,

,.. .
i
fh.t C. f't {)._$ -/��
Vll'\ 1'1/.

.,
'..
,. .

,,

THE RELATIONSHIP BETWEEN

EARNINGS' YIELD,
MARKET VALUE AND RETURN
FOR NYSE COMMON STOCKS:
!.;>

,>-l,
FURTHER EVIDENCE
. ).
} ,£. ,,_;;,­
l-. . •

By

S. Basu
Associate Professor

..!� ' INJJIS lJBRi�RY

U"
'

nc�§·uLI111THnub
11 " c•mn·
>, .

NnuiU
. �. I''·
.
;
"'
j

�" ,.fJ FACULTY OF BUSINESS

<:· : .. .•{
.
;\
."
. :·
I McMASTER UNIVERSITY
ilnnis
:·1a
' .
·
HAMILTON, ONTARIO, CANADA LBS 4M4
...':.·:� :";t\
;:'._,·
HB Research and Working Paper Series No. 180
'
!
.;,.:�·:. 745
:'
. z ,..
.
• November, 1981

.. . R
47
·

, '.- · no.180
:l
(
 THE RELATIONSHIP BEIWEEN EARNINGS' YIELD,

MARKEl' VALUE AND REl'URN FOR NYSE

CCMOCJN STOCKS: FURI'FIER EVIDENCE

by

Joe Basu (5,oJ\� J

Faculty of Business
McMaster University
Hamilton, Ontario
Canada, L8S 4M4

Version of: Please do not quote

September 1981 without author' s permission.
Comments welcome.
 THE RELATIOOSHIP BETWEEN EARNINGS' YIELD, MARKfil1

VALUE AND REl'URN FOR NYSE CCM-m' S'IOCKS:

--- ------
FURl'HER EVIDENCE

ABSTRAcr

The empirical relationship between earnings' yield, firm size and

returns on the common stock of NYSE firms is examined in this
paper. The results confirm that the common stock of high E/P firms
earn, on average, higher risk-adjusted returns than the common
stock of low E/P firms. This E/P effect is clearly significant
even if experimental control is exercised over differences in firm
size, i.e., the effect of size, as measured by the market value of
common stock, is randomized. The results also show that while the
common stock of small NYSE firms appear to have earned marginally
higher risk-adjusted returns than the common stock of large NYSE
firms, the size effect virtually disappears when returns are
controlled for differences in E/ P ratios. The evidence included in
this paper lends credibility to the view that, at least for NYSE
firms, the effect of differences in market value or size on common
stock returns is of secondary importance when compared with the
effect of E/P ratios.

1. INI'RODucrION

Recent empirical research on the relationship between earnings' yield,

firm size and common stock returns has revealed some anomalies with respect to

the pricing of corporate equities. In particular, the findings reported in

Basu [1977] for instance indicate that portfolios of high (low) earni ngs'

yield securities trading on the NYSE appear to have earned higher (lower)

absolute and risk-adjusted rates of return, on average, than portfolios

consisti ng of randomly selected securities. As noted by Basu, his results

suggest a violation in the joint hypothesis that (i) the single-period capital

asset pricing model (CAPM) has descriptive validity; and (ii) security price

behavior on the NYSE is consistent with market efficiency.

Similarly, Banz [1981] shows that common stock of small NYSE firms earned

higher risk-adjusted returns, on average, than the common stock of large NYSE

firms. This size effect appears to have been in existence for at least forty

years and, according to him, constitutes evidence that the CAPM is

misspecified. Moreover, relying on the work of Reinganum [1981] , Banz asserts

 2

that although the size and earnings' yield effects are related, the latter is

a proxy for the former, i.e., the earnings' yield effect is a proxy for size

and not vice-versa. This result, if correct, is an important one since it not

only provides an explanation for the earnings' yield anomaly, but also

suggests that in conducting tests of market reaction and/or efficiency

researchers need only control for firm size. Unfortunately, the study by

Reinganum contains sufficiently serious shortcomings that it would be

inappropriate to rely exclusively on its findings in this regard. l

The purpose of this paper, accordingly, is to re-examine the relationship

between earnings' yield (E/P ratios), firm size and returns on the common

stock of NYSE firms. Section 2 describes the data, sample and other

methodological considerations. The empirical results are then presented and

discussed in Section 3. Finally, some concluding remarks are provided in

Section 4.

1. At least three facets of the Reinganum piece suggest that it might be of

limited use in addressing the issue as to whether the earnings' yield effect
reported by Basu for NYSE firms can be attributed to the size effect observed
by Banz or vice-versa. First, the methodology employed in that work fails to
adjust for differences in risk. This can be observed by noting that despite
significant differences in the relative risk levels of the E/P and value
(size) portfolios shown in Table 11 of that paper, excess or "abnormal"
returns are computed as the difference between a given portfolio's realized
return and that earned by a market index. In other words, the determination
of excess returns is premised on the assumption that the risk levels of the
experimental portfolios are the same as that for the market index. Second, by
including both AMEX and NYSE firms in the sample, the author may have
inadvertently introduced a potentially serious confounding factor. To the
extent the earnings' yield and firm size effects are different for AMEX and
NYSE securities due to the presence of a separate "exchange effect", then the
results based on an aggregate AMEX-NYSE sample may have little significance,
if any, for statements to be made in connection with each of the two groups of
securities. Recall that the earnings' yield and size effects reported by Basu
and Banz respectively are for NYSE securities only. Third, notwithstanding
the use of the "research" COMPUSTAT database, Reinganum's findings may be
affected by retroactive, as well as other sample selection biases. While this
issue could easily have been tested, the study fails to have done so. Note
that by comparing the distribution of returns for two samples of firms
classified by market value of equity (size) and selected from the CRSP
database and the merged CRSP-COMPUSTAT counterpart respectively, one can
assess the effect of selection biases implicit in the latter. A priori, one
would expect the effect of selection biases to be more pronounced in the case
of AMEX firms because of their generally lower survival rates.
 3

2. DATA AND MEI'HOOOLCX3Y

----

The following general research design was employed to examine the

relationship between E/P ratios, firm size and common stock returns.

Initially, securities were partitioned into groups or classes on the basis of

their E/P ratios and the market value of their common stocks. These groups

were then combined to form (i) a set of earnings' yield portfolios, each

consisting of securities with similar E/P ratios but simultaneously belonging

to different market value classes� and (ii) a set of market value portfolios,

each consisting of securities with similar market values of equity but

simultaneously belonging to different E/P classes. In other words, the

earnings' yield and market value portfolios were constructed by controlling

for (i.e., randomizing) the effect of firm size and E/P ratios respectively.

The risk-return relationships of these portfolios then were compared and,

finally, the ir risk-adjusted returns were tested statistically in a

multivariate setting in order to determine the existence o f a significant

earnings' yield and/or size effects.

Data and Sample

The primary data for this investigation were drawn from two sources.

Accounting earnings per share, on a 12-month moving basis, for the years ended

December 1962 through 1978 were collected from an annually updated version of

the Compustat Prices-Dividends-Earnings (PDE) Tape. The updated version of

the PDE tape is analogous to the Merged Annual Industrial Compustat Tape

produced by CRSP. Security prices, returns and common share data were

obtained from the :rronthly stock return file of the CRSP tape .

To be included in the sample, a firm was required to have traded on the

New York Stock Exchange during the period investigated, as well as its

applicable return, market value and accounting earnings data must not have

been missing from the data bases described above. A total of about thirteen
 4

hundred firms satisfied these requirements for at least one year, with

approximately nine hundred qualifying for inclusion, on average, in each of

the 17 years investigated.

Portfolio formation and risk adjustment issues

From a methodological point of view, the earnings-price ratios and market

values of the common stock of all sample firms were computed as of December 31

of each year. While the market value of common stock was determined as the

market price times the number of shares outstanding, the E/P ratio was defined

as the most recent 12-month moving earnings per share, excluding extraordinary

items and discontinued operations, as of December 31 of a given year scaled by

the market price of common stock at that date.2

The computed E/P ratios for each year then were ranked in ascending order

and the quintiles from the distribution served as the basis for assigning

sample firms to one of five earnings' yield portfolios, i.e., lowest quintile

to portfolio EPl, next lowest to portfolio EP2 and so on. As such, portfolio

EPl includes firms with the lowest E/P ratios, while portfolio EPS includes

those with the highest E/P ratios. These ranking and portfolio assignment

procedures were repeated, but in this instance on the basis of the market

value of common stock variable, to form five market value (size) portfolios

with the smallest firms being included in portfolio MVl and the largest in

MVS. Some summary statistics pertaining to these two sets of portfolios are

included in Panel A of Table 1.

