Tarun Chordia Slides

Cross-Sectional Asset Pricing with Individual Stocks: Betas versus Characteristics

Tarun Chordia, Amit Goyal, and Jay Shanken

Main question

-Are expected returns related to
   – Risk/betas, OR
   – Characteristics

-If both, which is more important?

How to answer?

-Use portfolios
  -Helps mitigate EIV problem
    –Fama and MacBeth (1973)
  -But,
    –Less efficient Ang, Liu, and Schwarz (2010)
    –Method of grouping is important Lewellen, Nagel, and Shanken (2010)

-Use individual securities
  -But, EIV problem

What we do

-Use individual securities

-Correct for EIV bias
  -Litzenberger and Ramaswamy (1979), Shanken (1992), Kim (1995)

-Allow betas to change over time
  -Two year rolling regressions

-Quantify contribution of betas vs characteristics in explaining the cross-section of returns

Methodology

Taruns

Methodology

Taruns1

Intuition …

Taruns2

Relative importance

Taruns3

Simulation

Taruns4

Data

-All common stocks on NYSE, AMEX, and NASDAQ
-Sample: July 1963 to December 2013
-Price greater than $1 (for CSR)
-Sample of all stocks and non-microcap stocks (>20% NYSE percentile)
  -Monthly average of over 3000 stocks with about 1500 non-microcap stocks
-Fama-French (2014) five factor model plus Momentum
  -Mkt, SMB, HML, RMW, CMA
-Characteristics include size, B/M, past six month return (exclude last month), operating profitability and investment
  -Assumed to be available 6 months after fiscal year-end
  -Winsorized at the 99th and 1st percentiles

Data

Taruns6

Taruns7

Taruns8

Taruns9

Taruns10

Time variation in risk premia

Taruns11

Taruns12

Taruns13

Conclusion

-Reject all factor models
  -Rejection not news

-Characteristics more important than betas

-Risk premiums
  -Negative on SMB
  -Positive on RMW and CMA
  -No premium on HML or MOM
  -Less robust positive premium on Mkt


?>