Citation Context

...inear Factor Models with Time-Varying Coefficients We begin by imposing virtually no theoretical structure, appealing instead to a well-known existence theorem to motivate our empirical approach (see =-=Harrison and Kreps 1979-=-). This theorem states that, in the absence of arbitrage, there exists a stochastic discount factor, or pricing kernel, of R i,t�1, M t�1, such that, for any traded asset with a net return at time t t...

Citation Context

...Friend 1973; Fama and MacBeth 1973). But recently, evidence has mounted that the CAPM is inconsistent with numerous empirical regularities of cross-sectional asset pricing data (see, e.g., Basu 1977; =-=Banz 1981-=-; Shanken 1985; Fama and French 1992, 1993). Perhaps most damning, the CAPM has demonstrated virtually no power to explain the cross section of average returns on assets sorted by size and book-to-mar...

Citation Context

...of empirical work finds that expected excess returns on aggregate stock market indexes are predictable, suggesting that risk premia vary over time (see, e.g., Shiller 1984; Campbell and Shiller 1988; =-=Fama and French 1988-=-, 1989; Campbell 1991; Hodrick 1992; Lamont 1998; Lettau and Ludvigson 2001). Yet if risk premia are time-varying, parameters in the stochastic discount factor will depend on investor expectations of ...

Citation Context

.... A large and growing body of empirical work finds that expected excess returns on aggregate stock market indexes are predictable, suggesting that risk premia vary over time (see, e.g., Shiller 1984; =-=Campbell and Shiller 1988-=-; Fama and French 1988, 1989; Campbell 1991; Hodrick 1992; Lamont 1998; Lettau and Ludvigson 2001). Yet if risk premia are time-varying, parameters in the stochastic discount factor will depend on inv...

Citation Context

...tunately, these models have also proved disappointing empirically. The consumption-based model has been rejected on U.S. data in its representative agent formulation with timeseparable power utility (=-=Hansen and Singleton 1982-=-, 1983); it has performed no better and often worse than the simple static CAPM in 1 Liew and Vassalou (2000) provide a first step in establishing an empirical link between the size and book-to-market...

Citation Context

...it stock; and X(s,) is the sensitivity function specified in Campbell and Cochrane. Similar but more complicated expressions can be derived for internal habit formation models (e.g., Sundaresan 1989; =-=Constantinides 1990-=-) by taking a local linear approximation of the respective stochastic discount factors. To our knowledge, almost all the cross-sectional tests of the CCAPM to date have pertained to models that assume...

Citation Context

...uch as l(s t) in (8), their fluctuations should be well captured by suitable proxies for time-varying risk premia. In contrast to traditional, discrete-time derivations of the conditional CAPM (e.g., =-=Jagannathan and Wang 1996-=-), which produces just two betas (one for the fundamental factor and one for the risk premium), habit models imply the existence of at least one beta for the multiplicative “cross term” of the factor ...

Citation Context

...s returns on aggregate stock market indexes are predictable, suggesting that risk premia vary over time (see, e.g., Shiller 1984; Campbell and Shiller 1988; Fama and French 1988, 1989; Campbell 1991; =-=Hodrick 1992-=-; Lamont 1998; Lettau and Ludvigson 2001). Yet if risk premia are time-varying, parameters in the stochastic discount factor will depend on investor expectations of future excess returns. For example,...

Citation Context

...tion for asset i with constant betas given by ′ E[R i,t�1] p E[R 0,t] � bl, (4) where E[R 0,t] is the average return on a zero-beta portfolio that is uncorrelated with the stochastic discount factor (=-=Black 1972-=-), and b { ′ �1 Cov (f, ¯ ¯f ) Cov (f, ¯ R i,t�1) is a vector of regression coefficients from a multiple regression of returns on the variable factors. In the empirical analysis that follows, we focus...

Citation Context

... expected excess returns on aggregate stock market indexes are predictable, suggesting that risk premia vary over time (see, e.g., Shiller 1984; Campbell and Shiller 1988; Fama and French 1988, 1989; =-=Campbell 1991-=-; Hodrick 1992; Lamont 1998; Lettau and Ludvigson 2001). Yet if risk premia are time-varying, parameters in the stochastic discount factor will depend on investor expectations of future excess returns...

Citation Context

...o table 1. Dc denotes consumption growth. models with only financial factors. Such arguments for large zero-beta estimates have a long tradition in the cross-sectional asset pricing literature (e.g., =-=Black et al. 1972-=-; Miller and Scholes 1972). However, if the sampling error is not independently and identically distributed and if multiple factors are poorly measured, there is little that can be said about the dire...

Citation Context

... Blume and Friend 1973; Fama and MacBeth 1973). But recently, evidence has mounted that the CAPM is inconsistent with numerous empirical regularities of cross-sectional asset pricing data (see, e.g., =-=Basu 1977-=-; Banz 1981; Shanken 1985; Fama and French 1992, 1993). Perhaps most damning, the CAPM has demonstrated virtually no power to explain the cross section of average returns on assets sorted by size and ...

