A rgum ent A
QECONOMIC^
1-2 (18) *2006
Wrocław University of Economics
Wrocław 2006
TABLE OF CONTENTS
I. INAUGURAL LECTURE FOR OPENING THE ACADEMIC YEAR 2006-2007
Jan Kurowicki
ECONOMY AND ART....................................................................................................................
5
II. ARTICLES
Danute Diskiene, Birut'e Galiniene, Vaclovas Lakis, Albinas Marcinskas
CHANGES IN THE REAL ESTATE AND LABOUR MARKETS
DURING T H E PROCESS OF LITHUANIA’S INTEGRATION
INTO THE EUROPEAN UNION..................................... ...............................................................
21
Aclela Barahasz
ORGANIZATIONAL PERSONALITY AS A METAPHOR
FOR UNDERSTANDING ORGANIZATIONS..........................................................................
37
A kbarZ. Ali, Adam Szvszka
ETHICAL FACTORS IN CAPITAL MARKET.
SOCIALLY RESPONSIBLE VERSUS UNSCRUPULOUS INVESTM ENT........................
63
Dariusz Klonowski
UNDERSTANDING VENTURE CAPITALISTS’ DECISION
ENVIRONMENT: EVIDENCE FROM CENTRAL AND EASTERN EURO PE..................
89
Ewa Maluszynska
FIRM M IGRATIO N ..........................................................................................................................
107
Marzena Stor
CREATING VALUE WITH DIVERSE TEAMS IN TRANSNATIONAL
MANAGEMENT: DIVERSITY AS LIABILITIES AND ASSETS..........................................
131
Kataizvna Kowalska
GP-FUNDHOLDING EXPERIENCE IN POLAND:
HEALTH C A R E ACCESSIBILITY IMPROVEMENTS............................................................
153
III. REVIEW S AND NOTES
Krzysztof Jajuga, Marek Walesiak (eds.): TAKSONOMIA 12. KLASYFIKACJA
I ANALIZA DANYCH - TEORIA I ZASTOSOWANIA [TAXONOMY. VOL. 12.
CLASSIFICATION AND DATA ANALYSIS - THEORY AND APPLICATIONS];
TAKSONOM IA 13. KLASYFIKACJA I ANALIZA DANYCH [TAXONOMY. VOL. 13.
CLASSIFICATION AND DATA ANALYSIS] (Eugeniusz Gatnar)........................................
173
Agnieszka Stanimir: ANALIZA KORESPONDENCJI JAKO NARZĘD ZIE DO
BADANIA ZJAWISK EKONOMICZNYCH [CORRESPONDENCE ANALYSIS
AS A T O O L FOR STUDY OF ECONOM IC PHENOMENA] (M ałgorzata Rószkiewicz).... 175
Bożena Klimczak, Andrzej Matysiak (eds.): DZIAŁANIA ZBIOROW E
W TEORII I PRAKTYCE [COLLECTIVE ACTION IN THEORY AND PRACTICE]
(Stanisław R ud o lf).................................................................................................................................177
Jerzy Niemczyk: WYRÓŻNIKI, BUDOW A I ZACHOWANIA STRATEGICZNE
UKŁADÓW OUTSOURCINGOWYCH [DISCRIMINANTS, STRUCTURE AND
STRATEGIC BEHAVIOURS OF OUTSOURCING SYSTEMS] (Adam Stabryla)
\g |
Zdzisław Pisz (ed.): ZMIANY SYTUACJI SPOŁECZNEJ NA DOLNYM ŚLĄSKU
W LATACH 1998-2002 [CHANGES O F SOCIAL SITUATION IN LOWER SILESIA,
1998-2002] (Robert Rauziński)........................................................................................................
183
Jan Skalik (ed.): ZMIANA WARUNKIEM SUKCESU. PRZEOBRAŻENIA METOD
I PRAKTYK ZARZĄDZANIA [CHANGE AS A CONDITION O F SUCCESS.
METAMORPHOSES IN METHODS AND PRACTICES OF MANAGEMENT]
(Cezary Suszyń ski).............................................................................................................................
I ^5
W iesław Pluta (ed.): ZARZĄDZANIE FINANSAMI FIRM - TEORIA I PRAKTYKA
[FIRM ’S FINANCE MANAGEMENT - THEORY AND PRACTICE]
(Dariusz Zarzecki)..................................................................................................................................189
IV. HABILITATION M ONOGRAPHS..................................................................................... ....191
A R G U M EN TA OECONOMICA
N o 1 -2 (18)2006
PL ISSN 1233-5835
AkbarZ. A li
*»
Adam Szyszka
**
ETHICAL FACTORS IN CAPITAL MARKET.
SOCIALLY RESPONSIBLE VERSUS UNSCRUPULOUS
INVESTMENT
S o c ia lly R esp o n sib le Investment (S R I) fun d s have been show n to underperform , primarily
due to restrictin g their investm ents to a su b set o f the universe o f in v esta b le assets. Rapid
growth o f S R I
funds im plies that there is a growing segm en t w ithin the investm ent
com m u n ity w h o are w illing to accept lo w e r returns than the unrestricted investors. H owever,
it also f o llo w s that investors’ utility d erived from ethical in vestm en ts perhaps reflects an
added d im e n sio n , or an ethical prem ium , that com pensates them for this underperformance.
This research q u estio n s whether in vestors, on average, would rem ain com m itted to ethical
in vestm en ts in the face o f decreasing w ealth . W e attempt to an sw er this question, by lirst
ob serving the d ifferen ces between an eth ical portfolio and an (un jeth ical portfolio, crcated by
using a ssets that are deem ed u n in vestab le by ethical screens. U sin g market and style
associated risk filtered premiums, w e find that (i) increased dem and for ethical assets results
in a d ecrea se in dem and for non-ethical a ssets, and (ii) poor past m arket perform ance, that
leads to g en era l wealth decreases, resu lts in increased dem and for unethical assets and
decreased d em a n d for ethical assets.
K e y w o r d s : S ocia lly R esponsible In vestm en t (SRI), ethics in capital market, Vector
A u toregression (V A R ), and Variance D eco m p o sitio n (VDC).
J E L c la s sific a tio n : G i l , G 20,
INTRODUCTION
T he n eo-classical theory o f fin an ce states that the only criteria of
investm ent ch o ic e should be the relationship between ex p e c te d return and
system atic risk. This n o tw ithstanding, we observe that p e o p le also take into
account o th e r considerations w h ile m aking their investm ent decisions. For
instance, th e y look at ethical and so cial values of the co m p an y they intend to
invest in. T h e re is now a special seg m en t o f the asset m an ag em en t industry represented m ainly by so-called S o cially R esponsible Inv estm en t (SRI)
funds - th a t have been g row ing fa st in the United States as w ell as in other
*
Sch ool o f C o m m erce, University o f A d ela id e, Australia
**
P oznan U n iv ersity o f E conom ics
d e v e lo p e d capital m arkets (th e U .K ., France, G erm an y , and Sw itzerland).
A c c o rd in g to the Social In v estm en t Forum (2003), th e total value of assets
m an ag em en t in socially resp o n sib le portfolios in th e U.S. reached $2.16
trillio n in 2002 which w as 11.3% o f all assets u n d e r m anagem ent in the
U n ite d States. In 1984, w hen the first statistics w e re taken, the value o f
so c ia lly responsible investm en t was estimated to be around $40 billion, and
by 1995 it was already $ 6 3 9 billion. Between 1995 and 2002, it rose by
a n o th e r $1.52 trillion, sh o w in g the growth rate w as 40% greater than the
g ro w th o f total conventional assets under m anagem ent (2003 Report on
S o c ia lly Responsible In v estin g Trends in the U nited Stated, Social
In v e stm e n t Forum, w w w .socialinvest.org).
E th ical values in the S R I segm ent are not universal. V arious SR I
m a n a g e rs may use d ifferen t criteria for the screening. At first, m ainly
n e g a tiv e exclusionary screen s w ere in use. For in stan ce, SRI funds w ould
refrain from investing in co m p an ies obtaining th eir revenues from w eapons,
tobacco products, alcohol o r gambling. In the tim e o f apartheid, many SRI
m anagers also eliminated from their portfolios com panies with interests in South
A frica (Teoh, Welch and W azzan 1999). Later, som e additional positive screens
w ere added to the selection procedure, which look for good employee relations,
environm ental and sustainability responsibility, products benefiting society, etc.
T h e increasing popularity o f socially responsible investm ent raises a
q u e stio n about returns offered to investors in ex ch an g e for their good ethical
stan d ard s. From the point o f view of finance th eo ry , there are two m ain
re a so n s w hy we should e x p e c t SRI to deliver rath er low er than higher riska d ju ste d returns. Putting any additional constraints on portfolio selection
m ay o n ly lead to long-term underperform ance or - in the best case scenario
- sim ila r perform ance to conventional assets of the sam e risk characteristics.
