term and default risk as the dominant variables in bond returns; however, default is not an issue with US Treasuries. Flannery and Protopapadakis (1988) and Warga and Welch (1993) indicate that unlike stocks, the characteristics of bonds change through time as their maturity dates approach.
6 We dropped three bonds from the initial sample of 26 bonds because they are "flower bonds" with coupon rates below 5%. We dropped eight more bonds because unique matching noncallable bonds do not exist. See Table 1, footnote a, for a list of the sample bonds. We perform additional tests on a control sample based on the closest coupon rate and a control sample allowing non-unique control bonds. We find no material difference in results.
7 As described in Fabozzi (1993), accrued interest is computed in accordance with the Treasury's convention of "actual/actual" day count by using the ratio of the actual number of days in a month to the actual number of days in a coupon period. (The other convention for day count is 30/360, which means that each month is assumed to have 30 days and each year 360 days.)
8 We compute average bond returns for a 60-day comparison period ending 16 trading days prior to the WSJ articles on the Treasury call announcement. We also examine the arithmetic mean to determine the sensitivity of the event-study results and obtain similar results. A more complete discussion of the returns methodology is available by request from the authors.
9 The duration-control sample matches the maturity to first call for the callable bonds, and to the maturity on or immediately after first call for noncallable bonds.
10 We thank an anonymous reviewer for raising this issue.
Cowan, A.R., N. Nayar, and A.K. Singh, 1993, "Calls of Out of the Money Convertible Bonds," Financial Management (Winter), 106-116.
Crabbe, L.E. and J. Helwege, 1994, "Alternative Tests of Agency Theories of Callable Corporate Bonds," Financial Management (Winter), 3-20.
Dann, L., 1981, "Common Stock Repurchases: An Analysis of Returns to Bondholders and Stockholders," Journal of Financial Economics (June), 113-138.
Datta, S. and U.S. Dhillon, 1993, "Bond and Stock Market Response to Unexpected Earnings Announcements," Journal of Financial and Quantitative Analysis (December), 565-577.
Daves, P.R. and M.C. Ehrhardt, 1993, "Liquidity, Reconstitution, and the Value of US Treasury Strips," Journal of Finance (March), 315-329.
Ederington, L.H., G. L. Caton, and C.J. Campbell, 1997, "To Call or Not to Call Convertible Debt," Financial Management (Spring), 22-31.
Fabozzi, F.J., 1993, (ed.), Bond Markets, Analysis and Strategies, Englewood Cliffs, NJ, Prentice Hall.
Fama, E.F. and French, K.R., 1993, "Common Risk Factors in the Returns on Stocks and Bonds," Journal of Financial Economics (February), 3-56.
Flannery, M.J. and A.A. Protopapadakis, 1988, "From T-Bills to Common Stocks: Investigating the Generality of Intra-Week Return Seasonality," Journal of Finance (June), 431-450.
Handjinicolaou, G. and A. Kalay, 1984, "Wealth Distributions or Changes in Firm Value: An Analysis of Returns to Bondholders and Stockholders around Dividend Announcements," Journal of Financial Economics (March), 35-63.
Haugen, R.A., 1996, "Finance from a New Perspective," Financial Management (Spring), 86-97.
Jordan, B.D., S.D. Jordan, and R.D. Jorgensen, 1995, "A Reexamination of Option Values Implicit in Callable Treasury Bonds," Journal of Financial Economics (June), 141-162.
Kihn, J., 1996, "The Financial Performance of Low-Grade Municipal Bond Funds," Financial Management (Summer), 52-73.
Kim, J., K. Ramaswamy, and S. Sundaresan, 1993, "Does Default Risk in Coupons Affect the Valuation of Corporate Bonds? A Contingent Claims Model," Financial Management (Autumn), 117-131.
Klein, L.S., 1989, "Stock Distributions: A Review and Synthesis of the Literature," Journal of Accounting Literature, 165-180.
Lev, B., 1989, "On the Usefulness of Earnings and Earnings Research: Lessons and Direction from Two Decades of Empirical Research," Journal of Accounting Research (Supplement), 153-192.
Longstaff, F.A., 1992, "Are Negative Option Prices Possible? The Callable US Treasury-Bond Puzzle," Journal of Business (October), 571-592.
Masulis, R.W., 1980, "The Effects of Capital Structure Change on Security Prices: A Study of Exchange Offers," Journal of Financial Economics (June) 139-177.
Mauer, D.C., A. Barnea and C. Kim, 1991, "Valuation of Callable Bonds Under Progressive Personal Taxes and Interest Rate Uncertainty," Financial Management (Summer), 50-59.
Mitchell, C. and A. Murray, 1991, "Treasury Calls Bonds for Early Repayment," Wall Street Journal (October 10), C1.
Shanken, J. and C.W. Smith, 1996, "Implications of Capital Markets Research for Corporate Finance," Financial Management (Spring), 98-104.
Warga, A. and I. Welch, 1993, "Bondholders' Losses in Leveraged Buyouts," Review of Financial Studies (Winter), 959-982.
We thank Brent Ambrose, Ben Greene, Susan Jordan, Eileen Malitz, Joseph Vu, and the referees for helpful comments and suggestions.
Linda S. Klein is an Associate Professor of Finance at University of Connecticut. Dogan Tirtiroglu is an Associate Professor of Finance at Concordia University.
|Printer friendly Cite/link Email Feedback|
|Author:||Klein, Linda S.; Tirtiroglu, Dogan|
|Date:||Dec 22, 1997|
|Previous Article:||An international comparison of derivatives use.|
|Next Article:||Biases in arithmetic and geometric averages as estimates of long-run expected returns and risk premia.|