As might be expected, the size (MVl -MVS) and earnings' yield (EP1-EP5)

portfolios differ quite dramatically in terms of market value and the E/P

2. In the case of firms with a December 31 fiscal year-end, the earnings

measure represents the annual primary earnings per share figure reported in
the annual report to shareholders. For other firms, it represents the sum of
the primary earnings per share applicable to the four most recent quarters.
For an elaboration, see the COMPUSTAT PDE Manual.
 5

TABLE l

SELECTED VALUES FROM THE DISTRIBUTION OF

MARKET VALUES AND EARNINGS' YIELDS FOR PORTFOLIOS

Quartiles from the Distribution of :�/

Market Value (millions of $) E P Ratio
/
\
a
/ . .
CD
....
0 QJ
..-! ;:I MVl 17.4 27.7 40.5 0.053 0.090 0.146
.-!
.... MV2 54.4 75.5 100.9 0.060 0.091 0.150
0 <tS
:>
...
1-1 MV3 125.9 165.8 213.8 0.057 0.085 0.131
...
0
MV4 295.3 397.3 507.3 0.054 0.079 0.119
...
QJ
i:>..
....... <tS
lo< MV5 807.7 1127.l 2041.5 0.048 0.072 0.108
"Cl �
< QJ
....
"
..:i "Cl
� e .-!
0 EPl 58.l 230.2 774.3 0.018 0.033 0.050
....
QJ
�
i:>..
"Cl
c
;:.. EP2 l,04.3 264.5 629.6 0.048 0.062 0.099
<tS
1-1
c
- EP3 83.9 199.8 535.5 0.063 0.080 0.125
i:ll)
CD
z
0 EP4 56.7 137.2 378.6 0.077 0.097 0.155
i::
._
.... EP5 34.7 74.5 215.2 0.097 0.144 0.215
....
ti i::
lo<
Ul <tS
Ill I>;:!
i:Cl

QI
;:I
.-! MVl* 17.8 29.7 48.7 0.051 0.084 0.131
CD MV2* 50.3 85. 7 125.l 0.052 0.083 0.130
....
�I
0
MV3* 103.3 179.l 251.6 0.056 0.084 0.128
..-!
MV4* 249.4 402.0 557.4 0.056 0.083 0.128
....
0

. .
i:Cl
..:i
...
1-1
0
""'
�I
't!
MV5* 724.9 1083.0 2021.0 0.056 0.083 0.127

� "Cl .-!
EPl* 59.2 169.9 509.9 0.019 0.034 0.055
� ....
QJ QJ
....
"
P< ;:.. EP2* 61.5 168.2 510.5 0.048 0.062 0.101
e
-
EP3* 59.7 164.3 479.6 0.062 0.083 0.129
"Cl Ul
0
c i:ll) EP4* 57.l 161.8 483.9 0.076 0.103 0.155
....
<tS i::
� EP5* 58.5 166.5 497.9 0.094 0.133 0.208
i::
1-1
<tS
I>;:!

.e_/The basic (or nonrandomized) portfolios are formed by ranking securities on

market value or earnings' yield, as appropriate. The randomized market value
(earnings' yield) portfolios are formed by controlling_ for the effects of
earnings' yield (market value) •

.2/Based on pooled annual data (as of December 31) for the period 1962-78.
 6

ratio, respectively. More importantly however, the summary statistics in

Panel A indicate that these two variables appear to be negatively associated.

Observe from the north-east quadrant that smaller firms, on average, seem to

have somewhat higher E/P ratios than the larger firms. Conversely, the south­

east quadrant of Panel A reveals that the low E/P portfolios, on average,

consist of larger firms when compared with the high E/P portfolios.

Nonparametric analysis of variance {Kruskal-Wallis), moreover, confirms that

the null hypotheses of equality in E/P ratios for the five size portfolios and

the equality in market values for the five earnings' yield portfolios

respectively, can be rejected at the 1% level or higher.

In order to control for the confounding effects that might arise because

of the negative association discussed above, two additional sets of size and

earnings' yield portfolios were constructed by randomizing with respect to the

E/P and market value variables respectively. Consider initially the formation

of the earnings' yield J;X)rtfolios which are randomized in terms of firm size.

At the outset, all firms included in each of the five basic market value

or size J;X)rtfolios, MV1-MV5, were ranked annually from minimum to maximum on

the basis of their E/P ratios. The quintiles from the distributions

applicable to a given value class {J;X)rtfolio) then were used to assign firms

to one of five earnings' yields groups or sub-portfolios. Next, the lowest

earnings' yield groups relating to the five market value classes were combined

to form randomized portfolio EPl*. The firms included in the other four

earnings' yield groups were combined in an analogous manner to form randomized

J;X)rtfolios EP2*- EP 5*. Note that since these earnings' yield portfolios

include securities drawn from the entire set of market value classes, they can

be viewed as being randomized with respect to firm size.

The randomization approach described above was then employed to construct

five market value or size J;X)rtfolios, MV1*-MV5*, which are randomized in terms
 7

of the earnings' yield variable. Essentially, the market v_ a lues of firms

included in each of the basic earnings' yield classes {portfolios), EP1-EP5,

were ranked annually and the quintiles from the underlying distribution were

employed to assign firms to one of five market value or size groups.

Securities assigned to the ith size group {i.e., ith market value quintile)

.applicable to each of the five E/P classes then were combined to form

randomized portfolio MVi*(i=l, ,5).

••• Some summary measures relating to these

size portfolios, as well as to the earnings' yield portfolios EP1*-EP5* are

provided in Panel B of Table 1.•

As in the case of the basic portfolios, the randomized size (MV1*-MV5*)

and earnings' yield (EP1*-EP5*) portfolios differ quite significantly in terms

of market value and the E/P ratio, respectively. However by construction, all

of the size portfolios MV1*-MV5* have similar E/P ratios (about 8.3 -8.4% on

average), while the five earnings' yield portfolios EP1*-EP5* consist of firms

of similar size -- the market value of common stock of firms included in each

of these latter portfolios, on average, is about $160 -17 0 million. Indeed,

results of statistical tests indicate that neither the null hypothesis of

equality in E/P ratios for portfolios MV1*-MV5* nor the null hypothesis of

equality in market values for portfolios EP1*-EP5* can be rejected at any

reasonable level of significance.3 This, of course, suggests that confounding

effects attributable to the earnings' yield variable cannot be expected to be

present in comparisons involving the size portfolios MV1*-MV5*. Similarly,

assessments based on the earnings' yield portfolios EP1*-EP5* should be free

from any confounding effects stemming from the size factor.

3. Nonparametric analysis of variance (Kruskal-Wallis) was employed in this

regard: see Hollander and Wolfe [1973] or Conover [1980] for an elaboration.
The computed Kruskal-Wallis test statistics -- distributed approximately as a
chi-square random variable with 4 degrees of freedom and based on more than
16, 0 0 0 observations (pooled annual data) -- are 3.45 and 5.8 0 for the null
hypotheses pertaining to portfolios MV1*-MV5* and EP1*-EP5* respectively.
Note that these amounts are well below even the critical value at the 1 0%
level of significance, i.e., Pr [x 2 (4) > 0.90 ] = 7.7 8.
 8

The a n a l y s i s then e n t a i l ed the m e a s urement o f the ri sk-return

relationships for the various size and earnings' yield portfolios. First,

monthly portfolio returns were computed for the 17-year period 1963-1979 as an

arithmetic average of the corresponding returns for constituent firms. Next,

two measures of risk -- standard deviation of monthly returns and systematic

(nondiversifiable) risk -- were estimated for each portfolio. The systematic

risk measure in particular was determined in the context of the Sharpe-Lintner

version of the CAPM:

A A

(1) r S [r - rf,tl
P, t - rf,t 0P +
=
p m, t
where r ,t = return on portfolio p in month t; computed as the cross-
p
sectional arithmetic average of the realized monthly returns
on securities included in p.

rf t = return on "risk-free" asset in month t; measured as the

'
realized :roc>nthly return on 30-day U . S . treasury bills.

rm t
'
= ,t
return on the "market" portfo io in month t; measured by the
CRSP Index of all NYSE firms.

8P = differential or abnormal return for portfolio p (estimated

OLS intercept) .

Sp = s y s t e m a t i c or nondi v e r s i f i ab l e r i s k f o r portf o l i o p
(estimated OLS slope) .

Finally, the configurations of the risk-adjusted differential or abnormal

returns, 8 ' for the various market value and E/P portfolios were examined in
P

order to ascertain the presence of a size and earnings' yield· effects

respectively. More specifically, the null hypothesis of no size effect on

risk-adjusted returns was tested in the context of Hotelling's multivariate T2

A

methodology by assessing whether the vector of op applicable to the five size

portfolios MVl*-MVS* is significantly different from zero. 5 The null

4. Both the "equally-weighted" and "value-weighted" versions of this index

are employed in this paper in order to determine the extent to which the
results are sensitive to the choice of a surrogate for the "market" portfolio .
 9

hypothesis of no earnings' yield effect was tested in an analogous manner by

employing the 6p pertaining to the J?Ortfolios EP1*-EP5*.

3. EMPIRICAL RESULTS

Rates of return for size and E/P J;X?rtfolios

At the outset, consider some descriptive statistics pertaining to the

rates of return earned by the various size and earnings' yield portfolios.

Table 2 shows the mean monthly return, r , and related standard deviation,
p
CY(r ), for (i) the basic market value and earnings' yield portfolios (Panel
p
A); (ii) their randomized counterparts - MV1*-MV5* and EP1*-EP5* (Panel B);

and (iii) the equally-weighted and value-weighted versions of an NYSE-based

"market" index (Panel C). In addition, the mean return per unit of standard

deviation, rp/u(r p ), or the reciprocal of the coef fic ient of variation of

monthly returns for the var ious portfolios is shown in column (3), and the

differences between that amount and the corresponding values for the two

versions of the "market11 index, respectively, are shown in columns (4) and (5)

of Table 2.