S R I investm ent o p p ortunities are just a subset o f the total investm ent
u n iv e rse . Restriction to this subset may lead to underdiversification and
c o n stru c tio n o f sub-optim al portfolios. However, ev en if the amount o f SR I
o p p o rtu n ities is large and d iv erse enough to allo w proper levels o f risk
re d u c tio n , there may be still an o th er issue. If the financial strength of ethical
in v e sto rs is substantial, the increased demand for so cially responsible stocks
m ay m o v e the prices up. T h is is only true if we assu m e that there are lim its
o f ac tio n to unethical, but rational arbitrageurs, and th at supply of SRI stocks
is n o t perfectly elastic, i.e. com panies can sw itch to becom e ethical only
slow ly and gradually in th e response to the dem an d o f ethical investors.
H ig h e r current stock prices w ill mean lower expected returns for investors, but
also lo w er cost of capital fo r th e ethical company. H einkel, Kraus and Z echner
(2001) presented a formal model o f equilibrium in w hich som e investors have
additional non-financial criteria w hile making investment decisions.
E m p iric a lly , the perform ance o f socially responsible investm ent was
tested in th re e main areas o f research . Firstly, the returns o f SR I funds were
co m pared w ith those achieved by conventional m utual fu n d s. Studies in this
area in clu d e: H am ilton, Jo and Statm an (1993), S tatm a n (2000), Bauer,
K oedijk a n d O tten (2002), G eczy , Stam baugh and L evin (2003), Schroeder
(2003) [S ch ro ed er (2003) p re se n ts a more detailed literatu re review on
socially responsible investm ent b o th for the U.S. and E u ro p ean m arkets.],
and B ello (2005). Overall, th ere w ere no m ajor d iffe ren c es noticed in the
p erfo rm an ce o f SRI funds and co nventional funds both in the United States
and in E u ro p e. Therefore, th e hypothesis that the e th ica l constraint on
portfolio selection will lead to underperform ance w as n o t clearly proven.
H ow ever, resu lts from this type o f studies should be in terp reted with caution.
T here m ay be substantial d ifferen ces in the level o f risk a m o n g portfolios of
various fu n d s. A dditionally, d iffe re n t investm ent sty les and skills of fund
m anagers m ay blur the picture.
S om e o f the above m entioned draw backs could be av o id ed , if specially
co n stru c te d social indexes, such as the Domini 400 Social Index, the Calvert
Social In d e x , the Citizens Index, o r the Dow Jones S u stain ab ility Index, are
used to ap p ro x im ate the p erfo rm an ce o f socially resp o n sib le investm ent. A
num ber o f studies assessed d ire c tly the changes in levels in a social index
against th e general market in d ex es (Sauer 1997, D iB arto lo m eo and Kurtz
1999, S ta tm a n 2000, Statm an 2 0 0 5 , Schroeder 2003). A gain, generally
speaking, underperform ance o f S R I could not be proven, an d in some cases
the ov erp erfo rm an ce of the social index was observed. H ow ever, results
from a co m p ariso n o f indexes a g a in st the market should b e treated perhaps
' /ith ev en g re a te r caution than th o se from studies on re tu rn s achieved by SRI
funds. P erfo rm an ce o f social in d ex es seem s to be highly sensitive to their
co n stru ctio n m ethod and index inclusion criteria m ay d iffe r significantly
am ong in d e x e s and over time (S ta tm a n 2005).
T h e fin a l approach in testing th e SR I perform ance is to go directly to the
individual d a ta on stocks and to construct self-m ade p o rtfo lio s of ethical
assets th a t pass selected screens (D iltz 1995, G uerard 1997, Derwall et al
2004). K n o w in g characteristics o f stocks included in p o rtfolios helps to
un d erstan d w hat really drives th e returns of ethical a sse ts and if they are
different ju s t because of being eth ical or due to other characteristics. After
acco u n tin g for size, b ook-to-m arket, sectoral m om en tu m effects, prior
stu d ie s w ere generally u n ab le to distinguish any d ifferen c e in perform ance
o f eth ic a l stocks com pared to th e overall market.
A s the num ber of c o m p an ies that care a b o u t ethics and social
resp o n sib ility - or at least p u b licly declare to do so - dram atically increases,
th ey constitute a larger sh a re o f the total u n iv e rse of investm ent
o p p o rtu n ities. Therefore, it m ay be difficult to spot any significant difference
b e tw e e n the perform ance o f a social portfolio and general market, because
the pro p o rtio n o f purely u n eth ical assets in the m ark et is too small. T his
im p lies that the assets c o n ta in e d in social indexes are very sim ilar to assets
in o rd in a ry indexes, as o rd in ary indexes include m an y “ethically n eutral”
sto ck s. O ne should, therefore, com pare ethical assets w ith those which are
e x tre m e ly unethical, in o rd er to exaggerate difference in characteristics and
p erfo rm an ce. W e are not aw are o f any studies that do so.
In o u r research we sim u late the perform ance o f self-constructed unethical
in d ex or portfolio and co m p a re it against so cially responsible assets
re p re se n te d by the D om ini 4 0 0 Social Index (D S 4 0 0 ). In this way, w e
e lim in a te “ethically n eutral” com panies from the g eneral market from our
a n a ly sis. W e also focus only o n the top five com m o n ly accepted screens, in
o rd e r to pick up the m ost u n eth ical companies, n am ely assets connected with
a lc o h o l, tobacco, gam bling, w eapons, and th o se that are considered
en v iro n m en tally harm ful. O u r intention is to com pare possibly extrem e sides
o f the m arket in terms o f (u n )eth ical investment. If th e re are any differences
in perfo rm an ce to be n o ticed , it may be h y p o th esised that unethical
c o m p a n ie s are more likely to deliver higher re tu rn s. If there are m any
in v e sto rs with strong ethical beliefs that are not m et w ith unethical rational
arb itra g e u rs, the “bad” c o m p an ies should be p e n a liz ed w ith higher cost o f
c a p ita l. T his would also m ean h igher returns for th o se few who do not care
ab o u t ethics and agree to hold “ vice” stocks. W e look at characteristics of both
c ateg o ries o f assets (risk, size, book-to-m arket) and check for the m om entum
effect. T his research also analyses how the return spread between unethical
and ethical asset indexes (“unethical premium”) ch an g es over time and look
fo r th e factors that may influence the degree of in v esto rs’ morality.
1. METHODOLOGY
E v alu atio n o f how eth ical assets (as represented by the DS400 index)
b e h a v e sim ilarly (or d ifferen tly ) from unethical a sse ts is explored in three
w ays. F irst, we explore the sty les o f assets that are selected in each category
of in v estm en t, as well as observe ho w that style changes o v e r tim e. (Style is
analyzed in term s of grow th v ersu s value bias, and also in terms of
capitalizatio n .) Second, we o b serv e the inter-tem poral relatio n sh ip between
the tw o c la sse s o f investm ent. F in a lly , this research ex p lo re s factors that
affect th e retu rn s spread (that is a lso term ed as u n ethical prem ium ) of the
two in v e stm e n ts classes, and d e te rm in e if those factors are im portant over
time. A s m en tio n ed in the p rev io u s section, we use tw o sets o f indexes, the
first re p re se n ts ethical assets and is proxied by DS 4 0 0 , w hile the second
index, re p re se n tin g vice investm ent, is a value w eighted “u n e th ica l” sectors
(the secto rs are described in the n ex t section) from the m arket.
First, th e tw o types o f assets are evaluated for 3 -facto r risk prem ia over
the period o f evaluation. C o m p ariso n s o f risk prem ia are also m ade over two
periods sig n ifie d by low volatility and high volatility p e rio d s in the stock
m arket. T h e break in the sam p le period is obtained by observing the
graphical p lo t o f the m arket ind ex and then further tested using C how ’s
structural b reak test (the results a re not reported but ca n be provided on
dem and). W e utilize the follow ing regression to obtain the risk prem ium for
the F am a an d French 3-factor m odel
I i.l “ If,i — Ctj ^ P m [I’m.I " ff.t 3 + P sM B I '"Small C ap.t ~ l'i^arg>e C ap .i]-^" P h M L I r v a h i e . t — 1 Growth,t]~*~ Ci.t
Eq. la and lb
w h ere:
r ¡.,
is th e m o n t h ly returns on p o r t f o l i o ty p e i at tim e t; i c a n b e eith e r eth ica l
(e q u a tio n I a ) o r u n eth ica l (e q u a tio n 1 b ) a s s e t s .
r nu is th e m o n t h ly returns on W ils h ir e 5 0 0 0 at tim e t
r f , is th e m o n t h ly risk free rate o f retu rn at tim e t
r
Sm all. i
is th e m o n t h ly returns for W ils h i r e S m a ll C ap 2 5 0 I n d e x a t t im e t
‘‘ Large.i is th e m o n t h ly returns for W ils h i r e L a r g e C ap 7 5 0 I n d e x at tim e t
r value,i is th e m o n t h ly returns for W ils h i r e A ll V a lu e In d e x at t im e t
>' Growth.i is th e m o n th ly returns fo r W ils h ir e A ll G row th In d e x at t im e t
is th e e r r o r term for the r e g r e s s io n ( p l e a s e n o te that d e s p it e th e s a m e sy m b o ls ,
c o e f f ic ie n t s
a, , p m , pSMB , p HMI in e q u a tio n s 1-6 are th e r e s u lt o f d ifferen t
r e g r e s s io n s a n d th e r e fo r e h a v e d iff e r e n t v a lu e s ) .