Citation Context

...k market indexes are predictable, suggesting that risk premia vary over time (see, e.g., Shiller 1984; Campbell and Shiller 1988; Fama and French 1988, 1989; Campbell 1991; Hodrick 1992; Lamont 1998; =-=Lettau and Ludvigson 2001-=-). Yet if risk premia are time-varying, parameters in the stochastic discount factor will depend on investor expectations of future excess returns. For example, in models with habit formation, in whic...

Citation Context

... this argument for two reasons. First, second-stage GMM estimation is unsuitable in studies that have a small time-series sample relative to the cross-sectional sample size (Ferson and Foerster 1994; =-=Altonji and Segal 1996-=-; Christiano and Den Haan 1996; Hansen et al. 1996; Ahn and Gadarowski 1999). Our quarterly data yield far fewer time-series observations than in many previous 31 We use the term “second-stage” loosel...

Citation Context

...ates of the parameters in ̂t cay are “superconsistent,” converging to the true parameter values at a rate proportional to the sample size T rather than proportional to �T as in ordinary applications (=-=Stock 1987-=-).1256 journal of political economy TABLE 1 Fama-MacBeth Regressions Using 25 Fama-French Portfolios: l j Coefficient Estimates on Betas in Cross-Sectional Regression Row Constant cay ̂t Factors t�1 ...

Citation Context

...ow. The scaling variable is ̂cay.sresurrecting the (c)capm 1241 explaining the cross section of average asset returns (Mankiw and Shapiro 1986; Breeden, Gibbons, and Litzenberger 1989; Campbell 1996; =-=Cochrane 1996-=-); and it has been generally replaced as an explanation for systematic risk by a variety of portfolio-based models (e.g., Elton, Gruber, and Blake 1995; Fama and French 1996). Moreover, the CCAPM perf...

Citation Context

...M estimation is unsuitable in studies that have a small time-series sample relative to the cross-sectional sample size (Ferson and Foerster 1994; Altonji and Segal 1996; Christiano and Den Haan 1996; =-=Hansen et al. 1996-=-; Ahn and Gadarowski 1999). Our quarterly data yield far fewer time-series observations than in many previous 31 We use the term “second-stage” loosely here to refer to any GMM estimation in which the...

Citation Context

.... A large and growing body of empirical work finds that expected excess returns on aggregate stock market indexes are predictable, suggesting that risk premia vary over time (see, e.g., Shiller 1984; =-=Campbell and Shiller 1988-=-; Fama and French 1988, 1989; Campbell 1991; Hodrick 1992; Lamont 1998; Lettau and Ludvigson 2001). Yet if risk premia are time-varying, parameters in the stochastic discount factor will depend on inv...

Citation Context

...ggregate stock market indexes are predictable, suggesting that risk premia vary over time (see, e.g., Shiller 1984; Campbell and Shiller 1988; Fama and French 1988, 1989; Campbell 1991; Hodrick 1992; =-=Lamont 1998-=-; Lettau and Ludvigson 2001). Yet if risk premia are time-varying, parameters in the stochastic discount factor will depend on investor expectations of future excess returns. For example, in models wi...

Citation Context

...one “stage.”resurrecting the (c)capm 1277 empirical applications for which monthly data were available. Both the optimal GMM weighting matrix of Hansen (1982) and the secondmoment matrix of returns (=-=Hansen and Jagannathan 1997-=-) are likely to be poorly estimated in samples of the size encountered here. Indeed, for this reason, it is often argued that first-stage GMM should serve as a robustness check on the results of any s...

Citation Context

...nts of returns. A large and growing body of empirical work finds that expected excess returns on aggregate stock market indexes are predictable, suggesting that risk premia vary over time (see, e.g., =-=Shiller 1984-=-; Campbell and Shiller 1988; Fama and French 1988, 1989; Campbell 1991; Hodrick 1992; Lamont 1998; Lettau and Ludvigson 2001). Yet if risk premia are time-varying, parameters in the stochastic discoun...

Citation Context

...atrix is employed. We make this argument for two reasons. First, second-stage GMM estimation is unsuitable in studies that have a small time-series sample relative to the cross-sectional sample size (=-=Ferson and Foerster 1994-=-; Altonji and Segal 1996; Christiano and Den Haan 1996; Hansen et al. 1996; Ahn and Gadarowski 1999). Our quarterly data yield far fewer time-series observations than in many previous 31 We use the te...

Citation Context

...tic discount factor will in general not be constant. Although predictable movements in volatility may be a source of variation in b,, they appear to be more concentrated in high-frequency data (e.g., =-=Christoffersen and Diebold 2000-=-). Since risk-free interest rates are also not highly variable, the denominator of b, is not likely to vary much in monthly or quarterly data. On the other hand, a large empirical literature (cited in...

