|
Article Excerpt
Tim Hudson (*)
Abstract
This paper suggests an investment strategy which allows an investor to specify the desired return on investment to be equal to the expected rate of inflation plus a certain premium rate, and then helps the investor select those stocks which will provide the greatest chance of meeting that specified investment goal. (JEL 040)
Introduction
Although the incentive for holding stock investments over a long period may vary from one investor to another, such investments could become an effective hedge against inflation if stocks have been carefully selected. The feasibility of using stocks as an inflation hedge has been the focus of several previous studies [Oudet, 1973; Branch, 1974; Fama and MacBeth, 1974; Bodie, 1976]. In these studies, a stock is defined as an inflation hedge if its real return is independent of the rate of inflation.
In 1990, analysts at Merrill-Lynch conducted an experiment, which involved a portfolio composed of equal shares of the ten highest-yield stocks in the Dow Jones Industrial Average. Once a year, in January for instance, the portfolio would be adjusted to replace those stocks that no longer rank among the top ten. This investment strategy will be referred to as the practitioners' method. Over a 20-year period (1968-1988), such portfolio strategy yielded a total return of 1,557 percent. During the same period, inflation increased by 353 percent. This strategy would have rewarded an investor with a real return of about 1,200 percent over 20 years.
Based on the results of this experiment, we propose to offer an alternative definition of the term "inflation hedge" as applied to common stocks as follows: a stock is an inflation hedge as long as it will yield a return exceeding the rate of inflation. This definition suggests that an investor might be able to formulate a long-term stock portfolio by specifying the desired return on investment as equal to the expected rate of inflation plus a certain premium rate, and then select, according to some rules, those stocks which will provide the greatest chance of meeting that specified investment goal.
From an academic viewpoint, it would be interesting to examine if the practitioners' method will always yield the highest possible return. Additionally, it would be of value if a model could be devised which will systematically analyze stock return expectations, and thereby allow the selection of the optimal long-term stock investment portfolio. The purpose of this paper is to develop such a model, hereafter referred to as the Inflation-Hedge Stock Investment Portfolio Model, or simply the IHSI model.
The implementation of the IHSI model requires the use of information about the expected long-term return on stocks. Several prior studies have addressed the question of estimating stock returns [Sharpe, 1964; Nelson, 1976; Kwan, 1984; and Flannery and James, 1984], but only in the short-run, which makes none of the methods suggested by these studies suitable for a model with the specific goal of providing long-term protection against inflation via investment in stocks. What is needed then is a new methodology that will take stock return estimates beyond the immediate short-term basis. This methodology will be explained in the second section of this paper.
Historical data on Dow Jones Industrial Average stocks will be used in the third section to illustrate how the newly developed methodology can be used to provide long-term estimates of stock returns. The IHSI model will be presented in the fourth section. In the fifth section, an example will be given to show how the information generated in the third section can be used in conjunction with the IHSI model to formulate an inflation-hedge portfolio. Empirical testing of the model will be described in the final section. The paper will conclude with some remarks and suggestions for future research.
Term Structure of Stock Returns
The task now is to develop a model for estimating long-term stock returns. Long-term stock returns can be related to short-term returns in a manner similar to the way long-term interest rates are related to short-term interest rates [Chambers, Carleton, and Walsman, 1984; Hilliard, 1984; and Campbell, 1987]. In the interest-rate case the relationship is referred to as the term structure of interest rates. Similarly, the relationship between long-term stock returns and its relevant short-term determinants will be referred to as the term structure of stock returns. To examine the term structure of stock returns, a two-factor model similar to that suggested by Stone [1974] is used. The model is specified as follows:
[R.sub.j, t+1] = [[alpha].sub.0] + [[alpha].sub.1][R.sub.i, t+1] + [[alpha].sub.2][R.sub.m, t+1] + [v.sub.j, t+1], (1)
where [R.sub.j, t+1] is the expected return on the jth stock for the holding period t + 1; [R.sub.i, t+1] is the interest rate factor index for the same holding period; [R.sub.m, t+1] is the selected equity market index for the same holding period; and [V.sub.j, t+1] is the random disturbance.
To formulate a portfolio using (1), the user has to project the [R.sub.m] and [R.sub.i] over a specified period ending, for example, at time t + 1. Unless [R.sub.m] and [R.sub.i] can be obtained from outside sources, and are constantly available, it would be necessary to follow a systematic procedure to determine them before [R.sub.j] can be estimated.
It has been suggested that long-term returns may be extrapolated by a weighted distribution of historical short-term returns and other relevant factors measuring economy and financial market conditions [Bollerslev, 1987; O'Brien, 1988; Lindahl, 1992; Abbasi, 1993; Bamber, 1997]. Because the level of interest rates and stock returns are highly correlated [Ibbotson and Sinquefield, 1976; Bodie and Rosansky, 1980; Flannery and James, 1984; Campbell, 1987; Carnes and Slifer, 1991], one can be used as a proxy for the other. This explains the presence of the term [R.sub.i, t+1] in (1).
Among those non-interest rate factors that need to be considered when estimating [R.sub.j, t+1], the most commonly cited...
|
