You are on page 1of 34

1

Dynamic Interrelationships between Macroeconomic Indicators, Global Stock


Market, and Commodities Prices and Jakarta Composite Index (JCI)

Pro !dler Man"r"n#
Jos"a Pardede MS$%, MSc%
&omora Sitor"s MSc%

!bstracts
Our paper aims to analyze interrelationship between Jakarta Composite Index and
mcroeconomic indicators, global stock market, and commodities prices. We employ ector
!rror Correction "odel #!C"$ to in%estigate whether dynamic linkages exist between our
research %ariables. We &ind that there is co mo%ement between %ariable in our research.
"oreo%er, 'ow Jones, gold price, and oil prices dominantly a&&ect JCI mo%ement in the long(
run.

JEL Classification) C*+, ,1-, ,1.
Keywords) Cointegration, /tock market, Commodities, "acroeconomy, !C"


'% Introd"ction
0cademic community has long been interested in the connection between the
&inancial markets and the economy. 1he recent decline in global economic acti%ity due to
subprime crisis has again intensi&ied study on the relationship between both &inancial indexes
and macroeconomic per&ormance. 0gainst the background o& e&&icient market hypothesis
which denote that asset prices should &ollow a random walk or at least be undpredictable, the
existence o& predictability pattern o& stock market indexes has been considered &ascinating.
1he common approach in the dynamic analysis between &inancial market and
macroeonomy used ector 0uto 2egressi%e #02$ to study the e&&ect o& many
macroeconomic inno%ations in the pricing o& &inancial market assets. "ost o& the studies in
this category mainly assume a priori that the direction o& 3indicator power4 comes &rom the
macroeconomy and goes to the &inancial market.
Our paper &ocuses on the determinant &actors o& Jakarta Composite Index #JCI$. JCI is
a modi&ied capitalization(weighted index o& allstocks listed on the regular board o& the
Indonesia /tock !xchange. 1he index was de%eloped with a base index %alue o& 155 as o&
+

0ugust 15, 167+. 0t the end o& year +51+, JCI has gained 1+.5* percents to -+71.78 le%el,
compared to *7+1.66 &rom the pre%ious year.
1his study pro%ide se%eral contributions to the literature. 9irst, it estimates a relati%ely
wide range o& multi%ariate n(regime !C" models. 1he model is applied to a ten(%ariable
%ector that includes ,ross 'omestic :roduct, Consumer :rice Index, Interest 2ate, 'ow
Jones Index, ;ikkei Index, <ang /eng Index, Coal Index, Crude :alm Oil Index, ,old Index,
Oil, and 2eal !xchange 2ate. We use the longest a%ailable monthly database &or Indonesia
'ata &rom ;o%ember +555 to 'ecember +51+.
/econd, we pro%ide e%idence that that the best &it to the =oint density o& the data is
pro%ided by the !C" model. We interpret the high le%el o& 2(s>uare o& about 5.?7 as an
e%idence that dynamic linkage between &inancial markets and the macroeconomy existed in
Indonesaia during the period o& obser%ation.
1hird, our study contains new results on the interaction between /tock Indexes and
macroeconomic indicators. 1his &indings can be summarized in se%eral main scenarios.
/cenario 1 &eatures the importance o& 'ow Jones in leading to the de%elopment o& the
national stock indexes. /cenario + con&irm the depressi%e e&&ects o& Oil price on JCI during
the last decade. /cenario * test the conse>uences o& the ,old price to the &inancial market.
We show that &inancial market is not directly impacted by the %ariables related exchange rate,
in&lation and interest rate.
1he structure o& this paper is as the &ollowing. In /ection +, we pro%ide literature
re%iew on interrelationship between %ariables in our analysis. In /ection *, we brie&ly explain
our study based on the !C" methodology. In /ection -, we describe our %ariables and
illustrate the data set. In /ection ., we present our results, while /ection 8 we conclude.

(% )iterat"re &e*iew

a% Macroeconomic *ariable and Stock Prices
It is widely documented that macroeconomic %ariables ha%e explanatory power &or
stock returns. 1his relationship between both %ariables is a well(established result, see e.g.
9ama #1671, 1665$, Chen et al. #1678$, @arro #1665$, /chwert #1665$ and 9erson and <ar%ey
#1661$ who &ound that A./. stock returns and its aggregate real acti%ity are correlated. Other
paper using international market data also &ound the same results, &or instance) 0sprem
#1676$, @eckers et al. #166+$, 9erson and <ar%ey #166*$, Cheung et al. #166?a, b$, Chung
#1667$.
*

1he existing model explains the relationship between macroeconomic %ariables and
stock returns &rom the perspecti%e o& discounted &uture cash &lows earned by in%estor o& the
stock #Chen et al., 1678B :earce and 2oley, 167*B Cim and Wu, 167?$. In%estors incorporate
the in&ormation into their estimates o& the appropriate discount rate and the expected &low o&
di%idend &rom stocks, which in turn a&&ects stock price 9or instance, with the model :
5
DE
t

!#c
t
$F#1Gk
t
$
t
, changes in systematic shock in&luences the price o& the stock, :
5
, by the
expected cash &lows, !#c
t
$, andFor %ia the re>uired rate o& return, k
t
. ,=erde and /aettem
#1666$ argues the latter #k
t
$ is related to both the le%el o& discount rates and the term(structure
spreads across di&&erent maturities, in which the stock market is taken as endogenous.
1he macro %ariables that become the proxies &or measures &or aggregate economic
acti%ity in this paper are real gross domestic product #,':$, industrial production, real
money supply, real consumption, aggregate &oreign reser%es, interest rate, exchange rate and
energy price #W1I oil spot$, and commodity prices #Crude :alm Oil and ,old$. 1he ,': and
Industrial production mainly a&&ects the economic acti%ity by in&luencing the &uture cash
&lows o& the stock.
1he money supply can a&&ect the stock market in se%eral methods. One way is through
the port&olio balance model that works through money supply increase that leads to a
port&olio shi&t &rom non(interest bearing money to &inancial assets including e>uities. In
addition to that, "oney supply &luctuations can a&&ect the stock market through their e&&ects
on in&lation uncertainty. Ananticipated in&lation may directly a&&ect real stock prices
#negati%ely$ through unexpected changes in the price le%el. In&lation uncertainty may also
a&&ect the discount rate thus reducing the present %alue o& &uture corporate cash &lows. 'e9ina
#1661$ argued that high in&lation initially has a negati%e e&&ect on corporate income due to
immediate rising costs and slowly ad=usting output prices, sinking pro&its and there&ore the
share price. "andelker and 1andon #167.$, 0sprem #1676$, and 1horbecke #166?$ also
showed that real e&&ects o& monetary policy a&&ect the stock price signi&icantly. "ore
speci&ically, :atelis #166?$ propose that money supply shocks in&luence e>uity prices mainly
%ia the risk premium.
We hypothesize that interest rate relate negati%ely with stock prices. "easured as
opportunity cost, the change in nominal interest rate will moti%ate in%estor to substitute
e>uity shares &or other assets in the port&olio. 1his increase has negati%e e&&ect on stock prices
&rom asset port&olio allocation. <igher interest also means higher &inancing costs and lower
corporate pro&itability. 9inally, higher interest rate may also discourage mergers, ac>uisitions
and buyouts #Wongbangpo and /harma, +55+$
-