A survey of the results in Panel A indicates that consistent with

previously published findings, the common stock of small NYSE firms appear to

have earned, on average, higher monthly returns than the common stock of large

firms: the smallest market value quintile, for instance, experienced an

average monthly return of 1.5% during the seventeen years ending 1979, while

5. See, for ex� ple, Morrison [1967] for an elaboration on the properties of
the Hotelling's T test of means. The use of an analysis of covariance (i.e.,
cross-sectional 11Cho w11 test) framework to test the null hypothesis of equal o
was rejected because the critical assumption of equal variances was clearl ?
violated in the case of the size J?Ortfolios. Furthermore, the Hotelling test
can be viewed as a generalized version of the multivariate counterparts
formulated to test for the equality of coefficients within Zellner' s seemingly
unrelated regression framework; a description of the Zellner framework can be
found, for example, in Theil [1971] . Note that the Hotelling T2 methodology
simultaneously tests all possible linear combinations of 0 0, including the
"'
ones formulated in the context of the Zellner framework.
 10

TABLE 2

RATES OF RETURN FOR MARKET VALUE AND

EARNINGS' YIELD PORTFOLIOS : SOME SUMMARY STATISTICS

(Based on monthly data for the period 1963-79)

Summary Statisti c£/

{r Ja(r ) { r Ja<} >
p p p p
a - "' - "'
Portfol.icr-/ r
p
O' ( ';p) {rJa(r)
p p }
r
- EI/
a (r )
EI }
-r
VI/
a (r )
VI }

(l) (2) (3) (4) (5)

<1l
0 QI
::s MVl 0. 0150 0.0736 0.2044 0. 0106 0. 0392
M M
""'
0 t1S MV2 0.0120 0.0604 0.1991 0.0054 0.0339
....).I.
"" :>
....QI. MV3 0.0100 0.0549 0.1818 -0.0120 0. 0166
0 MV4 0. 0091 0. 0504 0. 1810 -0.0128 0. 0158
A.o .:.::
).I MV5 0. 0064 0.0431 0.1476 -0.0461 -0. 0176
"'
....... t1S
�
QI
<
..... N

"'
r:.:I ""'

� 0 M
e
A.o "' QI
"'"'
•
d EPl 0. 0083 0. 0628 0. 1326 -0.0611 -0. 0326
t1S :>-<
).I
- EP2 0. 0071 0. 0534 0. 1323 -0. 0614 -0.0329
0 <1l
d
EP3 0.0088 0. 0513 0.1722 -0.0216 0. 0700
z
'-'
co
d EP4 0.0127 0.0534 0.2378 0. 0440 0.0726
.....
C) d EP5 0. 0157 0. 0597 0.2632 0. 0695 0.0980
""' ).I
<1l t1S
t1S r:.:I
l"1

:i
QI
MVl* 0.0136 0.0719 0.1894 -0. 0043 0.0243
M
t1S MV2* 0.0115 0. 0608 0.1890 -0.0048 0.0238
<1l
:>
MV3* 0. 0097 0.0544 0.1790 -0.0147 0. 0138
0
<U
.....
MV4* 0.0097 0.0503 0. 1933 -0.0005 0.0281
M
""'
,:.!
MV5* 0.0081 0.0430 0.1882 -0.0055 0. 0230
........
0 ).I
t1S
�
).I
l"1 0
A.o "'
"' <U
..... ,_,
"'"'
r>l
EPl* 0. 0087 0.0632 0. 1368 -0.0570 -0.0284
� QI
N >t
EP2* 0.0559 -0.0454
A.o 0.0083 0.1483 -0.0169
�
0 EP3* 0. 0091 0. 0528 0.1717 -0.0221 0.0065
"'
·rn
Cll
d i:: EP4* 0.0121 0.0524 0.2308 0.0371 0.0657
.....
� d
).I
EP5* 0.0144 0. 0545 0.2645 0.0708 0. 0994
t1S
r>l

Equally Weighted
: "'"'
0
t,)
QI 0
.... M Index (EI) 0.0110 0.0568 0.1937 o.o 0.0285
r>l ....
.....
.:.::
).I .....
:ij Value Weighted
A.o � 0
t1S ).I
: r:i.. Index (VI) 0.0069 0. 0420 0. 1652 -0.0285 0.0

,!/The basic (or nonrandomized) portfolios are formed by ranking securities -0n
market value or earnings' yield, as appropriate. The randomized market value
(earnings' yield) portfolios are formed by controlling for the effects of
earnings' yield (market value) .

-b/- "'
r s mean monthly return on portfolio p; a (r ) = standard deviation of monthly
r�turn on portfolio p; and r· a(� ) • recipr8cal of the coefficient of variation
/
of monthly returns for portf8lio �.
Note subscripts EI and VI represent equally
weighted and value weighted indexes of NYSE firms respectively.
 11

the largest had earned about 0.64% per month.6 Similarly, portfolios of firms

with high E/P ratios seem to have earned higher rates of return than their low

E/P counterparts. Note, for example, that the highest earnings' yield

quintile earned aboi.it 1.57% per month versus about 0.83% earned by the lowest

quintile. More interestingly however, while the higher returns for the small

firms appear to be simultaneously accompanied by generally higher levels of

variability (risk), as evidenced by the a (r ) values in column (2) of Table 2,

p
this is clearly not the case for the high E/P portfolios. In fact, the

highest earnings' yield portfolio EP5 seems to have a marginally smaller

standard deviation of monthly returns than the lowest earnings' yield

portfolio EPl. As a consequence, while the dispersion in the mean return per

unit of variability measure, r pl a(rp), for the five basic E/P portfolios

largely parallels that for its unscaled counterpart (i.e, rp ), it is

substantially less in the case of the basic size portfolios.

Turning to Panel B of Table 2, one finds the results for the market value

and earnings' yield portfolios that were constructed by controlling for the

confounding effects stemming from differences in the E/P and size variables

respectively. Although the preceding remarks on the configuration of the

rates of return, by and large, are also applicable to these two sets of

randomized portfolios, an important difference should be noted. Observe from

oolumn (3) that the mean monthly returns per unit of variability, rp1a(rp>,

6. These results are based, as mentioned previously, on the sample of firms

drawn from the merged COMPUSTAT-CRSP data base. In order to test for the
existence of a survivorship bias, an additional five market value portfolios
were constructed by partitioning all firms contained in the monthly return
file of the CRSP tape. The rates of .return for these latter portfolios then
were compared with those for MV1-MV5 and the vector of differences between the
two sets of returns was tested for statistical significance. Results of
Hotelling's T2 test of means indicates that the difference in returns for the
two set of market value portfolios is not significantly different from zero at
any reasonable probability level. This, of course, suggests that the effects
of a significant survivorship bias on oommon stock returns is not present in
this study.
 12

earned by the portfolios of small firms '(MVl* and MV2*) are virtually

identical to the amounts earned by their large firm counterparts (MV4* and

MVS*). In other words, the higher mean monthly returns experienced by MVl*

and MV2* are accompanied by proportionately higher levels of variability in

returns so that their coefficients of variation (i.e., the reciprocal of

rp / (J < � p> shown in column (3 )) are similar to the levels reported for

portfolios MV4* and MVS*. It would appear, accordingly, that after

controlling for confounding E/P effects, the entire difference in the realized

returns for small and large NYSE firms can be explained by or attributed to

differences in risk (variability) levels. 7 The lack of homogeneity in the

r:p"' cr(rp) statistic for portfolios El?l*-EPS*, on the other hand, confirms that

the differential performance of the earnings' yield portfolios cannot be

explained along these lines, i.e., the difference in returns between low and

high E/P firms cannot be attributed to differences in variability (risk) or

firm size.

Finally, the relative performance of the experimental portfolios vis-a-

vis the equally-weighted and value-weighted versions of the NYSE index can be

discerned from columns (4) and (5). A comparison of the statistics reported

in these two columns reveals that the results are quite sensitive to the

choice of a "market" index. For instance, while all five randomized

portfolios MVl*-MVS* have earned rates of return (per unit of variability} in

excess of the value-weighted index, a diametrically opposite result is

obtained if the equally-weighted index was used instead. In this regard, note

that the r :p"'cr (rp) statistic for the equally-weighted index is about 17% higher

than the corresponding value for its value-weighted counterpart, i.e., 0.1937

7. The issue as to whether the difference in returns of small and large NYSE
firms can also be explained in terms of systematic (nondiversifiable) risk per
se is addressed in the next sub-section. Recall that the variability measure,
cr (rp>' includes both systematic and unsystematic risk.
 13

in the case of the former versus 0 .1652 for the latter (see Panel C) . 8

CAPM risk-return relationships

Although the preceding analysis provides some insights into the risk-

return relationships for the various experimental portfQlios, . it fails to

address an important issue. Essentially, to what extent are the risk-return

relationships observed for these portfolios consistent with the relationships

predicted by the Sharpe -Lintner capital asset pricing model? In order to

examine this issue, equation (1) was estimated for each of the size and

earnings' yield portfolios by employing ordinary least squares. The equally-

weighted NYSE index was assumed to be the surrogate for the "market"

portfolio9 and selected results pertaining to these asset pricing regressions

are shown in Table 3. Specifically, that table includes: (1) the estimated
"

systematic risk for experimental portfolio p, 13 pi (2) the coefficient of

correlation between the return on portfolio p, net of the risk-free rate, and

� �
the corresponding net return on the "market" portfolio, p (r , r ); (3) the
"

estimated abnormal or differential return for portfolio p, o ; and (4) the t-

p
"

value pertaining to the null hypothesis o p = O. Also shown are the results of

Hotelling's T 2 test performed on the vector of abnormal returns applicable to

the alternative sets of size and earnings' yield portfolios. While column (6)

contains the value of the F-statistic corresponding to the computed T 2

statistic, the F-values relating to the individual null hypotheses that the o
p

8. The results for portfolios MV1-MV5 and MVl*-MVS* suggest that the
difference in rr/ 0-(rn) between the two versions of the "market" index can be
attributed, more appfopriately, to the confounding effect of the E/P variable,
rather than firm size per se. Note that since E/P ratios and market values of
NYSE firms appear to be negatively associated, the value-weighted index can be
expected to have a somewhat lower weighted average earnings' yield than its
equally-weighted counterpart.

