factor portfolios.) If this were the case, and if in addition there was a liq-
uidity premium, then one might expect that the lower expected returns as-
sociated with having a lower factor loading would be offset by the return
premia associated with illiquidity.
Similarly, we might expect that the low factor-loading stocks are those
that did very poorly over the previous year and just recently entered the
high B/M portfolio. However, this would bias our tests toward supporting
the factor model, since the momentum effect would lower the returns of
these low loading stocks. In order to bias our results against the factor
model, the past years’ returns of the stocks must be negatively related to the
factor loading.
To explore these possibilities, we calculate the average turnover and the
average return over the twelve preformation months for each of the forty-
five portfolios. These are tabulated in tables 9.9 and 9.10. The findings re-
ported in these tables suggests that our lack of support for a factor model is
CHARACTERISTICS AND RETURNS 345Table 9.9
Average Turnover for Portfolios Sorted by Size, B/M, and HML LoadingPortfolios are formed based on size (SZ), book-to-market (B/M), and preformation
HML factor loadings. The numbers presented are the average turnover for each of
the portfolios, for the period 1973:07–1993:12, where turnover is defined as the
trading volume (VOL) for the month (as reported by Center for Research in Secu-
rity Prices (CRSP)) divided by the number of shares (NS) outstanding at the begin-
ning of the month, times 1000. The averages are value-weighted and ME is market
equity, so turnover in month tis defined as
Char Port Factor Loading PortfolioB/M SZ 1 2 3 4 5
11 75.6 66.1 60.3 60.1 58.7
12 96.6 61.8 50.6 47.4 56.3
13 52.7 39.3 37.9 41.1 45.8
21 61.9 43.7 40.2 40.2 46.2
22 62.3 42.5 38.4 39.9 42.8
23 44.6 39.3 37.6 38.7 39.9
31 49.1 38.6 38.2 42.0 45.4
32 56.2 43.1 44.7 44.2 51.6
33 49.0 41.5 38.8 45.9 44.5
Average 60.89 46.21 42.97 44.39 47.91
Turnover
MEVOL
NSME.
,,
,t ,
titit
i it=⋅it
∑
1
Σ