If F am a an d F rench’s three fa c to rs account for all risk, the residuals from
both re g re ssio n s should be o nly w h ite noise. H ow ever, w here the two
residual se rie s contain inform ation n o t accounted for by th e 3 risk factors, it
is im portan t to understand the relatio n sh ip with each o th e r as well as with
other m a rk e t factors.
S eco n d , after ensuring th a t the residuals series are not white noise, w e
can then proceed to evalu ate any inter-temporal relationship that may exist
b etw een the two investm ent styles. Therefore, the second stage o f this
research analyses w hether the tw o investment styles are cointegrated, as well
as if any sim ultaneous relatio n sh ip might exist. If any long-run relationship
d o es exist, it should be m o d elled while attem pting to capture any short-run
relatio n sh ip . As both types o f assets exist in the sam e econom y, long-term
and short-term relationship sh o u ld be observed. H ow ever, this research tries
to e v alu ate the relationship betw een the two asset types based upon the
d e m a n d s due to investors’ ethical preferences, o th er than known system atic
risk factors, such as broad m arket effects and investm ent styles. H ence to
m odel such a relationship th ese systematic and know n investment style
facto rs should be included as exogenous to the system . Also, since both
ty p es o f assets are priced concurrently in the m arket place, it is essential to
o b tain their relationship on a sim ultaneous b a sis by using a V e cto r
A u to reg ressiv e (VAR) m odel. O ther reasons for u sin g a VAR model include
its ab ility to use n on-stationary series without sacrificin g coefficient validity,
as w ell as being able to fo recast out-of-sam ple e ffec t of one endogenous
v a ria b le on the other.
T h e results from such a sim ultaneous system o f equations can then
p ro v id e a m ulti-dim ensional analysis of any re m ain in g relationship that m ay
ex ist. F or exam ple, we are ab le to gauge the speed o f adjustm ent of the tw o
assets to a long run average, if any long-term relatio n sh ip exist. If eth ical
assets adjust faster than u n eth ical ones, then it im p lies that there is less o f an
o v er-reactio n by ethical in v esto rs and that u n ethical investments are less
effic ie n tly priced. The ex iste n c e o f any short-term relationship betw een the
tw o investm ent styles will reveal how investors’ d em an d for the two types o f
a sse ts is affected by each oth er. For exam ple, if current returns from a
p a rtic u la r type o f in vestm ent w ere affected by its ow n lagged returns, it
w o u ld im ply that past in fo rm atio n has not been co m pletely incorporated by
th e investors and that inefficiencies exist. H ence, a system of equations is
also able to provide in d icatio n s o f any persistence o f returns for each index,
len d in g support for m o m en tu m based trading. A dditionally, there m ay be
e v id e n c e that either one ty p e o f investment affects th e other, or feedback o f
inform ation between the tw o asset types is revealed. It is of interest to evaluate
the sign of lagged coefficient to determine the specific characteristics o f any
short-term relationship betw een ethical and vice asset groups.
T o test for short run relationships between tw o series, Engle and G ran g e r
(1 9 8 7 ) have provided a V e c to r A utoregression (V A R ) specification o f first
differen ces. In our case, the tw o series are return p re m iu m s from ethical
(^Ethical -rr.t) and unethical (runethicaKf.i) indexes. As m entioned earlier, known
system atic and investment styles are accounted for by including them as
exogenous factors in the system o f equations. Hence, o u r V A R model is
expressed as follows:
n
'E thical.I " r f.t
n
a l ( rEthical.t-l “ >'f.l l ) + ^
/= l
w l ( lUnethical.l-l "
) +
P „ i [ l ‘m , - T f , ] +
/= l
P sM B lr.S m a ll C a p .f* Liirge Cap.t] ^ P lIM L [rV a liie .t-rC iro w th .t]
^ E lh ic a l.t
Eq. 2a
n
n
*Unethical.t _If, t=
Yl ( *Ethical, l-l “ *"f. I-1 ) +
/=1
Xl ( ^Unethical.l-I “ *"f.t-l )
Pmlrni.rlV.t ] +
1=I
PsM B[**Sm all C a p.fI"Liirg e Cap.t]
P lIM L [l"V alue .t_rG row th.t]
^U n e th ica l, I
Eq. 2b
w h ere:
a | a n d yi a r e t h e la g c o e f f ic ie n t t e r m s o f e th ic a l r e s i d u a l s a n d o)j a n d Xi a re th e la g
c o e f f i c i e n t t e r m s o f u n e th ic a l r e s i d u a l s .
T e stin g fo r G ranger causality o f one variable to a n o th e r is conducted
through th e jo in t test o f sig n ifican ce for co and y. If oo is sig n ifican t it reveals
that the ch an g es in unethical resid u als causes changes in ethical residuals,
w hile a sig n ifican t y show s th at ch an g es in cthical resid u als G ranger causes
changes in unethical residuals. T h e appropriate lag len g th 1 is obtained by
search in g fo r the optimal A k aik e (1974) Inform ation C rite rio n over various
intervals up to 4 lags. The results indicate that a lag o f 2 for both series provides
the optim al AIC. (We used E -view s software to d eterm ine the optimal lag
structure by optimizing AIC.) H ence, equations 2a and 2b look as follows:
2
2
*Elhical.rlf.t— ^ * <,
( r E,hicaU_| - Tff.) ) +
/=1
w, ( runethical.t-l _ *"f.t-1 )
Pm[Tm.rr|.(l
+
/=1
P sM B lT S m a ll C a p .f r LargeCap.t]
P lIM L [rV a !u e .t- r Growth.t]
^ E th ic a l.t
Eq. 3a
2
* U nethical.t- I i.t
2
Yl ( Tp-tlncaM-l “ ^ f.t-l
/= l
P s M b [^ Small C a p .fr ijn g e Cap.t]
)
50 ( * Unethical.t-l ” lf.t-1
)
Pnil**ni.t“ *'f.t 1
/= l
P lIM L [ l V a lu e ,fl"G io w tlu ]
^U nethical.t
Eq. 3b
T h e jo in t test o f sig n ifican ce for coj and (o2, and fo r yi and 72 provides
e v id e n c e o f existence and th e direction of causality. H ow ever, any lead-lag
re la tio n sh ip observed b etw een variables using G ra n g e r causality testing
re v e a ls only in-sam ple effects b u t is unable to pro v id e the dynam ic nature o f
re la tio n sh ip between these variables. Also, the m ag n itu d e and direction
o u tsid e the sam ple period c a n n o t be gauged. Sim s (1 9 8 2 ) has shown that for
a g iv e n system s of equatio n s, its reaction to a ran d o m innovation can be
o b se rv e d for each variables by Im pulse R esponse F u n ctio n (IRF). T hrough
this technology, one is able to observe the transitory as well as perm anent
e ffe c ts on each variable in th e system due to a ra n d o m shock originating
fro m o n e o f the variables w ith in the system. G rap h ically , one can observe
the p a th o f one o f the variab les due to a one standard deviation shock w ithin
the sy stem .
F u rth e r, Sim s (1982) h as show n that if th e fo re c a ste d error of each
v a ria b le and for each tim e perio d can be a ttrib u te d due to its ow n
in n o v a tio n s and those d u e to th e other v a ria b le s in the system . T h is
m e a n s th at each v a ria b le ’s fo recasted variance c a n be decom posed to
p ro v id e u n d erstan d in g o f its future d ire c tio n through V a rian ce
D e c o m p o sitio n (V D C ). F o r exam ple, if e th ic a l a sse ts had a la rg e r
in flu e n c e on unethical a s s e ts , then the ethical in v e stm e n ts’ forecasted
v a ria n c e w ould prim arily b e d u e to its own in n o v a tio n s, but the v ariance
o f th e vice investm ent w o u ld show a much la rg e r im p ac t due to e ffe c ts
fro m th e ethical in v estm en t innov ations. V D C is d e riv e d from a m o v in g
a v e ra g e representation o f th e o riginal V AR e q u a tio n . (F o r further d e ta ils
se e S im s (1980, 1982).)
