Many equity market researchers assume conventional three-factor (market return, size, book-to-market) and four-factor (plus momentum) models as standards of comparison for discovery of new sources of abnormal returns. Are they the best standards? Could they be derivatives of more economically fundamental sources of differences among individual stock returns? In their March 2007 paper entitled “Too Many Factors! Do We Need Them All?”, Soosung Hwang and Chensheng Lu seek to identify the minimum number of economically fundamental factors needed to explain why different stocks generate different returns. They investigate 16 factors (12 firm characteristics and four macroeconomic measures) that others have found to explain such return differences. Their principal test is to measure returns from zero-cost portfolios that are long stocks with high (top third) values and short stocks with low (bottom third) values of evaluated factors. Using data for a large sample of non-financial stocks during 1963-2005 and contemporaneous macroeconomic data, they conclude that:
- Liquidity has the largest average monthly hedge portfolio return (-0.92%), with illiquid stocks exhibiting clearly higher returns than liquid stocks. Momentum (past returns) has the second largest average monthly return (0.85%).
- Results for many factors correlate significantly with results for others (e.g., liquidity and size), indicating redundancy. Some empirical factors may be invalid due to data mining bias.
- For most subperiods, market return and liquidity used together explain differences in returns among stocks.
- A model combining market return, liquidity and coskewness (contribution to the skewness of market returns) explains individual stock returns in 35 out of 40 years (all except 1981-1985). Macroeconomic
factors (short-term interest rate or credit spread) account for asset returns during the early 1980s but explain little in other subperiods. - Adding a fourth factor does not improve the performance of this model.
- This model works as well as, and is economically better motivated than, the widely used four-factor model combining market return, size, book-to-market and momentum. Conversely, the traditional four-factor adequately incorporates all other factors except liquidity.
For extremes of illiquidity and coskewness, think works of art and lottery tickets, respectively.
In summary, there is considerable redundancy and invalidity among the many factors used to explain differences in returns among individual stocks. Three factors may be necessary and sufficient, with liquidity the most influential.
The paper offers definitions and some historical context for the factors considered.
For practitioners, the alternative factor set proposed by the authors is more abstract and less easily put into action than size, book-to-market and momentum.