9. The results corresponding to the use of the value-weighted NYSE index as

a surrogate "market" portfolio are introduced at a later point.
 14

�
SOME CAPM RESl!LTS FOll MARKET VALtJ! AND

EARNINGS' YIELD PORTFOLIOS : EQUALLY WEIGRT!D NYSE INDEX

(Bat1ed on monthly data for the period 1963-79)

CAPM Statisti J?/ c/

Hotelling 1 s Test Results-
/ \
1 A A

P (r" i5
p. t"" )
Pordouo!. s c(iS ) F(o ) F*<o )
P m p P p P

(1) (2) (3} (4) (5} (6)

..
Q MV1 l.253 0.967 0.0024 1.82 0.65 l.06
.. <II
.... :I 0.23
MV2 l.051 0.987 0.0007 l.07
"'
Q ....
... 0.26
.. > MV3 0.954 0.987 -0.0007 -l.lS
.. 0.857 0.96 5 -0.0010 0.20
MV4 -l.02
""'
Q ..
<II
MV5 0.681 0.897 -0.0026 -l.91 o. 72
-
""
"'
..
""
�
<II ::i:
"
..: ..
1:1 a
Q �
�
""'
� ....
EPl l.055 0.955 -0.0030 -2.31 l.05 4.88
"'
c <II
..
.. >< EP2 0.911 0.968 -0.0034 -3.56 2.49
-..
c
Q EP3 0.889 0.983 -0.0014 -2.lS 0.91
z
..... 00 EP4 0.923 0.981" 0.0022 3.00 l.77
c
u .. EPS 1.023 0.972 O.OQ46 4,66 4.27
.. c
"'
.. ..
.,
= :.l

<II
:I MVl* 1.230 0.972 o.oou 0.95 0.18 0.61
"'
....
>
MV2* l.062 0.992 0.0001 0.16 o.oo
.. .. MV3* 0.945 0.987 -0.0009 -1.50 0.44
0
... MV4* 0.864 0.974 -0.0004 -0.47 0.04

ii MVS* 0.923 0.13

...
'
Q 0.700 -0.0010 -0.82
...
t!Q i-
= "" �
....
... � "
1:1 <II ..
" >< EPl* l.074 0.965 -0.0028 -2.43 1.16 4.85
�
""' -..
...
a EP2* 0.963 0.979 -0.0025 -3.06 l.83
Q
� :Q EP3* 0.919 0.987 -0.0014 -2.36 1.10
a a

�,
.... EP4* 0.907 0.983 0.0017 2.49 l.21
a
"'
.. EPS* 0.934 0.972 0.0039 4.31 3.64
:.l

!I The basic (or 11011randomized) portfolios are formed by ranking securttie.s on

!Dllrket value or earnings' yield, as appropriate. The randomized market vaJ.ue
(earnings' yield) portfolios are formed by controlling for the effects of
earnings' yield (market vaJ.ue) •

.!?/8P • estimated systematic risk for portfolio p; p(r' r") • coefficient of

p, m
correlation bet:ween the return on portfolio p (net of the risk-free rate), r",
• p
and that on the market index (net of the risk free rate), r'; d s differential
m p,
return -- estimated intercept for OLS regression of r' on r'; t(o ) s c-vaJ.ue for
• P m P
cS • 0.
P

_£,/Results for liotelling's test of.means.!2 Sho� are the F(S,199) - statistics
pertaining to the hYt>othesis that o • O; F*(cS ) represents the F - value
p p '
corresponding to the T s tatis tic . Selected fractiles from the t (n,d}
2

distribution are:
0 . 90 0 . 95 0.99

'if (S,120) l .9 0 2.29 3.17

� (5, ..) l,85 2.21 3. 02

 15

for a g iven portfolio is equal to zero are shown in column (5). Note that

these latter F-values can be viewed as the multivariate'analog s of the

univariate t-statistics reported in column ( 4 ), and selected fractiles from

the theoretical F-distributions are included in note (c) of Table 3.

Consider initially the results for the market value portfolios. It will
A

be readily noted that the level of systematic risk (Sp) declines quite

dramatically and in a monotonic way as one moves from the portfolios

consisting of small firms to those consisting of the larger ones•. Since the

correlation coefficients reported in column (2) suggest that these portfolios

are equally well diversified (at least approximately)' the difference in s

p
can be attributed principally to the difference in the standard deviation of

returns (see cr (r� in Table 2)•10 Moreover, consistent with the discussion in

the previous section, size portfolio MVl seems to have earned a positive

abnormal return of about 0.24% per month, while its large firm counterpart,

MV5, experienced a negative abnormal return of about 0.26% per month. The

magnitude of op for these two classes of firms, however, is considerably

smaller in the case where the effects of differences in E/P ratios are

controlled. Observe that the abnormal returns experienced by MVl* and MV5*

amount to only about 0.11% and -0.10 % per month, respectively. Results of

both the univariate t-test and the Hotelling's multivariate T 2 test, moreover,

indicate that the iSP for portfolios MV1*-MV5* are not statistically

significant. In other words, the estimated abnormal returns for the five size

portfolios, as well as all linear combinations thereof, are not stochastically

different from zero. This result, of course, is consistent with the

hypothesis that market value or firm size per se did not have a significant

10. Recall that, by definition, s = Cov (r ,r )/cr 2 (r ) = p (r ,r ) cr (r )/o (r )·

p m
As such, the difference between the systema tlic risks o� two P8
rt�olio;? can be
m
explained in terms of the differences in the p (r ,r ) and cr (r ) parameters.
p m p
 16

effect on the risk-adjusted returns of NYSE firms during the period 1963-79.

An examination of the c.APM results for the earnings' yield portfolios, on

the other hand, leads to an entirely different conclusion regarding the effect

of E/P ratios on performance. At the outset, observe from columns (1) and (2)

of Panel B that not only are the randomized portfolios EPl*-EPS* equally well

diversified, but they also have similar levels of systematic risk, at least

when compared to their market value counterparts. These similarities

notwithstanding, it would appear that the five earnings' yield portfolios have

earned abnormal returns that are by no means homogeneous. Note in this

connection that the abnormal returns experienced by these earnings' yield

portfolios range from - 0.28% per month for EPl* to 0.39% per month for EPS*.

Consequently, an arbitrage portfolio that had a "long" position in EPS* and,

simultaneously, a "short" position in EPl* could have earned about 0.67% per

month (or about 8% per annum) more than a randomly selected portfolio of

equivalent risk. In addition, the Hotelling' s T 2 test confirms that from a

statistical viewpoint the vector of 8 for portfolios EPl*-EPS* is significant

p
at the 1% level or higher. Multiple comparison test procedures suggest that

this result can be attributed primarily to the highest earnings' yield

A

portfolio EP5*; the op for the other four randomized portfolios, EP1*-EP4*,

are not significantly different from zero in a multivariate setting (see the

F-statistics in column (3) of Panel B). 11 In short, these findings are

consistent with the statement that E/P ratios did, in fact, have a significant

effect on the risk-adjusted returns of NYSE firms during the 17 year period

11. In contrast, the t-values shown in column (4) lead to the inference that
all five earnings' yield portfolios (EPl*-EPS*) have earned abnormal returns
that are stochastically different from zero. Unfortunately, the statistical
significance levels associated with these results can be expected to be
overstated since they are based on univariate confidence intervals. It is
important to recognize that the structure of the experiment and related
hypothesis tests entail the adoption of a multivariate testing perspective
and, as such, the construction of joint or simultaneous confidence intervals.
Note that the test involves five variates (i.e., five E/P or size classes)
rather than just one.
 17

ending December 1979.

The CAPM results presented hitherto have been based on the equally-

weighted NYSE index. In the light of Roll's [1977, 1978] criticisms of

empirical tests of the CAPM, ooupled with the evidence included in Table 2, it

seems appropriate to test the sensitivity of those results to the use of an

alternative surrogate for the "market" portfolio. A survey of Table 4, which

presents the CAPM findings based on the value-weighted index, reveals that the

oonclusions stated above are not altered in any substantive way. Nonetheless,

at least three facets of these results should be highlighted.