T h e th ird aspect o f th is re se a rc h evaluates th e re tu rn spread betw een
u n e th ic a l and ethical a sse t in d e x e s, and facto rs th a t help explain such a
s p re a d o v e r various m ark et co n d itio n s.
*Unethical.f* iilhicnl.t
"^PsMBlrSmall Cap.i'H^ige Cap,t]~^PllMLl rvalue.t*l Growth.
Eq. 4
w here:
• unethical.! ■ rEihicai.i is
th e
m o n t h ly
p o r tfo lio
return
d if f e r e n c e s
b e tw e e n
e t h ic a l
p o r t f o l i o s and u n eth ical p o r t f o lio s .
T h is research’s prim ary h ypothesis is that a d e cre ase in the return spread
b e tw e e n unethical and ethical asset indexes may be d u e to a higher dem and
for ethical assets. Demand for eth ical assets may be due to tw o reasons; first,
if in v e sto rs can accept low er re tu rn s in lieu o f fe e lin g “ g o o d ” about
e n c o u ra g in g ethical corp o rate b e h a v io u r and in v estm en t. S u ch a trade-off
by in v e s to rs w ould m ost lik e ly tak e place w hen th e investors feel
c o n fid e n t a b o u t their w ealth. T h e seco n d reason for th is d em an d may be
if in v e sto rs are aw are o f fu tu re m a rk e t conditions and are a b le to gauge if
“ vice” a s s e ts will under p e rfo rm in future. An a lte rn a tiv e reason for a
d ecrea se in uneth ical-eth ical re tu rn s spread m ay b e th at socially
re sp o n sib le (S R ) fund m an ag ers m ay have g a in e d -e x p e rie n c e over tim e
and h e n c e a re able to pro d u ce b e tte r returns fo r th e ir in v e sto rs. To test
the first th e o ry , we analyze if u n e th ic a l-e th ica l sp re ad is related to past
m arket c o n d itio n s or c o n te m p o ra n e o u s ethical retu rn s, b y perfo rm in g the
fo llo w in g reg ressio n :
A [l"lli)c th ic a l.t ” 1 E th ica l.t]— l^ i~^Pni-Lag ^ [ r n u - l
]"^PElhical_premiuin ^ [ ^ E T h i c a l. f * m . l] ^ ^lt
Eq. 5
w h ere:
A [I'uneiiiicni.t - rEthical.il is the m o n th ly c h a n g e o f the sp re a d b e t w e e n u n eth ica l and
e th ic a l p o r t f o lio s ,
A [rm,(-i - iy ,. i ] is the m o n th ly c h a n g e o f th e m ark et p rem iu m ,
A[iEThicai.rrm.i]r m.i is the m o n th ly c h a n g e o f the eth ical p r e m iu m .
If Pm-uig in the regression a b o v e is positive, it p ro v id e s evidence that
unethical assets improve returns in the period fo llo w in g higher market
returns in th e previous period, w h ile a negative sig n ific an t coefficient
indicates th e opposite. W e h y p o th esize that the beta sh o u ld be negative if
in vestors’ a sse t holdings m ove fro m ethical to unethical a ssets based upon
negative ch a n g e s in market co n d itio n s and hence in v esto rs’ w ealth. Thus a
negative co efficien t indicates that investors are ethical (in th a t they choose a
low er retu rn on ethical assets to v ice assets) if they hav e higher levels of
wealth. H o w ev er, it may sim ply b e the case that contem p o ran eo u s returns
from eth ical assets are high, and h en ce a negative sig n ific a n t PEthicaLpiemium
coefficient.
O ur se c o n d hypothesis relates th e experience o f eth ical fund managers
with the p erform ance of ethical fu n d s. Since socially re sp o n sib le investment
style is a re c e n t style of investm ents, it could be h y p o th esiz ed that ethical
fund m a n ag ers do not have th e sam e level o f ex p e rie n c e in fund
m anagem en t, as would m anagers o f other styles, and th is in turn implies
u n d erp erfo rm an ce of SR I fu n d s. This underperform ance may lead to a
d e c re a se in investors’ cash flo w s to such funds and hence a decreased
d e m a n d for ethical assets u n d e r m anagement. T his d e cre ase in demand m ay
p e rh a p s be one of the factors that may have lead to a tem porary negative
p re ssu re on asset prices. H ence, if ethical fund m an ag ers gain experience and
th e ab ility to better m anage ethical portfolios over tim e, then the unethicaleth ical spread should also d ecrease over time. T o ev a lu ate this hypothesis,
the fo llo w in g regression is estim ated :
' Unethical.I “ * Ethical.! ~
+ P lim c t i m e + (p t
Eq. 6
w here:
t im e is th e n u m b er o f m o n th s s in c e M a y 1990.
In th e above regression, if our hypothesis holds, then Prime should be
n eg a tiv e , indicating d ecreased unethical-ethical sp read over time. It is
n e cessary to point out that changes in asset p rices in response to fund
m a n a g e r decreased holdings should only occur if the fund m anagers are the
m a jo rity shareholders for th at asset.
2. DATA DESCRIPTION AND PRELIMINARY RESULTS
O u r proxy for a portfolio o f socially responsible com panies is the D S400
Ind ex , w hich was initiated in M ay 1990 by K inder, Lydenberg, Domini &
C o p m a n y (KLD). It is a capitalization-w eighted index that consists o f 400
c o m p a n ie s: approxim ately 2 5 0 o f them are large-cap stocks that are also
in c lu d e d in the S & P 500 Ind ex , there are a b o u t 100 non-S & P 500
c o m p a n ie s that are selected to provide proper in d u stry representation, and
th e re are approxim ately 5 0 non-S& P500 c o m p a n ie s with particularly
s tro n g ethical ch aracteristics. T h e exclusionary sc re e n s elim inate from the
c o m p o sitio n o f the D S400 Ind ex any com pany th a t d e riv es any revenue at
all fro m the m anufacture o f alcohol or tobacco p roducts, or from the
p ro v is io n o f products o r se rv ic e s related to g am b lin g , o r firm s that derive 2
p e r c e n t or m ore o f its re v e n u e from sales o f w e ap o n s. B efore 1993 there
w as a ls o an additional screen that elim inated c o m p a n ie s with interests in
S o u th A frica.
On th e o th e r extrem e we c o n stru c t a portfolio that in cludes unethical
assets se le c te d on the basis o f th e five most com m on screens, namely
com panies associated with a lc o h o l, tobacco, g am bling, w eapons, and
env iro n m en tal harm. Our u n eth ical portfolio is capitalization-w eighted,
rebalanced m onthly, and c o n sists o f all co m p a n ies included in
D istillery & V in tn ers, Brew ers, T o b acco , G am bling, D efen c e, Forestry,
M ining, an d O il& G as E xploration sectoral indexes o f th e US m arket, as
provided by D ataStream .
O u r p ro x y fo r the m arket p o rtfo lio com es from th e p e rfo rm a n c e o f the
Dow J o n e s W ilsh ire 5000 T o ta l M ark et Index, w h ich p ro v id e s broad
m arket re p re se n ta tio n o f all c a te g o rie s o f stocks. W e c a lc u la te the size
p rem iu m (S M B ) by su b tractin g th e retu rn on the W ils h ire T o p 750 Large
C o m p an y In d e x from the re tu rn o n the W ilshire S m all C a p 250 Index.
T he W ils h ire Sm all Cap 250 In d e x is a subset o f the W ils h ire Sm all Cap
750 In d e x . It is a m arket c a p ita liz a tio n -w e ig h te d in d e x o f 250 stocks
u sing p ro p rie ta ry sam pling a n d co n stru ctio n te c h n iq u e s to m inim ize
tu rn o v e r a n d liq u id ity pro b lem s w ith o u t altering the p e rfo rm a n c e pattern
o f sm all c a p stocks. A m ore d e ta ile d description o f all in d ex es used in
this
s tu d y
can
be
fo u n d
on
the
W ils h ir e ’s
w ebpage
w w w .w ilsh ire .c o m /in d e x e s. W e calc u la te the b o o k -to -m a rk e t value
p re m iu m (H M L ) by su b tractin g th e return on the W ils h ire All G row th
Index fro m th e return on th e W ils h ire All V alue In d e x . W e check for
c ro s s -c o rre la tio n betw een o u r S M B and HM L and o b s e rv e that they are
sta tistic a lly in d ep en d en t. A ll th e a b o v e sty le-in d ex es a re su b sets o f the
Dow J o n e s W ilsh ire 5000 C o m p o s ite Index, are c a p ita liz a tio n -w e ig h te d
and w ere ta k e n on a m onthly b a s is fro m D ataS tream . T h e risk free rate is
assu m ed to be the US 1 3 -w eek T reasu ry Bill rate , a s provided by
D a ta S tre a m .
