Recently the standard market model has been used to examine holding period returns of corporate bonds. These studies have involved issues such as: the impact of accounting earnings data on bond price behavior [5]; the relationship of bond betas and ratings [19, 21]; the effect of ratings changes on bond prices [27]; the relationship of bond betas to duration and yield [3, 13, 15]; bond performance of bankrupt and nonbankrupt firms [26]; and tests of the Capital Asset Pricing Model based on bond returns [7]. While the empirical appropriateness of applying the market model to common stock returns has been demonstrated, similar tests have not been conducted with regard to long-term corporate bonds. Section II of this paper will examine the assumptions of the normal error regression model when used in the form of the market model and applied to a sample of long-term corporate bonds during the early years of their lives. The issue of systematic changes in the regression parameters will be addressed in Section III. Lastly, conclusions will be presented in Section IV.

Two methods for deriving efficient sets involve either the Markowitz [3] approach, where every security can be viewed as being related to an index unique to itself, or the Sharpe [4] single-index model, where every security is related to the same index. Given the extreme differences between these models, Cohen and Pogue [1] developed two intermediate models. They found that the efficient set derived from the Sharpe single-index model came closer to approximating the Markowitz model's efficient set than their models when empirically tested on a sample of common stocks. Subsequently a similar test was performed by Wallingford [6] which yielded contradictory conclusions.

The general framework employed in analyzing diversification among securities involves the mean-variance theory of portfolio selection described by Markowitz [8]. Observation of the securities comprising the efficient set indicates which financial sssets possess attributes (i.e., expected return and covariances) making them worthwhile components of an optimally diversified portfolio. This paper will be concerned with forming an efficient set from four security classes—common stocks, preferred stocks, corporate bonds, and U.S. government bonds (hereafter denoted CS, PS, CB, and GB, respectively). The first objective will be to derive and analyze an efficient set from a sample of these securities in order to determine which securities have potential benefits for diversification.

This paper examines the problem of deriving Black's (1972) minimum-variance zero-beta portfolio. Long's (1971) methods, used by Morgan (1975), are briefly mentioned. Then the complementary pivot algorithm of Lemke (1965), which has been shown to be capable of deriving the optimal solution to certain quadratic programming problems that are subject to a non-negativity constraint, is described. Finally, Lemke's algorithm is shown to be capable of deriving the minimum-variance zero-beta portfolio efficiently from samples of risky assets where both long and short positions are allowed by reformulating the problem so as to avoid the difficulties encountered by having a non-negatively constraint.

The effect of short selling on the composition and location of the efficient set has been analyzed in a variety of ways. However, the situation typically facing investors where the initial margin requirement is less than 100 percent and the riskfree interest rate that is paid on the short proceeds is less than the rate paid on initial margin has not previously been considered. The Elton‐Gruber‐Padberg algorithm (1976, 1978), subject to certain modifications, is shown here to be capable of identifying the efficient set under such conditions.

In their article, “Beta as a Random Coefficient, ” Fabozzi and Francis [1] present evidence which suggests that beta is a random coefficient for a “significant minority” of NYSE stocks. They obtained their evidence first, by characterizing the market model as a random coefficient model of the type described by Theil and Mennes [7], and second, by estimating its parameters for a sample of NYSE stocks over the period December 1965 through December 1971. This paper describes weaknesses in Fabozzi and Francis' implementation of the estimation procedures of Theil and Mennes [7] and Hildreth and Houck [3]. Improvements are suggested and utilized in an analysis of the returns of 683 NYSE stocks over the period January I960 through December 1971. The results of the analysis indicate that Fabozzi and Francis have overstated the case for beta being a random coefficient of the form described by Theil and Mennes.

Beta coefficients were initially defined by Sharpe [11] as the slope term in the simple linear regression function where the rate of return on a market index was the independent variable and a security's rate of return was the dependent variable. As indicated by Brenner and Smidt [4], accurate estimation of beta coefficients is important for at least two reasons. First, they are important for understanding risk-return relationships in capital market theory. Second, they are important for use in making investment decisions. Some confusion has appeared, however, in recent research regarding both the optimal estimation interval and the intertemporal stability of beta coefficients. The purpose of this paper is to examine this confusion and present new evidence on the estimation and stability of beta.

Financial intermediaries often use stress testing to set risk exposure limits. Accordingly, we examine a model with an agent who faces stress testing constraints and another who does not. Three results are obtained. First, when there are K* binding constraints, the constrained agent's optimal portfolio exhibits (K*+2)-fund separation. Second, the effect of the constraints on the optimal portfolio is identical to that of an adjustment in the expected payoffs of the risky securities that tends to lower them. Third, a security's equilibrium expected return depends on both its systematic risk and its idiosyncratic returns in the states where the constraints bind.

The U.S. Securities and Exchange Commission is committed to having exchanges fully implement decimal pricing by April 9, 2001, and is also considering revising the Uptick Rule. We consider the likely impact of the pending smaller tick size associated with decimalization on the efficacy of this rule by examining the execution quality of NYSE short-sell orders immediately before and after the tick size was reduced from 1/8th to 1/16th in 1997. We conclude that, in general, short market orders will receive better execution after decimalization, but at-the-quote limit orders will receive worse execution, and suggest revisions to the Rule. Journal of Economic Literature Classification Numbers: G18, K22.