First, as might be expected g iven the weighting scheme underlying the

market index, the correlation coefficients reported in column (2) indicate

that the level of diversification for large firm portfolios is considerably

higher than that for the small firms. On the other hand, this characteristic

is not shared by the earnings' yield portfolios EP1*-EP5* because they are

randomized with respect to firm size. Second, estimates of abnormal returns,

o ' seem to be particularly sensitive to the use of the alternative versions

p
A
of the NYSE index. For example, the value-weighted index yields a op of

0.0 0 7 2 for portfolio EP5*, which is about 85% more than the 0.0 0 3 9 estimate

obtained by using the equally-weighted index. Finally, some caution should be

exercised in interpreting the results pertaining to the relative performance

of the size portfolios MV1*-MV5* in particular. Since the Hotelling's T 2 test

indicates the vector of 6P for these five portfolios is not stochastically

different from zero, the most appropriate inference is the returns earned by

both small and large firms are statistically indistingu ishable from the

corresponding returns predicted by the CAPM.12

12. Reliance on the t-test, Aon the other hand, seems to lead to a somewhat
different conclusion. The t(op) statistics shown in column (3) indicate that,
with the exception of MV3*, the other four size portfolios have 8 which are
significantly greater than zero. Additional tests, moreover, sugg�t that the
abnormal returns earned by small and large firms are not sta tistically
 18

TABLE 4

SOME CAPM RESULTS FOR MARKET VALUE AND

EARNINGS' YIELD PORTFOLIOS : VALUE WEIGHTED NYSE INDEX

1
CAPM Statisti c£/ Hotelling's Test Result s.£
A

Portfoli � p(r' r') cS t(o ) F(cS ) F*(cS )

BP p, m p p p p

(l) (2) (3) (4) (5) (6)

Cll
0
-.-1 Ill MVl 1.400 0.802 0.0071 2.30 l.04 l.54
..... ::l
0 ..... MV2 l.262 0.881 0.0044 2.21 0.96
""' "'
.., :> MV3 l.206 0.925 0.0025 l. 73 0.59
1-1
0 .., MV4 l.146 0.957 0.0018 l.78 0.62
"" Ill
� MV5 1.014 0.990 -0.0006 -l.45 0.41
,... 1-1
< ""
,.J
Ill
N
�
l>l -.-1
� g
""
""
"" .....
i:: Ill EPl l.379 0.926 0.0005 0.27 0.01 5.05
"' -.-1
1-1 l>-1 EP2 l.192 0.940 -0.0004 -0.27 0.01
=
EP3 l.14
z
0 - l.125 0.923 0.0016 0.25
.,
.._, bO EP4 l.181 0.892 0.0054 3.20 2.01
i::
t.I -.-1 EP5 l.206 0.851 0.0083 3.75 2.76
-.-1 =
Cll 1-1
"' "' .
""
l>l I
Ill MVl* 1.388 0.814 0.0057 l.95 0.75 l. 77
::l
..... MV2* 1.295 0.897 0.0038 2.03 0.81
Cll "'
0 ::> MV3* 1.202 0.9.32 0.0023 l.65 0.53
-.-1
..... .., MV4* l.141 0.954 0.0024 2.31 l.05
0 Ill
""' � MV5* l.007 0.986 0.0011 2.25 LOO
.., 1-1
1-1 "'
"" 0 :El
""
,.J
l>l "" ""
Ill .....
:a N
-.-1
Ill
-.-1
"" EPl* l.377 0.918 0.0008 0.44 0.04 5.25
e l>-1
0 EP2* l.227 0.923 0.0008 0.53 0.05
"" -
Ill EP3* 1.161 0.926 0.0017 l.23 0.30
;
gl
bO
= EP4* 1.126 0.905 0.0048 3.09 l.87
-.-1
i:: EP5* 1.137 0.878 0.0072 3.90 2.99
1-1
"'
l>l

a/
- The basic (or nonrandomized) portfolios are formed by ranking securities on
market value or earnings' yield, as appropriate. The randomized market value
(earnings' yield) portfolios are formed by controlling for the effects of
earnings' yield (market value).

1?/e estimated systematic risk for portfolio p; p(r' r') coefficient of

P
m �
p, m
correlation between the return on portfolio p(net of the risk-free rate), r',
A p
and that on the market index (net of the risk-free rate), r'; t(cS ) = t-value
A m p
for o = O.
p

2
£/ Results for Hotelling1s T test of.means. ShoWI) are the F(5,199) - statistics
pertaining to the hypothesis that cS = O; F*(o ) represents F - value
p p
2 !II
corresponding to the T statistic. Selected fractiles from the F(n,d)
distribution are:

0.90 0.95 0.99

� (5,120) l.90 2.29 3.17

� (5,...) l.85 2.21 3.02

 19

Results on interaction effects

To summarize, the empirical results pres�nted above confirm the presence

of a significant earnings' yield effect on the NYSE during the period 1963-79.

But, was this effect homogeneous across alternative market value classes? In

other words, to what extent did the E/P effect vary between small and large

NYSE firms? An examination of this issue should permit one to determine

whether or not there existed an interaction effect between earnings' yield and

firm size.

Actual rates of return and selected CAPM results for earnings' yield

portfolios pertaining to each of five market value classes are presented in

Table 5. These earnings' yield portfolios were constructed by ranking

securities included in a given market value class (i.e., size portfolio MVl -

MV5) on the basis of their E/P ratios.

In general the E/P effect, which was observed in the case of the

aggregate sample of NYSE firms, also seems to be present in each of the market

value categories. To see this more clearly, Figure 1 contains a scatter

picr (rp)
h

diagram of F and op versus market value for the alternative sets of

earnings' yield portfolios. It will be readi].y noted that in all five size

classes, the common stock of high E/P firms have experienced higher risk -

adjusted returns than the common stock of their low E/P counterparts. A

closer examination of the configuration of 8 in Figure 1, however, suggests

p
that the earnings' yield effect becomes somewhat weaker as one moves from the

smallest size class (MVl) to the largest (MV5), i.e., the difference in the

different from each other. Accordingly, it can be argued that although a

significant size effect is not observable in the context of the univariate
methodology, the estimated risk-return relationships for both small and large
firms are inconsistent with CAPM predictions based on the value-weighted NYSE
index. The propriety of this conclusion, however, should be questioned
because of the serious shortcomings associated with relying on the t-test
methodology in a multivariate setting (see note 11 above).
 m!&.2.

ACTUAL RE'CURNS & SELECTED CAPH RESUJ,TS FOR EARNINGS' YIELD PORTFOLIOS CLASSIFIED BY MARKET VALUE
(Based on 1Wonthly data for the period 1963-79)

Summary Statiatic .,!!./

r "
Portfoli.,.!!./ Actual Returns Equally Weighted NYSE lndex Value Weighted NYSE Index
"
HV
/ . /.
EP /. "' "' --- -=--- ---- _,
r r ) II p(r' r') 6 t(6 ) F*(6 ) ll 4 t(6 ) F*(6 )
Class Class p p/ p p p, .. p p p p m p p p
a(}p ) a (} �a·p,'h
(1) (2) . (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)

1 0.0124 0.0886 0.140 1.449 0.932 -0.0016 -0.68 4.99 1.587 o. 755 0.0040 0.98 5.44
2 0.0135 0.0751 0.180 1.256 0.951 0.0009 0.52 1.417 o. 796 0.0056 l. 74
MVl
3 0.0131 0.0691 0.189 1.167 0.960 0.0010 0.73 1.326 0.809 0.0053 1.87
4 0.0170 0.0712 0.238 1.180 0.941 0.0048 2.82 1.338 0.792 0.0092 3.00
5 0.0193 0.0737 0.262 1.2ll 0.935 0.0069 3. 77 l.331 0.762 0.0116 3.45

1 0.0104 0.0728 0.142 1.223 0.955 -0.0021 -1.39 4.82 1.471 o. 852
' 0.0023 0.84 5.33
2 0.0082 0.0625 0.131 1.061 0.964 -0.0032 -2.74 1.288 0.868 0.0005 0.25
MVZ 3 0.0111 0.0583 0.191 0.990 0.965 0.0002 0.16 1.202 0.869 0.0037 1.81
4 0.0140 0.0571 0.244 0.959 0.953 0.0032 2.63 1.151 0.849 0.0066 3.12
5 0.0165 0.0612 0.270 1.021 0.948 0.0054 3.92 1.120 0.827 0.0091 3.75

1 0.0075 0.0661 0.114 1.074 0.923 -0.0039 -2.21 3.76 1.381 0.881 -0.0003 -0.16 4.35
2 0.0069 0.0567 0.121 0.949 0.951 -0.0038 -3.07 1.199 0.891 -0.0006 -0.31
HV3 3 0.0088 0.0544 0.161. 0.921 0.960 -0.0017 -1.63 1.176 0.909 0.0014 0.86 N
4 0.0127 0.0533 0.239 0.903 0.962 0.0024 2.34 1.117 0.883 0.0055 3.14 0
5 0.0140 0.0555 0.252 0.923 0.945 0.0035 2.75 l.158 0.879 0.0067 3.60
-- -
l 0.0074 0.0607 0.123 0.916 0.858 -0.0030 -1.38 l. 76 1.311 0.911 -0.0003 -0.16 2.50
2 0.0069 0.0533 0.130 0.851 0.907 -0.0031 -1.96 1.164 0.921 -0.0004 -0.28
MV4 3 0.0081 0.0500 0.162 0.829 0.941 -0.0018 -1.49 1.089 0.918 0.0010 0.68
4 0.0104 0.0506 0.205 0.836 0.938 0.0004 0.34 1.083 0.902 0.0032 2.09
5 0.0128 0.0514 0.249 0.855 0.943 0.0027 2.28 1.082 0.886 0.0056 J.37
--·

l 0.0055 0.0529 0.105 0.705 0.151 -0.0035 -1.46 2.81 1.133 0.902 -0.0017 -1.08 2.76
2 0.0059 0.0470 0.126 0.699 0.843 -0.0031 -1.75 1.069 0.956 -0.0012 -.1.23
MV5 3 0.0043 0.04H 0.096 0.686 0.869 -0.0047 -3.00 l.Oll 0.953 -0.0027 -2.81
4 0.0065 0.0428 0.152 0.658 0.872 -0.0023 -1.55 0.940 0.925 -0.0003 -0.26
5 0.0096 0.0424 0.226 0.659 0.883 0.0008 0.56 0.916 0.910 0.0028 2.29

.!!/Portfolios are formed by ranking securities on the basis of their earnings' yields (EP) in a given market value (MV) class.