O u r a n a ly s is com prises a 1 5 -y ear period from M ay 1990, w hen the DS
400 In d ex w as initiated, to th e e n d o f A pril 2005, w hen w e concluded this
study. T h is perio d is partitioned in to two sub-periods. T h e period from
M ay 1990 to July 1998 is o f ste a d y m arket g row th a n d low er price
v olatility (3 .9 % per m onth fo r m ark et, 4% for e th ic a l and 4.2% for
unethical a sse ts), while the se c o n d p eriod is c h a ra c terise d by higher price
v olatility (5 % per m onth fo r m a rk e t, 5.14% for e th ic a l and 5.68% for
unethical a sse ts).
Table I
D escrip tive Statistics
D escrip tiv e statistics for the m arket rate o f return (rm) proxied b y W ilshire 5000, riskfree
rate o f return (rf) proxied by the 9 0 day treasury bill rate, S M B ob tain ed from the d ifference in
returns b etw een W ilshire 7 5 0 and W ilsh ire 250, HML obtained from the difference in returns
b etw een W ilshire All Growth Index from the return on the W ilsh ire All Value Index, return
on eth ical assets (relhieai) proxied by D S 4 0 0 Index and return on unethical assets calculated by
form in g a value-w eighted from 4 se cto r indexes - D istillery & V in tn ers, Brewers, T ob acco,
G a m b lin g , D efen ce, Forestry, M in in g , and Oil&Gas E xploration. A ll returns are on m onthly
b a sis o v er the w hole sam ple and the tw o sub-sam ple periods: L o w V olatility period (M ay
1 9 9 0 - July 1998) and High V o la tility period (August 1998 - April 2 0 0 5 ).
S a m p le P e rio d : M ay 1990 - A pril 2005
I’m
rr
r,„-rr
SM B
HML
•"lithkul
ncthUnl !*f fUnetlik-ul rura.i1 ilt.il " Tf Fl.’nrtlikill-rfctliiciil
M ea n
0.0071
0.0034 0.0037
0.0011
0.0022
0.0104
0.0070
0.0111
0.0078
0.0000
M ed ia n
0.0128
0.0038 0.0109
0.0047
0.0014
0.0108
0.0082
0.01 12
0.0072
-0.0008
S td . Dev. 0.0429
0.0015 0.0429
0.0324
0.0328
0.0447
0.0447
0.0482
0.0481
0.0431
Low V o la tility Period: M ay 1990 - Ju ly 1998
rm
rr
r m-rf
SM B
HM L
ri:ihkal
rivihiiui - rr rutKiiiUui fUnrthk-ul *
H IM'Illk.il"rKihlcul
M ea n
0.0102
0.0040 0.0062 -0.0028 -0.0001
0.0149
0.0109
0.0100
0.0060
-0.0054
M ed ia n
0.0146
0.0042 0.0104 -0.0006
0.0009
0.0180
0.0139
0.01 12
0.0072
-0.0058
0.0010 0.0389
0.0175
0.0408
0.0408
0.(>4I6
0.0415
0.0292
S td . Dev. 0.0390
0.0260
H igh V o latility P erio d : A ugust 1998 - A p ril 2005
HML
rKthlfjil
0.0011
0.0026 -0.0015
0.0050
0.0046
0.0028
0.0003
0.0110
0.0085
0.0062
M ed ia n
0.0122
0.0022 0.0114
0.0083
0.0046
0.0028
0.0014
0.0083
0.0059
0.0044
S td . Dev. 0.0509
0.0015 0.0510
0.0395
0.0449
0.0514
0.0515
0.0568
0.0568
0.0551
r<
r m-rr
SM B
M ean
rm
rKihkui - r r FUnetlili-ul ruiwihu») -
S o u rce: self-com puted with E -v ie w s software
T ab le 1 presents detailed statistics. Over the 15 years, both the DS400 Index
and our unethical portfolio on average outperformed the market, delivering not
only higher absolute risk prem ium s, but also offering better Sharpe ratios. T he
unethical portfolio performed, on average, slightly better than DS400 and ethical
assets. Interestingly, the first sub-period shows that ethical assets outperformed
unethical assets by about 50% (1.5 times) on a monthly basis. The second period
show s that unethical portfolios outperformed the ethical portfolios by alm ost
400% (4 times). (However, statistically the difference between ethical and
unethical risk premiums in the analyzed period and the tw o sub-periods is not
significant at the 5% and 10% significance levels.)
3. DETAILED RESULTS AND DISCUSSION
Past research on ethical fu n d s and assets has been restricted to the
evaluation o f risk prem ium s u sin g various m odels. H ow ever, it is of
im portance also to observe tim e v ary in g risk prem ium fo r not only ethical
but also a sse ts that will be e x c lu d e d using basic screens. A ssets that reside
on the o th e r extrem e of the ethical scale due to the natu re o f their business
can p ro v id e an insight into th eir characteristics, as w ell as tim e varying
behaviour. A dditionally, d ifferen ces in com position and m arket behaviour
betw een the tw o types of assets are also revealed. As m en tio n ed previously,
this re se a rc h also focuses on tw o o th e r aspects - first, th e nature of inter­
tem poral relatio n sh ip betw een the tw o types o f assets, b o th long term and
short term in nature. Second, this research also ev aluates factors that affect
the return sp read between unethical and ethical assets.
W e first evaluate the types o f investm ents in term s o f F am a and French
3-factor m o d el, over a 15-year p e rio d , as well as over tw o sub-periods of low
and high m ark et volatility. T he re su lts are presented in T a b le 2 below:
T a b le 2
Fam a and French 3-Factor m od el for Ethical and U n eth ical assets
T h is ta b le sum m arizes the risk prem ia for ethical and unethical a ssets over the w hole
sam ple period - A ll (M ay 1990 - April 2 0 0 5 ), as well as for tw o su b -sam p le periods: Low
V olatility p eriod (M ay 1990 - July 1998) and High Volatility period (A u gu st 1998 - April
200 5 ). E thical portfolio returns (rclhita|) p roxied by D S 400 Index and unethical portfolio
returns are ca lcu la ted by forming a v a lu e-w eig h ted from 4 sector in d ex es - D istillery &
' intners. B rew ers, Tobacco, G am b lin g, D efen ce, Forestry, M in in g , and Oil&Gas
E xploration. T -sta tistics are provided in parentheses.
i,t "I'M — ®i
Pm
" l"f,l 1^" PsMIl l**Small Cap,t ~ ri,arge Cap.tl"^ PlIML l*"Valtic,t — ^Growth.tl"*’ ii.l
where:
r ,, is the m o n th ly returns on portfolio typ e i at tim e t; i can be cither eth ical (right hand side
panel) or n o n -eth ica l (left hand sid e pan el) a ssets,
r 11U is the m o n th ly returns on W ilshirc 5 0 0 0 at tim e t,
r (A is the m o n th ly riskfree rate o f return at tim e t,
r small,i is the m o n th ly returns for W ilsh ire S m all Cap 250 Index at.tim e t,
r Luryc.i is th e m o n th ly returns for W ilsh ire L arge Cap 750 Index at tim e t,
r Value,! is th e m onth ly returns for W ilsh ire A ll V alue Index at tim e t,
I Gruwllu is the m on th ly returns for W ilsh ire A ll Growth Index at tim e t,
Q., is the error term for the regression.
In d e p e n d en t
V a ria b le
In te rc e p t
r m-rr
Risk P rem iu m fo r E th ical Assets
All
Low V o latility H igh Volatility
0.004 “
0.004 11
0.00311
0.004
0.001
0.006
(5.02)
(5.028)
(2.490)
(1.327)
(0.214)
(1.339)
1.012"
0.767"
0.831 “
0.689"
(39.603) (11.382)
(11.279)
(7.028)
1.032“
(58.58)
SM B
HML
Adj R-squarcd
R isk P re m iu m for U nethical Assets
All
Low V olatility High Volatility
1.03911
(48.654)
-0.2194
-0 .1 9 0 “
-0.225 “
-0.101
-0.114
-0.073
(-9.90)
(-6.032)
(-7.140) (-1.2(H))
(-1.047)
(-0.605)
-0.0401
-0 .1 8 4 “
-0.022
0.677 “
-0.153
0.799 “
(-1.76)
(-4.033)
(-0.762)
(7.796)
(-0.974)
(7.284)
0.96
0.96
0.96
0.45
0.60
0.50
L ev el o f significance is sp ec ified as “ for 1% ,b for 5%, and c for 10%
S ource: self-com puted with the u se o f E -view s softw are
T h e results reveal d ifferences in characteristics betw een the two types o f
assets. F o r example, ethical assets have a beta c lo se to 1, which is not
surp risin g since 250 o f the 4 0 0 com panies in the D S 4 0 0 are also included in
S & P 500. Unethical assets have lower levels o f m arket risk suggesting that
m ost assets are from m ature and stable industries. In term s o f size bias, ethical
firm s are larger, though no such bias was found for unethical assets. In term s
o f value or growth bias, both types o f assets were very different. Ethical assets
w ere grow th orientated, as th e betas were negative and significant (except
d u rin g the high volatility period) while vice investm ents are statistically value
biased (except during the low volatility period). R eaders should note that the
A D F test for unit root on the residual for all regressions was rejected in every
case, and shows that the residuals are not white noise. H ence, we conclude that
all risk factors have not yet been accounted for.