1he exchange rate has indirect impact to stock prices. 0ccording to ,an et al. #+558$,
I& the domestic currency depreciates against &oreign currencies, the export price will go down
and, conse>uently, the %olume o& the countryHs exports will rise, assuming that the demand
&or this product is elastic. "ukher=ee and ;aka #166.$, 0chsani and /trohe #+55+$ con&irmed
this positi%e relationship existed in Japan and Indonesia. 0dditionally, 0=ayi and "ougoue
#1668$ also showed that an increase in stock price has a negati%e short(term e&&ect on
domestic currency %alues but in the long term this e&&ect is positi%e, while currency
depreciation has a negati%e short and long(term e&&ect on the e>uity prices.
On the basis o& economic theory, we ha%e two scenarios with regard to the sign o&
relationship between macro %ariables and stock prices, &irst the II&lowHH scenario, which is
based on the II&lowHH approach to exchange rate determination, and the IIstockHH scenario,
which is based on the port&olio approach to exchange rate determination. 1he II&lowHH
scenario depends on two well(documented relationships, the relationship between the real
exchange rate and economic acti%ity #see e.g. Cornell, 167*B Wol&&, 1677$, whereby a &all in
the real exchange rate increases the competiti%eness o& domestic goods %ersus &oreign goods
and the le%el o& domestic aggregate demand and outputB and the relationship between
economic acti%ity and stock markets #see e.g. /chwert, 1665B 2oll, 166+B Cano%a and
'e;icolo, 166.$.
1he IIstockHH scenario depends on the port&olio approach to exchange rate
determination. 0gents allocate their wealth amongst alternati%e assets including domestic
money, domestic bonds and e>uities, and &oreign securities. 1he role o& the exchange rate is
to balance the asset demands and supplies. 1hus, any change in the demand &or and supply o&
assets will alter the e>uilibrium exchange rate. 9or example, a rally in the A/ stock market
will cause Indonesia stock market to rise as a result o& intensi%e trading between Indonesia
and A/ companies. 1hat in turn will increase wealth and the demand &or each o& the assets in
Indonesia #wealth e&&ect$. 1he excess demand &or IndonesiaHs currency will cause interest
rates to go up and a substitution &rom &oreign securities to domestic assets resulting in an
appreciation o& the domestic real exchange rate. 1he economic currency exposure &or
indi%idual &irms will depend on the currency structure o& its exports, imports and &inancing.
'e%aluation can either raise or lower a &irmHs stock prices depending whether the &irm is an
exporting or importing &irm.

(%( Commodities Prices and Stock Prices
.

1he linkage between oil price and &inancial markets appears to be natural. "ussa
#+555$ argues that oil price %olatility in&luence economic acti%ity, corporate earnings,
in&lation and monetary policy. 1hus, an increase in the oil price has implications &or asset
prices and &inancial markets. Chen et al. #1678$ also used oil prices as a measure o& economic
risk in the A./. stock market. 9or Japanese market, <amao #1677$ and @rown and Otsuki
#1665$ also support the &indings that oil price play an important role in pricing e>uities. Asing
17 national e>uity markets data, 9erson and <ar%ey #166*$ &ind that changes in A./. crude oil
prices keep contributing to %olatility o& global economy.
In contrast with abo%e, <uang et al. #1668$ argues that oil price shocks do not
in&luence the aggregate economy. <owe%er, Ciner #+551$ contests the &indings o& <uang et
al., and argues &or &urther research to produce e%idence &rom international e>uity markets to
support the robustness o& results. In addition to that, "alliaris and Arrutia #166.$ pro%ide
e%idence o& a negati%e reaction o& share prices to the :ersian ,ul& crisis. Other papers
examine the impact o& an oil price &actor in explaining the systematic in&luence on prices in
the stock markets #e.g. <amao, 1676B Caneko and Jee, 166.B 9a&& and @rails&ord, +555$.
Commodity prices has increasingly important role in explaining e>uity markets #see
e.g. Wen et al. #+51+$$. @uyuksahin and 2obe #+515$, /il%ennoinen and 1horpy #+515$ and
1ang and Kiong #+515$ documented how &inancialization o& commodities a&&ects the linear
correlations between di&&erent commodities or the correlations between commodities and
&inancial assets. !%en &urther, @uyuksahin et al. #+515$ showed that commodity and stock
markets could mo%e like a 3market o& one4 in times o& extreme e%ents. In recent papers,
@uyuksahin and 2obe #+515$ and 1ang and Kiong #+515$ both &ound that the commodityL
e>uity correlations signi&icantly increased a&ter /eptember +557.

(%+% Global Stock Market and Stock Prices
@eside macroeconomic %ariables e&&ects, co(mo%ement among the world stock
markets with domestic stock prices has long been suggested. 2ipley #16?*$, :anton et al.
#16?8$, 'wyer and <a&er #1677$, and !un and /him #1676$ using &actor analysis, cluster
analysis, unit root tests, and %ector autoregression respecti%ely already in%estigated this
relationship and came to the conclusion that markets o& %arious countries are related
signi&icantly. Jeon and Chiang #1661$ using uni%ariate and multi%ariate approaches showed
that a common stochastic trend existed in the system o& stock prices in the ;ew Mork,
Jondon, 1okyo, and 9rank&urt exchanges.
8

9urthermore, empirical studies between the well(de%eloped e>uity markets and the
0sian markets are well documented 0ggarwal and 2i%oli #1676$ and Cheung and "ak
#166+$, who obser%ed the day(to(day co(mo%ement o& the A./. market and %arious 0sian
markets. 1his market interdependence relationship has also been examined using daily data
by other researchers #e.g., "aldanado N /aunders, 1671B !rrunza N 2osenberg, 167+B
:hilippatos et al., 167*, !un and 2esnick, 167-$.
1he idea that macroeconomic %ariables explained stock prices has &aced many
contrary arguments recently. /ome researchers argued that stock market mo%ements since the
mid 1665s could not be explained by economic &undamentals. 1his idea has emerged a&ter
A/, !uropean and 0sian markets witnessed unprecedented highs in the mid 1665s but then
&ollowed by sharp re%erse in the +555s as a result o& excessi%e speculation. 0ccording to
Carlson and /argent #166?$ and /hiller #+55.$, the e>uity prices during the second part o& the
1665s in the A/ does not happen due to the change in &undamental %alues such as pro=ected
earnings growth or di%idends but because o& exogenous shocks andFor market irrational
beha%ior. Jee #166., 1667$ and Chung and Jee #1667$ supported this %iew and argued that
&undamental %ariables like discount rates, earnings, di%idends and industrial production did
not explain price mo%ements. In this case, stock return %ariation a&ter 1665s may not be
explained by the notion that stock market is the main indicator o& real economic acti%ity
#9ama, 1665$.
@inswanger #+555, +551, +55-$, gi%ing support to the stock(market bubble hypothesis
to explain the breakdown in the linkage between stock returns and real economic acti%ity in
the A/ in the second part o& the 1665s was. <is paper &ound that real economic acti%ity
explained only a small &raction o& the %ariation in real stock returns in the Anited /tates,
Japan, and the aggregate !uropean economy during the 1665s relati%e to the 1685s and
16?5s. 1he latest paper by Jaopodis #+558$ in%estigated the dynamic linkages among stock
prices, interest rates, in&lation, and economic acti%ity &or the Anited /tates since the 16?5s,
also reports that there was no consistent relationship between real acti%ity and stock prices &or
the last three decades.
1his study intends to examine the impact o& macroeconomic %ariables and market
interdependences, and main commodity prices toward Indonesian stock indices. 1he e%idence
&rom the pre%ious studies &or ma=or markets cannot be replicated on Indonesian data, &or
some reasons as &ollows) 9irst, IndonesiaHs economy is %ery sensiti%e to world market prices
o& its natural resources. 1he structure o& its industry, &ull o& company which has limited
processing ability o& raw materials into end products, ampli&ies this price dependency. 1he
?

companies listed in the exchange comprise o& a large number o& export(dependent industry
that rely on international business cycles and, there&ore, ha%e high &luctuations in their
pro&itability.
/econd, a small e>uity market is also prone with manipulation opportunities that do
not exist in more de%eloped markets, and may draw speculators inside, thereby increasing the
possibilities o& the market reacting inappropriately to new in&ormation. 1he obser%ation could
be explained by both trading noise and the substantial commodity price risk.
1he !C" &ramework we used in this study has ad%antages and shortcomings in
modelling the relationship between stock prices and macroeconomic %ariables. Wen et al.
#+51+$ argues that this model is e&&ecti%e because they are &lexible and are able to control &or
serial correlation in asset returns, and it is easy to obser%e the interactions between %ariables
%ia an impulse(response &unction. <owe%er, !C" also ha%e some disad%antages. 1hey
o&ten assume that asset returns &ollow normal distributions. "oreo%er, they cannot capture
non(linear dependence between %ariables.


+% Data , Methodolo#y
+%' -ariables
1he data used in this study is monthly series data during the period o& January +555 to
'ecember +51+, obtained &rom @loomberg, C!IC, @ank Indonesia and @:/. 1o answer the
research >uestions, we use these &ollowing data as &ollows)


No. Variable Description Source
1 LJCI Jakarta Composite Index Bloomberg
2 LD! "eal ross Domestic !roduct Badan !usat Statistik
# LC!I Consumer !rice Index Bank Indonesia
$ I" !olic% "ate &SBI'BI "ate( Bank Indonesia
) LDJI Do* Jones Index Bloomberg
+ LN,- Nikkei 22) Index Bloomberg
. L/SI /an Seng Index Bloomberg
0 LC12L Coal prices at t3e Ne* Sout3 4ales Bloomberg
5 LC!1 Coal prices at 6ala%sia Bloomberg
17 L1LD old !rice Bloomberg
11 L1IL Crude 1il &4est 8exas Intermediate( !rice Bloomberg
12 L"99" "eal 9::ecti;e 9xc3ange "ate Index C9IC
8able 1
"esearc3 Variables
7

+%(% Descripti*e !nalysis

1he Jakarta Composite Index #JCI$ is a modi&ied capitalization(weighted index o& all
stocks listed on the regular board o& the Indonesia /tock !xchange. 1he index was de%eloped
with a base index %alue o& 155 as o& 0ugust 15, 167+. 0s o& today, it includes more than -55
companies listed in the stock exchange.

.i#"re '% Jakarta Composite Index (JCI) *s% .orei#n Stock Market

9rom the &igure abo%e, it shows that, by 'ecember +51+ Jakarta Composite Index
#JCI$ grows by approximately eight times o& its original le%el in January +555. JCI grows up
16O in the past 1+ months to a historically high -6?7 points le%el. 1his is the most signi&icant
increase compare to other ma=or stock indexes such as 'ow Jones Index #'JI$, ;ikkei ++.
Index #;CM$, and <angseng Index #<I/$ which recorded less impressi%e increase with only
about less than twice during the same time.







7
277
$77
+77
077
1777
J
a
n
<
7
7
J
u
n
<
7
7
N
o
;
<
7
7
2
p
r
<
7
1
S
e
p
<
7
1
=
e
b
<
7
2
J
u
l
<
7
2
D
e
c
<
7
2
6
a
%
<
7
#
1
c
t
<
7
#
6
a
r
<
7
$
2
u
g
<
7
$
J
a
n
<
7
)
J
u
n
<
7
)
N
o
;
<
7
)
2
p
r
<
7
+
S
e
p
<
7
+
=
e
b
<
7
.
J
u
l
<
7
.
D
e
c
<
7
.
6
a
%
<
7
0
1
c
t
<
7
0
6
a
r
<
7
5
2
u
g
<
7
5
J
a
n
<
1
7
J
u
n
<
1
7
N
o
;
<
1
7
2
p
r
<
1
1
S
e
p
<
1
1
=
e
b
<
1
2
J
u
l
<
1
2
D
e
c
<
1
2
JCI vs Foreign Stock Market
(Jan 2000=100)
JCI DJI N,- /IS
6

.i#"re (% Jakarta Composite Index (JCI) *s% Global Commodities

We take monthly obser%ations on the mo%ements o& JCI along with ,lobal
Commodities #Oil, C:O, ,old and Coal$ in &igure +, and &ind that these indexes are mo%ing
along together %ery closely, &or period spanning January +555 to 'ecember +51+, with JCI
and ,old as the ones that seems to ha%e the highest correlation, especially a&ter the period o&
global sub(prime mortgage crises. JCI and Oil also seems to ha%e a %ery signi&icant
relationship due to the &act that Indonesia is oil(producing countries and &ormer members o&
O:!C. Indonesia, a resource(based economy #the main producer o& C:O and Coal$, is highly
in&luenced by the &luctuation o& C:O and Coal prices in both short(run and long(run.

7
277
$77
+77
077
1777
D
e
c
<
5
5
J
u
n
<
7
7
D
e
c
<
7
7
J
u
n
<
7
1
D
e
c
<
7
1
J
u
n
<
7
2
D
e
c
<
7
2
J
u
n
<
7
#
D
e
c
<
7
#
J
u
n
<
7
$
D
e
c
<
7
$
J
u
n
<
7
)
D
e
c
<
7
)
J
u
n
<
7
+
D
e
c
<
7
+
J
u
n
<
7
.
D
e
c
<
7
.
J
u
n
<
7
0
D
e
c
<
7
0
J
u
n
<
7
5
D
e
c
<
7
5
J
u
n
<
1
7
D
e
c
<
1
7
J
u
n
<
1
1
D
e
c
<
1
1
J
u
n
<
1
2
D
e
c
<
1
2
JCI vs Global Commodities
(Dec 1=100)
JCI 1il C!1 Coal old
15

.i#"re +% Jakarta Composite Index (JCI) *s% Macro $conomy


1he data &rom &igure * shows that Indonesia market capitalization to ,': is still
below .5O, and there&ore indicates that Indonesia stock market is under%alued. It is worth
noting that Indonesia ,': growth a%erages around ..* O between January +555 and
'ecember +51+ and are bene&iting &rom the low le%el o& in&lation. 0s shown in the &igure,
o%er the period o& last decade, the central bank has consistently ad=ust the policy rate to
smooth business cycles and shield the economy &rom external shocks &rom global markets.

+%+% Methodolo#y

1his study use ector 0utoregression #02$ model to capture the linear
interdependencies among multiple time series %ariables we describe in section +. 02 was
introduced by Christopher /ims #1675$. 0 uni%ariate autoregression is a single(e>uation,
single %ariable linear model in which the current %alue o& a %ariable is described by its own
lagged %alues. 0 02 is an n(e>uation, n(%ariable linear model in which each %ariable is in
turn explained by its own lagged %alues, plus current and pre%ious %alues o& the remaining n(
1 %ariables.
0ccording to /tock and Watson #+551$, this &ramework pro%ides a systematic way to
capture rich dynamics in multiple time series. In determining the %ariables included in 02,
7
17
27
#7
$7
)7
+7
7
)
17
1)
27
2)
J
a
n
<
7
7
J
u
n
<
7
7
N
o
;
<
7
7
2
p
r
<
7
1
S
e
p
<
7
1
=
e
b
<
7
2
J
u
l
<
7
2
D
e
c
<
7
2
6
a
%
<
7
#
1
c
t
<
7
#
6
a
r
<
7
$
2
u
g
<
7
$
J
a
n
<
7
)
J
u
n
<
7
)
N
o
;
<
7
)
2
p
r
<
7
+
S
e
p
<
7
+
=
e
b
<
7
.
J
u
l
<
7
.
D
e
c
<
7
.
6
a
%
<
7
0
1
c
t
<
7
0
6
a
r
<
7
5
2
u
g
<
7
5
J
a
n
<
1
7
J
u
n
<
1
7
N
o
;
<
1
7
2
p
r
<
1
1
S
e
p
<
1
1
=
e
b
<
1
2
J
u
l
<
1
2
D
e
c
<
1
2
! !
JCI vs Macro "conom#
In:lation D! !olic% "ate 6arket Capitali>ation to D!
11

economic theory and institutional knowledge is re>uired to sol%e the identi&ication as shown
in &igure -.
.i#"re /% &elationships 0etween -ariables


1he standard in 02 is to analyze the results &rom ,ranger(causality tests, impulse
responses and &orecast error %ariance decompositions. In our study the computation o& this
statistics are done using !%iews so&tware.
1o identi&y the best 02 model, we &ollowed standard identi&ication procedures in
&igure .. We identi&y the problem in our study, create rele%ant hypothesis, and collect data.
1hen, we check the stationarity o& our data and per&orm cointegration to select between 02
or !C" models.










1+

.i#"re 1% -!& Identiication







Problem
Identification
START
Hypothesis
Data
Collection
Unit Root Test
(ADF Test)
Data Stasionary
Transformation
Cointe!ration
Test
"ariables
Cointe!rated
"AR #odel
$stimation
"$C #odel
$stimation
Inno%ation
Acco&ntin!
$'D
'o
(es
'o
(es
1*

+%+%'% 2nit &oot 3est
!stimation o& time series econometric models will yield an un(meaning&ul conclusion,
when the data contains a unit root #not stationary$. ;on(stationary series will result in a
spurious regression. 1his condition is characterized by high coe&&icients o& determination,
+
R
, and signi&icant t statistics, but the economic interpretation o& this relationship would be
misleading #!nders, +55-$.
/uppose
t
Y is a stationary time series i& it satis&ies the &ollowing conditions)
1. # $
t y
E Y = time(independent
+.
( )
+
+
# $
t Y t y
Var Y E Y
(
= =
(

time(independent
*. co%# , $
k t t k
Y Y
+
= time(independent

/tationary tests can be conducted by %arious methods such as graphics, correlogram
and the unit root test. 1he unit root test is employed in this research. 1wo methods o&
unit root test that are commonly used are the 0'9 #0ugmented 'ickey 9uller$ test
and the :: #:hillips :eron$ 1est.

+%+%(% !"#mented Dickey ."ller 3est
1he rationale o& 'ickey 9uller test is to examine whether, or not, a time series is a
random walk.
Consider 02#1$ model
( )
1 1
*.1
t t t
Y Y

= + +


I&
1
1 = , then the model abo%e becomes random walk. 2andom walk is one &orm o&
non(stationary time series. In e>uation #*.1$ subtract
1 t
Y

to both sides, thus,


( )
1 1 1
1
1
1
#*.+$
where, 1
t t t t
t t t
Y Y Y
Y Y



= + +
= + +
=

1-

9rom the e>uation abo%e, I can the make hypothesis

5 1
1 1
< ) 1
< ) 1

=
<

1he techni>ue is by making regression between
1
and
t t
Y Y

. <ence, we ha%e
coe&&icient o& regression, =
)
*
* ), where then signi&icance tests are employed based on that
hypothesis. 1he 'ickey 9uller

statistic is as &ollows)
P 1
P std. error# $

= . I& the absolute %alue


o& the

statistic is larger than absolute %alue o& 'ickey 9uller critical %alue, then the null
hypothesis is re=ected, implying that time series data is stationary. I& the absolute %alue o& the


statistic is smaller than absolute %alue o& 'ickey 9uller critical %alue, then null hypothesis
is not re=ected implying that the data is not stationary.
1he 'ickey 9uller test has been expanded on the model o& 02 #p$, re&erred to as the
0ugmented 'ickey 9uller test.
1 1
+
#*.*$
p
t t i t i t
i
Y Y Y
+
=
= + + +


1
where ... 1

p
p
i j
j i


=
= + +
=



9rom e>uation #*.*$ we can be build hypothesis
5
1
< ) 1
< ) 1
p
i
i
p
i
i

=
<


1he 0ugmented 'ickey 9uller test could be conducted by &orming regression between
1 1
, and , +,..., .
t t t i
Y Y Y i p
+
= Qthus, obtaining the coe&&icient regressionQ
P
P 1
p
i
i
=

, and
per&orming a signi&icance test based on hypothesis abo%e.
1he test statistic, 'ickey 9uller

statistic is obtained by)
1.


P
1
'ickey 9uller.
P
std. error# $
p
i
i
p
i
i



I& absolute %alue o&

statistic is larger than absolute
%alue o& 'ickey 9uller critical %alue then re=ection o& null hypothesis implies that time series
data is stationary, meanwhile i& absolute %alue o&

statistic is smaller than absolute %alue o&
'ickey 9uller critical %alue then null hypothesis is not re=ected.

+%+%+% 0i*ariate -!& system with order p
02 with order p o& bi%ariate system or two %ariables
t
y
1
+
=
| |
|
\
t
t
y
y
can be de&ined as
t 1 t 1 p t p t
...

= + + + + y y y e #*.-$
where

1
+

| |
=
|
\
is two(dimension %ector,
11, 1+,
+1, ++,
, 1, +,...,


(
= =
(

i i
i
i i
i p is ( ) + + coe&&icient matrix and
e
1
+
t
t
e
e
| |
=
|
\
is a white noise %ector. In other words)
1$
t
e has zero mean, | |
t
E e = 4
+$
t
e has constant %ariance,
t
,
T
t e
E e e t ( = E


*$ and e
t s
e are not correlated, &or t s = .
!>uation #*.1+$ can be written as &ollows)
11, 1+, 1 11,1 1+,1 11,+ 1+,+ 1 1 1 1 + 1 1
+1, ++, + +1,1 ++,1 +1,+ ++,+ + + 1 + + + +
...
p p t p t t t t
p p t p t t t t
y y y y e
y y y y e




( ( ( ( ( ( ( ( (
= + + + +
( ( ( ( ( ( ( ( (


1wo(dimension random %ector
t 1 t t 1
..., , , ,...
+
y y y is a stochastic process %ector. 0
stochastic process %ector is stationary i&
18

1$ | | ! , t =
t
y
+$
( ) ( )
( (
co%# , $ ! R # $, t dan 5,1,+,... = = I = (
t t h t t h
h h
y
y y y y
Implication o& the second property is that &or hD5,
t
y has the same co%ariance matrix,
that is
( )( ) R ,
t t h
E t

= E (


y
y y .

+%+%/% Cointe#ration 3est
JohansenHs cointegration test is based on the 02#p$ model o& non(stationary
%ariables. 9or simpler Johansen test procedure, 02#1$ model will be used. 2emember that
02 #1$ model is noted in matrix notation)
1 1 t t t
Y Y

= H +

In JohansenHs cointegration test, analysis o& %ariables is not only &ocused on the result
o& 02 e>uation system #Impulse 2esponse 9unction and ariance 'ecomposition are the
most commonly used, as pre%iously disucssed$, but also considered a stepping stone &or the
next cointegration test, whereby reparameterization need to be done &rom 02#1$ model to
"odel ector !rror Correction #!C"#1$$.
1he ,ranger theorem ensures the existence o& an error correction representation in a
cointegrated regression. @ased on this theorem, e>uation 02#1$ can be represented in the
&orm o& !C" as &ollows)

1 1
1 1 +
where)
and
t t t
t t
Y Y
Y Y Y I

= H +
= H = H


1his !C" #1$ &orm contains in&ormation about short(run and long(run changes stated by
parameter and . 1his "atrix will be &urther used to determine whether regression
system is cointegrated. 1his is the core o& Johansen test procedure in analyzing the
cointegration relationship between obser%ed %ariables.
9or instance, a component o& %ector 5
t
is a &irst order integration or written as I#1$,
then H HH HY
t-1
is a linear combination o& %ariable Y
t-1
I#1$. In order to estimate all combination
i
I H H
1?

possibilities &rom M
t(1
which results in close correlation with Y
t-1,
a stationary element,
Johansen uses H HH H matrix characteristics as &ollows)
'% I& 2ank#H HH H$D5, then, there is no cointegration between %ariables
(% I& 2ank#H HH H$Dm #m ) the number o& %ariables in 02 model$, then all %ariables are
cointegrated
+% I& 5 S 2ank#H HH H$ Sm, then 2ank #H HH H) states the number o& %ariables that are
cointegrated between 5 and m.

"atrix H HH H can be decomposed to H HH H =
T
where is speed o& ad=utsment and is long(run
coe&&icient matrix so that
T
5
t6'
up to m(1 combinations is a cointegrated relationship which
ensures that 5
t
reaches long(run e>uilibrium. 9urther, 2ank #
T
$ can be determined by
calculating eigen%alue &rom
T
.

+%+%1% Imp"lse &esponse ."nction
0n impulse response &unction aims to obser%e the e&&ects o& a one standard de%iation
shock to one o& the inno%ations on current time %alues and the &uture %alues o&
endogenous %ariables included in the model.
!nders #+55-$ illustrates impulse response by employing bi%ariate 02 as &ollows)
1 1 1 11 1+
5 + + + +1 ++

=
( ( ( (
= +
( ( ( (

i
t t i
i t t i
y e y a a
y e y a a

1he residual can be expressed as)
( )
1 1 1+
+ + +1 1+ +1
1
1
1 1

( ( (
=
( ( (


t t
t t
e b
e b b b


1hen, by combining two e>uations abo%e we can ha%e "o%ing 0%erage
representation as &ollows)
11 +1
1+ ++
1 1 1
5 + + +
# $ # $
# $ # $
t t i
i t t i
i i
i
y y
i y y


=
( ( (
= +
( ( (

(
(


17


1he &our coe&&icientsB
11 1+ +1 ++
# $, # $, # $ dan # $ i i i i are called impulse response
&unctions.

+%+%7% 3he Cholesky Decomposition
1he Cholesky decomposition or commonly known as %ariance decomposition
pro%ides in&ormation about the relati%e importance o& each %ariable in the 02 system
according to the shocks. 1he Cholesky 'ecomposition method is another strategy to describe
the dynamic system contained in 02 by collecting estimates o& error %ariance o& a %ariable,
or, the di&&erence amount between the %ariance be&ore and a&ter shock. @oth shocks
originated &rom that %ariable itsel& and shocks o& other %ariables. ariance decomposition is
used to predict the %ariance percentage contribution o& each %ariable due to changes in certain
%ariables in the 02 system.
!nders #+55-$ demonstrates mathematically the mechanism o& %ariance
decomposition by &irst building n(step ahead &orecast error as &ollows)
11 11 1 11 1 1+ 1+ 1 1+ 1
#5$ #1$ ... # 1$ #5$ #1$ ... # 1$
t n t t n yt n yt n yt zt n zt n zt
y E y n n
+ + + + + + + +
= + + + + + + +

1he %ariance o& the n(step ahead &orecast error %ariance o&
t n
y
+
can be obtained

+ + + + + + + + +
11 11 11 1+ 1+ 1+
# $ #5$ #1$ ... # 1$ #5$ #1$ ... # 1$
y y z
n n n ( ( = + + + + + + +


@ecause all
+
# $
jk
i are nonnegati%e then %ariance o& the &orecast error increases as
&orecast horizon n increases. :roportions o&
+
# $
y
n due to shocks in
{ }
yt
and { }
zt

se>uences are

+ +
11 11 1 11 1
+
#5$ #1$ ... # 1$
# $
y yt n yt n yt
y
n
n

+ + +
( + + +


and
16

| |
+
1+ 1+ 1 1+ 1
+
#5$ #1$ ... # 1$
# $
z zt n zt n zt
y
n
n

+ + +
+ + +

/% &es"lts !nalysis
/%'% 2nit &oot 3est
1he test employed is 0ugmented 'ickey 9uller #0'9$ test.
1he model used in 0'9 test is)

1 1
+
p
t t i t i t
i
Y Y Y
+
=
= + + +


9rom the abo%e mode, a hypothesis can be &ormulated)
;on(stationary data B stationary data
5
5
< ) 1
#;on stationary data$
< ) 1
#/tationary data$
p
i
i
p
i
i

<


/tatistical 1est)
P
1
D 'ickey 9uller
std. error
p
i
i
p
i
i

| |
|
\


/igni&icance) D .O
Decision r"le in !D. testin#8
I& the T statistic %alue is smaller than the 'ickey 9uller critical %alue, then the null
hypotesis is re=ected, indicating that the time series data is stationary.









+5

3able (% 2nit &oot 3est (!"#mented Dickey ."ller)


9rom 1able +, we can say that we cannot re=ect the null hypothesis that all %ariables
contain a unit root at the .(percent signi&icant le%el, suggesting that the natural logarithm o&
all %ariables in our study are I#1$.

/%(% )a# )en#th Determination in -!&
In this section, the 0IC and /IC criterion are used in determining the optimal lag
length in a 02 model.
( )
( )
# $ ln +
# $ ln ln# $
1 D number o& obser%ation
//2 D /um />uare 2esidual
k lag length
> number o& regressor k 1 number o& estimated parameter
!!R k
"IC k T #
T
!!R k
!C k T # T
T
| |
= +
|
\
| |
= +
|
\
=
= = + =

'etermination o& optimal lag used by the researcher in order to estimate a short run
e>uation is based on 0kaike In&ormation Criterion #0IC$. 1he criterion o& optimal lag
in&ormation can be seen in 1able + below.
Intercept
8rend ?
Intercept
None Intercept
8rend ?
Intercept
None
1 LJCI 7.5255 7.175+ 7.501# 7.7777 7.7777 7.7777
2 LD! 7.5551 7.)+55 1.7777 7.7777 7.7771 7.2)#)
# LC!I 7.0.77 7.$0+. 1.7777 7.7777 7.7777 7.#10#
$ I" 7.)01. 7.10#5 7.#202 7.7777 7.7777 7.7777
) LDJI 7.$#+7 7.$011 7...1. 7.7777 7.7777 7.7777
+ LN,- 7.11)$ 7.$2$2 7.#22# 7.7777 7.7777 7.7777
. L/SI 7.+5.1 7.$#52 7.07$7 7.7777 7.7777 7.7777
0 LC12L 7.$#$. 7.7)15 7.0.)5 7.7777 7.7777 7.7777
5 LC!1 7.+))) 7.#0+1 7.0.17 7.7777 7.7777 7.7777
17 L1LD 7.5.+2 7.7111 7.5550 7.7777 7.7777 7.7777
11 L1IL 7.+15$ 7.1).) 7.0)2+ 7.7777 7.7777 7.7777
12 L"99" 7.)771 7.11)5 7.02#5 7.7777 7.7777 7.7777
No. Variable
Le;el =irst Di::erence
@nit "oot 8est &! Value(
+1

3able +% Comparison between se*eral model selection criterion

0ccording to 1able * abo%e, it can be seen that the optimal lag based on 0IC is lag 7.

/%+% Cointe#ration 3est
1he purpose o& cointegration test is to assess similarities o& mo%ement and
relationship stability between %ariables in a long(run. When a data series contains a unit root
and integrated to the same order, cointegration test can be per&ormed to assess the existence
o& cointegration. In this research, the JohansenHs Cointegration 1est method is employed. 0n
in&luential relationship can be seen &rom the cointegration that exists between %ariables.
When a cointegration exists between %ariables, this implies that in&luential relationship
occurs throughout %ariables and in&ormation is parallelly distributed.
1he JohansenHs Cointegration 1est indicates that a cointegrating %ector exists, or at least a
linear independent combination exists &rom the %ariables contained in the model. 1he
conse>uence is that alternati%e hypothesis which states the presence o& cointegration
relationship can be accepted.






$ag $og$ $% F&" 'IC SC ()
7 1770.701 N2 2.119<21 <1#.))212 <1#.#70 <1#.$)25#
1 #1+7.)2+ #52$.10+ 2.009<## <$7.0..5 <#...7$#0A <#5.)00$.A
2 ###).055 251.75)) 1.5)e<##A <$1.#7$.) <#).2710# <#0.02)7.
# #$)0.210 10#.7+1. 2.0.9<## <$1.775.+ <#1.5..$$ <#..##50#
$ #)5).1. 102.+720 #.079<## <$7.51#0. <20.5)21$ <#+.7)#+5
) #..$.#7. 275.+7#2 #.2)9<## <$1.#515$ <2+.)770 <#).#$1)1
+ #5)7.1+$ 1...7)2)A #.)#9<## <$1.02)#) <2$.77$02 <#$.)0$+0
. $170.0$1 1##.0)7# +.$)9<## <$2.72)7) <21.2.)11 <##.)5$12
0 $#25.#5. 1)7.7#0) ...+9<## <$#.7+++#A <15.#0.25 <##.$$)$+
++

3able /% 3race 3est &es"lts

1his cointegration in 1able -. test is based on trace test, since the %ariables do not
ha%e normal distributions. 1he Johansen Cointegration 1est o& %ariables indicates the
existence o& ele%en cointegration e>uations. ;ot all %ariables are stationary in le%elB
there&ore, there is a cointegration among %ariables. 1hus, estimation model by !C" can
generate stationary estimation and errors. Cointegration test result indicates that research
%ariable has long(term relation. It can be concluded that the next step o& analyzing short(run
analysis between research %ariable in long(term can be executed.
/%/% )on#6&"n Model
In the long(run #with the use o& cointegrating %ectors interpretation$, the &ollowing
model in 1able - can be constructed. We also compute error correction %ariable &rom the
cointegrating relationship between the %ariables.











None A 7...)$)$ 1701.#+) ##$.50#. 7
2t most 1 A 7..$52$. 0+#.2001 20).1$2) 7
2t most 2 A 7.+5).00 ++1.#20# 2#5.2#)$ 7
2t most # A 7.)$5.11 $0..)0#. 15..#.75 7.7771
2t most $ A 7.)27).) #.1.75)$ 1)5.)25. 7
2t most ) A 7.#..+1 2+#..+70 12).+1)$ 7
2t most + A 7.##+7+# 15$.)252 5)..)#++ 7
2t most . A 7.2)22#0 1#$..#2# +5.01005 7
2t most 0 A 7.227721 52.25$#+ $..0)+1# 7
2t most 5 A 7.10.77+ )+.71)7+ 25..5.7. 7
2t most 17 A 7.1)01$5 2)..00$1 1).$5$.1 7.771
2t most 11 7.77$$.1 7.+)$225 #.0$1$++ 7.$10+
@nrestricted Cointegration "ank 8est &8race(
8race
Statistics
7.7)
Critical Value !rob.AA 9igen;alue
/%pot3esi>ed
No. o: C9&s(
+*