'!!_/r• mean monthly return on portfolio p; ) • standard deviation of monthly return on portfolio p, • estimated systematic risk for portfolio p;
p p p
a(;- B
p(r' r') • coefficient of correlation between the return on portfolio p (net of risk-free rate), r', and that on the market index (net of risk-free
p, m • p
- • • •

rate), r'; 6 m differential return -- estiiuated intercept for OLS regression

of r' on r'; t(6 ) • t-value for 6 • O; and F*(6 ) • F(5,199) - value
w p • p "' p p p
2
.
correspon,Ung to Hotelling' e T test statistic for 4 • Selected fractiles from the F(n,d) distribution are:
p

0.90 0.95 0.99 0.90 0.95 0.99

F (5,120) 1.90 2.29 3.17 F (5,m) 1.65 2.21 3.02

 /\ FIGURE 1
Plots of i /c dr ) and 8P for Earnings' Yield Portfolios by Market Value Catego ry
p p
0.30 0.0 1 2 0.012 J:------,
ACTUAL RETURNS EQUALLY WEIGHTED INDEX VALUE WEIGHTED INDEX
, ----
-----
5 ------------- 5
______
____ 5
0.25 0.008 0.008
___
i---- -,,,- ___________
1
-- 5 ________
_____ 5
-
____________ 4_____________ ,
t ,-----
--
5 __---
:-------------5._______
-
--- -- �--
-
-- 4 ------
-
-
--- 5 ----
-
-- --- 5 ----
-
I ---4 ,
-
--- .
----
...
............ .....
-
-- .. -
... ..
-
..
-
0.004 0.0041.,.
I ,,....,....... -. --
'
-
--
- 4
-
................-. - ----,
---
-
,
4...........---
0.20 3: ____________ ---5_______ -- 5 __ '5
- --
c.
, 4
3--- - -- - 3 --
---
_
- -
.
4 ----- - -- -- - ----
-
2' ,
-------
', --- -- -
-4 - -----:\..._ - ' .,___
-
I
,, _
' ·- ' -
--- --
1
--- - .___
', 3___ - ...5 -- 1
- -3-------------3 --
c.
--- --
..
l
'.....
'
...
•,
.....
...........3------------- 3', ........
2.... -..................._
3 ......
. ... ...
..
- -- Q.
(C-0
� ''\,2 ........... -. ... ... ...
.....
... 4 ...
........
... --
-
1
.
..!;::
- - '
:k<· ,
'•,
'4 (C-0 0 0 L -, ·------...-=-------1 ------------
0 . 15 1----------�--1- ,, · -------- · 4-- _----- 0 .0 ---2------------ ___ -_ ..
b I ---- ....,.____ 2
·
-2 -3- -- -- - --- - - I
_
_ _
------- �-:.:
1
-_
:C. I · .:::���-;,----2------------- 2---����-,---- 2 '.::-:-.,.. �-1 -._
I �__ ·-,
_ :;,.. ---�����4 i
I _,. ...
1 ' ...
,, -- --
f
-
... - �
--- - -- -
-- ----- -- __ ' 2------ ----2 ----- _,<::.--- 2--------------�
-- 1
-- -
-
...... =1
-
0.1 0 -0.004 - -0.004
- ---- ---...�-� �
I-
t
o. os -0.ooa 1 1 1 I ..
-0 .oo e ---....--------------...

MV1 MV2 MV3 MV4 MV5 MV1 MV2 MV3 MV4 MV5 MV1 MV2 MV3 MV4 MV5
(27.7) (75.5) (1 65.8) (397.3) (1 1 27.1 ) (27.7) (75.5) ( 1 65.8) (397.3) (1 1 27.1 ) (27.7) (75.5) (1 65.8) (397.3) 11 1 27.1 )

MARKET VALUE CATEGORY

(Millions of Dollars)
 21

abnormal returns between the high and low E/P portfolios seems to be smaller

for size category MV5 when compared to MVl for instance. This inference, in

fact, is confirmed by the Hotelling's T2 test results, which are reported in

columns (8) and (13) of Table 5. In particular, note that the vectors of

abnormal returns for only size classes MVl to MV3 are significant at the 1%

level or higher. Additionally, multiple comparison tests performed in the

context of the Hotelling framework indicate that while the a for the highest
p
E/P portfolio is significantly different from that for low E/P portfolios 1

and 2 for MV1-MV3, this is not the case -- even at the 1 0 % level of

significance - for the two classes which include the largest NYSE firms.13

It would appear, therefore, that the earnings' yield effect is not entirely

independent of firm size.

Further evidence on this latter point is provided in Table 6, which shows

more directly the effect of varying firm size per se on the performance of

securities included in each of five mutually exclusive E/P categories. As

before, the size portfolios in Table 6 were constructed by partitioning firms

included in a given earnings' yield class (i.e., portfolios EP1-EP5) on the

basis of the market value of their comron stock .

With the exception of EP5, the abnormal return vectors for the other four

earnings' yield classes are not significantly different from zero at the 1%

level or higher. Furthermore, the normalized vector of weights associated

with the maximum T2 statistic for category EP5 reveals that the rejection of

the null hypothesis can be attributed to the abnormal return performance of

not only small high E/P firms, but also the larger high E/P firms included in

13. In the case of size category MVS , the abnormal return for the highest E/P
firms, however, is significantly larger than that for E/P p:>rtfolio 3 at the
5% level.
 TABLE 6

ACTUAL RETURNS & SEl.ECTED CAPM RESULTS FOR MARKET VALUE PORTFOLIOS CLASSIFIED BY EARNINGS ' YIELD
(Based on monthly data for the period 1963-79)

S1111a
11 ry Statis tics!!/
/ �
Portfo li .,!_/ Actual Returna Equally Weighted NYSE Index Value Weighted NYSE Index .
'
MV "'
. -� / .
EP
/ / . "' "' . \. . "'
r p p ( r ' r') Ii t ( li ) F* ( li ) ll p(r' r') Ii t (li ) F* ( 5 )
Class Class p p p, m p p p p p, m p p p
rp o (�p) /o(�p)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) ( 1 1) (12) (13)

l 0 . 0131 0.0894 0 . 1 46 1 . 449 0.922 -0. 0009 -0 . 36 1 . 45 1.577 0.745 0.0047 1.12 0 . 78
2 0.0099 0 . 0759 0 . 130 1.250 0 . 931 -0. 0028 -1.48 1.574 0 . 875 0.0015 0.59
EPl l 0 . 0066 0.0664 0.099 1.038 0.888 -0.0047 -2.18 1.409 0 . 894 -0. 0014 -0.65
4 0.0058 0.0576 0 . 101 0.889 0.87 7 -0.0044 -2.28 1 . 251 0.915 -0.0017 -1.06
5 0. 0063 0.0502 0 . 125 0.649 0 . 733 -0.0025 -1.02 1.083 0.907 -0. 0009 -0.59

1 0.0098 0.0706 0.138 1.186 0 . 955 -0 . 0024 -1.65 2.87 1.388 0.829 0.0019 0.68 0.73
2 0 . 0075 0 . 0588 0 . 127 0.985 0.951 -0.0034 -2.69 1.248 0.895 -0.0001 -0.04
EP2 3 0 . 0061 0.0546 0 . 112 0.903 0.939 -0.0042 -3.23 1 . 184 0.91.4 -0.00ll -0 . 82
4 0 . 0062 0 . 0512 0.121 0.811 0 . 899 -0 .0036 -2.29 1.133 0.932 -0.0011 -0.84
5 0 . 0058 0 . 0442 0.131 0 . 668 0 . 857 -0. 0031 -1.91 1 . 008 0.959 -0.0012 -1.35

1 0 . 0108 0 . 0656 0 . ) 65 1 . 106 0.958 -0.0009 -0 . 68 1.87 1 . 282 0 . 824 0.0032 1.21 1 . 25
2 0 . 0093 0.0572 0. 163 0 . 970 0.963 -0 . 0015 -1 . l 7 1.200 0 . 884 0.0019 1.00
EP3 3 0.0081 0.0529 0 . 152 0.894 0.959 -0.0023 -2.15 1 . 14 3 0 . 910 0 . 0008 0.49 N
4 0 . 0097 0.0494 0 . 195 0 . 821 0 . 942 -0.0002 1.073
N
-0.15 0.914 0.0025 1. 79
5 0 . 0063 0.0421 0 . 151 0.652 0.879 -0. 0024 -1 . 69 0.925 0 . 926 -0. 0004 -0.38

1 0 . 0150 0 . 0705 0.2ll 1 . 181 0 . 952 0.0029 1.88 2.09 1.334 0 . 798 0.0073 2.44 2 . 29
2 O . Ol l7 0 . 0572 0 . 239 0 . 965 0 . 958 0 . 0029 2.52 1 . 161 0 . 855 0.0063 3 . 04
EP4 3 0 . 0124 0 . 0538 0 . 230 0.905 0 . 955 0.0020 1 . 80 1 . 119 0.871 0.0052 2.84
4 0 . 0125 0.0523 0.239 0.865 0.940 0.0024 1 . 88 1.091 0 . 880 0.0053 3.04
5 0.0099 0 . 0483 0 . 227 0.699 0.905 0 . 0009 0.68 0.952 0.915 0.0031 2.50

l 0 . 0194 0 . 0750 o . 259 1 . 221 0 . 930 0 . 0069 3.57 4.50 1.357 0.763 0.0116 3.40 3.00
2 0 . 0171 0.0649 0.250 l.lltO 0.946 0.0052 3 . 34 1.292 o. 796 0.0094 3 . 24
EP5 3 0 . 0155 0 . 0590 0.262 0.986 0.949 0 . 0046 3 . 50 1 . 15 7 0 . 826 0.0082 3.49
4 0.0145 0 . 0561 0 . 259 0.932 0.944 0.0040 3.03 1 . 156 0 . 869 0.0072 3.68
5 0 . 0121 0.0507 0.239 0.830 0 . 929 0.0022 1.69 1 . 069 0 . 887 0.0051 3 . 04
--·�-· -·

.!!l roccfolios are formed by ranking securities on the basis of their market value (llV) in a given esrnings ' yield (EP) class .