A fter filtering out m arket and style effects fro m the returns prem ium s,
re sid u a ls from equations la an d lb are tested to o b serv e if any relationship
e x ists betw een ethical and unethical assets. F or instance, it is of interest if
th e re does exist any long run relationship and if one ty p e of investment leads
th e o ther. Since Fam a’s 3 -fa c to r risk model has not accounted for all the risk
in e ith e r indexes, we will p ro ceed with the next asp e ct o f this research in
e v a lu a tin g the relationship b etw een the types o f investm ents, as well as the
facto rs that account for the unethical-ethical spread. T o analyze the inter­
tem p o ral relationship b etw een the two assets, both sh o rt and long term , it is
im p o rtan t to ensure if any co integrating relationship that may exist betw een
th e tw o investm ents is taken into consideration. C ointegration tests on the
tw o index series, after filterin g out known sy stem atic risk factor and
in vestm en t styles, show that th ere d o es not exist any lon g -term relationship
betw een u n eth ical and ethical investm en ts during the sa m p le period or any
su b -sam p le periods (we tested fo r the existence o f unit ro o t using A D F test
and there w as none). The resu ltin g relationship is p ro vided in T able 3 below:
T ab le 3
V ector A u toregression (VAR) R esults
T h is ta b le provides results from V A R m odel that includes p rem iu m s from Ethical (rElhjta| )
and N o n -eth ica l (rUnclhiaii ) asset in d ex es as endogenous variables and market risk premium
(rm-rf ), H M L (rVaiUo - rGrowih ) and S M B (rSmaii cap Cup ’)• T -sta tistics are provided
underneath co effic ie n t estim ates. R esu lts are reported over w h o le sa m p le period: All (M ay
1990 - A pril 2 0 0 5 ), as well as for tw o su b -sam p le periods: L ow V o la tility period (M ay 1990 July 1 9 9 8 ) and High Volatility period (A u gu st 1998 - April 2 0 0 5 ). S ig n ifica n ce levels are
reported at 1, 5 and 10%.
Low
Period: Volatility
Period: All
D e p e n d e n t V ariable
ri.thicui.t-i
rKlhkuU-2
I'llnethli'ul.M
Funelhk;i!,t-2
In te rc e p t
r m- r f
HML
rKlhkuLt
rUnclhkoLI
rKthlo.Lt
High
P e rio d : V olatility
rilndhkuU
rKthlcuLt
n'nclMul.l
0.0332 e
-0.1518 h
0.0418
0.0730
0.0280
-0.2105 h
(1.7456)
(-2.0830)
(1.2555)
(0.6054)
( 1.0676)
(-2.1568)
-0.0296
0.0364
-0.0488
-0.0586
-0.0295
0.1004
(-1.6216)
(0.5202)
(-1.5867)
(-0.5262)
(-1.1450)
(1.0484)
-0.0190
0.0608
-0.0387
-0.0258
-0.0089
0.0586
(-1.1028)
(0.9220)
(-1.3334)
(-0.2458)
(-0.3758)
(0.6625)
0.0408 h
-0.0411
0.0674 h
0.0554
0.0272
-0.0788
(2.3571)
(-0.6198)
(2.2927)
(0.5205)
(1.1446)
(-0.8925)
0.0065"
0.0075 h
0.0077a
0.0037
0 .0 0 5 4 “
0.0089c
(8.6097)
(2.5678)
(7.9443)
(1.0619)
(4.2849)
(1.8746)
1.0255“
0 .7 8 0 7 “
1.0327“
0.8 542“
1.0045“
0.7117“
(56.9827)
(11.3189)
(46.3467)
(10.5791)
(37.8568)
(7.2127)
-0.0501 h
0 .7 0 4 6 “
-0.1841“
-0.1602
-0.0326
0.8523"
(-2.1373)
(7.8384)
(-4.0527)
(-0.9737)
(-1.0627)
(7.4810)
-0.2271“
-0.0828
-0.1995“
-0.1706
-0 .2 2 9 2 “
-0.0430
(-9.8764)
(-0.9398)
(-5.8266)
(-1.3754)
(-6.9701)
(-0.3513)
R -s q u a re d
0.96
0.47
0.97
0.60
0.96
0.53
A d j. R -s q u a re d
0.96
0.45
0.96
0.57
0.95
0.49
SM B
L ev el o f sig n ifica n ce is sp ecified as 11 for 1% ,b for 5%, and c for 10%
S ou rce: self-com pu ted with the u se o f E -v iew s software
T h e results in Table 3 sh o w that the m agnitude and signs of risk prem ia
fo r th e m arket, SMB and H M L are similar to those w hen each investm ent’s
risk prem iu m was determ ined (see Table 1) over each sam ple period. T able 3
a lso rep o rts how investm ent-specific excess returns are affected by their ow n
and th e ir counterpart’s lagged excess returns. W e d e fin e investm ent-specific
e x c e ss returns as returns fro m either ethical, o r vice investm ents, after
filte rin g out returns due to m arket and style factors. O u r results show that
n e ith e r investm ent types are affected by their own p ast returns and hence no
e v id e n c e o f persistence in retu rn s is revealed. As m o n th ly returns on indexes
are u se d in this research, it is not surprising th at investm ent-specific
au to co rrelatio n of returns is absent. However, there is evidence that excess
re tu rn s from one type o f investm ent affects th e other, though this
re la tio n sh ip is not stable o v er tim e. Results from T a b le 3 provide evidence o f
eth ic a l excess returns bein g positively affected by tw o period lagged
in fo rm atio n from the vice investm ents during the overall sample period as
w ell as periods o f low volatility. During periods o f high volatility, returns
fro m ethical investm ent are not affected by its o w n , or unethical, lagged
retu rn s. Interestingly, one p e rio d lagged ethical e x ce ss returns had a negative
in flu en ce on unethical in v estm en ts, during the co m p lete sam ple period and
also d u rin g periods of high volatility.
S in c e results o f the co m p le te sam ple may be an aggregated effect of w hat
really occurred during su b -p erio d s, it is necessary to confirm if the observed
relatio n sh ip s were sig n ifican t in nature. We co n d u cted block causality
te stin g to reach any c o n clu sio n with regards to the relationship betw een
e x c e ss returns from the tw o investm ent classes. R etu rn s from assets are due
to sy stem atic and n o n -system atic factors. After acco u n tin g for Fam a’s (1993,
1996) system atic and style factors, excess returns from the two indexes
sh o u ld be due to id iosyncratic sources. Since o u r analysis is related to
in v e s to rs ’ preference based indexes, firm specific risk is diversified across
firm s and industries. T hus investm ent-specific excess returns should be due
to in v e sto rs’ preference fo r th at particular type o f investm ent. This further
im p lies that if investors’ preference for e ith e r ethical or unethical
in v estm en ts does not ch ange, there would be no resu lta n t change in excess
re tu rn s for either asset typ es. Alternatively, if m ajo rity investors’ affinity
(d islik e ) for a certain type o f investm ent increases, its return w ould also
in c re a se (decrease) relative to its counterpart. T a b ic 4 provides results o f
sh o rt-te rm relationship below :
T ab le 4
B lock C au sality Test Results
T his table provides results from B lo ck Causality Tests using the V ector Autoregressive
model (T ab le 3). Results are reported o ver w h ole sample period: A ll (M a y 1990 - April 2005),
as w ell as for tw o sub-sample periods: L ow V olatility period (M ay 1990 - July 1998) and High
V olatility period (August 1998 - April 2 0 0 5 ). Significance levels are reported at 1, 5 and 10%.
D e p en d e n t variable: raim-ui.
Excluded
All
fUncihkul 6.3757“
Low Volatility High V olatility
7.2767“
1.3719
D e p en d e n t v ariable: runcuiicii. i
Excluded
All
rEUikai
4.4097
Low Volatility High V olatility
0.6332
5.3187*-'
is specified a s a for 1% ,b for 5%, a n d c for
10%
S ou rce: self-com pu ted with the use o f E -v ie w s software
G ra n g e r causality test results fro m Table 4 indicate th at excess returns
from e th ic a l investm ent are cau sed by excess retu rn s from unethical
in v estm en ts during our sam ple p e rio d o f 15 years. W e d o observe causality
in the o th e r direction as well but at 11% significance level, w hich we feel is
too high to be judged as c o n c re te evidence. H ow ever, once the sample
period is sp lit into low and high volatility periods, this unidirectional causal
relatio n sh ip is only observed d u rin g periods of low v o latility . D uring a high
volatility p erio d , the direction o f cau sality between the tw o investm ent types
is rev e rse d in that unethical in vestm ents are affected by ethical returns.
