Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.633243
Title: Modelling consumption asset pricing models : empirical evidence from the UK
Author: Sherif, Mohamed A
Awarding Body: University of Manchester
Current Institution: University of Manchester
Date of Award: 2005
Availability of Full Text:
Access through EThOS:
Abstract:
This thesis adopts a range of different methodologies in an attempt to evaluate the performance of consumption-based asset pricing models. In particular, it sets out to investigate the relationship between asset prices, consumption and investment decisions. Different utility functions are used in an attempt to examine their roles in pricing assets, reducing pricing errors and solving the equity premium puzzle. First considered is whether the value of relative risk aversion can be changed by using parametric tests and different utility functions. Whereas the power utility model introduces a basic learning framework of the relation between consumption and asset returns, there is general consensus that there is evidence against the model as an asset pricing tool and for its ability to resolve the equity premium puzzle. Within the context of representation of agent models, various studies have attempted to introduce more general preferences. In this study, tests are made of the traditional CCAPM, the Epstein and Zin (1989, 1991) model, and two external habit formation specifications using GMM on a quarterly data set spanning 35 years. In a novel approach the models are estimated for both the whole economy and four separate industrial sector groupings. The structural stability tests advocated by Hall and Sen (1999) are used to test the models further. There is little evidence found to support the traditional CCAPM and the recursive preferences model of Epstein and Zin (1989, 1991). There is highly supportive evidence for the performance of the habit formation models, particularly the Campbell-Cochrane specification. Importantly, the analysis of the four sector groupings shows that estimated levels of risk aversion are similar across these groupings, conforming with theory. In line with previous studies, the models show signs of sensitivity to choices of asset return data, consumption measures, and particularly, instrumental variables. Second, a non-parametric framework is used in an attempt to investigate the performance of the models. In particular, investigation of the pricing errors of consumption asset pricing models by estimating the vertical and minimum distances to the Hansen-Jagannathan bound. Additionally, bootstrap experiments are conducted in a further attempt to examine the performance of the models. The power utility model produces high pricing errors associated with higher values of relative risk aversion. The Epstein-Zin (1991) recursive preferences model and the Abel (1990) formulation manage to reduce the pricing errors. However, these pricing errors are associated with higher values of relative risk aversion. In the bootstrap experiments, in the majority of simulations, the models violate the Hansen and Jagannathan bound, and the distance is an order of magnitude larger than the IMRS volatility. However, the lower risk aversion coefficient associated with the lower distance is obtained from the Campbell-Cochrane model. Third, investigation of the performance of the consumption-based asset pricing models in pricing bonds. Additionally, an investigation of whether the term structure in the UK can reflect information about consumption growth. In this study, the methodology adopted by Harvey (1988) is used to test the relation between term structure and the consumption asset pricing models. Also used are the structure stability tests of Hall and Sen (1999). The models perform better with longer horizons. The results suggest that the estimates of the coefficient of relative risk aversion/curvature parameter are negative and insignificant, very often with the models that no longer consider the habit formation specification. The empirical evidence based on the regression are broadly supportive of efficiency in using lagged consumption and yield spread as predictors of consumption growth, rather than other lagged stock returns. The final study investigates the recent explorations of asset pricing in the UK, using the OLS and GMM methodology to revisit the conditional Sharp-Linter CAPM model with fixed and time varying parameters. Additionally, an examination is made to the performance of the consumption-based asset pricing models and the conditional asset pricing model with the Famma-F'rench factors, 8MB and HML. Following Harvey (1988), a specification is estimated that allows for the time variation in conditional covariance, conditionally expected returns, and the conditional variance of the market and it is found that the restriction can not be rejected. However, there is evidence that the conditional Sharpe-Linter model with fixed parameter is unable to capture the dynamic behaviour of asset returns. Additionally, the estimation of the conditional CAPM that allows for time variation performs better with the inclusion of BMS and HML. As for the consumption-based asset pricing models, the Campbell-Cochrane specification still best characterises the UK market.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.633243  DOI: Not available
Share: