On the closed-end funds discounts/premiums whithin the context of the investor sentiment theory

ON THE CLOSED-END FUNDS DISCOUNTSIPREMlUMS IN THE

CONTEXT OF THE INVESTOR SENTIMENT THEORY.

By

Ana Pmlla Can'al/ID do Mon/~ Alljoim Pmf. of Busi"ess Adnrini.ftmtiml £,fCO/t' S'lperior de Tecrr%gio c dt Ge.(ldo de 8mgollfo

CamfIlI.! d~ Soma Apofallia -Apar/lIl/a 1 J4 5301-8578mglllrru Pm111Kal Td: +351-273-3(/-3/OZ Ft~r.-+35/-273·11·1051 IlpmontC@IJlb.Jlt

on"

lffr1llrrl!! Jose d" RocfmArlNm{o Professor of Fina/~r

£fCOlt, do' ECtl/lOmia e Gr!sTt;tl - Unin:rsidm/e JtJ Mi"fw

Guuflar47/fI BRAGA

PORTUGAL TE1,.-+ 351-153-6"·4555 FAX:

+

351-151-18-17J9

rnrmllda@teg.uminlto.pt

ON THE CLOSEI)-EN(} FUNDS DlSCOUNTSfPREMIUMS IN TIlE

CONTEXT OFTIIE INVESTOn. SENTIMENT THEORY.

ABSTRACT

l11c c.~iS1el1ee of dosed-cod funds di~countsJ"remium5. n!though an i~~ue l~rgel}' ~tudied. it is sWI puuling both ncnd~mics os well 05 practitioners. As it is well known. ,hllY rcsult fmm the

difference between Ihe value of tin: slmes of the fund. dctemlined hy thll market, 11l1d their Ilet

nsset yalue (the mar]..et value nfthe securities held by the fund. less the liobilities). To"-in!). into pecolltU thot the dosed-end fund slmres ore trotled 011 the stock exehan&:e. os well as the asscts

included on their portfolios. no di~crep~"eie~ would be expected (~t lcust theuretically) belweeu the m:ITkct Y:llue of the funds IInd their net :Issei vnlues. since the nwrket should lie able 10 adjust ~nd eurTect the prices. due III the fad that the infornmlloo is widely dirru~cd.

111 allempt tu uplnin thi~ "pu::/c" ~evernl theories h:l\'e been ~uggested. On olle hUlld. those based on rational faclors. such as: pOlentiallll.,( liabililies due 10 lInrcalised capilal gains.

the dividend policy. the fund portfolio composition. agency costs and lIIanagem~1I perfonuance and. 011 th~ utllcr hand. tllOse based IlII bdmviollr.d faclOrs. such as .he ill\'CSlor scntimtnl

'heory. TIlis Intter framework, lit kll~llheore'icllll)' is. in our view, the one 'h~t ~ecm~ to hcttt!!" explain nlmost nil the fea'urcs of ,ht "pu::Ic:"·. trying oot only

'0

e);pl~in thc c);islence of discounts but 11150 the existence of premiums IInd their beh~viul1r ~mnn£ Ihe funds themselves IInd over time.

Inlhis COnle);l, we devdoped our research Irying lu e:o;:p!~il1 the e.~;Slcnce IInd persis,ence of

Ihe discounlsfprcmiums. Wc al~o iUYl!StiJ;olc Ihe correlation hclwcen the di~countslrremiums of

those funds IImuns themsclves nnd ench other over time. the menn rcvef5ion of the discounts/premiums. ns well liS the predictobility power !If Ihe fund slmres and of Ihe net nsset \'31111: rCIUnls. It was ~Iso our objective 10 search for the releYllncc of the investor scntiment theory in order 10 explain the discounts/premiums. 50 that wc used Bmner's (11)93) methodology IInd Ihe signlll exlrncliun lechniqlle of Frend. IInd Roll (1'J1I6). Wc nlso e~fTied 0111

(as for os we know. for the first lime) a panel dn'~ nnnlysis in order tn check hnlY Uluch (lfthe discounts/premiums Y~riability is dl1e to the presence of ""noise lI"IulcI".I"··.

TIlls resellrch W1l5 b~sed on 11 sllmple of Nonh-Amcricllll c1oseu·~nd fund~, which iovcsI m~inly on stocks and/or brlllds tmdcd on the !'lYSE or on the AMEX. during 'he period from Jonunry 1987 10 Junc 1999 {inclusivc}. TIle dDt~ was collected from Ihe Wic.!crr/oel"[!cl" d3talmse. From the resuhs that we got. we noticed that there 5ccms 10 exist on indicl1tion of 'he presence of--nnise trnders'" on the closed-cnd funds market whidl. in turn. secms In conliml 'he

~~5umptions of Ihe investor sentiment th~ory: tlte discoullt5l'premiunl~ wcre po~itively

curtelated. were menn r~ver1ed and hnd some predictnbility power in tcrms of fund shllft: returns

bUI nOI so much in rd~tion to their net nsset Y~ltle returns. Ncyertheless. we obscrved Ih~1 the estinmled pmponicn ofllle voriunce of slond~rdised weekly discounl~ dl1mse.~. explained by the

inycstor sentiment on Ihe total period stutli\.-d. was only 8.6~ •. Alsu. the rt:~lIlls frnm Ihe panel data analysis seem 10 sllggest the ,ehllively loIY imponance of Ihe investor sentimtn\ theory 10 e:o.plnin thuse discount5l'premiums.

Key-wurds: Discounts/Premiums: Clllsed·cnd Funds: hwestnr Sentiment 'nlcnry: Bch~vinur~1 Finauce; Pnn~1 Dnt:! AlIDlysi~; Unit Root tests.

ON THE CLOSED-END FUNDS DISCOUNTS/PREMIUMS IN THE

CONTEXT OF THE INVESTOR SENTIMENT THEORY.

I. INTH.ODUCfION.

As the price oflhe closed-cnd fund shares oud its net nsset value (NAV) ore delennined independently, they clln diverge from each other ~nd. as a consequence. wc can observe either diseoullls or premiums, being the first ones the most commun and persistent form. in r..,ct'nt years (v.g.: Dimson ond Minio-Kuserski: 1998; E1ton, Gmber and BU5se, 1998; Klibnnoff. Lamont and Wizman. 1998). Ne~ertheless. laking into accounlthat the closed-end fund shores arc traded on the stock c:o;changc, as well as the assets on their portfolios. no discrepancies would be expected (at lenst theorctically) between the market value of the funds and their net asset values. Besides the filct tltnt thc funds nre frequently 1raded at discount. they var)" from fund to fund. in the same period. nnd along Ihe time. TIle flotntion nfthe discounts follow. in geneml and closely. the markct cycles and the lounch of new funds happens in general during the phase in which the majority ofthc existing ones nre at premiwn or ut diseountl. TIlese new funds, usually. arc plnced in the market at premium but. surprisingly, this is going to be diluted. becoming at discount or at reduced premium (Lee. Shleifer & 111l1ler, 1990).

:\.l1otlter intriguing ospect is the behaviour of the discounts when an open·ending up<:r.ltion is announced. In gencfill. ill such situJtion, it can bc observed thnt the price of the fund tends to converge 10 its NAY, deeply reducing the discount. Aftcrwards, and until the

up~r.ltiun effectively takes place. the discount still red\lce~. ~pprODching zero, in must thl'

cascs1• We should point out. however, thM the conflicts between the administration and the stockholders can obstruct and inlpl'de nn open.ending opetntion (v.g.: Brnucr. I1J84; Oarclay, Huldcmess & Putttiff. 1993). In cnsc the manogers do not 11\\11 significant participations in the fund, they will tl'nd to resist to the operDtion because they can loose their jobs or certoin privileges (for e.~ample: pecllniary benefits). 111e l~rge stockholders, and the blockholders. cnn also resist, even if the uperlltion con bencfit them in tenns of excess returns, because they may prefer to maintain their privote benefits (v.g.: Barclny, Holdemess & Pontiff, 19~3).

On the other hand. several author.; (omong olhers, for example: Cheng. Copclond e O'Hnnlon. 199~; Pontiff, 1995; Arak e T~ylor. 1996a; Sias, 1997b,) detected e:o;cess returns in relution tu funds sold at high diseount/prcmium os a result of implemented certain strategies having concluded that the discounllprcmium is mean reverting. at le3~t in the short period. Givcn this apparent market inefficiency, it would be e:o;peeted that the rational investors would tl)' 10 take udvantage of this opportunity and would implenll'nt orbilr.lgc strategies. However. Puntiff rl995) cottdlldcd thatthuse investors did lint ~Ul:ceed to implement complctely effective 's ... f," •. "mpk:

W.",.

1989 olld Le" •• 1: ·lh<>m ... t.",

'''',u,rltng 10 lir;,l..1<) .~ S,h,llh,;m 114~~llh, lernpon1 ""I<m o(d", d«tin<: un Ih. ui"o""";" p,,·b.th du' 10 Ih •• <,h.m"" nf

,I"

It"".,,,;nl) .hoolll1,

.m";..,,",,, or,h, op<rolion.

i..~ Iflh' fund w~llo' will I'''') h" r<"m<,ut<u.

strategies. Consequently. the llctUlll discount/premium of the fund conmining informatiun un the future diseounllpremiurn lVould also contain infonnation which would alluw forecasting fund retuOls (Pontiff. 19~5; Chcung, f\.w:mg & L\<t:, 1997).

In attempt to ~"plain the "pume" sume theories lVere suggested. On onc haml, pt'rhaps the most common type of e.~planations, arc those tlmt arc based upun rational fllctors. such as, the potential tax lillbilities for the fund's unrealised capital gains. the dh'idcnd policy. the composition of the fund's portfolio. agency costs and management pcrfonnance. alT10ng others. On the other hand, arc those which ore bosed upon behaviournl factors like the i!lve~tor sentiment thenry.