Citation Context

...5) has long been a pillar of academic finance, and early evidence seemed to favor the theory’s central tenet that the market portfolio be mean-variance efficient (see Black, Jensen, and Scholes 1972; =-=Blume and Friend 1973-=-; Fama and MacBeth 1973). But recently, evidence has mounted that the CAPM is inconsistent with numerous empirical regularities of cross-sectional asset pricing data (see, e.g., Basu 1977; Banz 1981; ...

Citation Context

...; Fama and MacBeth 1973). But recently, evidence has mounted that the CAPM is inconsistent with numerous empirical regularities of cross-sectional asset pricing data (see, e.g., Basu 1977; Banz 1981; =-=Shanken 1985-=-; Fama and French 1992, 1993). Perhaps most damning, the CAPM has demonstrated virtually no power to explain the cross section of average returns on assets sorted by size and book-to-market equity rat...

Citation Context

...es consumption growth. models with only financial factors. Such arguments for large zero-beta estimates have a long tradition in the cross-sectional asset pricing literature (e.g., Black et al. 1972; =-=Miller and Scholes 1972-=-). However, if the sampling error is not independently and identically distributed and if multiple factors are poorly measured, there is little that can be said about the direction of bias. Procedures...

Citation Context

...he pricing errors are generated using the Fama-MacBeth regressions in table 3 below. The scaling variable is ̂cay.sresurrecting the (c)capm 1241 explaining the cross section of average asset returns (=-=Mankiw and Shapiro 1986-=-; Breeden, Gibbons, and Litzenberger 1989; Campbell 1996; Cochrane 1996); and it has been generally replaced as an explanation for systematic risk by a variety of portfolio-based models (e.g., Elton, ...

Citation Context

...re significantly, several investigations have found that these tests, which rely on a consistent estimate of the variance-covariance matrix of pricing errors, have poor small-sample properties (e.g., =-=Burnside and Eichenbaum 1996-=-; Hansen, Heaton, and Yaron 1996). The small sample size of the Wald tests exceeds the asymptotic size, and this is especially true when the number of moment restrictions is large relative to the time...

Citation Context

...itable in studies that havesa small time-series sample relative to the cross-sectional sample sizes(Ferson and Foerster 1994; Altonji and Segal 1996; Christian0 and DensHaan 1996; Hansen et al. 1996; =-=Ahn and Gadarowski 1999-=-). Our quar-sterly data yield far fewer time-series observations than in many previouss'' We use the term "second-stage" loosely here to refer to any GMM estimation in whichsthe matrix used to weight ...

Citation Context

...t stock; and l(s t) is the sensitivity function specified in Campbell and Cochrane. Similar but more complicated expressions can be derived for internal habit formation models (e.g., Sundaresan 1989; =-=Constantinides 1990-=-) by taking a local linear approximation of the respective stochastic discount factors. To our knowledge, almost all the cross-sectional tests of the CCAPM to date have pertained to models that assume...

Citation Context

...; Fama and MacBeth 1973). But recently, evidence has mounted that the CAPM is inconsistent with numerous empirical regularities of cross-sectional asset pricing data (see, e.g., Basu 1977; Banz 1981; =-=Shanken 1985-=-; Fama and French 1992, 1993). Perhaps most damning, the CAPM has demonstrated virtually no power to explain the cross section of average returns on assets sorted by size and book-to-market equity rat...

Citation Context

...tic discount factor will in general not be constant. Although predictable movements in volatility may be a source of variation in bt, they appear to be more concentrated in high-frequency data (e.g., =-=Christoffersen and Diebold 2000-=-). Since risk-free interest rates are also not highly variable, the denominator of bt is not likely to vary much in monthly or quarterly data. On the other hand, a large empirical literature (cited in...

Citation Context

... significantly,sseveral investigations have found that these tests, which rely on a con-ssistent estimate of the variance-covariance matrix of pricing errors, havespoor small-sample properties (e.g., =-=Burnside and Eichenbaum 1996-=-;sHansen, Heaton, and Yaron 1996). The small sample size of the Waldstests exceeds the asymptotic size, and this is especially true when thesnumber of moment restrictions is large relative to the time...

Citation Context

...5) has long been a pillar of academic finance, and early evidence seemed to favor the theory's central tenet that the market portfolio be mean-variance efficient (see Black, Jensen, and Scholes 1972; =-=Blume and Friend 1973-=-; Fama and MacBeth 1973). But recently, evidence has mounted that the CAPM is inconsistent with numerous empirical regularities of cross-sectional asset pricing data (see, e.g., Basu 1977; Banz 1981; ...

Citation Context

...ggregate stock market indexes are predictable, suggesting that risk premia vary over time (see, e.g., Shiller 1984; Campbell and Shiller 1988; Fama and French 1988, 1989; Campbell 1991; Hodrick 1992; =-=Lamont 1998-=-; Lettau and Ludvigson 2001). Yet if risk premia are time-varying, parameters in the stochastic discount factor will depend on investor expectations of future excess returns. For example, in models wi...