•
b/- "'
- r • mean monthl y return on por tfolio p;
o (r ) • standard deviation of wontbly return on portfolio p. ll • estimated systematic risk for portfolio p ;
p p p
p ( r ' r ' ) • coefficient o f correlation between the return on portfolio p (net o f risk-free rate) , r' , and t h a t o n the market index (net of rlak-free
p, m • • p • •

rate), r'; 6 � differential return -- estimated intercept for OLS regression of r' on r ' ; t (li ) • t-value for Ii • O; and F•(6 ) • F ( 5, 199) - value
m p • p 11 p p p
2
corresponding to Hotelling' e 1' t:e e t statistic for a • Selected fractiles from the ii(n,d) distribution are:
p

0.90 0.95 0.99 0.90 0.95 0.99

F ( 5 , J.20) 1 . 90 2 . 2!1 3.17 F (5,-) l . 65 2.21 3.02

 23

size groups 2-4. 14 In addition, multiple comparison tests lead one to infer

that the abnormal returns experienced by the smallest firms (i.e., group 1)

are not stochastically different -- at even the 1 0 % level -- from the

corresponding returns for firms included in any of the other four size

portfolios (i.e., groups 4-5). This remark applies to all five earnings'

yield categories.

In short, these results lend credibility to the view that for NYSE firms

the effect of differences in firm size on common stock returns is of secondary

importance when compared with the effect of E/P ratios.

14. The weights t.ll1derlying the maximum T 2 statistics for EP5 are as follows:

Normalized Weights for EP Class �

.
Size Portfolio

Market Index 1 2 3 4 5

F.qually-weighted 0 . 273 0 . 106 0 . 244 0. 313 0 . 064

Value-weighted 0 . 220 -0. 178 0 . 260 0 . 549 0 . 148

Note that about 56% of the weight underlying the T 2 statistic for the equally­
weighted case can be accounted for by size portfolios 3 and 4. The comparable
figure for the statistic pertaining to the situation involving the value­
weighted index is 69% .

Incidentally, the quartiles from the distribution o f market values

(pooled data) for each of these five size portfolios applicable to EP class 5
are as follows:

Market Value Size Portfolio for EP Class 5

(millions of .fil_ 1 -- 2 3- 4
- - 5

I...c:Mer quartile 13 . 7 34. 2 57 .5 102 . 3 316. 3

Median 21. 5 44 . 0 74. 9 161.3 574 . 7
Upper quartile 29 . 5 55. 1 99 . 4 239 . 3 998 . 6

A comparison o f the above with the distribution o f market values for the
entire NYSE (see Table 1) indicates that size groups 3 and 4 cannot be said to
contain small firms, i.e., those belonging to the lowest market value quintile
(MVl) on the exchange. On the contrary, they include firms with market values
canparable to those contained in the second and third quintiles of the NYSE.
 24

4. SG1E CONCIIJDING REMARKS

The empir ical evidence presented in this paper indicates that, at least

during the 1963-79 time period, the returns on the common stock of NYSE firms

appear to have been related to earnings' yield and firm size. In particular ,

the common stock of h igh E/P firms seem to have earned , on average , h igher

r isk-ad j usted returns than the common stock of low E/P f irms. This E/P

effect, furthermore, is clearly significant even after experimental control

was exerc ised over d ifferences in firm size, i.e. , after the effect of s i ze ,

as measured by the mar ket value of common stoc k , was randomized across the

high and loo E/P groups .

The results also indicate that while the common stock of small NYSE firms

appear to have earned marginally higher risk-adjusted returns than the common

stock of large NYSE firms, the size effect virtually disappears when returns

are controlled for differences in E/P ratios. Accordingly, it would appear

that the size effect observed by Banz for NYSE firms may, in fact, be a proxy

for the earnings' yield effect rather than vice-versa. Further analysis for

possible effects of interaction between E/P ratios and market values of common

stock , however , suggest that a somewhat different interpretation is more

appropriate. Essentially, the evidence indicates that firm size may have an

indirect or second-order effect on the returns of NYSE common stocks : the

strength of the earnings' yield effect seems to vary inversely with firm size.

More specifically, the results show that the E/P effect is sufficiently weak

for larger than average NYSE firms that from a stochastic viewpoint it either

is not s ignificant or , at best, is marginally sign i ficant. In other words ,

the common stock of high E/P firms seems to have experienced significantly

higher r isk-adj usted returns than the common stock of the ir low E/P

counterparts in all mar ket value categor ies other than those which include

these large firms .

 25

These empirical anomalies are consistent with the hypothesis that either

the one-period capital asset pricing model is misspecified due to the omission

of other relevant factors and, therefore, does not adequately represent market

equ ilibr ium or that the NYSE is not completely effic ient, or both. To the

extent one believes the former interpretation is appropriate because of the

longevity of the E/P anomaly, then the results reported here imply that, as a

minimum, E/P ratios are correlated with the set of missing factors that are

relevant to the pricing of NYSE cormnon stocks .

 26

REFERENCES

Basu , s., 1977, Investment performance of common stocks in relation to their

pr ice-earnings ratios : A test of the efficient market hypothesis ,
Journal of Finance , June , 663-682.

Banz , R.W., 1981, The relationship between return and market value of common
stocks, Journal of Financial Economics, March, 3-18.

Conover , W.J. , 1980 , Practical nonparametric statistics (Wiley , New York) .

Hollander , M. and D.A. Wolfe, 1973, Nonparametric statistical methods (Wiley,

New York) .

Mor r ison , D.F. , 1967, Multivar i ate statistical methods (McGraw Hill , New
York) .

Reinganum, M.R., 1981, Misspecification of capital asset pricing: Empir ical

anomalies based on earning s ' yields and market values , Journal - of
Financial Economics, March, 19-46.

Roll , R. , 1977 , A cr itique of the asset pr icing theory ' s tests , Journal of
Financial Economics , March , 129-176.

Roll , R.,1978, Ambigu ity when per formance is measured by the secur ities
market line , Journal of Finance , September , 1051-1069 .

Theil , H . , 1971, Principles of econometrics (Wiley , New York) •

 Faculty of Bus ines s
McMaster Univers i ty

WORKING PAPER SERIES

101 . Torrance , Geo rge W . , "A Generalized Cos t-effec tivenes s Model for the
Evaluation of Health Programs , " Novemb er , 1970.

102 . Isb ester , A. Fras er and Sandra C . Cas tle , " Teachers and Collect ive
Bargaining in On tario : A Means to What End ? " November , 19 7 1 .

103. Thomas , Arthur L . , " Trans fer Pri ces o f the Multinational Firm : When
Will They b e Arbi trary ? " (Rep rinted from : Abacus , Vo l . 7 , No . 1,
· June , 1 9 7 1 ) .

104 . S z endrovit s , Andrew z . , "An Ec o nomic P roduction Quantity Model with

Holding Time and Cos t s of Work- in-proces s Invento ry , " March , 1974 .

· 1 11 . Bas u , S. , " Inves tment Performance o f Common S t ocks in Rel ation t o

their Price-earnings Ratios : A Text o f t h e E f ficient Market
Hypo thesis , " March , 1 9 7 5 .

112 . Trus co t t , Wil liam G . , " S ome Dynami c Extensions o f a Dis crete Location­
Allocation Problem , " March , 1976.

113 . Bas u , S . and J . R . Hanna , "Accoun t ing for Changes in the General
Purchasing P ower of Money : The Impact on Financial S tatement s o f
Canadian Corp ora tions for the Period 1 9 6 7- 7 4 , " April - 19 7 6 .
(Reprinted from Cos t and Managemen t , January-February , 1976) .
·'

1 14 . Dea l , K . R . , "Veri fication o f the Theoretical Consis tency o f a

Dif ferential Game in Advertising , " March , 19 7 6 .

114a. Deal , K . R . , " Op timizing Advertising Expenditures in a Dynamic Duo p o ly , "

March , 19 7 6 .

115 . Adams , Roy J . , "The Canada-United S tates Labour Link Under S tress , "
[1976 ] .

11 6 . Thomas , Arthur L . , "The Extended App roach t o Jo int- Cos t Allocation :

Relaxation of Simplifying Ass ump tions , " June , 197 6 .

117 . Adams , Roy J . and C . H . Rummel , "Worker ' s P articipation in Management

in Wes t Germany : Impact on the Work , the En t erprise and the Trade
Unions , " Sep tember , 1 9 7 6 .

118 . S zendrovi ts , Andrew z . , "A Comment on ' Op timal and Sys tem Myopic
Policies for Multi- echelon Pro duction/ Inventory Ass emb ly Sys tems ' , "
[1976 ] .

119 . Headows , Ian S . G . , " Organic Structure and Innovation in Small Work
Groups , " Octob er , 1976 .

Continued on Page 2 . . .
 - 2

120. Basu , S . , "The Effect o f Earnings Yield on As ses sments o f the

As sociation B etween Annual Accounting Income Numbers and Security
Prices , " October , 1 9 7 6 .