The first set oftheuries tries to explain the c:o;ish.'llce llfthe discuu11ls. but uot :llway5 the premiumsl1r of the bdl:lViuur of these at the time oflln initi:ll public nffering (lPO) or of an open.ending operation. HOlVever.there are some farturs, so called rational. which 5eem to have ecunumic and statistical rclev~nee for c:o;plaining the nO!.5·s~ctional behaviour of the discounts/premiums, such as the characteristics of the fund's portfolio cumposition {the c:o;islence of restricted assets, i11iqllid or foreign}. the dividend policy, the unrealiscd cnpitlll appreciation. and agency costs.

TIle uther sct of theories. the im'estor s~l1Iime1ll11wory (for e.,mnplc). scems to consider allllust all the pieces of the "puzzle" trying, nut only, to e:o;plai1l1hc e:o;i~tcnce ofdiscount~ but nlso of the premiums as well as their behaviour among funds nnd along lime. Never1heless. it is 0[50 not ..,.,etopt uf criticism. the mnin ones being relmed with the fact of being IIblc to fully e:o;plain the vnrintion of the discountsJpremiums uf the clused·end funds, nn indicator of the investor's sentiment. nnd if this is (er not) II fllcter of systemlltic risk.

Within this context, we carried out UUT research trying tll explain the e.,istence and persistence of the discountslpr~rniums, having as starting point Dc Long, Shldfer. Summers & Woldmann Model (1990). Having Il dcep relotion with our main objective. we investigated the corrciation among the discuuntslpremiums and of thl!.~e for I'ach fund over time. thl' e.~istence (ur nut) ofmeau reversion for discounts/premiums as well liS of their weekly changes.nnd also their predictive power on the fund's ~hare returns os well as of their NAV returns. finally. in order to investigate about the rclev~nee of this theory for the explanatioll of thc discounts/premiums, wc npplied the BrIluer'~ (l993) methodology a~ well as the signal extraction technique of frcndt and Roll (l'J86). cUIl)'ing out nlsn a pnncl data analysis (to the best of our knowledge for the first time) for checking how much of the variabilh,· of the discounts/premiums is due to the presence of noise Ir~ders.

111is paper is structured in the fullowing way: first we e:o;pose the investor sentiment theory and its applicableness to the closed-cnd funds, evidencing the implications of this in the behaviour und explanation ufthe di~cm111tslprcmiums. After, we e:\hibitthe mcthodolugy IIs~d to deternliue the relevnnce of this theol)' in the explnniltion of the variability of the

diS«luntslpr~miums. as \Vdl as the severnl methods pnd Sllltislicallechniques such OIS the panel d313 analysis. On the rourth sec1ion, we describe the sllmJ>l~ IIml define the "lIrillbl~ used in Ihe study. On the finh sec1ion, we e.,rosc l111d an:llyse the resulls and, <11 ]:1Sl, wc present the eonclusions us well us some ~ui!!!estions are indicated for future researcll.

2. THE INVESTOR SENTIMENT THEORY AND ITS APLICABlLITY TO

TilE CLOSED-ENO FUNDS,

111e investor sentiment theory i~ b05cd on the notions of rntionol ond informed inv~5tor

vcrSU5 non r~tionDI amlle5S infimllell inve~tnr (Ihe "noise traders") liS well as the way tlmllhis

loiter type or investors affecls the DSsel prices. Dc Lung. Shleifcr. Summers IInd Wnldmon"

(I 99U) fnrmolised such a thcrrry.llllvillg presented a mood ofosset voluolion IIpon these type or investors which do notllcl in D rntion~1 woy, crenling oddilionol risk ror the IISsels they hold.

According to these lIuthlJfS. Ihere arc two Iypes of invcstnrs ill the market: the rntiunlll investors mid the noise lraders. The first 0I1CS. which Ihey dcsignaled by sophisticated

invcSlors', fom, their ellp«lations based 011 infonnation concerning Ihe illtrinsie valu~ of the

:Isseis. 11ley ronn rationol c."'[pt'ct:llioIlS. TIle secol1d ones. the noi~e traders. who do not havc :lCI:CSS 10 inside inronnation, frequently fonn biased e.'{peclalions about the IISSel prico,:s. Is o,:an e'·en be said thal Ihey aq in an iITlltionol way. :IS their expectations are not bosed on the os.sets'

fundam~ntal vllluc out nn Ihe misleilding infonnalion ohlained from pseudo-signals of the mllrket~. TIlc uptimistic or pc~~imi~lje noise trnders' opinion makes the a~sets resale price

hecnme unellpeqed. As 0 rcsult. the luset~ con he undervalued or o\'ervalued. This

unpredict~biJity is wnrsened hy the fnct Ih~tthc noise Irnders' opinion could change (or become

even more elltreme) during the p~riod ortlle ~trnteID' implementalion. cre~ting an Ildditionnl risk

to rntional inve~orsl -the "noise trntl~rs risk". which limit thdr perfonnnnc~ as ~rbilrllgellfS.

Consequently, asset prices can diverge significntllly fmm Iheir inlrinsic values albeit there is nu

fundamental risk (De Lons. Shleircr. Summers & Woldrnattn, 19'JO).

In this moller. Dc Lons, Shleifer, Summers & Wllldmann (1990) consider thM the

investors can invest in Iwo asset types: ossets without risk. that pay 1I rc~l fi:\ed dividend. ,..

which supply is perfectly elnstie nnd risky "~sets which also pay 0 real filled dh'idenll, r.like the

riskless a.~sets bul which supply is inelastie, i.e .• the supplied quantity is filled and nnnnaJised al OI1C unil. Ihat is why their price "anes along time. According wilh Ihe lIuthors, the varialion in

'th ...

"""l:Slrn

h .. e .«en fxll~.''''' lo.n ."",et) nf"lfer"""",," in<bdin!; Ibr

""

,

de

i~{,..,.,'i.,... ~"'''''' rnf""",,,cn. , .... " "1r!I ,b.if ,",\""e<wiom.,. UIlI.>W.d

.

,,,.

,

<

im"e<l .... (""a ,hell <'I"'tmi ... on"1 ) tu.ud~I'rn.11 &un! t<d .. ,cat ",,"~m. O""U.tu~.fI '" OCIln"",,< <1IIeultll1ll. aro/ irntl<>n:illJ b ... it .... tlut 100< !i",.J. Cllfy ""1IImol;",,' ,1 .. ,,1 d~ .. "I>' ,IIItlllfk ,"'ue IDc u.nl'. Shk;r ... S",n", • .., It.

\\.loIm:um. .~I'D 10&~ h. '~rI''':cn.

,

bc-i

,

np«tIl;1If\I ~'r IU.lnI ond. in "."'''\.1<00<1 rh."

"fII"''''''Of

P""~';l'" III ,.1.11", .. l~lho::.,,,,,pn=

'lhe", InHlI1II. <""On •• I<ff

'fl''''''''

>r"'I~~! IIrl!'~;CS 10 ..

,,,b.

rn. ""~.

1IIl1r.rit,ngl:u! du. ro rh.n""" rr.od,,,· ml '/Icy

Jr.

"0I3~1< 10 driv<..., .. Fri«"0 tho::" rund",,,,",,' ... I ... ,'«.WO. I" p".tIl.lI .. ,· .,. ruk .n,., ..

the risky ilsset prices is ffiilinly due tll the misevalulItinll mode by nnise traders. So, Ihe risky :lSS<ts priox is gi\'~I, in equilihrium. by 1111: equalion:

r.",

1 + JI(p, -p) + p

p _

{2Y},u:

u,:

1+,. ,.

r

(l+rY

\\-1,ere:

1'

,,,

The rrice of Ihe risk)' n.'~m. ill reriod r: r .. Fi:\cd re~1 dividend:

r

"

Ai"lsulutr ~""mdent ofris\. oy~rSi(lH: Ir '" ref~cnt~!!C of"np/se "'udr!"s" ill lire 1111l1kct:

(I,

1\ '" nun informed invc~ll1r -··n"i .. e trader ... · senriment. ,hal i~. rhdr ol'rimi5m or pc!Isimism Idnlilcly

to Ihe 1I.lsel

rTice

.

~.

\Ilrere: p, -

N

(P.rJ";

,

)

This equation represenls Ihe ~qllililJrilll11 risky ~ssets price where the prke depends on e~ogenous pDrametets of the model (the technologicDI parameters. r. mid the bchnviolltal pammelcrs, r) plus Ihe public inrllfTllMion regarding the m:tual and future senlimclltlJr,he nnisc traders, Le., the mispcrccl"lIion (scnlimenll nooul the~e a~scl prices by the noi~c tr:rder~. I'. al moment ,.

TIlis model C:lII be applied 10 Ihe closed-eud fUllds· pricillg and il may c,'(p[aill Ihe

e~istellec, "ariam:c lml persiSlence of Iheir diseounL>;/prcmiums. if \I e con.~itler lit:lt Ihe clo~ed.

cnd rund's Nc\V is equh'alcnt 10 Ihe riskleM IIssel of the model. since its fundamental ,·alue is

easily cDlculoted~, :I/Id Ihe fund shanos IIrc cllui\,aICf,ttu:l risky :I~sct. In Ihis m~nm:r. Ihe nOlion

of noise traders CIIO explain the ··closed·end runds' pllzzle" ilS the rund slmres are sllhjeet 10 the ·'nuise troders sentiment·· whidl is systematic and correlated ~mong Ihe fUlllls' (De Lun.\:,

Shleifer, SWllIners k Woldmann. 1990).

When inveslors OfC uplimistie OhOlIl closet/-cnd funds. they force the fund pfiees to 11 superior levcl relative to its rundament~l value. therefore the discounts deneasc or ho:come premiums. When the investors orc pessimistic. the oppositc is ohserved. discounts incrense. Then. im'eSlors ure subjeclto two typl!.~ of risk: the risk from holding fund's portfolio (similnr to

the rundamental risk) and the risk of the res~le price ~eql1ivalenl to the "noise trader risk"). The

resale price risk epIl1e5 froln lire uncertainty in relution 10 Ihe im'e~tor sentiment AI th~ mOr11el1l

that they need to sell the fund. L c .• at tlmt mOlllcntthe discount might widen. If the scntimcnt is

systemAtic. thnl is. if it offects all funds nnd other ~sseIS. Ihen Ihe aSSl1ciated risk should he reworded. Due to this. closed-cnd runds have to sdI. nn average, ot discount to re\l'ord Ihe

:lssocialeu noise Imucts' risk (tee. Shleirer & Th~ler. 1990: Shleifer &. Summers. 19901.