T hese c a u sa lity results provide som e evidence that so c ially responsible
in v estm en ts are in dem and d u rin g g ood times but during a period when vice
in v estm en ts provide higher retu rn s, ethical in v estors’ dem and changes.
D uring p e rio d s o f low volatility, ethical assets provide significantly higher
returns th a n both market and u n eth ical investm ents (1 .4 9 % against 1.02%
and 1% p e r m onth respectively). D u rin g this period, c au sa lity was observed
from th e seco n d lag o f unethical specific excess returns to ethical specific
excess re tu rn s and the relatio n sh ip was positive. D u rin g high volatility
period, eth ical investm ents retu rn s were more than tw ic e o f the market
(0.28% ag a in st 0.11% per m o n th ) but significantly u n derperform ed vice
in v estm en ts that returned 1.1% p e r month. D uring th is period, 1 period
lagged e th ic a l excess returns n eg ativ ely affected u n eth ica l excess returns.
W e take th is as evidence that d u rin g bad times in v estors’ d em and for ethical
assets d ro p p e d in favour o f vice investm ents over the next period.
A n a ly sis of the system s o f equations is fu rth er conducted by
d e c o m p o sin g the errors fo r ea c h investment type. T a b le 5 show s the results
o f V a ria n c e D ecom position (V D C ) below:
Table 5
Results o f Variance D eco m p o sitio n o f Non-ethical and Ethical Indexes
T h is table reports the results o f V ariance D ecom position o f Ethical and Unethical risk
filtered residuals using equation la and lb on a monthly basis for th e com p lete sam ple period.
All (M a y 1990 - April 2005), as w ell as for two sub-sam ple p eriod s: L ow Volatility period
(M ay 1 9 9 0 - July 1998) and H igh V o la tility period (August 1998 - April 2005).
Variance Decomposition of 4Kiiik«i.i
Low Volatility
All
M onth
1
2
3
4
5
10
Çllnlhiail. All
0.0000
0.5256
2.6866
2.7210
2.7276
2.7283
^Onvthk-ul, l.ow
^Klhliul. All
KM).(XXK)
0.0000
99.4744
1.9210
7.2160
97.3134
97.2790
7.2168
97.2724
7.2168
7.2169
97.2717
H igh Volatility
Çi:tiii<»i.i.uw
^sUnetlik'iil. Hlch
0.0000
100.0000
0.1089
98.0790
92.7840
1.0471
1.0611
92.7832
1.0697
92.7832
1.0704
92.7831
^Kllilml. Illuli
100.0000
99.8911
98.9529
98.9389
98.9303
98.9297
Variance Decomposition of 4iiwihi.»i
Low Volatility
All
M onth
1
2
3
4
5
10
ÇlJnlhlral. All
99.9942
99.8418
99.8390
99.8380
99.8380
99.8380
^Klhlcul. All
0.0058
0.1582
0.1610
0.1620
0.1620
0.1620
^Unelhh'ul. l.'tw
99.9834
99.9415
99.9145
99.9144
99.9144
99.9144
H igh Volatility
ÇkiIiIvhI.I.»w
^(liM'lhkul. Illch
0.0166
98.8232
• 98.4382
0.0585
0.0856
98.3586
98.3499
0.0856
0.0856
98.3495
0.0856
98.3493
Çuthunl, IIlull
1.1768
1.5618
1.6414
1.6501
1.6505
1.6507
S ou rce: self-com puted with the u se o f E -view s software
T h e results show d ifferen ces between the tw o investm ent types o ver
v ario u s periods. For ex am p le, ethical investm ents d eriv e 100% of their
variab ility o f returns from th e ir own innovations d u rin g the first period,
w h ile unethical investm ents d o n ’t. However, the a d ju stm en t process is m uch
faster d u rin g the low volatility period (4 periods fo r b o th investm ent types)
than d u rin g the high volatility period (6 periods fo r both investm ent types).
D uring a low volatility period, ethical investments derive much of their
inform ation from unethical assets (more than 7% after the 2nd period) while vice
investm ents derived less than 1% from socially responsible investments. High
volatility period VDC results show that only 1% o f ethical returns variance is
determ ined by unethical assets variations. The results are similar for unethical
investm ent returns, with about a 1.7% variation due to ethical returns.
F ig u re s 1 (whole sam ple period), 2 (Low v olatility period) and 3 (H igh
vo latility period) present re su lts from impulse resp o n se functions (IRF) o f
the V A R estim ates in T ab le 3.
R esponse of Ethical returns to One S.D. Innovations
Sample Period - All
-----EtIicaJ returns ------Unethical returns
R esponse of Unethical returns to One S.D. Innovations
Sample Period - All
-----Ethical returns ------Unethical retirns
F igure 1. R esu lts o f Impulse R esp o n se Function for the w h ole sa m p le period
Source: o w n analysis with the u se o f E -v ie w s software
The graphs a b o v e provide impulse-responses o f 1 standard deviation shock o f ethical and unethical
returns to ethical returns (upper graph) and unethical returns (lower graph). E x cess market returns,
SM B and H M L are exogenous factors. T he sam ple period is ALL (M ay 1990 - April 2005).
Response of Ethical returns to One S.D. Innovations
Sample period - Low Volatility
----- Ethical returns
------- Unethical returns
Response of Unethical returns to One S.D. Innovations
Sample period - Low Volatility
----- Ethical returns
-------Unethical returns
F igu re 2. Results o f Im pulse R esp o n se Function over L o w V o la tility period.
S ou rce: own analysis with the u se o f E -view s software
T h e graphs above provide im p u lse-resp on ses o f 1 standard d eviation shock o f ethical and
u n eth ical returns to ethical returns (upper graph) and unethical returns (low er graph). E x c ess
m arket returns, SM B and H M L are exogen ou s factors. T h e sa m p le period is Low V o la tility
period (M ay 1990 -J u ly 1998).
Response of Ethical returns to One S.D. Innovations
Sample period - High Volatility
—
Ethcal returns —
Unethical returns
Response of Unethical returns to One S.D. Innovations
Sample period - High Volatility
—
Ethcal returns —
Unethical returns
Figure 3. R esu lts o f Impulse R esp on se F un ction over High V o la tility period.
Source: o w n analysis with the u se o f E -v ie w s software
The graphs a b o v e provide im p u lse-resp on ses o f 1 standard d eviation sh o ck o f ethical and
unethical returns to ethical returns (upper graph) and unethical returns (lo w e r graph). Excess
market returns, S M B and HML are e x o g e n o u s factors. The sam ple p erio d is High V olatility
period (A u g u st 1998 - April 2005).
E a c h figure provides a grap h ical representation o f the effect of a one
sta n d a rd deviation shock to e a c h variable by itself and by the other variable
(th e returns were not orth o g o n alized . W e used one stan d ard deviation shock
to see the response on the d ep en d ent and independent variable.). The two
v a ria b le s under consideration are excess returns from ethical and unethical
in v estm en ts. As m entioned earlier, excess m arket retu rn s, SM B and H M L
are exog en o u s variables in th e V A R estim ation to ex tract excess returns
fro m each investm ent type. IR F during a low volatility period (Figure 2)
sh o w th at a one standard d ev iatio n shock to ethical returns has a negative
effe c t in the first period that ad ju sts to a positive effec t in the second period.
H o w ev er, during the sam e sam p le period, a one stan d ard deviation shock to
u n e th ic a l assets by ethical assets produces very little deviation. A high
v o latility period shows a slig h tly different effect. E th ical assets response to
an u n eth ical investment sh o ck is not as much p ro n o u n ce d , but the response
o f v ice investm ents to eth ical shock starts with a n eg ativ e effect over tw o
p e rio d s. This is consistent w ith our proposition that ethical investm ents are
n eg ativ ely affected during b ad tim es (high volatility period).
W e next analyse the factors that affect the return spread between unethical
and ethical assets. The tw o facto rs are one-month lagged excess market returns
(ab o v e the riskfree returns) and contem poraneous ethical premium, defined as
the returns of ethical investm ents above the riskfree rate. Equation 5 is used to
p ro v id e the results as follows:
Tabic 6
Factors that A ffect Non-ethical Return S preads
T h is table reports the regression results o f changes in u n eth ical return spreads (A lrUne,hic.,|
- rpjhicail) as the dependent variable and changes in lagged m arket risk premium (A [rm - r,J)
and ch a n g es in ethical premium (A IrEclhjca| - rm]) using E quation 5 on a monthly basis. T h e
sa m p le periods are All (M ay 1990 - April 2005), Low V o la tility period (M ay 1990 - July
19 9 8 ) and High V olatility period (A u gu st 1998 - April 2 005).