12 1 . Agarwal , Naresh C . , "Labour Supply Behaviour o f Marri ed Women - A

Model with Permanent and Trans i tory Variab les , " October , 1 9 7 6 .

122 . Meadows , Ian S . G . , "Organic S t ructure , Sat is faction and Personality , "
Octob er , 19 7 6 .

12 3 . B an t ing , Peter M. , " Cus tomer Service in Indus trial Marketing : A

Comparative S tudy , " O ctob er , 1976. (Reprinted from : Europ ean
Journal of Marke tin g , Vol . 10 , No . 3 , Sunnner , 1 9 7 6 ) .

124 . Aivazian , V . , "On the Comparative- S tatics o f As s e t Deman d , "

Augus t , 19 7 6 .

125 . Aivazian , V . , " Cont amination by Ris k Recons idered , " Octob er , 1976.

12 6 . S z endrovits , Andrew Z . and Geo r ge 0 . Wesolowsky , "Variat i on in

Optimizing Serial Multi- S tate Produc tion/ Inventory Sys tems ,
March , 19 7 7 .

12 7 . Agarwal , Naresh C . , " S i ze-S tructure Relati onship : A Further

Elaboration , " March , 1 9 7 7 .

12 8 . Jain , Harish C . , "Minority Workers , the S tructure of Labour Marke ts

,•
and Anti-Di s criminat i on Legis lation , " March , 1 9 7 7 .

12 9 . Adams , Roy J . , " Employer S o lidarity , " March , 1 9 7 7 .

130 . Gould , Lawrence I . and S tanley N . Laiken , "The Ef fect of Income

Taxation and Inves tment Priori ties : The RRSP , " Marc h , 19 7 7 .

131 . Callen , Jeffrey L . , " Financial Co s t Allocatio"ns : A Game- Theo retic

Approach , " March , 1 9 7 7 .

13 2 . Jain , Harish C . , "Race and Sex Dis crimination Legis lation in North
Ameri ca and Britain : Some Les s ons for Canada , " May , 1 9 7 7 .

13 3 . Hayashi , Kichiro . " Corporate Planning Practi ces in Japanese

Multinational s . " Accep t ed for publicati on in the Academy o f
Management Journal in 1 9 7 8 .

134 . Jain , Harish C . , Neil Hood and Steve Young , " Cross-Cul tural Aspects of
Pers onnel Po li cies in Mul ti-Nationals : A Cas e S tudy o f Chrys ler . UK" ,
June , 19 7 7 .

135 . Aivazian , V . and J . L . Callen , "Investment , Marke t S t ructure and the

Co s t of Capi tal" , July , 1 9 7 7 .

Con t inued on Page 3 . . .

 - 3 -

136 . Adams , R . J . , " Canadian Indus trial Relations and the German Example " ,
October , 1 9 7 7 .

137 . Callen , J . L . , "Production , E f ficiency and Welfare in the U . S . Natural

Gas Transmi s sion Industry " , October , 19 7 7 .

138 . Richards on , A . W . and Wes olowsky , G . O . , " Co s t-Volume-Prof i t Analysis

and the Value of Information" , Novemb er , 19 7 7 .

139 . Jain , Harish C . , "Labour Market Problems o f Native People in Ontario " ,
Decemb er , 19 7 7 .

140 . Gqrdon , M . J . and L . I . Gould , "The Cos t o f Equity Cap i tal : A Reconsid­
eration " , January , 1 9 7 8 .

141 . Gordon , M . J . and L . I . Gould , "The Cos t o f Equity Cap ital with Personal
Income Taxes and Flotati on Cos t s " , January , 1978.

142 . Adams , R . J . , "Dunlop Af ter Two Decades : Sys tems Theory a s a Framewor.k
For Organi zing the Field o f Indus trial Relations " , January , 1978.

143 . Agarwal , N . C . and Jain , H . C . , "Pay Dis crimination Agains t Women in

Canada : I s s ues and P o licies " , February , 197 8 .

144 . Jain , H . C . and Sloane , P . J . , "Race , S ex and Minority Group Dis crimination
Legis lati on in North America and Britain " , Mar ch , 19 7 8 .

145 . Agarwal , N . C . , "A Labour Market Analys is o f Executive Earnin gs " ,

·

' June , 1 9 7 8 .

146 . Jain , H . C . and Youn g , A . , "Racial Dis crimination in the U . K . Labour

Marke t : Theory and Eviden ce" , June , 1 9 7 8 .

147 . Yagil , J . , " On Alternative Methods o f Treating Ris k , " Sep t emb er , 1 9 7 8 .

148 . Jain , H . C . , "At titudes t oward Cmmnunication Sys tern : A Comparison o f

Anglophone and Francophone Hos p i tal Employees , " S ep tember , 1978 .

14 9 . Ros s , R. , "Market ing Through the Japanes e Dis tribution Sys tem" ,
Novemb er , 1978 .

150. Gould , Lawrence I . and S tanley N . Laiken , " Dividends vs . Cap i tal Gains
Under Share Redemp tions , " December , 19 7 8 .

151. Gould , Lawrence I . and Stanley N . Laiken , " The Impact o f General
Averaging on Income Reali zation Decis ions : A Caveat on Tax
Deferral , " Decemb er , 19 7 8 .

152 . Jain , Harish C . , Jacques Normand and Rab indra N . Kanungo , "Job Mo t ivation
o f Canadian Anglophone and Franc ophone Hos p i tal Employees , Ap ril , 1 9 7 9 .

153 . S t idsen , Ben t , " Communi cations Relations " , Apri l , 1979 .

154 . S z endrovi ts , A . Z . and Drezner , Zvi , "Op timizing N- S tage Production/

Inventory Sys tems by Transporting Di f f erent Numb ers of Equal- S i z ed
Ba tches at Various S tages " , April , 1979 .
Cont inued on Page 4 . . .
 - 4 -

155 . Trus co t t , W . G . , "Alloca tion Analysis o f a Dynamic Dis tribution

P rob lem" , June , 1979 .

156 , Hanna , J . R . , "Measuring Capital and Income " , November , 1 9 7 9 .

15 7 . Deal , K . R . , "Numerical So lution and Multip le S cenario Inves tigation o f

Lin�ar Quadrati c Dif ferential Games " , November , 1979 .

158 . Hanna , J . R . , "Pro fessional Accounting Education in Canada : Prob lems

and Prospec t s " , Novemb er , 1979 .

159 . Adams , R. J . , "Towards a More Competent Labor Forc e : A Training Levy

S cheme for Canada " , December , 1 9 7 9 .

160 . Jain , H . C . , "Management of Human Resources and P roductivity" ,

February , 1 9 80 .

161 . Wensley , A . , "The Efficiency of Canadian Foreign Exchange Marke t s " , ·

February , 1980 .

162 . Tihanyi , E . , " The Market Valuat ion of Deferred Taxes " , March , 1 9 80 .

163 . Meadows , I .S . , "Quality of Working Li f e : P ro gres s , P rob l ems and

Prospec ts " , March , 1 980 .

164. S z endrovits , A . Z . , "The Effect o f Numbers of S tages o n Mul ti- S tage

Production / Inventory Models - An Empirical S tudy" , April , 1980 .
,•

165 . Laiken , S .N . , " Current Action to Lower Future Taxes : General Averaging
and Anti cipated Income Models " , Apr il , 1 9 8 0 .

166 . Love , R . F . , "Hull Properties in Lo cation P roblems " , Apri l , 1 9 8 0 .

167 . Jain , H . C . , " Dis advantaged Groups on the Labour Market " , May , 1 9 80 .

168. Adams , R . J . , " Training in Canadian Industry : Res earch Theory and
Policy Imp lications " , June , 19 8 0 .

16 9 . Joyner , R . C . , "App lication o f Process Theories t o Teaching Uns tructured

Managerial Decis ion Making" , Augus t , 1 9 80 .

170. Love , R . F . , "A S t opping Rule f o r Facili ties Lo cation Algorithms " ,
Sep tember , 1980.

171 . Abad , Prakash L . , "An Op timal Con trol Approach to �1arketing - Production
P lanning" , October , 1980.

172 . Abad , Prakash L. , "Decentrali zed Planning With An Interdep endent

Marketing-Production Sys t em" , Octob er , 1980 .

173 . Adams , R . J . , " Indus trial Rel ati ons Sys tems in Europe and North America" ,
Octob er , 1 9 80 .

Con tinued on Page 5 • . .

 - 5 -

174. Gaa , James C . , " The Role o f Central Rulemaking In Corporate Financial
Rep ortin g " , February , 1981.

1 75 . Adams , Roy J . , "A Theory o f Emp loyer At ti tudes and Behaviour Towards
Trade Unions In Wes t ern Europe and North America" , February , 1 9 8 1 .

176 . Love , Rob ert F . and Js un Y . Won g , " A 0-1 Linear Program T o Minimi ze
In terac tion Cos t In S cheduling" , May , 1981 .

177. Jain , Ha r ish , " Employment and Pay Dis crimination in Canada : Theories ,
Evidence and P olicie s " , June , 19 8 1 .

178. Basu, S. , "Market Reaction t o Accounting P o licy Deliberation : The

Inf la t ion Accounting Cas e Revis ited" , June , 19 8 1 .

179 . Basu, s,. , "Risk Information and Financial Lea s e Dis closures : Some
Empiri cal Evidence'' , June , 1 9 81 .
(1� \ · o�
I
L. h 'd
S 'nL
CO H
;)3�
<; ��'Jc-