Stock return and inflation risk: the disaggregate model.

Author:Yi, Taihyeup David

    Does stock return risk exist and, if so, what are its determinants? Although a substantial amount of research has focused on inflation and aggregate stock returns, these specific questions about inflation risk have not been thoroughly investigated in the literature. Moreover, most of the empirical research finds that aggregate stock returns are negatively related to a measure of either actual or expected inflation, even though, as claims against real assets, stocks should satisfy the Fisher hypothesis, i.e., the expected inflation coefficient should equal unity (Fama (1981), Geske and Roll (1983), Kaul (1987), and Kaul and Seyhun (1990)). More recently, Boudoukh, Richardson, and Whitelaw (1994) recognize that inflation risk may be an important explanatory factor for the inflation-stock return relationship. They examine industry nominal stock returns and find cross-sectional differences in the covariance between expected inflation and nominal returns. Their results suggest that the stock returns of non-cyclical industries co-vary positively with expected inflation, while the reverse holds for cyclical industries. In a recent extension of Boudoukh, Richardson, and Whitelaw, Pilotte (2003) examines the impact of capital gains and dividend yields on nominal stock returns and finds a significant differential effect, which can be attributed to the presence of excess returns (i.e., an inflation risk premium).

    Analyzing the covariance is not intuitively obvious. On the one hand, this covariance may depend on purely nominal factors such as whether a firm or industry is a net nominal debtor or creditor. However, the results of Boudoukh et al. suggest that real factors influence the covariance because an industry's status as cyclical or non-cyclical is an attribute of the real economy. While the Boudoukh et al. view of the inflation risk factor is macroeconomic in scope, our aim is to determine whether real microeconomic factors also contribute to inflation risk. We use both accounting data and relevant market data to construct the potential microeconomic factors. To be more precise about our microeconomic framework, consider that inflation risk is the risk that future stock returns will be low when future inflation is high. Low returns can occur for many reasons, one of which is associated with adverse cash flow performance. As shown in the work of Campbell and Mei (1993), a positive inflation shock lowers expected future cash flows. One potential effect of this expected-cash-flow change is a shift in the probability distribution of future cash flows towards the negative cash flow region. By itself a greater future negative-cash-flow exposure would not necessarily induce a risk premium. However, if the changed exposure (as it interacts with other information) signals a reduction in the future viability of the firm, a present-day risk premium would be required since the firm's future stock return could be low when inflation was high.

    In our view, as indicators of the relevant characteristics of the firm, market participants use available accounting data such as asset turnover, leverage, and earnings, and market data such as market capitalization and the BE/ME ratio in forming their assessment of the future viability of the firm. The interaction of these real microeconomic variables with expected inflation induces a certain perception of financial risk and thus influences the size of the inflation risk premium embedded in stock returns.

    To compute monthly inflation, we use the consumer price index which is available from St. Louis Fed's website, Economic Data--FRED[R] over the sample period of January 1927 to December 2007. Our sample of firms includes those listed on the NYSE, AMEX, and NASDAQ. The sample consists of 48 industry portfolios. Stock return data are retrieved from Kenneth R. French--Data Library. Disaggregate stock return equations are estimated based on Fama and French (1993) three factor model and Jegadeesh and Titman (1997) momentum factor model. Examining them on portfolio-by portfolio basis, the standard statistical tests of significance provide only a modicum of support for the Fisher inflation hypothesis while the SMB/inflation and HML/inflation coefficients are significantly different from zero at least for about 50 percent of industry portfolios. These findings indicate the influence of inflation on the real economy, and that the interactions between SMB and inflation and between HML and inflation need to be incorporated into the capital asset pricing model.

    The paper is organized as follows: Section 2 presents a theoretical framework for stock return risk and its empirical model. The sample data and empirical findings for disaggregate stock returns are presented in Section 3. Section 4 provides a brief conclusion.

  2. MODEL

    To provide economic content for this risk premium, we assume that it depends on the conditional covariance of expected inflation with selected economic variables, [X.sup.j.sub.t],

    Risk [premiun.sub.i,t-1] = f[[Cov.sub.t-1] ([E.sub.t-1] [TT.sub.t], [X.sup.j.sub.t])] (1) where Et-1 is the conditional expectations operator. Boudoukh, Richardson, and Whitelaw (1994) take a macroeconomic perspective and argue that the left-hand side of (1) co-moves with the covariance of expected inflation and the business cycle and that the extent of the co-movement depends on the correlation of the industry's nominal cash flows with both variables. More recently and also from a macro viewpoint, Pilotte (2003) infers that the risk premium is non-zero in cyclical industries because of the presence of a significantly differential impact of dividend yields and capital gains on total nominal returns, which is consistent with the existence of excess returns. In contrast to Boudoukh, Richardson, and Whitelaw (1994), we use a microeconomic approach to provide a rationale for the f function in (1). We discuss a microeconomic source of risk premia and relate it to the firm's accounting and market data. This part tells us how various microeconomic characteristics of the firm can create low stock returns. Our basic hypothesis is that the covariance between financial distress and expected inflation determines the nominal stock return premium in (1). We assume that the product of the conditional expectation of financial distress and expected inflation proxies for the conditional covariance in (1),


To continue reading