D e p en d e n t V ariable; Unethical S p re a d
P e rio d
Intercept
M arket Risk p rc m iu in (-l)
Ethical M ark et p re m iu m
A djusted R -sq u ared
Low V olatility H ig h Volatility
All
O.(KK)
0.000
-0.001
(-0.042)
(-0.098)
(0.076)
-0.171"
(-2.716)
-0.736“
(-3.288)
0.078
-0.101
(-1.112)
-0.084
(-0.258)
-0.008
-0.176h
(-1.940)
-1.074"
(-3.346)
0.147
L ev el o f significance is sp e c ifie d as “ for 1% ,h for 5%, a n d c for 10%
Source: self-com puted with the u se o f E -view s software
T h e re su lts show that lagged e x c e ss market returns and ethical prem ium
are not sig n ifican tly related to retu rn spread betw een u n eth ic a l and ethical
assets d u rin g good times (low v o latility period) but are significantly and
negatively related during bad tim es (high volatility period). T h is provides yet
additional ev id en ce o f investors c h an g in g their preferences d u rin g bad times.
O ur last concern with the results obtained may be that S R I fund m anagers
may hav e un d erg o n e a learning c u rv e since this is a rece n t developm ent in
term s o f an investm ent class. T h is implies that re tu rn s from ethical
investm en ts w ould im prove o v er tim e. W e em ploy eq u a tio n 6 and the results
are p rese n te d in Table 7 below:
T a b le 7
L earning ab ility over time
This table p resen ts results o f the fo llo w in g regression:
ri:nc(hicul.t" fEthical,! —Oj Punic time + (Pt
where:
ruiH.-ihiciii.i- rEihicjii.i >s
monthly p ortfolio return differences betw een ethical portfolios and
unethical p o rtfo lio s,
time - is the num ber o f months sin ce M ay 19 9 0 ,
(p is the error term for the regression.
In tcrccp t
t-stat
p-value
-0.009
-1.265
0.208
l-stai
p-value
0.689
1.582
0.115
P'rime
A dj R -s q u .a re d
0.008
S ource: self-com p u ted with the u se o f E -v ie w s software
O ur re su lts show that there is no evidence that re tu rn s from SRI have
im proved o v er time. A d d itio n ally , SRIs have underp erfo rm ed vice
investm en ts during recent periods.
CONCLUSION
O u r research asks if it is p o ssib le that investors c h o o se to invest on
ethical p rin cip les rather than retu rn s. Our em pirical ev id e n ce suggests,
ov erw h elm in g ly , this is not the c ase. O ur research show s th at during periods
of low m a rk e t risk, investors rem ain ethical, though d u rin g periods of high
risk in v esto rs are more con cern ed with regards to th eir w ealth . W e explore
th is hypothesis in a th ree-step process, by usin g returns from extrem e
investm ent-preference based indexes that are ad ju sted for known risk
facto rs. First, the two in v estm en t types are check ed for any cointegrating
rela tio n sh ip , and account fo r such a relationship if it d id exist. W e also check
fo r contem poraneous sim u ltan eo u s relationships b etw een the two classes o f
in v estm en ts. O ur results show no long run relationships but there is evidence
o f a sim ultaneous relationship betw een ethical and vice investm ents. T his
re la tio n sh ip changes from low m arket risk period to h igher risk period and
p ro v id e s som e evidence o f investors changed b eh av io u r with regards to their
e th ic a l preference. W e also use impulse resp o n se function (IRF) and
d eco m p o sitio n of variance (V D C) of residuals from sim ultaneous
re la tio n sh ip to provide fu rth er evidence of this c h an g e d behaviour.
T h is research then ex p lo res the factors that co n trib u te to this changed
in v e sto r preference. W e find that past market co n d itio n s, which perhaps
lead s to changes in their w ealth position, is an im p o rtan t factor to changes in
in v e sto rs m aintaining ethical investments. Y et an o th er aspect that is
e x p lo re d is the m anagem ent ab ility o f ethical inv estm en ts over time. W e do
not find any evidence o f im p ro v ed m anagement ab ility over time.
O u r research points to the fact that ethical in v estm en t may be a fad o ver
th e last decade. If this is true, perhaps further research in asset pricing should
in c lu d e this ethical prem ium to provide better fo rec astin g ability.
A cknowledgem ents
We would like to thank Meir Statman fo r providing us with his data o f DS400 Index.
Adam Szyszka acknowledges the support o f The Foundation fo r Polish Science.
REFERENCES
B auer R ., Koedijk K., Otten R., International Evidence on Ethical Mutual Fund Performance
and Investment Style, “C EPR D isc u ssio n Paper”, No. 3 4 5 2 , 2 0 0 2 .
B e llo Z ., Socially Responsible Investing and Portfolio Diversification, “Journal o f Financial
R esea rch ”, V ol. 28, 1, pp. 4 1 -5 7 , 2 0 0 5 .
D erw a ll J., Guenster N., Bauer R ., Koedijk K., The Eco-Efficiency Puzzle, “F inancial
A n a ly sts Journal”, V ol. 6 1, p p .51 -6 4 , 2005.
D iB a r to lo m eo D., Kurtz L., M an agin g Risk Exposures o f S o c ia lly Screened P ortfolios,
N o rth field Information S erv ices, w w w .n orth in fo.com . 1999.
D iltz J., The private cost o f socially responsible investing, “A p p lied Financial E con om ics”,
V o l. 5 , pp. 69-77, 1995.
F am a E ., Efficient capital markets: II, ‘T h e Journal o f F in an ce” , V o l. 4 6 , No. 5, 1 5 7 5 -1 6 1 7 ,
1 991.
Fama E ., M arket Efficiency, Long term returns and behavioural fin a n ce, “T he Journal o f
F in an cial E co n o m ics”. V ol. 4 9 , 2 8 3 -3 0 6 , 1998.
Fama E. and French K. R., Common risk factors in the returns on stocks and bonds, “The
Journal o f Financial E conom ics”, V o l. 3 3 , 3-56, 1993.
Fama E. and French K. R., Multifactor explanations o f asset pricing anom alies , “The Journal
o f F in a n c e ” , V ol. 5 1 ,5 5 - 8 4 ,1 9 9 6 .
G ranger C . W . J., Investigating Causal Relations by Econometric M odels and Cross-Spectral
M ethods, “E conom etrica”, 3 7 ,4 2 4 - 4 3 8 , 1969.
G eczy C ., Stam baugh R., Levin D ., Investing in Socially Responsible Mutual Funds,
“W harton
School
Finance
W ork in g
Paper”,
SSR N
E lectro n ic
Paper
Collection:
http://.ssrn.com /absiract=416380. 2 0 0 3 .
Guerard J., Is There a Cost to Being Socially Responsible in Investing, “Journal o f Investing” ,
V o l. 6 , 2 , p p .1 1 -1 9 , 1997.
S ., Jo H., Statman M ., Doing Well While Doing G ood? The Investment
Performance o f Socially Responsible Funds, “Financial A n alysts Journal”, V ol. 49, 6, pp.
H am ilton
6 2 -6 6 , 1993.
H einkel R ., Kraus A., Zechner J., The Effect o f Green Investment on Corporate Behavior,
“Journal o f Financial and Q uantitative A n alysis”, V ol. 36, 4 , pp. 4 3 1 - 4 4 9 , 2001.
Johansen S ., Estimation and Hypothesis Testing o f Cointegration Vectors in Gaussian Vector
Autoregressive Models, “E conom etrica”, 59, 1 5 5 1-1580, 1991.
The Impact o f Social-Responsibility Screens on Investment Performance:
Evidence from the Domini 400 Social Index and Domini Equity Fund, “R eview o f
Sauer D .A .,
F in an cial E con om ics”, V ol. 6, pp. 2 3 -3 5 , 1997.
S ch roed er M ., Socially Responsible Investments in Germany, Sw itzerland and the United
Slates, A n A nalysis o f Investm ent F unds and Indices, C entre for European Econom ic
R esea rch , D iscu ssio n Paper N o. 0 3 - 1 0 , 2003.
Sim s C ., Macroeconomics and Reality, “E conom etrica”, V ol 4 8 . 1980.
Sim s C ., Policy analysis with econometric models, “Brookings P apers on E conom ic A ctivity
1” , 1982.
Statm an M ., Socially Responsible M utual Funds, “Financial A n alysts Journal”, V ol. 56, 3, pp.
3 0 -3 9 , 2 0 0 0 .
Statm an M ., Socially Responsible Indexes: Composition and Performance. SSR N Electronic
Paper C o llectio n , http ://ssm .com /abstract= 705344. 2005.
T eoh S .W ., W elch I., Wazzan P., The Effect o f Socially Activist Investment Policy on the
Financial Markets: Evidence from the South African Boycott, “Journal o f B usiness”, V ol.
7 2 , l , p p . 3 5 -8 9 , 1999.
Received: July 2006, revised version: January 2007