Hence, closed· cnd funds scll. on nvernce. tit discount because discounts nuclunte and iO\,C'Stors

require an additional relum for bearing the risk of l1ucttl:lting discounts (Shlcirer & Slimmers:

199U). Lee, Shlcifcr and TImler (1991) point oul that disoonnlsfpremiums cbanges renee! nnt Ihe

-1..1vlnl.rp1 [ull<h!'llT I dj,idmd <1jtJ"aI./II ro d .. sum B[I!,: d"i:klld.l rooJ by lit< ",Id •• In III """r""~;I ... -nc: 11,. mIll .. pnc. nflll. funJ .!w><>1d I~ .~U\v.t .. ~ ID~" ... "nfl\O 1"",IbI,o ''''''~''

1'''''

"

.§

E

"

'

-I '

'~ ..,,7.

"

,

..

-<1

d,

_

t

•

niScOlml"'f~mi"m ofthc period.

The expression (1) is the opemtional form of the De Long. Shleifer. Summers aod

Waldmann model (1990) prediction aboot the \'Ilrian~e of discounts changes o\'er lime.

According 10 Bmoer (1993) the :tssessmelll of this appruach requires the estimnle of two

statistics

E{F,.)

Dnd

1'0/'

(

;;,),

for 11 closed-end fund. Assuming thllt the fund's NAV is random

wnlk. Le .. it is idemical allll independently distributed (Li.d.). then

E(;:;;)

cnn ~ estimnted hy

the mean weekly mte of return on the fund's portfolio. which is represented byi't:. 111e estinmte

of I

"or(;;,).

the \'anllnce of the mte of tetum oflhe fund's slocks price due to 1I0ise Ifoding. call

he achic\'ed. Ileeording to the Duthor.lhtouch the signnl extraction technique used by French nnd

Roll (1986).

According to French and Roll (1986), [onl: !cnlt returns arc less affected by the noise

lraders sentiment Ihan short ltnn returns beeau$c the miscvaloalioll induced by the noise Imllers

octivity leuds 10 re\'ert (it is mt."dn reverting) suom:r or later. So, if daily returns (Ire independent.

the returns for longer holding periods should be equal to the aCQlmulated daily returns within 1I111t period. DUI if daily retur1\S lire cOlTelDted to ~elt othn IIlld across time" due to noi!IC

tTllllers' :ldiv;ly. subsequently the longer hoilling periods' variance will be slImller than the

\'arinncc of Qlmulntcd daily tctumslfrench & Roll: 19&6).

TIle Freoch lInd Roll ~ignnl e)(1rnetion teehniquel4 Ihnt intend 10. specifically. identify Ihe

frnction of Ihe dnily return vnrinnee enused by the infonnntion. which they represent lIS V ••

consists in severn! steps. First, cOn1[lllle the nvernge dpily mle of re III m for each sUbpcriod of

two years of the sample (Ihe lIuthnrs use 11 Intnl period of Iwenty years). The seconll step

compriscs the sum of the squared devi(ltiun~ around thnt overuge. Afler, under Ihe nS5umption

thnt the daily returns dre serially indepcmlcnt, estimale the implied six-month voriance hy

dividing the sum of the squnttd uevjlltions by four. since each twu year subperiod contains fllur

semester.;. Firml1y, divide the nctu111 varillnct." of six-lIIollth return for the subperiod by the

implied varinnce. In orller to men~ure how much Ihe unify return is affected by noisl', the (I~sumptioll of the s~nal illllependcllce will be viol(ltcd by the pn:.<cuce of negative

aUlowrreiation anll Ihe observed ~ix-mnlllh vnndnrc will be smaller Ihan the cumulated dail)·

variance fBrauer. 1993). Fr~nch IInd Roll (1986) u~lennine the lowcr and upper bound of the

rclolive vDriance error. 'I1le upper bound" is given hy the differcn«: bctween one Dnd the rulin

of six.·month Vl1fillflcel• (Ihe ratio between the variance oftht daily infomtJllion eomponmt nod

I' 1f'''<I~''

...

"ir,,~ ."'o."n~b.I~~ .""'nccbil)'.r:o ... lhmn 11 nol. ~h"''''''ly. md'l"' .... '"

"F'." .. ' ... RDII tI~1I61 .... Irsnl tIq .. p<lf,i~" Ib ....

m

In .. pt...Q Ill< ... ''''''''''~, Ill< doily •• 1. "r .. """ ortbo: U,cU. d", .. ; II:>fm>I ""'''''''' b""" >Bd ooQ nf Ihill'ltn!~t •. Ill. Ih«'n" gr inf""""'IOII. ,h. IItrnv uf """0 tndrnr..,d lIIe tbtM) ftr "'d<2'~

;r:~::U!~~h~I~:~:= :;:.~~~l7;~'::!',~ ~:tltr~·~n;~.:::;"'h·

."et

I'm., ",10111111'

IF,

o

n<b ,.

Rod. l?!~) "~.pt<os""" v~ .. ..., Ju"o .hud,·mrnll"""d.

'

"

Ihe ((11~1 daily return variance). nle lower bound" is given b} a third of the difference between onc and lhe ralio of Ihe six-momh \'3riance plus Iwo thirds of the first onler 3ulocorrcl:llion of

Ihe daily return.

Whim, Brnucr (1993). applied this methodology 10 vnri~ncc: 0(1/)1.' (Jj~coul1ly'prC:nlium5

change of the closed-eud funds using "crkfy ralcs of return of the fuuds and considering the

ratio of the ne/lInt·implied six·month variance as reprc~enlilli\ c of the fractiOIl uf lhe weekly

return variance tine to the incorporation n[Ihe infomlalion. having represented it as ": . ,'nking

intn u(cuunt Ilm!. the Dc LC>II&. Shlcifer. Summers nnd Wnhlmmm model (I'J<)fl) clninu

Ihals. '" fr,

+

U,. when! ~"is the weekly return orthe fhod shares prices. ns n result. nccnrding

to Bmuer (1993). ((I_I";

)rilr.k

\

)

I

represenls on estimate of "m'(ir,) for the weekly return. Thu.~.

considering equation (2). the estiJlwted v~rilluce oflhe discounts/premiums chnn!:c is !:i\'l,~n by:

o( -)_

('-

I';!

,

-",

(

,J

1'111' LW, ". {

I

+;

1:

f

(.1)

3.2 Addltlon:!1 delinc:!tion nrlhc rcsearch.

Bellring our objt'1:tivc in mind. we willlry to c;tplaln the e;'I;islence Dnd persislence of the

discounts/premiums in Ihe eOnlc:'t1 of Ihe inveslor scnliOlcnl Ihcory. leSling some of ils

implicntions in Ihe behaviour of c1osed-t-nd fillld lliscountslprcmiums (n~mcly: the correlation among Ihc discnuntslpremiums ncruss funds and alnng limc. the discountsJpremiums nnd ils

weekly chanYII mean reversion. :IS well a.~ ils predicti\'e power on the fund shnrc price return

and its NAV). It is also our gOIl! tll investignte uhout Ihe rele\'ance of the inVc.~lIlr s~ntin\l:n1

theory for Ihe c.~[llanatjon of Ihll discounts/premiums variohi!ity.

Thus. we desisn our rcsc~reh in two phases. In the lirst phuse. \\e stnrt hy determining the

correll1tion of the Weekly discounts/premiums levels ~Cru5S the funds in the smll]1ie. the

currelntion ur the weekly discounts/premiums changes. npplying the P .. arson currelatiun

cocfficiL,u, attempting to lest Whelher the discounts/premiums nre positi\'ely currdoll:ll. Ne.~t. we anlllyse the behnviour of Ihe lim~ ~crics of the uiscounts/prcmiu11Is levels Alld ils weekly changl:S 10 cunlinn ifthl!S~ nre 5lnliollnry. For this purpose. we will enrry QU1the t1nssic tests of unit ranI, i.e .• Ihe stDtinnarity tests: AUl!n1ented Dickey-Fuller Test and I'hillip~-rerrnn Test on th~ varinhles. TIle nllmber of logs is determined by Akoikc Information Criterion lAIC),

Finally. in IIrtler In Icsl Ihe prellielive pm,,"r .. f Ih ... disct)ulllsipremiunls" we ennskler the

reblionship belween Ihe Ctlmlllalell shore price return oflhe filnds and disco1lnts/premiums. ItS

well as Ihe cllmulalell NA V return IIf Ihe funds ami its disflluntsfpremiums. Ihrough Ihe

following regressions:

" "''''''''"£'/ul~'' ,Lt,tt«''''" "",,,,,,,,"nu'l< ",;,UI rmkp<oo<n1 ,f .. ",h h Rd1. tQ.l<\,

Where;

.

,

I,nFNlh_

'

",

o

f

,

+ p,~DlS(".., +t;,~

'

_,

.

1

LRJJ'Iu_1 =on;} t p,~DlSC" +t;,~

..,

{;Rf

·

FV

u•t '" TI,c curnul~tellslmlc price IctUn1S oftllC funll j;

.

BRrPL'J" .. TIle cUllmiMcd NAV returns of the fund I; DISC,., :; n~ lIi~eoum/prernium oflhc funll i. in the period I:

a; c fI." 11le funll·specilic inlcrccpl and slop~ cucmdents. r"$p.:dil·ely.

11

(5)

(6)

~. '" The cutllulath'e fctum horizon of in\'c,tmcnl (as suggested by HarllOUI·elis. La Porta and WizmalL 1993. hmi1O!1S of onc. four arnJ thirtecn weeks are used).

To correct the heteroskedasticity problem wc use the melhod of White (1980) for one week investment horizon \If return. !lnd the method of Newey·West (l'}87). fllr investment horizons or four IInd thinecn WeekS

'9.