3able 1% -$CM )on#6&"n Model














$JCI(*1) $GD&(*1)
C <2#.)7)1$ 5.####$2
LC!I&<1( 2.1.+.#0 7.#+0).
B<..#)2+0C B<22.17$+C
"289&<1( 7.72+.+1 <7.77##)2
B<#.7712#C B +.+.$+$C
LD14&<1( 2.7)2)) 7.7++#25
B<17.))+.C B<+.7).7)C
LNI,,9I&<1( <1.$5$+72 <7.72$15)
B ..02++0C B 2.2$5+2C
L/2N&<1( 1.2$17+5 7.71$75#
B<0.#5#51C B<1.+52#$C
LC12L&<1( 7.2+21#$ <7.7)2#.
B<$.2+$#+C B 1).12+#C
LC!1&<1( <7.#$.275 <7.77$+)5
B 2.0$155C B 7.+..11C
L1LD&<1( <1.7#.#). 7.27021)
B +.27.$5C B<22.1227C
L1IL&<1( 7.#+2+22 <7.7#)5$.
B<).701#+C B 0.5$#$.C
LD"289&<1( 2.21)12$ <7.17.)+)
B<5.17)2)C B ..0)7##C
+-

/%1% Gran#er Ca"sality 3est

3able 7% Gran#er Ca"sality 3est &es"lts

U ;ull hypothesis re=ected at 1O signi&icance le%el
UU;ull hypothesis re=ected at .O signi&icance le%el
UUU ;ull hypothesis re=ected at 15O signi&icance le%el

1he ,ranger(causality test is conducted to study the lead(lag relationships between
JCI, macroeconomic %ariables, global stock markets and commodity prices. 1he results are
reported in 1able 8 abo%e. "acroeconomic %ariables, namely, JC:I and J,': are &ound to
be the most important %ariables in determining the JCI per&ormance when they were
considered in pairs with the JCI using the ,ranger causality test. 1he results also indicate that
the pricesH &luctuation o& global commodity such as ,old, Oil, C:O and Coal does not
signi&icantly a&&ect the per&ormance o& JCI. 9urthermore, the mo%ement o& global stock
markets such as 'ow Jones, ;ikkei and <ang(/eng signi&icantly determine the &luctuations
o& JCI.
Null /%pot3esisE =<Statistic !robabilit%
LJCI does not ranger Cause LC12L +.1..0$A 7.772+$
LC12L does not ranger Cause LJCI 1.5++0 7.1$#$.
LJCI does not ranger Cause LC!I 1.2).. 7.20.2)
LC!I does not ranger Cause LJCI $..5721A 7.775+1
LJCI does not ranger Cause LC!1 +.2127)A 7.772))
LC!1 does not ranger Cause LJCI 7.#)5+1 7.+50)$
LJCI does not ranger Cause LD14 1..51.1 7.1.710
LD14 does not ranger Cause LJCI ).15#$$A 7.77+)0
LJCI does not ranger Cause LD! 2.012##AAA 7.7+#15
LD! does not ranger Cause LJCI 2..5$$#AAA 7.7+$25
LJCI does not ranger Cause L1LD 2.1.500 7.11+)0
L1LD does not ranger Cause LJCI 2.72+$$ 7.1#)#)
LJCI does not ranger Cause L/2N $.25025AA 7.71)20
L/2N does not ranger Cause LJCI ..2..7+A 7.7775+
LNI,,9I does not ranger Cause LJCI ).22).5A 7.77+$1
LJCI does not ranger Cause LNI,,9I 7.#.520 7.+0)71
L1IL does not ranger Cause LJCI 1.+.$#5 7.1570+
LJCI does not ranger Cause L1IL 0..#+#$A 7.7772+
LD"289 does not ranger Cause LJCI 1.$$$) 7.2#570
LJCI does not ranger Cause LD"289 5.1.)20A 7.7771.
"289 does not ranger Cause LJCI 1.$12#$ 7.2$+.)
LJCI does not ranger Cause "289 2.12.02 7.122+#
+.

/%7% Imp"lse &esponse ."nction
0n impulse response &unction states the e&&ect o& one standard de%iation shock to one
o& the inno%ations on current time %alues and &uture %alues o& endogenous %ariables. 0 shock
&rom endogenous %ariable directly in&luences the %ariable itsel&, which then in&luences other
endogenous %ariables through the dynamic structure o& 02 and !C. I29 pro%ides
direction and magnitude o& the e&&ect between endogenous %ariables as it demonstrates the
in&luence o& one(standard de%iation endogenous %ariable shock on other endogenous
%ariables and the %ariable itsel&. 1here&ore, with new in&ormation coming up, any shock that
occur in a %ariable, will a&&ect the %ariable itsel& and other %ariables in a system. Impulse
2esponse 9unction on research %ariables &or 15 upcoming period is presented below.

.i#"re 7% Imp"lse &esponse ."nction

1o obtain additional insights into the mechanism o& transmissions o& stock market
mo%ements, we now examine the pattern o& dynamic responses o& JCI to inno%ation &rom
each %ariable. 0s can be seen &rom the table abo%e, the impulse response o& the JCI to a ,':
+,-.
+,-/
,--
,-/
,-.
)- 0- 1- /- 2- 3- 4- .- 5- )--
Response of 67CI to 68DP
+,-.
+,-/
,--
,-/
,-.
)- 0- 1- /- 2- 3- 4- .- 5- )--
Response of 67CI to 6CPI
+,-.
+,-/
,--
,-/
,-.
)- 0- 1- /- 2- 3- 4- .- 5- )--
Response of 67CI to RAT$
+,-.
+,-/
,--
,-/
,-.
)- 0- 1- /- 2- 3- 4- .- 5- )--
Response of 67CI to 6D9:
+,-.
+,-/
,--
,-/
,-.
)- 0- 1- /- 2- 3- 4- .- 5- )--
Response of 67CI to 6'I;;$I
+,-.
+,-/
,--
,-/
,-.
)- 0- 1- /- 2- 3- 4- .- 5- )--
Response of 67CI to 6HA'8
+,-.
+,-/
,--
,-/
,-.
)- 0- 1- /- 2- 3- 4- .- 5- )--
Response of 67CI to 6C9A6
+,-.
+,-/
,--
,-/
,-.
)- 0- 1- /- 2- 3- 4- .- 5- )--
Response of 67CI to 6CP9
+,-.
+,-/
,--
,-/
,-.
)- 0- 1- /- 2- 3- 4- .- 5- )--
Response of 67CI to 6896D
+,-.
+,-/
,--
,-/
,-.
)- 0- 1- /- 2- 3- 4- .- 5- )--
Response of 67CI to 69I6
+,-.
+,-/
,--
,-/
,-.
)- 0- 1- /- 2- 3- 4- .- 5- )--
Response of 67CI to 6<RAT$
Response to Choles=y 9ne S,D, Inno%ations
+8

implies that JCI responds most strongly to ,': on month 15 and +5 when the ,': shocks
occurs. 1his may re&lect a low degree o& economic and &inancial integration and the &ree
in&ormation between the real and capital sectors. 1o some extent, JCI also reacts to ,':
without lag. @ecause the two %ariables operate with a lag, this result is as expected.
9urther examination o& &igure abo%e re%eals some interesting patterns. 0s can be seen
&rom the &igure, JCI responds positi%ely to 'ow Jones and ,old while JCI responds
negati%ely to <ang /eng, C:O, oil, in&lation and gold. 0lthough the responses are somewhat
sluggish between 15(+5 months, but JCI continue to react noticeably to these %ariable shocks
a&terward.
1he inno%ation analyses suggest that the JCI does not interact with their own key
macroeconomics &actors. "ost o& the %ariations in the index could not be captured by
inno%ations in macroeconomic indicators such as ,': growth, interest rate. 1he causal
relationships that macroeconomics %ariables granger cause stock prices are not >uantitati%ely
supported by the inno%ation analyses.
1he increase in global commodities prices is responded negati%ely by JCI, with
declining JCI per&ormance. 0ny global commodities #coal, oil and C:O$ price increase will
tend to ha%e negati%e e&&ect on stock markets. Oil price increase will lead to higher cost o&
productions, as oil is one o& important production &actors. 1he increase cost will be
trans&erred to the consumers, which will, in turn, lead to lower demand and thus consumer
spending, due to higher consumer prices. Jower consumption could lead to lower production
and thus might increase unemployment. /tock market would react negati%ely in such case.
<owe%er, the ele%ation o& gold price is positi%ely responded by JCI.







+?

