Empirical Asset Pricing
Models and Methods
Wayne Ferson
The MIT Press
Cambridge, Massachusetts
London, England
2019 Massachusetts Institute of Technology
All rights reserved. No part of this book may be reproduced in any form by any electronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from the publisher.
This book was set in Times New Roman by Westchester Publishing Services. Printed and bound in the United States of America.
Library of Congress Cataloging-in-Publication Data
Names: Ferson, Wayne E., author.
Title: Empirical asset pricing : models and methods / Wayne Ferson.
Description: Cambridge, MA : MIT Press, [2019] | Includes bibliographical references and index.
Identifiers: LCCN 2018020617 | ISBN 9780262039376 (hardcover : alk. paper)
Subjects: LCSH: StocksPrices. | Rate of return. | Econometric models. | Moments method (Statistics) | Estimation theory.
Classification: LCC HG4636 .F47 2019 | DDC 332.63/222dc23 LC record available at https://lccn.loc.gov/2018020617
d_r0
Contents
6Applications of m-Talk
15.1Motivation and General Setup
23.6Predictive Panel Regressions
30.3Vector Autoregressions
Contributed by Davidson Heath
List of Figures
An event tree.
Triality in empirical asset pricing.
Minimum variance boundaries. Draw the tangent on this figure.
UE(Z) portfolio weights.
A Hansen-Jagannathan bound.
A minimum variance bullet.
The equity premium puzzle for bonds.
Figure 10.4
TMBM bounds.
HJ bounds with conditioning information.
HJ bounds with bias adjustment.
MBA-level argument for APT.
The four basic cases of GMM problems: two dimensions (exactly identified or overidentified) two parameter types (linear or nonlinear).
Consumption as the sum of increments.
Weighting kernels for GMM matrices covariance.
Cross-sectional regression of returns on betas.
Diff-in-diff.
A mixture of alpha distributions.
Production-based asset pricing.
List of Tables
Optimal orthogonal portfolios.
Precision of the bounds.
Evaluation of asymptotic standard errors for HJ bounds.
GMM moment condition table.
A latent variable model.
A conditional CAPM with time-varying covariances.
A K-factor expected return model.
A model with co-skewness and co-kurtosis.
Analysis of predictable variance.
A stationary long-run risk model.
Holdings-based performance evaluation.
Comparison of four asymptotically equivalent tests of ex ante efficiency of a given portfolio.
CSR is GMM.
Forecasting regressions.
Preface
This book is designed for PhD students in finance or advanced masters students interested in empirical asset pricing and as a reference for researchers. The main goal is to help readers build and understand a tool kit for empirical asset pricing research. As asset pricing has matured as a field, it has developed more breadth. This is a long book, but it leaves out many advanced topics. The aim is to present the material at an advanced introductory level. I try to provide enough references that a scholar can use the book as a jumping-off point for a research program. The structure is motivated from my experience in teaching this material in doctoral programs at more than a half-dozen universities where I have served as a regular or visiting faculty member.
The book has four main sections. The first section of the book, parts IIII (chapters 114), covers the theory of empirical asset pricing. This section is structured around three central paradigms: mean variance analysis, stochastic discount factors, and beta pricing models. I describe a relation among the three paradigms, which I call triality. I develop the role of conditioning information in the implementation and interpretation of asset pricing tests. The second section of the book, part IV (chapters 1524), examines empirical methods. I start with the generalized method of moments, emphasizing how to implement the technique in various types of problems. There is a long chapter on panel regression methods and a chapter on bootstrapping for inferences about cross sections, where multiple comparisons must be considered. The third section consists of part V (chapters 2528) on fund performance evaluation and is perhaps the most comprehensive review of this important area that is currently available. Throughout the book, I combine classical foundations with more recent developments in the literature, and I relate some of the material to applications in investment management. The final section of the book, part VI (chapters 2934), presents selected applied topics on empirical asset pricing. These include a long chapter on predictability in asset markets, including predicting the levels of returns, volatility and higher moments, and predicting cross-sectional differences in returns. There are shorter chapters on production-based asset pricing, long-run risk models, the Campbell-Shiller approximation, the debate on covariances versus characteristics, and the relation of volatility to the cross section of stock returns.
I owe a debt of gratitude to many people who helped me in the writing of this book. Several years of doctoral students at the University of Southern California suffered through being guinea pigs for early drafts. I am grateful to Wayne Chang, Suk Won Lee, and Ben Zhang for comments. Allison Kays and Min Kim found many typos in early drafts. Min Kim went beyond the call of duty, finding not only typos but several instances where my arguments were unclear or had holes in them. Davidson Heath contributed Matlab code for the appendix. Cyd Westmoreland did some of the most impressive copyediting that I have ever seen. And my wife, Nancy Ferson, deserves credit for putting up with my incessant typing far into many evenings, as I pursued this obsession.
Introduction
This book has four main sections. This introduction briefly describes the main sections and then goes through each part in more detail, including some background and suggestions for using the book in a PhD or masters level course. I then discuss how the material in this book compares with some of the competition.
The first section of the book, parts IIII, covers the theory of empirical asset pricing in 14 chapters. This is structured around three central paradigms: mean variance analysis, stochastic discount factors, and beta pricing models. The stochastic discount factor approach, which I call m-talk, is a beautiful way to integrate and illustrate some of the main concepts and models in asset pricing. But it is only one of the three paradigms. I describe a relation among the three paradigms, which I call triality. This idea has evolved from review chapters that I published in 1995 and 2003. This useful perspective is not emphasized in other books. I develop the role of conditioning information in the implementation and interpretation of asset pricing tests and in the triality relation in more depth than is done in other available texts.
The second section of the book, part IV, examines empirical methods in ten chapters. I start with the generalized method of moments (GMM), emphasizing how to implement the technique in various types of problems. I include material on some empirical methods that are becoming important for asset pricing research but dont get much coverage in other texts. This book has three chapters on regression methods, including a long chapter on panel regression methods and a shorter one on bootstrapping for inferences about cross sections, where multiple comparisons must be considered.
Next page