اوراق قرضه شرکتی ایالات متحده آمریکا: مطالعه ناهنجاری های بازار بر اساس گروه های صنعتی وسیع

We examine three major U.S. corporate bond market indices for calendar-based anomalies over the period 1982–2002. The analysis covers the entire corporate bond market and two broad industry classes: industrials and utilities. We find mixed support for the weekend effect in the overall bond index and the industrials index and to a lesser extent in the utilities index. We also show strong evidence of a January effect. This paper not only updates the study of corporate bond market anomalies through the period 2002 but also is the first examination based on broad industry classes.

Over the past three decades numerous studies have shown that there exists a “weekend effect” in the stock market and several other markets wherein Monday returns tend to be significantly negative when compared with returns of other days of the week. Recent studies of Brusa, Liu, and Schulman (2000) and Mehdian and Perry (2001) show that there is a reversal in this effect whereby Monday returns in the U.S. stock markets tend to be positive and greater than other days' returns after 1988. Evidence is also provided in further studies by Brusa, Liu, and Schulman (2003a) that this “reverse weekend effect” is a unique feature of the U.S. stock market. They show that, while the U.S. stock market shows a reverse weekend effect, foreign markets show a traditional weekend effect or no effect at all. In another work, Brusa, Liu, and Schulman (2003b) show that the reverse weekend effect exists not only in broad indices but also in most industries. They find that Monday returns tend to be positive in the post-1988 period for both broad market indices and industry indices. Their conclusions are valid even after considering the influence of the month-of-the-year and week-of-the-month effects. In the present study, we examine whether there exists a reverse weekend effect in the U.S. corporate bond market. We also examine for industry effects based on two major industry groupings, industrials and utilities, and check for the month-of-the-year and week-of-the-month effects. Our motivation stems from the fact that, soon after the existence of the original weekend effect was found in stock markets, it was found that other markets appeared to be affected in the same way. Based on this reasoning, there is a possibility that U.S. bond markets also exhibit a reverse weekend effect like the stock markets. In this study, three separate bond indices are examined. While there appears to be no weekend effect in the bond returns based on the model developed by French (1980), we find that Monday returns in the post-1987 period are significantly less than the average returns for the rest of the week based on the model developed by Connolly (1989). We also find evidence of the turn-of-the-year effect in bond returns. While there appears not to be enough significant difference between industries to draw any important inferences, some effects appear to be isolated based on industry. Seasonal patterns like the day-of-the-week effect and week-of-the-month effect have been found in long-term debt markets.2 Some of the major studies have examined the impact of seasonalities in the bond markets. Schneeweis and Woolridge (1979) examine the existence of seasonality in U.S. monthly bond holding period returns over the period 1952–1977. Using both parametric and non-parametric tests they find significant differences in the mean returns by month of the year. They report high rates of return in January and October and also report a shift in the later years of their sample period. Using data over the years 1963 through 1982, Chang and Pinegar (1986) document a January seasonal in the U.S. market for corporate bonds that becomes more evident as the bond rating declines. Smirlock (1985) examines the January effect in a variety of debt instruments over the period 1953–1981. He reports seasonality in low-grade debt instruments in that they produce substantially higher returns in January than in any other month. This was not the case with Treasuries and high-grade corporate bonds. Flannery and Protopapadakis (1988) examined intra-week seasonality in three stock indices and Treasury bonds with seven different maturities. They report that significant intra-week seasonalities exist in Treasury returns as well as stocks. They show that return seasonals are not uniform across securities, and they also show that negative Monday returns may be subject to a unified explanation across all securities. Jordan and Jordan (1991) show that, for the period of 1963–1986, corporate bond returns exhibit January, turn-of-the-year and week-of-the-month effects but no significant day-of-the-week or turn-of the-month effects. In another interesting study, Singleton and Wingender (1994) reproduce the Treasury and stock returns from the sample period of Flannery and Protopapadakis (1988), and they use the Connolly (1989) method for trimming outliers. They find that the weekend effect disappears when outliers are eliminated. Recent works include that of Starks, Yong, and Zheng (2003), who document a January effect for the municipal bond closed-end funds. They also provide evidence that the observed January effect can largely be explained by the tax-loss selling activities at the end of the previous year. Since Jordan and Jordan's (1991) study, no other major study has looked at the weekend effect in corporate bond markets. We feel that there is a need to re-examine this issue using recent data, especially in light of what has been found in the stock markets after 1987. This serves as motivation for our study. In a remarkable study, Reilly, Kao, and Wright (1992) mention three major reasons as to why the analysis of bond market indexes is important. Their reasons are that (1) the bond portfolios of both pension funds and individuals have grown in recent years; (2) bond index funds have become increasingly popular because those who monitor the performance of numerous bond portfolios have discovered that, similar to equity asset managers, most bond portfolio managers have not outperformed the aggregate bond market; and (3) there will be substantial growth in research related to the bond market. Their reasons mentioned in 1992 are more important today than they ever were before. Of the total amount of $6223 billion bond market instruments outstanding in 1994,3 only 40.2% or approximately $2501.65 billion were corporate bonds. By 2004 the size of the bond market instruments outstanding had increased to $11,626.6 billion of which corporate bonds had a share of 57.9% or approximately $6731.80 billion. These data indicate not only the magnitude of increase in the size of the bond markets but also the growth of the corporate bond market into the major part of the total bond market instruments outstanding. Thus, a study examining anomalies of the corporate bond market would be well received by both the academic and bond investment communities. This is especially true because of the significant increase in the number of bond mutual funds over the past decade and a half. The number of bond mutual funds was only 170 in 1980 and grew to 1046 by 1990 and to 2041 by 2004.4 Considering the competition among bond mutual funds and the size of the corporate bond market, it is important for us to study bond market anomalies. In the current study, we examine bond market indices as well as check for industry-specific effects. A study of this nature would be helpful for bond portfolio managers to make decisions with regard to the impact of market anomalies on their buy and sell decisions. This might enable them to better time their buy and sell decisions while reducing risk and increasing return. There is another reason for checking industry-specific differences and bond market anomalies. Elebash (1994) notes that state and local government pension plans, like investors in general, show less enthusiasm for bond index funds than they show for stock index funds. It is possible that a study of bond market indices similar to the numerous ones that examine stock market anomalies might generate further interest in bond market funds and their indexing. In this study we examine three different bond indices, two of which are industry-specific. This is the first study to compare and contrast the market anomalies in different industry-specific bond price indices. It is apparent from earlier studies that the quality of bond ratings affects the returns on bonds. However, no earlier study has examined industry-based differences on bond returns. A study that checks for differences and similarities across various bond industry indices would thus help bond portfolio managers to better time their trades. In addition, this study is the first major study on the corporate bond markets since Jordan and Jordan's (1991) study. Thus, it can provide us an update on the status of seasonalities in the U.S. corporate bond market as well as help us to understand whether there exist any similar or dissimilar pattern(s) in two industry-based bond indices. Since our data span nearly two decades, we can glean evidence of any change in the bond markets' anomalies over this period. The rest of the paper is organized as follows: We describe the data in the next section, followed by empirical evidence in Section 3. We conclude in the last section.

In the present study we examine daily returns on three separate bond indices for market anomalies over the period 1982–2002. We find that Monday returns are not significant using the French (1980) model. However, we find that Monday returns are significantly lower than the average returns for the rest of the week using the Connolly (1989) model. This is different from the results of Jordan and Jordan (1991), who found no significant day-of-the-week effects, using data prior to 1986. This is possibly due to changes in the nature of the equity market and its impact on the bond market. This explanation is lent credence by recent studies which show that, as information alters expectations in one market, it causes traders to adjust their holdings across markets, producing an information spillover and thereby leading to cross-market hedging. We also report turn-of-the-year effects consistent with the Jordan and Jordan (1991) study and other earlier studies. The most important reason for this could be tax-loss selling, especially considering that the bonds represented in the Dow Jones Indices are among the more popular ones in the industries they represent. While there appears to be some differences in market anomalies based on the nature of the industry, they are not different enough to draw significant conclusions. The only industry-based effect that appears to be significant is that there is less evidence of a turn-of-the-year effect in the utilities index as compared with the composite and industrial indices. It appears that the day-of-the-week effect was related to the week-of-the-month effect prior to 1988 in two of the bond indices under study, but the effect seems to have disappeared in the post-1987 period. This study contributes to existing literature in several ways. It updates the market anomalies observed in corporate bond returns since the Jordan and Jordan (1991) study. The data for that study covered the period 1963–1986. The data used in this study cover a more recent time period. As seen in the study, there appears to be evidence of a traditional weekend effect in the corporate bond markets using the Connolly (1989) model. We also find that the turn-of-the-year effect first reported by Jordan and Jordan (1991) continues to be apparent in the bond indices. The practical implications of this study for bond fund managers are that it helps them to time their purchases and sales of bonds based on the day-of-the-week and month-of-the-year. Investors and fund managers alike can use this information to benefit from the market anomalies. For example, if the stock market continues to exhibit the reverse weekend effect and the bond market continues to exhibit the traditional weekend effect, hybrid fund managers can set up their trades to simultaneously benefit from both markets. Considering recent evidence presented by Fleming et al. (1998), and Connolly et al. (2005), managers are indeed looking for cross-hedging opportunities.