/%9% -ariance Decomposition

3able 9% -ariance Decomposition o JCI &es"lts

1able ? reports the %ariance decomposition results &or the e&&ect o& %arious shocks
in%ol%ed in the !C" model on the JCI. 0ll !C" models include 7 lags. 1he numbers
reported indicate the percentage o& the &orecast error in each %ariable that we can attribute to
each o& the structural inno%ations at di&&erent horizons #&rom 1 month to 155 months$. We
report the percentages &or selected &orecast horizons o& multiple ten.
1he decomposition results show that e%en in the long(run #i.e the 155(month &orecast
horizon$, ,':, C:I and Interest 2ate contribute a relati%ely small share to the %ariation o&
JCI. On the other hand, 'ow Jones and Oil :rices are the shocks that produce the highest
%ariation in the JCI. In the short run #e.g., 1(month &orecast horizon$, JCI, 'ow Jones, and Oil
:rices produces -?O, 1?O, and +5O o& the %ariation o& JCI, respecti%ely. !xtending the
&ocus to the long(run #e.g. 155(month &orecast horizon$, JCI, 'ow Jones, and Oil :rices
generates 16O, ++O, and 17O o& the %ariation o& JCI.

/%:%.orecastin#

1his study computed multistepLahead &orecast o& JCI by iterating &orward the reduced
&orm !C". @ecause the ultimate test o& a &orecasting model is out(o&(sample per&ormance,
this study &ocus on out(&o(sample per&ormance o& JCI &rom period January +51* to 'ecember
+51*. 0ccording to the &orecast, JCI will increase steadily to .**1.. in June +51* and .?+6.7
in 'ecember +51*.

!eriod S.9. LJCI LD! LC!I "289 LD14 LNI,,9I L/2N LC12L LC!1 L1LD L1IL LD"289
1 7.7).$50 177 7 7 7 7 7 7 7 7 7 7 7
17 7.#1$2$2 $..+$)1) 7.)1#5)$ 1.0#5..2 7.+)$)$2 1..252$# 1.75#$1) 1.$7#275 1.$.21.$ 2.5.)#2) 2.#+$02. 27.+.1#) 2.7.#0)+
27 7.$+.52# 2$.$.7$+ 1.2$$.$$ 1..$5#.) 7..0770 21.00..1 1.$0$#55 $..7.77+ 1.557)70 12..1+$5 +.227521 21.)$12. 1.27.7#0
#7 7.)))).0 22.2#7#) 1.17$52. 2.#+.21 7.+7+1+2 2#.1)+1+ 1.2#0$10 ).07.$)+ 2.5272## 11.2.15+ 5.22$+22 15.152$1 7.00775)
$7 7.+$271. 22.1)5+ 7.0.5+)0 2.1)0$72 7.)).$20 22.7...1 7.501.7$ +.07#20# #..$#0# 5..5111. 17.+1##2 15.)20+) 7..7)#70
)7 7..11+2. 21.$7)52 7.57$751 2.2+.22. 7.)7$+01 21.$0$22 7.07)27$ ..)$2+22 #.52...2 17.77#.) 17.507+$ 15.).++1 7.)5.2+#
+7 7...#20+ 27.$$251 7.0$75#$ 2.)1)$$5 7.$)57)) 21.0057. 7.+5175+ ..5#+7++ #.05272) 17.#17+$ 11.215$ 15.25).# 7.)7.+22
.7 7.0#27+5 27.770#2 7..$5)#) 2.+1+)2# 7.$#1#0) 22.2)20. 7.)50))) 0.1#.7.) #.5.0$0) 17.727+1 11.+01+. 15.70+25 7.$#0+5+
07 7.00027$ 15...172 7.+00++$ 2.+1).10 7.$10+11 22.27#0) 7.)2)+.2 0.#07705 $.1#2$12 5...+.+2 12.7)71) 15.7$+$+ 7.#57)51
57 7.5#55.$ 15.)#.## 7.+$..52 2.+$2))5 7.$7)7.5 22.12$0+ 7.$+5... 0.+7.)7. $.21$#$) 5..+1#$2 12.225## 15.77)+ 7.#)$$0
177 7.500.15 15.25.50 7.+12#2 2.+5#$20 7.#57.$2 22.15#). 7.$2)2+0 0..))#$) $.2#5++1 5...212 12.#)272 10.5$)#2 7.#2222)
C3olesk% 1rderingE LJCI LD! LC!I "289 LD14 LNI,,9I L/2N LC12L LC!1 L1LD L1IL LD"289
Variance Decomposition o: LJCI
+7

.i#"re 9% JCI .orecast &es"lts


1% Concl"sions remarks
Our paper pro%ides the latest examination o& the e&&ect o& macroeconomic %ariables,
global index, and commodity prices on JCI. Asing the !C" methodology, we compute 15
di&&erent structural shocks &or the JCI.
1he results show that the impact o& JCI, 'ow Jones, and Oil price &actor shocks play a
signi&icant role in explaining the ad=ustments in JCI. 9urther, the ,ranger temporal causality
tests suggest a strong role &or idiosyncratic C:I, 'ow, ,':, <ang /eng and ;ikkei shocks
leading to JCI, whereas the Coal, Interest 2ate, !xchange 2ate shocks do not lead the JCI.
Our results &orecast that JCI will keep increasing e%en until 'ecember +51* period.
9uture research e&&orts could also in%estigate the e&&ect o& shocks on JCI across
di&&erent sector &or a panel o& countries. 1he empirical &indings will pro%e use&ul to in%estors
who need to understand the exact e&&ect o& macroeconomic %ariable, global stock market, and
commodities prices changes on JCI across industries.





7
1777
2777
#777
$777
)777
+777
.777
J
a
n
<
7
)
J
u
l
<
7
)
J
a
n
<
7
+
J
u
l
<
7
+
J
a
n
<
7
.
J
u
l
<
7
.
J
a
n
<
7
0
J
u
l
<
7
0
J
a
n
<
7
5
J
u
l
<
7
5
J
a
n
<
1
7
J
u
l
<
1
7
J
a
n
<
1
1
J
u
l
<
1
1
J
a
n
<
1
2
J
u
l
<
1
2
J
a
n
<
1
#
J
u
l
<
1
#
JCI Forecast
+6

7% &eerences

0ggarwal, 2., N 2i%oli, :. #1676$. 1he relationship between the A/ and &our 0sian stock
markets. 0/!0; !conomic @ulletin 8#1$, 115L11?.
0sprem, "., 1676. /tock prices, asset port&olios and macroeconomic %ariables in ten
!uropean countries. Journal o& @anking and 9inance 1*, .76L81+.
@arro, 2.J., 1665. 1he stock market and in%estment. 2e%iew o& 9inancial /tudies *, 11.L1*1.
@eckers, /., ,rinold, 2., 2udd, 0., /te&ek, '., 166+. 1he relati%e importance o& common
&actors across the european e>uity markets. Journal o& @anking and 9inance 18, ?.L6..
@inswanger, "., +555. /tock market booms and real economic acti%ity) is this time di&&erentV
International 2e%iew o& !conomics and 9inance 6, *7?L-1..
@inswanger, "., +551. 'oes the stock market still lead real acti%ityV 0n in%estigation &or the
,(? countries. Journal o& 9inancial "arkets and :ort&olio "anagement 1., 1.L+6.
@inswanger, "., +55-. <ow important are &undamentalsV !%idence &rom a structural 02
model &or the stock markets in the A/, Japan and !urope. Journal o& International
9inancial "arkets, Institutions and "oney 1-, 17.L+51.
@rown, /., Otsuki, 1., 1665. "acroeconomic 9actors and the Japanese "arkets) 1he C0:"
:ro=ect. In) !lton, !.J., ,ruber, ".J. !ds., Japanese Capital "arkets. <arper and 2ow,
;ew Mork, pp. 1?.L16+.
@uyuksahin, @., <aigh, "., 2obe, "., +515. Commodities and e>uities) e%er a 3"arket o&
One4V J. 0ltern. In%estm. 1+, ?.L6..
@uyuksahin, @., 2obe, "., +515. /peculators, commodities and cross(market linkages.
Working :aper.
Cano%a, 9., 'e;icolo, ,., 166.. /tock markets and real acti%ity) a structural approach.
!uropean !conomic 2e%iew *6, 671(151..
*5