Regording the underlying IIdvantages of using p~ncl dolo analysis. in pnniculnr to this type of surveys (despite ils limitaliults!o). which eharacteri~e the pheuomenon considering its individual and time·series charactcristics. w~ test as well the predictive power of Ihe di~count!ilpremiums on the fund returns and NAV relurns using this technique, as fllr :IS wc know for the first time in this type of rescarch. Wc achievc the following procedures: test the

homog~neity uf the coefficients _ to test whcthcr the parametcrs of the model arc homogeneous

(ocros5 individuals and t!trough the time), covarianee anlllysis nnd model ofcrror component is used· to 11.'51 Ihe heterllgencity of the interception and homogeneity of the slope. If iI is not rejected. it is still ~pplicd the enndition:ll test of till! interccption homogeneity liS the slope i.~ homogeneOllS. i.e .. the Hansmanl

! test is applied lu determine if we shuuld use the Ii.~ed effeCL~

mlldcl or the mndom effects nlllde!.

Aner having tested these implicatiuns of the invc.~tor sentiment theory in the b~h:lviuur or closed-end funds discounts/premiums (and of its changes). wc pass to the second phase of the research where wc investigote on the relevance of [his theory in the explanation of the discounts/premiums vllriahilily. Consequenlly wc npply the melhmlology ofBrauer (1'}'}3) and the sigrml c.~traction tedmique of FrclH:h nnd Roll (1986). itS rollows:

1st step _ Calculate the standardised we~kly discounts/premiums changes. eonsideriog the e:\pressino (3), oud compute its descriptive stntistics lI5 mean. standard d~\·intinn. skewness,

"

1\1

.

,

.

mo:ll~ldl c.n he • .nom'l;",,"}" .rpll..t"~ Ih. unil .. tiOlt "ttoS.!lon I< coru;d,,,rl h)'lh, t ... .I1 "tu". m:1hod. ""'"~ Ih. J101i'~<:tl pr""::< En<1I:l' The l!Il1bod of 1,I,11i,. I t9Wl I"."'I'P"''''' Ih:l1lh. ,.!i~\I:ili of lilt ffiimmd e<[1"toa~ ore mi.l!)

"ncou.lolcd. "hn. n." ••.

w."

m"h,d I t9~7) ptc~<S a oo,."nonc. ltI01ti, .. 1im,lot IlLlI i. c""",lom "nit Ih. Ft ... I1C"C "fOOth

t"I,""l.d:utlCll~ .lIJ '"lllttl",t,"",,~f""l:Il",,"l1 fLl/m. Fer tn'·"1mrnloon:.o!U of"". p",;OO.ihOl. mrthod • .,o ,iknl1<at.

,.. Alootll Ihi, "~~«I

"<'.

for ."O'l't'.Jm~. (1'1'I7).lht"!ll1 t'l'l$).ll>lIo (t~Ml ,nJ ll"""",n .n~1'Jtor I tQ~t I

"So<. ~ ~ l1>u",wo. IQ7~:H""'m.0 .todTa,lor. tq~t: 11011)'. tq~2; Il,i.". tq~6. A .. II,,",. (Qq.l.rn\ flIt,,&,. 1"'11

"

kurtosis. ~uloc(lrrd~tiun with l~gs of Olle, two, three nut! fmlr weeks ~mllh~ ~UInLlnnJ "rmr of the 1st order :lutocorrc\ation.

2nd step _ Delcnnine the menn weekly fund price rclunlS for c~ch two )"caTS suhpcriod of111l: twelve yc~r5 sample tolnl period {which rcprcsc!\l si:-; ~lIbpcriod5 in tin: lotal) and the ~um orlhc square dcvjatio/l!i around each mean. in each ~lIhpcriml and for ench fund .

3rd step - Estimate the implied si~-monlh variance by dividing the sum of the ~C]uare devintions by fou~ .

'Ith step _ Compute the aclua) - implied six-month "arillllce ratio through the division o[lh" Delllnl (observed) six-month \'nriHncc by the implied six-month variance, liS cnlculmed in (he previous step.

5th step - Cnlculnte (he "grea( mean" by averaging (he actual-implied variance rn(iu among

011 the funds. in each period, and divide this by the totnl number of 5uhperinds tavernge fur period of the impli~d variance avernge ratio for fundl.

6th step -Compute the estimated standardised weekly discounts/premiums challg~s variance, regarding expression (.1), lIS

V:

represents the "grem menn". calculnted in the previous step; if.

is the weekly mean fund NAV rC\IIrn in the subperiod or period Dnd V<lr(~ •. I is the \'ariallc~ of the fund slmr.:..s price relum (trude·lo-trade) in the subperiod or period.

7th 5tep· Calculule the ~stimuted propot1ion of the stomlnrdised wcekly discout1!s/premiums changes variance explnincd hy the pr~scnce of noisc trad~rs (and. consequently. for th~ investor sentiment theory), dividing the estimated variance (calculated in the pre\"iou~ ~tep) by the observed standardi~ed weekly discounls/premiums changes variance in the sample subperiod or pniod.

To test the rohuslness of the results. wc analyse the relationship h~twcclI the ohserved and cstimMed vnrilmce of standardised weekly discounts/premiums changes as wcll :IS the relationship betwcen the observed variance of the standardised weekly discounts/premiums dt~nges and the spccilic-v~lue of (I . V~' ) for cadl rund". '1111' generic linear regression models arc. respectively;

Wher~:

)

'

<11

'

(6.0,1..

==

r

o;

.

,

+

PII

.

,Vw~{

,

~D,

l'

J

+

c

,

.

,

far(w, L

=

~'o,

..

+

~'",

(I

~

r

~

'

L

+

c,

.

,

I,/J

'

{"'O,

L

..

AduJI "~li~l\cC ufth~ ~tund~tdi~cd weekly discuuolS1prcmiums dWllg~.,: rm:(AD,l'J :::; EMinmtcd "ariancc of (ho ~Ianllardi~cd wcekly di.lconoLI'prcmiums challr.c~:

(7)

IS)

{I-I':)" G I'rnpUllion of the vJrianc~ of the shares plice return of Ihc fund nnriltuted Ut Ih~ nni~~

twdinJ!. s~cific In c~ch fund;

" n"

to Ihe numbtr nf

"mmm

In ~" 1""0 'C'" ,,~,..noo

:',h,.

,·.1", t<1"c>enl!

'he

plc!","

"'

.

"flhe {" .. !

,

10

"

..

,.'urn

,

·

.n""". du.

ID Ih' !>I"l< rmd)nc 11< n " ....

"

,,",,,ed

~, ,10, ""',,,

Il

Pili., : PH.' I: rpOu : PH" ;:: Lincar r~J;lessiul1 cucmdcnL~: &'.1

=

RD.l1dnm error.

We use panel dillB analysis to improve Ihe resulting infonnalion of these line~r regressions (3S

best of our knowledge for .he first time for this errect).

4. DESCRIPTION OFTHE SAMPLE AND DEFINITION OF TilE VARIABLES.

n,e data wns obtnined from Wiesenberger closed-end funds database. which wc selected n samplc of 41 closed-cnd invcstment fuods tmded in the NclY York Stock Exchange (NYSE) nml/or ill the Americnn Stock Exchnnge (MIEX). thnt invcst mainly in North Amcricon. specialiscd nr diversified stocks nnd/or bonds - excluding the ·country funds~ nnd the

-municipal blinds ftlllds". tluring the period of Jllnun!)' 2. 198110 Junc 18. 1999 !inclusive). with a minimum of 10·1 ohservations. The funtl to be included should have 11 repor1 of. at least. two years of publication either of ilS quotations (price oflhe fundI or of its N;\ V. Nunetheless.

it is not a primordial fe~turc for Ihis rcsearch. ~s we h~vc no intention of evaluating the

perfurmDnce of these limds. the selected s~mple dotS nut suffer of the survivorship bias problem because the smllple hns funds that stopped existing during thc sample pcriod ontl others that

~t3ned in the meantime.

On lablc I. in the nprendix. wc lisl the fuuds eOlllnim:d in the sample. With reference to the empirical research done by other authnrs. namely Brouer (191101) oml I'eavy (11)1)0), wc mode two adj~lm('nl5 in the lime-series of the sample. Uenee. the firsl 24 obsen·utions were not eoosidered (elluh'alent to silt montlu) after the dule of 11'0 or the fund bC'Cau5e, as Peuvy 11990) tlaims. discounts tend to increase in the weeks after the fund IPO which could cause a hillS in the results. On tht: other hond, wc did not include those observations nf the sill. months previous

to the open-ending operntion date because discounls lend 10 decrease during the period pn:violls

to the 0pclI-endins operation announcement. liS supported hy Brouer (19114) research:". The number of observotion$ 01. tohle I. in the IIppeuuix. i~ rc1l1th'e to the effeclive !lumber or observations. previously corrected by Ihese adjullmcnt5.

The variahles in study nre defined likl' this:

RF.VIJ, -111l' price return oflhe limd shares. ill continuous capitalisation tlmt is calculoted ru.:

RF.VD, "lu[l: +

lI,

)

-

In[I:

.

•

J

Where:

1',::0 Fund Price. at the cml nrthe \\"cck,. (clnsinR price):

D,::::: Tntal dividctlll di.<trillllted by tht: rund ut t"~ cllIl of ,hc wc~k r (-distribution of im;()m~ Utl~

c"l'itnll!~illsJ:

RVPL,- NAY rcttlnts of the funt.l. in continuous capitalisation. specified as:

Rt"f1., :: In[t·, ~ {),J-tn[I~_1

J

''It'''L .. ' .... r ... tOr! ... Dhoa> OIie ... , .. hu •• :I)UIfm ...

"

Wh~re:

r',_ Net n[....e\ I'alue. althe end oflhe Ileek I;

n

,

"

Total <lil-jdcnd di.stribul~d hy lh" CS~t5 Ihut cumpus~s the fwnl punrb!iu. ut the cnd urrlic "~.,,k I.

TIle cumulated Rlum, for fcur and thil1cCII weeks. either for fund return or its net ':lSsel \-i1luc. is c:dculalcd by tlh~ sum ohhe weekly rcturu IIIlhe I'ropn~d (umulated invCSlmt'lllllOrizoll.

DISC, •

Discount/

premium

.

in

percenln

ge

:

_DISC

_,

_'"

_{':'~I: _lOO}

tJ.D1SC'J" • Discollnt/premillm change. which ;5 given as: MJISC,~ =DJSC;, -DlSr;,.,

Although we do nut plan to lIuolyse the results in this sectinn. "e think Ihot is impnrtant to emph~size (sce table 2. in the appendix) thal dis~out1ls/pf<:miums VDry o\'cr this pniod. The higher meDII discOlJt1I was at 1'1118 (_8.45%) nntl the higher premium il' 11)Q2 (fl.44~~).

5. TUE RESULTS AND THEIR ANALYSIS.

5.1 Tbe discounts/premiums correlation.

Onc or the assumptiolls or the investllr sentiment tlll.'O~·. applied 111 the closell-entf funds

discounts/premiums. claims that discuunlslrrtllliul1ls arc positivcl)' correlated among th~m (Lec. Shldfer &. Thaler. 11)1)0). nlCrcCme. wc le511he nul! hypothesis th~1 the closcd-cnd funds discounls/premiun1s arc not correlnted.using the Penrson coefficient ofdctenninntiou.

We have confinncd Ihnt mOle thou hnlf!7(%J orthc cnrrclntinTl coeffidents nre positi\"e (sce talde J. in the appendi:"!). and about 88% of these are statistically si!;nilicanl . . 4.5 a cllIIsequencl!_ this is suggestive thut tlte dnsed-cnd fuolls t.list:Ountslpremiums lire enm:lated to each other Dnd they tend 10 move 10l:ethl:l". 111t: samc result is found in rehllion tn the weckly

discounts/premiums change:".

These result:; are consistent with tht results IIf Lee, Shleircr IInd Thnlcr t 191) I) as well as

Cheuns. Kwan and Lee (11)91). Siuce these results seem to confirn1 onc of the implications of

the investor sentiment theOT)' which predicts tlmt clused-end fuuds discUlI11lsll'reU1illlllS (om.l discounts/premiums changc) arc driven by the inve5tnr sentiment. ~nd so they arc likely to vary together.

5.2 The slationnrily orlhe discnunls/I,remiums.

The Slotionnrity an~lysis becnmes imponllnt because if the discounts/premiums ren':ct the

investnr sentiment. since. couceming the inve.~tor sentiment Iheory. Ihe selllimcnt is mean revertiug. theu the discounts/premiums shoult.l Ile menn rc\"crtiug ns well. '1,c hypothesis of unit root of the discounts/premiums Icvd "5 well ns wcekly t.liscounts/prcmiul1ls chnngl' is tested

"

existence and f1e~iSlcnce of the discounts/premiums, assuming th31 il explains a part of the

discountslprcmiumx " .. riante.

For this study. wc the 1I1ll11ysetJ period from 2101/1987 ID ]111211998 and we only

included [unds in which the difference between Ihe number of observatiollS of the lime series of the fund return and 1111: ones of the net 35sct value return is less than I % of lIu: lolnl

observations. So, we eliminaled in these series Ihe observations thnt lire nol syndlronized1D

• Wc impose these limitations since the Brlluer (lI)!]3) methodology as well as the French and Roll (\986) technique imply ilml identical number o[nbservnlions is used in the fund return ns of

NAV return.

Cunsidering the cltpressiuII (3).lhe standardised discounts/premiums change is computed

:md i13 menu. st:ulIl:lrd deviation. 5kewne~s, kurtosis. autoenrrelntion with Ings of one. two. duee

amI four weeks as wdl as the stnndurd error of the ht order Dutocorrdation'l are calculated.

Notice that all funds of the sOImpl1!: h311e neG:llives IsI order :Iutrn::orrdntion coefficients'!. which

is consistent with thl!: rnults of Brouer tll}!))) and Bonser·Nenl. Bnmer. Neal nnd WhcDtlcy

(1990). According to these authors. GiYen a negative 1 sI onler Dutoronelation coefficient of

weekly discounts/premiums change litis is an indicator of the presence of noise lraders. like

French and Rull (19116: 15) S<!y".

Aflerwllrds. wc clllculale the "GrCIlI menn" -I','. which result was ofO,929. what means

thM. according 10 Ihe signal e}\lraction lechnique of Frendl and Roll. 92.1)!. of Ihe dosed·end

fund slmres rate of return variance is due to the rntiollnl answer of the il1\'estors to the

infonnation emittcd for Ihe morkel and 7.1 ,;, of that variance i~ just due to the 1I0ise. For Ihis

reason. a small portion of the funtl shllres vDriance c~n just he explained by the investors' ilTDlinnal behaviour and. a.~ eou~equcnee, investor sentim~nt is likely to have luw signifieont

wcight in tlte mispricing of the assets, Our results are slightly superior to the Brnuer (1993)

resultli which WIIS fF~' · 0.9·17 Dud tire porlion of the reUlm voriane~ explained by the noise of

5.3~~.

At lust. wc calculate the estimated· observed weekly standnrdised discollnts/premiums

change variance ratio. The estimated weekly stanunrdised discOlll1ts/premillms clmnge v.,rinnce

is calculated by the e:lprcssinn (4). having used thc rnlio between the weekly menn fund NAV

return and the fund shores return vllrinnce. These results lire summarised on toble 11, in Ihe

.. Ai. K.U1It •• h. f.~IQ"'r, fimd ... d"", ... 'ed lium Ihe in~iol un"I. 0(41 n..d.· AUo!.CET.CI~1. CNN. en. EGX. EIs. rr.

!{I'.IIS. Itn. MRF. PEn. RIF 'Qd VUI

., lha. ,,:sdl> nn h. ol""Md dj.n")' hy Ib •• 1J!ben

• , " .... nl'D~ ID U ... Off}t ond IJorc.tHI."I.IIII ... I'."t ond WM>d~ {1990lllr. n<t;2I"" lot cnJ ...

,,"

.

<

....

I .... a ... !he

.... U)·1hc: duct'UIII::l"'.,.,i ... i. d ...

'

I>

n .. _ )11<"""'0,," ",dllle b ... fun:! ""d ... d IMf ~rfCtl> ... orukll)t:I£

a.

,...,.

• ".,. .• ,11 '" Ih. snunu. ",d. ff."",,1IOJ or ~Ir <:k>If'd .. "d (""~I .... lo .

.. In llO"=mIf "i'h!he im .. II". 1Mim<nI d"'~,y 4q"" .. lb. \.)f'OII ... <If ",n _c grna'" ,""till. fimd shlo .. "~ ,Ir.ouIJ ""

'IlIDC'Ond ... d. "I,;t.~. Fti'~<fm" IIc..ld b. mntn,d in I""'n ... b.s<mon<l"'R> .. ....td t<l>:r.IIe<><pl'" ""OCIIf"I.nic ...

4Fl<n<h '" R'~l. t9U 151 On d .. oU", h'nd.l<ffIl~,"; 10 It=~ a'J~""'. do. la ,110 &<1 rh" •• m d",,,,; lad. mo>' I ... nf.N.1

.n) (111<0 "ilhin 11" \"'l ... l .I"""d. Ih. ~rp""" h. M~" ~md:PiOln <>n be inda<N b,' nu.!}". armi.prinn!', IlUinly inhcc

m .. """" ... , ... In<I<prnlknl "am ~')' 'a day

"

appendix. TIle lirsl \'lIlue of Ihis table conesponds 10 Ihe estim~ted \'ari:mee in r..t:ninn to

obsCT"cd weekly slnnd~rdised discounlsiprcmimll5 change variance I1Ilio. The second value

corresponds to Ihe number of ooservDtions used in the caJcIlI~tion oflhc previous r~lio. for each

fund ~uli in c~c1t pcriod, Tlrc 135ttwo liuC!S ofthi! table eon~pond. re~pl'(!ti\'l'I}'.to the weighed

aye-rage of the estimQled to observed variance ratio. weighed by the munher of oJ,sef\'alions in

ellch period. for all the funds nllflto the numher of loIn I ohscrvations of the pcriod.

TakinC the last yalue of the mlio on tahle 11. in the append;'" corTespondinc: 10 tlu: tntal

period ilImlrscd. Ihe pcrcentage IIf the ubserved \Ve~kly stllndanlised discuuntsfprcmiu1II5

change yarionce nver tinu:. cxploined by the investor sentiment theory. for 011 the funds in the

sample is. around. 8.55%. slishtly superior tn the 6.77% achieved by Bmuer (1993:211).

Dividing the lulnl periud in study io thrc~ Sllhperiods (1987 10 19CJU: from 1991 h. 1994

ond from 1995 to 1998). the period from 19117 to 1990 was the one thot showed th~ I~rge$t

percentnge (9,65~ •. ~ppmxim~ltly) nf the obseT"cU "cekly standardised discounts/premiums

change Yorimlce. e.~pl~ined b) the investor s~ntiment theory. Anal) sing fund to fund. it is shown

Ilral the mod~1 can explain. in the hest of hypotllcscs. about I:!% of the weekly 51:IIIdllrdiscd

discmmtypreminms change \·ariance of Ihe funds ADX. GAB. SBF and TV, TIle remaining

funds prcsenl I!:qual nr lower pereenla:cs than II.S!". Il3ving found Ihe snloJleSI pereenla!;e in

the CIIse oflhe fUlld BKT (5.61)%). in the period flom 1957 tn 1998, Analysing by subperind.

and across funds. it is rc.1lised that thc model ~lIows Ihe explainlltion.llllhe mosl. AtOu"d 15'l~

oflhe weekly stand~rdiscU di~eountslpremiums change vorinnee ofthc fund GAB. in 1987·911.

But in the period of 1991·1),-1. Ihe highest percentuge th~tthc model ~ou explnin is unl)' II),H%.

appmllimateJy (fur the ease oflhe fund 'fYJ.

These re5ulL~ SUGGest thn\. nlthough the presence uf noise tmrlers in the elosc(r·end Iill1rls

sector cun innn~ncc ond justify the discounts/prcmiums clIistence. no more than 9~., roughly. of

the variance uf the discounts/premiums cll31tges would he ~:lpluined by presence of noise trnders, Like Br-mer 119931. wc ean not conlinll Ihm the inye~tor sentiment has 11 101 of

relevance on the explanlltion oflhe discounts/premiums variability.

Using the panel oflhe observations conllfcted with the 26 funds in the sample. during the

thr~e subpcriods in that the time horizon was distributed (1987·90: 1991·9-1 and 1995·9H). wc

test the robustness of our results, Conseque-ntly. we analYH the rel:llionshiJl between lilt

ohs~rved stantlllrdis~d distollntYllfemiums weekly ch~nge v~rianec und the eslimated onc, set

oul oflhe generic linellr regression model (7) lIS ddinl'd in Ihe section 3.2. Based on the Generic

linl'lIr rC!;Jession 1II0del 18}. 11'1' nnalyse thc relalionship belwecn Ihe stondardised

discounts/premiums weekly cllan:;e variance (ohseT"ed) and the specific value of I'ilch fnnd (I •

"

The results presented in the previous sections allow us 10 confirm some orlhe investor sentiment theory underlying implicntions concerning 10 the discounts/premiums bchnviour,

requiring !hnt noise traders nre prl'scnt in the closed-cnd funds market. However, when wc

nnnlyse the relevance of this theory to the I!,'(planntion of the weekly stondnrdised

discounts/premiums change varinnet!, wc verify that the investor sentiment theory CDII only

explnin 8.6% of that v .. rillllce. TIlis result is cOllfinncd by the panel linIn 011:IIY5i5. Wc have

nnalysed the linear correlation bCh~ccn the observed weekly standardised discollnts/prcmillm~

c1mngc vaTinnce ond the cstimntcd varion!:!: os well 115 between the observed "orionce ond the

prnponion of the fund return vnrinnce explnined by the noise tl1lding. Like Brnlle(s (1<:19;:1)

5tudy. we do not find cmpiric~1 evidence th~t ~l1ows us to mmOllncc, definitely. th~t investor

sentiment theory explains greilt p~n of the discounts/premiums variancc (reg~rdJcss of our

results being slightly morc uptimistic).

Though. this research presents as main limitation the fact of using a methodology.

50-called indirect. to dd~nninc tIll: innucncc of the noise tl1lders in the discounts/premiums

variance, which is based on the I1Itio of the implied to observed fund return variance as indicalor of the investor sentiment. resulting from this, us it is evident, deviations in rclation to what

would be c.~peeted. Tlle fact that the results were not vcry satisf.1ctory regarding what would be

C:"I:pectcd might have to do with the limitntion of the npplied methodology, '6 even though wc found some evidcnce that noise traders nrc present in this morke\. Thesc results also imply Ihot Ihere must be nther fnctors thnt joint wilh Ill(: onc now nnnlyscd will bctter c:"l:plnin this puulc. Onc of Ihose fnc[ors may be relalell with limits of nrbitrage nnll Ihe performance of Ihe

arhilrageur.;. Sblcifcr nnd Vishny (1997) [W5tulnte thnt there arc limit~tioos to the nrhitrage

(such os, tl1ln5o~tion C051S, financinl restrictions, agency costs. among otberl that lIimeult the arbitragcll!"S a~livity. Tlli5 inhibits them of implem~nting pnfect slTlltegil!S nnd, as a consequence, the quick olljustmcnt of the markct prices is not verified, Such limits might allow that the irnttional investors affect fund prices for more time than desired. As the discount!ifpremiums persist, it would be intcrestiog 10 study why the arbitrageurs ore notahle to implement with to[al ~uccess its stl1ltegies as well as if they arc rcally present in Ihe closed-cnd funlls market. It would still be imponant to stully if there is n rclationship bc[ween the nrbitmgeurs Himpotellcc" in lending tu converge the fUlld priecs 10 NAV (that m~y be one of the

reasons of the c1nsell-end funds lIisconnts/premillms persisteoee) nod the iovestor sentiment.

Also to he nolicell that the studies Ihat, up to the moment. were disclosed concentrme the :l1Iention in a restriclcd group of factors, studying them a lul of times in all isolated way, and they rarely make the study with the combinntion of scvcrnlllnd different fUC1ors, nbove nil the

" tlu "n, •• n oIct11. "k;"~ ''''0 <O",t"""OIion lh. ,..,,11> <>bIoi .. d m ,h. )11","1 nU~)' nlYl in ",1\ .. 1<1<>I'b<, I" ~. r""t ,\: Wh •• I1<)", t9~~. Eho". (j,uh,,'\: Du"," .. 1998. r,.nJ,;,[ '" SdUIlU;J". 19%. 1I1101nu.elil. \..a C.fIi<l &; WiLfrlUl. t993. Dro"".

t'l'!J: Ch:n .. Kan.1: !.tin ... IQ~l; Oor .... I .. «,!; Shlei{ .. ,t'l'll;

L".

St.I.;{ .. &: TII:II." t'l'll: t.n. Sh[ei{ .. &. 11o,I ... tQO(I.:urnm~ ",hml,h>! Mbe ''''Ik" 11. in 'he m.,l" .nd 'hey I"n",n« ;,.I""",,,,i"ll pm", ,nl."",)· I";' Rn",.". '''''"I

~2

faclors ofrntinnnl nature with nnn rntionlll ones - said behavioural faclors.l'uture developments

might try to find methodologies tltal combine Ihese t\\O factor types, namely the investor

scmimcnt theory anll the limits of arhitrage, as explanation nf thc discoUltlsfpremiunlS. Another invesliglltion path will he to mtolyse the in\'~stor sentiment comparing 10 lite inve.<IOr.<'

nppetencc 11 for open-end investment funds (polentinls compctitors of the dosed-cnd funds) with

thc time that new funds are placed olllhe market, null the evolution oflhe seasoned olle (namely

Ihe b~hnviour of its discollllts/prcmiums) since it has been \erified that new funds. that try to

offer tlilTerent iovcMmenl policy {objectives) from the 5en~oncd one. nppearetl when the uther

closed·enll funds were ot premium or:lt reduced disconot.

7, REFERENCES.

Ahralt.1tll. Ahr~ltatll; Elnn, Don: amI MJrcll~. Alan L I ')<)3; Dues S~nliment Explnill L"1llled-eml I'uml

Dis\'ullltl5'! Evidence FHJUI BI'U" Fumls. FII/ll/lclnl R~I'I~H' 211(.\1: (,1l7·6 16.

Arak. Marcel1e and Taylor. Ihmt. 1.1)%a. Optimal 'Irading With lIicnl1·Hc\cl1ing I'rices: S>lilching Belll"een Foreign Stocks aud CI(Jscll·cnd Country Funds . . lp/ll1ed EcmlnJllics 2~(l1): 11)67-1 U7-1. Arnk. Marccllc and Ta~'lor, Deall. )·J%h. Risl: and Retum ill Trndin/! C)o~d - En" COlll11ry Fund,: Can Ttndin/! Bent Holdio/! FOIci!!11 Stocks'!. n,r (}1II1>"1rrT)' Rerirll' (JfF.l"fJIJ(l/uh·.~ aud FiuuJ/cr 36/21: 21 '1-23 1 Arcllano. M. 1993. On th~ Testinp or Curnlatcd Err~C1~ with Panel Data. Jmrma/ nf F.C"lmmlfCIl·ic •• 5'1'

fl.7-97.

Ualtagi. Uadi 11. 1995 ~lCtric Ao"IHi~!.Il:I.!.!!: CIIichcstcr: John Wilcy &. Sons.

Bnrdn~. lIIiehnc1 J.: Hllldemess. Cliffurd G.: Hnd Pllntin: Jdrrc}. 1~,)3.1'ri\alc Ikodits ffllm IlIlld,

O ... nership Dnd Discuunts nn Clnsed·end Funds.. Juurtlal of Fimmd,,/ EC"UIJr)Nli<'s 33: 26.1·2111. Bon~cr·Ncnl, Cntherinc; Bmller, (jreggor}~ Ncnl Rnlll:n: nlld Wileat!ey. Simnll. 199f1. In!cl1lati"II~1

1r1l'cstm~nt /{estrictiulls m1<l Clo~ed·elld Country Fllnoll'rices. IlJc Jmmml 01 Fillw"e .1512): 513-5.17. Bonlun/m. J. N.: Kim. D.: and Lee. Charles /11. C. 1'195. ("[l)~cll-cl\ll COllmr) FIlIIlh lInJ 1).5. /IIarket SclllimenL R~"ie'" .yFimlllci,,/ Swdics H: H7\1-9IH.

ilrJucr. ([Ic!;,cory A .. 19R·I. Ol'cn-cn,HnC Clllscd'cnd Funds. JnflrIJu/ nf FiJ/{lIIrial £rmmmiC".< 13: -I'JI-507.

Br-mer. ([Ieggory A .. 198R. Closed·~nd Fund Shald Ahlll'nn~ll{elllm~ ~nd the Inrimn~tioll Cutlteut or f)iscollnts and Plcmiums. 7/1<' JOllmul of FiJlum .. 'c -I3! I): 113·1

n.

Urnllcr. Greggory A. 1993. lr1\."c~tor Sentiment "lid the (,lo~cd'<'lld Flmd Pl1ule: a 7 l'crcclI[ Snlutilll1.

Jnllnml nfFillUllcial &,,"iccs Rt'sran:h 7(3): ]'l9·2]6.

Drick!ey, A. lames; and Scballheim. lames S ... IQR5. unio!; the Lid on Closcd-eod In'cstlllcnt Comr;mie~: A Cllse llf Abnormal Retllrn~. J",,"'ol of nllOlJdolalllf Q":mljla/il'~ '/IIo(\\'i .• :!P! 11: 107_ 117.

Bro ... n. Gcgnry \\". lQ99. V"lntilit~. Sentinlcl1t. and Noise Traders. FiJlmK';u/ Alkr~"lls .iolfrJIul 55121:

82·90,

Ch~o, Nni-FII: !\..an, ita)mond; Miller, and M~n"l1 11 .. 199~. Arc th~ Di'CII1I1!1S I>n Clu!cci-cnd Funds n Sentimcntlmlcx·!". 17,~ J",mUlI n(FiIUlJlCf -\11/2): 79S-RUn.

Ol~ng, A.: COf'L'lnod L.: und O'Hynlul1, J .. t'lQ-l. Il1w.stlllcnts Tfll.11 Discn\lll1~ ond Ahn"rrol1l RctllTlIs: UK EI·id~llc~. Jarrllllll of Busines.r Fimmcr ,';, ,'/ccnulllilJ): 2H"I; RI3-831

"lh. 'nl"'u," "1'I'<1<IIte re", "1"'"...,,,01 fund, nh, oIem,mJ (U, <1'<"...",.J {u"d,)"", r., m .... ",d by 'he li'I"'oI .'''I"",''on "r

23

Chcung. C. Shmnnn: Kllnn. Clnrcncc C. Y.: and Lcc.la:;nn. 1997.1111: Noise Trader Hyp()lho:si~: ne C:l!;O: nrClo~cd _ End Counlry fumls. R~.~,dl ill Fimmce (Edilcd hy AndrcII" H. Chen) 15: I1 S-136.

Choprn. Navin: Lee. Ch~rl~$ M.C.; Shleifer. Andrei: amll1J~1er. Richaru H •. 1993. Ye~. Uiscounts on CIlI!'cu-eml Funlls Are 0 Semirnent Ind .... ". H.e JOIJllIal of Fi/w"cc -18(2): II01-HUS.

D~ Len!!. B.: Shldf~r. A: Summers. L; and Wo!dmunn. It. 19QO. Nui~e Truucr Ri$k in I'inando!

Markets. Jrmma/ of polincal EC(III<Inry 'IN: 703-73S.

1l~ long. J. 8.; nnd Shkif~r, A .. 1992. Closeu-cnd Fund Di~cllullIs. nlr Jnlll'lll1i of /,orUoJjn

Managrnr~nI.: ·16-53.

llimson. E.: and Minio-Kmersld. C. I <)')R. Cl05ed-~nd Funds: A Survey. LBS, fonhcmning in FiIJar,d,,/

.1fl1r~tlS. In.rlilurirms,'; /oslrumrlJls: 71).

Ellon. Edwil1 J.: Grul>er. MlIJ1in J.; ond Busse. JelTrc' A.. 1998. Do Investors Core Ahout Scntiment. Hie

.ltwnrn/ nffJlIsil1es.< 7\(4): 477-500.

-Frnnlwl. Jelfrcy A.; nnd S~hmuk[~r. Sergio L.. 1996. Country Fund lli'tounts, Asymmelric Informntion Dnd 1he Mc:tican Crisis or 199-1: nid Local Residcnts Turn Pessimistic ht!rlirc international In\C5IOr5'1.

NBER Working l'uper n' 5714: 43.

French. Kcnnelh: aud Rnli. Richard, 1986. Stock Return Vmianc.es -The Arri\'ol of Inrornmtion mill /lcactiull orTrutlers. JOlll1lol nf Financial EcrlJlom/cs 17:. 5-26.

Ilardou,·clis. Gihs ,\.: Ln Porta. Raruei; and Wizman. 111ierry A .. 1993. What Mo,"es the lliscollnts lln

Country Equity Funds? NBER Working Paper n~ 4571.

HausJI1on. J. A .. 1978. Specifh"lltion Tcsts in Econometri"s. EcrnlrJltlelrko 46(6\: 1251-1271. Hou.lmml. kITY A.; ond Ta),lur. Wi1!imn E. 1\1111. Pond dotu und Ullubscrl'ublc Individual ElTects.

ErrmmnClrka ·19(6): [J77-Jj'J8.

Huny. A. 1982. A Remark on Hnusmun's Spccilicoliun Tesl. Ecaname/rim 30jJ[: 7<19·759.

11$ioo, C. 1986. Annlv~i~ orl'nnd Duln. Econometric SIJdcty "Ionugraphs Nll. 11; (omhridgc iJnivcrsity

Press. Comhridge, 19R(i.

J(1rge. SI~~3na M. F. 1997. neterminan1c~ do F.ltmtl1ra de CM[1itoi~: HID (.0.10 Portllgu~s -1<)91l D 1<)<)5.

llissefln~~n dc Mcstrndo. Brngu: Eseolo de Ecunnmia e Cicstnn da lJnivcrsit!ude dn Minim.

Klibanllff. P.; I...:Imon!. O.;onll Wizman. T. 1998. In~c~or Re3ction 10 Salient NClls in Clo~cd-Clld

Cnuntry Funds. T1r~ Jowl/ol of Finrlllu H(2): 673-69.'>.

Krnmcr. Charles and SIlIilh. R.. Todd. 19'>5.Rcccn! TIIIOIllil in Emerging Mn,kds nnd 11re Be1mvior of

Countr)-Fund UiSCullnts: Rcne\\jng !lIe i'u7.1.1e of Pricing of Closed-End Mutual Funlls. IMI' Working I'npt:r. \\'1'/95/68.

L~~. Clr~rl~5 M.

c.; Slrleifer.

Audlti; nnd l11akr. Ilklmrd H .. 1990. Cln~ed-cnd MllIual FlJn~5. J,,,,rnol ,,(

Ecnn"',,;c /'rrspcrrl,·cs 4 (4): I~J-164.

Lee. Cl",k~ M.

c.

:

Sldcircr, Andrd; ond Uralcr. rucb~rd H. 11)<) I.ln,·es(r;;r Sentiment ond th" Clo~II-"nd

funt! Plluk T1,r J<1IIMII1I q(Fillllllrr 46(1): 75-109.

Lcvi~. Mnrio nnd Thomas. D~[~n C. 1<)95. In\'cstments Trust [PO's: Issuing Bchnvior nod Pri~e

Pcrformnncc Evidence frnm London Stock Exc1wn!l~. JnUTII"/ nf Ranklo/!'" Finorrc~ I<JPI): 1-I37-1·1SH. Monte. Ana I'Jula C. 201)0. Snhrc O! Discnllnt~rcmillm§ dns Funt/os de Invc~timento Fedr3dn~ no Cllntc.\1I1 d~ Tcnri~ dn Scntimentn do Invc~tidnr. I1is5Crtn~~n dc MCSlTudl,. BrJya: Escnlll de Econnmia c

ficslIlo r!~ lIni\"crsidadc dn Minhn.

Monte, Ann Pmllo C ant! Arm~d:t. M. J. dn Roch~. 20nO. SohTt' no; [)i~count<;il'Tt'rnium~ dn. Funri,," de Inve<;1iOlcntn Fcc1mdn,,' Hmp dil'Ic."~n .'<llhre u<; nllndrm cl1I1ccnl1mi<; n"e e~nlic:lm a sun e~i~tCncia.:

WorkinG poper n" 7/201l0; Nt"Idco de cslotins cm Cic<;lfin; BrogD: Universid3dc dll Minh",

Ne\\c)'. Whitncy K. ami Wcs\. Kcnrrcth D. 19H7.A Sirnpk. Pusitiw Semi-Definite. IIete1l1skcdasticity ond ;\utocOlTClalion Consistcnt Co,'nrianl'C Motri.'l. £CmroJ/l~lriw 55(3): 703-708.

Pen,·y. J. [9'l0.Returns ollinitbl Public OlTcrio!!s of Closed-end Funds. Re,.;!,,"

of

F;I1<lIIc;,,1 Slut/its 3;

~.!

Pind)"ck, RohclI S.: nlld Itubiufcld. Daniel L 1998. I'cnnm!!f![ic "loddS oml ECOJl(ll1!ic FOIccam: 4th Edition. Economic Scries. McCirol\-1 1i11 Intcmntinnal Edition ••

PontilT. J. 1995. Clns~c.1-end I'remia Jnd Returns [mplicatiuns for J'inanci~l f\l3rht Equilihrium. JOllm,,1

of FiJUIllc/u/ EC(}//lJruic~. 37: 3-11-370.

Shl~ifer. Andrd; and Summers. LU\lltll~~ 11 .. 19yn.Th~ Nni.o;~ Tl1Id~r Apprwch tu Finml<. .... lmll"nol of Econr",,;c J'rrsprc/il'rs 4(1) 19-3J.

Shkifer. A.; ollll Vishny. R. 1997. Lhnils of AlhillOpe. JOllma/ ,,(Fimmrr. 5133·55.

Sias. Richard W.; 19~7n. Pricc Prcs.~urc and the Rule uflnstitulionolln,estOIS in Closed-cnd rund.l.

n,

c

Jnunw/ <If Fil1am:io/ Rtscarch ZIHZ): ZI 1-12').

Si:l.S. Richard W. 19971>. TIre Sensilivi1y uf Indi"i"u~1 mlll Institutional Imcsturs E:\I'~c1ati(>l1S To Changing Marke1 COIll!i1ions: E~idcnce fmm Ch'st:l!-End Fund.1.. !lel'iell' uf (llmlt/il<uhe FiIlr1lJ,'c ","l

AccO/m/llrg H: 245-2fi9.

St"ort. J.: ond Gill. L. 199B. Econnmctric.<. 2th Edition. O:..rmd: l'remi.·c I I~II Europe.

WciS);. Kathk~n. 1989. 'nle I'ost-Offcrill!; Price Perfonnallcc nr Clm.cd-end i'llIIds. Fillaudal

Managcl1telll Aut<uun: 57-65.

White. H. 19KU. A Iidcrnshdastidty-Corui5tcut Cov~liance M~tri~ F.5timalm and 0 Direct Te~t fllr

"'

~

"

"

0

i

..

~

~

§

]

~

~

0

·1

0

,

0

.

.

~

_J

~

_~

_~

_•

0

U

j!

_~

0

..

~

0

,

,

.;;;~ ~

_!

§

_~

"

§

_~ ~'~'

·

E

,

_oS

.

,

~

_~

~

-j

~

~

if

t

_•

~

~~

;;

_.~

•

·

_a

_ji

""

~

_!

"E

"

-

! '

~ ii.~ 0

.~

_it

,

,

~

:i"

,

~

~

0

,

~~

;;

•

,

.E

~

_~

,

,

,

u

_,

~

i

~

~ 0

"

₀

~

"

0

;;.

_,

"

~

E

z

w

.. ..

"

~

,

"

!

0 ~

1

]

!

,

0 0 ~ ~

"

E

,

~ ~ ~

...

