The substitutability of equities and consumer durable goods: a portfolio-choice approach.

• Author: Knapp, Dale J.

Introduction

The stock market crash of October 1987 led many economists to lower their forecast of GNP growth for 1988. The rationale was that the loss of wealth that resulted from the crash would force asset holders to reduce their consumption expenditures. Yet a major slowdown in economic growth was not realized in the months following the crash.

Garner |11~ explains this fact in terms of the life-cycle hypothesis of consumption. He argues that the potential adverse effect of stock market losses on planned consumption would be spread out over the entire lifetime of asset holders so that the immediate effect of the crash on the level of economic activity would be insignificant. The existing empirical evidence tends to support this view.

Brayton and Mauskopf |3~ find that a one dollar decrease in stock market wealth results in a decrease of only 5 cents in consumption spending. Garner |11~ reports that other studies obtain similar results with estimates ranging from three to seven cents. Runkle |22~ suggests that the insignificant impact of the stock market crash on the real sector may be a result of investors considering the wealth accumulated early in 1987 as temporary. He suggests that investors' current attitudes about their wealth are generally reflected in their spending on durable goods. According to this argument, if the stock market gains of the early 1987 were perceived to be permanent, the tremendous rise in equity prices would have manifested itself in a significant increase in purchases of durable goods.

Runkle |22~ suggests that in order to examine whether the stock market gains of early 1987 were considered temporary, the demand for housing in the post-crash period should be examined. Inherent in Runkle's suggestion is the presumption that common stocks and durable goods in general, and housing in particular, are highly substitutable in investors' asset portfolios. Casual observation tends to support this contention.

Using data from 1987 and 1988, Runkle |22~ points out that in the first half of 1987 spending on durable goods was below the level consistent with rising stock prices. Immediately after the crash, spending on durable goods actually increased by over 20 percent. Thus it appears that as stock prices rose, consumption of durable goods declined and as equity prices declined, durable goods consumption increased. This is consistent with the notion that stocks and durable goods may be substitute assets. However, anecdotal evidence can be misleading as it does not take into account the effect of other factors.

While a significant portion of the empirical literature is devoted to estimating the degree of substitutability of alternative assets, the vast majority of the studies are concerned with the relationship between money and various near-monies.(1) What is lacking is an examination of the degree of substitutability between common stocks and consumer durable goods in a multi-asset portfolio choice framework.(2) Such an analysis is the objective of the present paper.

Using the analytical approach made famous by Chetty |4~ and quarterly data covering the period from 1963.4 through 1991.3, we estimate elasticities of substitution between common stocks and residential housing and between stocks and government bonds, Treasury bills, money, the sum of savings and time deposits, and corporate paper.(3) We also test whether these elasticities changed following the 1987 stock market crash.

We find that there is virtually no substitutability between stocks and other financial assets. Moreover, we find no evidence that asset holders are willing to substitute between stocks and housing. This last finding contradicts Runkle's suggestion that as stock returns decline, consumers may move into housing, or other durable goods. In fact, it appears that individuals consider equities to be a requirement in their portfolio, and are not willing to use other assets as substitutes. We also find that, with one exception, the stock market crash of 1987 did not have a significant impact on the substitutability between common stocks and the other assets. The only exception is that, following the crash, stocks and Treasury bills actually became complements.

The remainder of the paper is divided into four sections. The next section provides a brief review of the approaches that have been used to measure the substitutability of alternative assets. This is followed by section III where the methodology of our empirical analysis is presented. The data and the estimation results are described in section IV. The paper concludes with section V which summarizes this work and offers some suggestions for further research in this area.

2. Background

   Two approaches have been used to measure the degree of substitutability of different assets. One approach has been to specify the demand for various assets as a function of the asset's own rate of return and the yields on alternative assets and use it to estimates the (partial) cross-price elasticities. An early study of the substitutability between stocks and other financial assets using this approach is that by Hamburger |12~. Using quarterly data from 1952 to 1960, he found that fluctuations in the value of stocks completely overshadow the short-run movements of other financial assets. He also found that the dividend yield on stocks is not significant in the household demand function for liquid assets. In a later study, Hamburger |13~ found that equities are poor substitutes for the liabilities of financial intermediaries.

   A disadvantage of the asset-demand approach is that the resulting cross-elasticity estimates may not be symmetrical. For example, based on the demand for demand deposits, Feige |7~ finds that these deposits are either independent or weakly complementary to savings deposits and loans. However, based on the demand for savings deposits and loans, he finds that they are weak substitutes for demand deposits. Feige and Swamy |9~ also report this type of asymmetry. Moroney and Wilbratte |20~ point out another disadvantages of the asset-demand approach. They argue that because various interest rates, representing the asset's own yield as well as the yield on alternative assets, are highly correlated, the estimates of cross-elasticities obtained from this approach are typically unreliable.

   Recently, Aivazian, Callen, Krinsky, and Kwan |1~ used portfolio analysis and annual data covering the period 1951-1973 to estimate various elasticities between financial assets held in the household sector. They found that the expected-return elasticities of stocks are independent of all other financial assets in that demand for stocks is not significantly affected by changes in the expected returns on other assets. They also found that the variance elasticities of stocks are dramatically different from all other assets. An increase in the riskiness of stocks leads the household sector to move out of stocks and all other asset classes and into mortgages and long-term government bonds. This suggests that most financial assets are not good substitutes for stocks.

   The second approach to measuring the degree of substitutability of alternative assets is based on the methodology first used by Chetty |4~. In this approach one estimates the elasticity of substitution between different assets by estimating parameters...