Carlson, J.@., /argent, C.<., 166?. 1he recent ascent o& stock prices) can it be explained by
earnings growth or other &undamentalsV 9ederal 2eser%e @ank o& Cle%eland !conomic
2e%iew **, +L1+, +nd >uar.
Chen, ;., 2oll, 2., 2oss, /.0., 1678. !conomic &orces and the stock market. Journal o&
@usiness .6, *7*L-5*.
Cheung, M. J., N "ak, /. C. #166+$. 1he international transmission o& stock market
&luctuation between the de%eloped markets and the 0sian(:aci&ic markets. 0pplied
9inancial !conomics +, -*L-?.
Cheung, M.W., <e, J., ;g, J.C., 166?a. Common predictable components in regional stock
markets. Journal o& @usiness and !conomic /tatistics 1., *.L-+.
Cheung, M.W., <e, J., ;g, J.C., 166?b. What are the ,lobal /ources o& 2ational ariation in
International !>uity 2eturnsV Journal o& International "oney and 9inance 18, ;o. 8,
pp.7+1(7*8,166?
Chung, <., Jee, @.(/., 1667. 9undamental and non&undamental components in stock prices o&
:aci&ic(2im countries. :aci&ic(@asin Journal o& 9inance 8, *+1L*-8.
Ciner, C., +551. !nergy shocks and &inancial markets) nonlinear linkages. /tudies in
;onlinear 'ynamics and !conometrics . #*$, +5*L+1+.
Cornell, @., 167*. 1he money supply announcement puzzle) re%iew and interpretation.
0merican !conomic 2e%iew ?*, 8--(8.?.
'wyer, ,. :., N <a&er, 2. W. #1677$. 0re national stock markets linkedV 9ederal 2eser%e
@ank o& /an 9rancisco !conomic 2e%iew ?5#8$, *L1-.
!rrunza, ., N 2osenberg, @. #167+$. In%estment risk in de%eloped and less de%eloped
countries. Journal o& 9inancial and Wuantitati%e 0nalysis 1?, ?-1L?8+.
!un, C., N 2esnick, @. #167-$. !stimating the correlation structure o& international share
prices. Journal o& 9inance *6, 1*11L1*+-.
*1

!un, C., N /him, /. #1676$. International transmission o& stock market mo%ements. Journal
o& 9inancial and Wuantitati%e 0nalysis +-, +-1L+.8.
9a&&, 2., @rails&ord, 1., +555. 0 test o& a two(&actor Imarket and oilH pricing model. :aci&ic
0ccounting 2e%iew 1+ #1$, 81L??.
9ama, !.9., 1671. /tock returns, real acti%ity, in&lation, and money. 0merican !conomic
2e%iew ?1, .-.L.8..
9ama, !.9., 1665. /tock returns, expected returns, and real acti%ity. Journal o& 9inance -.,
1576L1157.
9erson, W., <ar%ey, C., 166*. 1he risk and predictability o& international e>uity returns.
2e%iew o& 9inancial /tudies 8, .+?L.88.
9erson, W.!., <ar%ey, C.2., 1661. 1he %ariation o& economic risk premiums. Journal o&
:olitical !conomy 66, *7.L-1..
,an, C., Jee, "., Mong, <. <. 0., N Xhang, J. #+558$. "acroeconomic %ariables and stock
market interactions) ;ew Xealand e%idence. In%estment "anagement and 9inancial
Inno%ations, *#-$, 76(151.
,=erde, Y., N /aettem, 9. 1666. Causal relations among stock returns and macroeconomic
%ariables in a small, open economy. Journal o& International 9inancial "arkets,
Institutions and "oney, 6#1$, 81(?-.
<amao, M., 1677. 0n empirical examination o& arbitrage pricing theory) Asing Japanese data.
Japan and the World !conomy 1, -.L81.
<uang, 2.'., "asulis, 2.W., /toll, <.2., 1668. !nergy shocks and &inancial markets. Journal
o& 9utures "arkets 18, 1L+?.
Jeon, @. ;., N Chiang, 1. C. #1661$. 0 system o& stock prices in world stock exchanges)
common stochastic trends &or 16?.L1665V Journal o& !conomics and @usiness -*, *+6L
**7.
*+

Caneko, 1., Jee, @./., 166.. 2elati%e importance o& economic &actors in the A./. and
Japanese stock markets. Journal o& the Japanese and International !conomies 6, +65L*5?.
Cim, ".C. and C. Wu, 167?, "acro economic &actors and stock returns, Journal o& 9inancial
2esearch 15, 7?(67.
Jaopodis, ;.1., +558. 'ynamic interactions among the stock market, &ederal &unds rate,
in&lation, and economic acti%ity. 1he 9inancial 2e%iew -1, .1*L.-..
Jee, @.(/., 166.. 9undamentals and bubbles in asset prices) e%idence &rom the A/ and
Japanese asset prices. 9inancial !ngineering and Japanese 9inancial "arkets +, 86L1++.
Jee, @.(/., 1667. :ermanent, temporary and non&undamental components o& stock prices.
Journal o& 9inancial and Wuantitati%e 0nalysis **, 1L*+.
Jee, J., N !nders, W. #+55-, 0ugust$. 1esting &or a unit(root with a nonlinear 9ourier
&unction. In !conometric /ociety +55- 9ar !astern "eetings #;o. -.?$. !conometric
/ociety.
"aldanado, 2., N /aunders, 0. #1671$. International port&olio di%ersi&ication and the inter(
temporal stability o& international stock market relationships. 9inancial "anagement 15,
.-L8*.
"alliaris, 0.,., Arrutia, J.J., 166.. 1he impact o& the :ersian gul& crisis on national e>uity
markets. 0d%ances in International @anking and 9inance 1, -*L8..
"andelker, ,., 1andon, C., 167.. Common stock returns, real acti%ity, money, and in&lation)
/ome international e%idence. Journal o& International "oney and 9inance -, +8?L+78.
"ussa, "., +555. 1he impact o& higher oil prices on the global economy. International
"onetory 9und +555. 7 'ecember, 0%ailable at) www.im&.orgFexternalFpubsF&tFoilF+555F
:anton, '. @., Jessig, . :., N Joy, O. ". #16?8$. Co(mo%ements o& international e>uity
markets) a taxonomic approach. Journal o& 9inancial and Wuantitati%e 0nalysis 11, -1.L
-*+.
**

:atelis, 0.'., 166?. /tock return predictability and the role o& monetary policy. Journal o&
9inance .+, 16.1L16?+.
:earce, '.C. and . 2oley, 167*, /tock prices and economic news, Journal o& @usiness .7,
-6(8?.
:hilippatos, '.@., Christo&i, 0., N Christo&i, :. #167*$. 1he inter(temporal stability o&
international stock market relationships) another %iew. 9inancial "anagement 1+, 8*L86.
2ipley, '. #16?*$. /ystematic elements in the linkage o& national stock market indices. 1he
2e%iew o& !conomics and /tatistics 1., *.8L*81.
2oll, 2., 166+. Industrial structure and the comparati%e beha%iour o& international stock
market indices. Journal o& 9inance -?, *(-1.
/chwert, ,.W., 1665. /tock returns and real acti%ity) 0 century o& e%idence. Journal o&
9inance -., 1+*?L1+.?.
/chwert, ,.W., 1665. /tock returns and real acti%ity) a century o& e%idence. Journal o&
9inance -., 1+*?(1+.?.
/hiller, 2.J., +55.. Irrational !xuberance, +nd edition. :rinceton Ani%ersity :ress.
/il%ennoinen, 0., 1horpy, /., +515. 9inancialization, crisis and commodity correlation
dynamics. Ani%ersity o& 1echnology /ydney, 2esearch :aper.
/ims, C. 0. #1675$. "acroeconomics and reality. !conometrica) Journal o& the !conometric
/ociety, 1(-7.
/tock, J. <., N Watson, ". W. #+551$. ector autoregressions. 1he Journal o& !conomic
:erspecti%es, 1.#-$, 151(11..
1ang, C., Kiong, W., +515. Index in%estment and &inancialization o& commodities. Working
:aper.
1horbecke, W., 166?. On stock market returns and monetary policy. Journal o& 9inance .+,
8*.L8.-.
*-

A/, Japan and !urope. Journal o& International 9inancial "arkets, Institutions and "oney 1-,
17.L+51.
Wen, K., Wei, M., N <uang, '. +51+. "easuring contagion between energy market and stock
market during &inancial crisis) 0 copula approach. !nergy !conomics.
Wol&&, C.C.:., 1677. !xchange rates, inno%ations and &orecasting. Journal o& International
"oney and 9inance ?, -6(81.
Wongbangpo, :., N /harma, /. C. +55+. /tock market and macroeconomic &undamental
dynamic interactions) 0/!0;(. countries. Journal o& 0sian !conomics, 1*#1$, +?(.1.

You might also like