{

3

1!

l

,

g

•

~

•

1

,

]

I

-"

~

f

0 §

"

,

~

~ ~ .~

"

.2 ~ 0

~

~

0

"

~

.

~

..

_E

_'5

"

,

~

!!

~ ~ 0 0

,

1

E

,

•

,;;

_,

.~

,

_.

_,

,

"

~

_'"

,

~ ~

1

1

:I

~I~

I

I

H~

I

~

I

~

I

~

I

'

I

-I

~

·

1

;

1

1

1

1

.

~

.

1

~

.

.

I

§

r

I

~

I

~

I

"

1

0

I

~

I

~

I

~I~

I

~

I

~

.

1

g

I

~

I

r

•

..

"

_I

_"

~

I

~

_"t

0

•.

~

'"

j

t

·

,

E

,

~

l-'-t-Hrn

*'fM

~

li

..

_~

,

~

_{L...L-'-'_-'---,}

"

Tnhle 7 -Summ:n)' of the ruuU, or the pand tlnl:! "onlYIIs -hypo/he,ll tnls.

11,'0,"'(11"'······°.\· /i"1 IF, II' ~alU<')

K....

IF, 1/1 t"fue) /i .. 1J 1/', 1/1 ,."fur) 11,: 0, "01'" ....•

u"

Ii-I IF"J 11> ,-al"r) K~

I

F,

Ip rlll,"l /j-U IF: ,1/1 rul .... \

11,:"

,

=u: " ...

_

-/

1"

/i .. \ IF, III m/ur} K,"", IF, "";"d ,... .. u IF, Ip \'01,01') A 1\

p,.Pz"'

... -

P.y

117.91 ID.OO) Il..~~~l (0.00, 11~,60 w.nn) t\

p.

-

P;

"

·· ...

•

P

.\"

·

·2.20 -2.30 -132 dnOO fI,=p:e ... =p" .. 1611.J~ IO.M} l."~,~, 111,(0) 247.l6 (0

on

11.,.., A-i\bllUl run.! ,e,u,n; 8 _Abo., IIAV rcl"",

u

117.03 ,0.001 111.'6 (0.00) 75.22 !D.OIll ·IJJ .J 22

..

'"

!J6.M UIJIUl 229.7S ,0.00) 159.10 (O.Q!!,L I]dn~ 100 IC~I ~Inlislic~ G;'~n b)~ F (.tm""-SSll,,rr)I(,\"-lf(.\f+11

.e

.'iSR,.,r/.\'T-A'(Mtll F_ p (SSR,n -5.1'11"./1.1/(,\' -11 - SSnl.cINT-NIM+I) F, '" (SSRm-SSRm,1tN-!J SSR'uIlNT-N-M) wl",,~;

:"'>R,,._II.,,dlUl .!IIn nf,'[U>, .. orll ..

'",

.. """"e

nIOIJd.

.wt",. R .. id,d ,om oh'l'w" afll!< "m;>lrmodd.

SSI~" .. R .. idu:d "111\ or''l''''r<> oClh. lIIul,,,Y.!J\ldlUl,

"""'<I

.v. , .. "b", offonch on Ib< .. mrlt;

M,. !I'm ... ' or,l", u~l.ruklry ,..mhl .. :

NI" .. ntlm1", ofllll1.1 M.b ob,m"I,.n •.

Tllhk 11-t-'hell eff.,ls l\Intld (hornn!:'nril)" nOhe slnpe and h.lrrollrnrily nftht Inlen:rpllnn). .\bnuI fUDd rrlurn AboUI N,\" r~larn

< <

1(-1

L

_,

_.,

RFND f)_I"

a,'

•

fI (1~'19 DISC 'J

L

R,"'L ,,_I. a"-00171 DISC '.'

Il612'11

,

.

,

(-f;~97~J

~=onll6'l _100000J H; ::41006)

₁

00000

₁

<

.

I{ ....

6

IIFND, ,_I "',..'

+11

031A DISC 'J ~, 1/1"1". ,,_. _

If:

_

n nl74

'ux

"

~ ",OOOH (4.7902) R';:OOI20 (-~ Iml

1000001

r

,

.

oo,.

1

,

_•

K-IJ L RF.VD'J_' ..

,.!

+0 fiSH I DfSC" L RI'I'L'J_' _n' _11 InrlJ rJISC'J

,.,

_(4.02411

,.

,

f_l.'llo"

,e",00108 [OOIllJ1] n;-00280 [UOllIHI]

-

-Uu~

a;' • a

;

'

'''1''''' ...

Ill.

'

'''

'

'

''<f

'

"''' ...

er ... cr 11 .. "' ... ~ .. ,·m.blt " ... limd '" 1is>J. 11", '"""'" a _ amod I"'''''''''''

• ...,.'f"I'I'I.

It>!he HUt;";';:>! of,l< . .... me;.'" _ ,b< "rhr ""'"~''''' '0 IU p-, .. Iu, T"hr.1} -iI~'polhQU ·I·.sl~ fnurillnrr Inpl)"J;J,

hlft'H""l"": !~" n'H ... lnn, /11: TMnl 'm"'~ttll.lI)' ~fthf rtlf'.DIrl~"t:I

FI J.I'IAAI! 1.16761 OIH'llH.' p."~",, 011'11.11 112: S'~l'r ilomltJ:mt/ry ,,"si inltrr'1'r/on ",'trn/:rrr,iry Whetc: ~

r[

M,

L

'"

1'",.1 +

1\

,

.

,

\'nr(hD, )'J +C" 171 ,..,,0/..: Fl 1.1: i.l0 1111)67 o IJUlY

liJ: Int,rul'lit>lIhnmoll.",lfI·:/"tn /h, ""pr/mm"/l.",lty

"'tilr'"

\":1,~,

=

~')"

+

f'lt,~-I:L+r"

II1

f.I 1.1: ioU

p .... ,,"" 0"\5~11 O.1llJH

Taille 10- Unenr Res:rcuinnl relath'u In Ihe I!.wcrnl mn,lt)s!11 c (,11). ht RtEr .. .!,,";

Filed drub Modrl (!lort hnmos:c"dl~' nnd

inttrnptiun helnll.=:tnt)ty)

2nl' n.~re.

..

inn:

Simple ;\Iodd -

r

n,,

'

l'lI

"'!rrf'S!i"" 1101;11

homnl!.n~;ly "Clh. cn.mti~nl'J

"

vor

(

L\13

,

le

"'PIILI + 12.20241 viir(W,t

vos{M5,l..

;

O

.

000286+0

,OOII

77~

(I

-V:

L

R-'-O.90H327

S~nmi-l,II/E-llII

113~9n)

11J0000] /i-f).IllMII

SSl!m' _1.16E.l/j

11 J(,:!.51

(025HI

tll'l •. ;",J' 1 .. " ... 1> 11", m .... "'ol<n"1'l''''' t""flj" . . ~ , .. , .. !>le flom f'Old I~ "'od lh" ,>I", ""'''''~ tU"Np;l<nlb .. "

.""e.p<:D<\j ID I"" I.U.o",nI 01 'he ."<fli«.,,, .. 01 lb. "{Ill " ... h.". I .... ",.coI ••

Table 11-'11" .,Iimnt.d prnpnHlon (rntill) nr Ih. ulimalcll \"nrlftnu in rdntinn 10 "bJer ... d "'.tkl~· .Inndftrdh.d

dhtuuntJ'p~n1I"ml .hll" e \'Ula"Of QUlle tJQltd_end in\"l:.1tmtnl C,,"dJ. for Ihe .rlod IIf 21111987 a JI/IUI'I1J8.

roM. ".,_"1 I"'I"~ 19';':-'11 I?n.,~ r.

".,_"

1

I"'I"~ 19';':-'11 I?n.,~

.

\iiX

,\.'IF n[.'1 iiKF

...

1l11~O ,~ IlD1H

..

nlliiJ 2D6 n IJol 201 "QJ\~

..

~ ~os n0l">19 ,~ o!lllll ,~ !I(){,19

""

011}.1 201 un6Jl

".

O1lli1~

"'.

00')(11

'"

1111'7 . " 00615

."

o I'II,?,~ (,!J U09U

.u

an,,,,

](10 517 IILU 009].1 n nH~ t:L~1 U,'i" CoAlI If,.[ S~I 01l1~1 OD7]~ 1- 0 I!~A OO~1J

'

"

ODIB

,

"

"U'6 1111 101 '0014! !OJ

'"'"

,~

,

..

D05~ lUB 01

,

00'

..

,

..

on717

'"

OIIH ItH

CoAM Ol~~- O!l'I~~ --oi~- OlU5

I

2 0 _ ; 1 1 ' 1 ~U! r,~s

I

Col"

I

00::: oo;;~ pn;~! oo:.~~

I

1

illS

I

Qom OO~~; OO~~~ oo:~

"'

,m

1.!lF

'

"'

rur ['.\If

".

sur n01.fil 101

,,

-•

"~,

"

00110

"

oMU 1.11

"

-

,

•

01410 ~ I--·m~ ~ - - I _ oo,,~ ""(.11 ii"iil4l 10] ;>{IS J09 OO~9; nw" n(1l)67 ;!O'I JG' J~tI OO'" 00167 a~(..t~ J01 o DJ~1 ~o~

.

_

,

143 1I0~.'7 ~Ol

o

c,

if

'"

O'I(,.I~

M.

I) IDf 4'/1 o ~llJ ~, !OII6H '!t I) II~~ I'I~ ~rJ 5011. nnol7 on.u n 115\ n:." t.F Ztf un ... i~bnl !fiR 6B o 0'l26 n 101" I) JbJr - -[- I) 11!l 11H 10'1 JOI

o IHI OU71~ on?~

I'" In~ OGIU 00IY7 JO~ onffi

'

"

IIOJ16

'"

00(,71

"

0.0965 11~~

'"'

I)I1J17 ~o~ nU/i1l lO~ oow~

'"

1l.1Il!"J H.1i Jq. o n",,~

""

n"l~

_,.,.

!l01.!"

'"

OQ6~ lOI

•

.

-

,,~ (,11 ".!O'J!

.u

nnH~ 6!5 ofli~1

'"

t''''11 SH O!J!>.U >00 ll.illl. .... ~ HH~