Description
This paper aims to empirically examine how shocks to monetary policy measures (the
short-term nominal interest rate and broad money supply) affect macroeconomic aggregates, namely,
output growth of the economy, national price levels and the nominal exchange rate.
Journal of Financial Economic Policy
The response of macroeconomic aggregates to monetary policy shocks in Pakistan
Abdul Rashid Zainab J ehan
Article information:
To cite this document:
Abdul Rashid Zainab J ehan , (2014),"The response of macroeconomic aggregates to monetary policy
shocks in Pakistan", J ournal of Financial Economic Policy, Vol. 6 Iss 4 pp. 314 - 330
Permanent link to this document:http://dx.doi.org/10.1108/J FEP-04-2013-0016
Downloaded on: 24 January 2016, At: 21:50 (PT)
References: this document contains references to 33 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 228 times since 2014*
Users who downloaded this article also downloaded:
Dogga Satyanarayana Murthy, Suresh Kumar Patra, Amaresh Samantaraya, (2014),"Trade Openness,
Financial Development Index and Economic Growth: Evidence from India (1971-2012)", J ournal of Financial
Economic Policy, Vol. 6 Iss 4 pp. 362-375http://dx.doi.org/10.1108/J FEP-10-2013-0056
Tess DeLean, J oseph P. J oyce, (2014),"Stock markets and the costs of banking crises", J ournal of Financial
Economic Policy, Vol. 6 Iss 4 pp. 342-361http://dx.doi.org/10.1108/J FEP-01-2014-0003
David Walker, (2014),"Market vs. administered Federal Reserve policy rates", J ournal of Financial Economic
Policy, Vol. 6 Iss 4 pp. 331-341http://dx.doi.org/10.1108/J FEP-05-2014-0032
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The response of macroeconomic
aggregates to monetary policy
shocks in Pakistan
Abdul Rashid
International Institute of Islamic Economics (IIIE), International Islamic
University, Islamabad, Pakistan, and
Zainab Jehan
Department of Economics, Fatima Jinnah Women University,
Rawalpindi, Pakistan
Abstract
Purpose – This paper aims to empirically examine how shocks to monetary policy measures (the
short-term nominal interest rate and broad money supply) affect macroeconomic aggregates, namely,
output growth of the economy, national price levels and the nominal exchange rate.
Design/methodology/approach – Johansen’s (1995) cointegration technique and error correction
models are used to explore the long-run relationship among variables. To investigate how
macroeconomic aggregates respond to a one-standard deviation shock to the underlying monetary
measures, the authors estimate impulse response functions based on error correction models. The
study uses quarterly data covering the period 1980-2009.
Findings – The results provide evidence that there is a long-run stable relationship between the
authors’ monetary measures and the underlying macroeconomic aggregates. They also fnd that
the industrial production adjusts at a faster speed relative to commodity prices and the exchange
rate over the examined period. Further, they show that the short-term interest rate has relatively
stronger effects on output as compared to broad money supply, whereas prices and exchange rates
adjust more quickly to their long-run equilibrium when money supply is used as a measure of
monetary policy. Finally, the authors fnd signifcant evidence of a price puzzle regardless of
whether they consider a closed or an open economy case. However, an initial appreciation
of exchange rate is observed in response to a one-standard deviation shock to money supply,
indicating the overshooting hypothesis phenomenon.
Practical implications – The fndings of the analysis suggest that the interest rate-oriented
monetary policy is more effective when the monetary authorities’ objective is to enhance the output
growth of the economy. However, in case of infation targeting, the broad money supply seems a
more appropriate instrument. Our fndings also suggest that the monetary policy has a signifcant
role in stabilizing both real and nominal sectors of the economy.
Originality/value – The main value of this paper is to examine the signifcance of monetary policy
for a developing and relatively small open economy, namely, Pakistan. The authors use the error
correction model, which improves the estimation by accounting for the long-run association. They
also take into account the world oil prices by including the world commodity price index as a
control variable in their empirical investigation. Finally, they utilize quarterly data rather than
annual, and they cover a relatively recent sample period.
JEL classifcation – C3, E4, E5
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1757-6385.htm
JFEP
6,4
314
Journal of Financial Economic Policy
Vol. 6 No. 4, 2014
pp. 314-330
©Emerald Group Publishing Limited
1757-6385
DOI 10.1108/JFEP-04-2013-0016
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Keywords Monetary policy, Convergence, Error correction model, Impulse response function,
Macroeconomic aggregates, Long-run relationship
Paper type Research paper
1. Introduction
Since the seminal work by Friedman (1968), the role of monetary policy in
macroeconomic stabilization is an inconclusive issue. Besides the development on
theoretical grounds, a substantial body of empirical literature has contributed to the
ongoing debate by providing signifcant evidence on how monetary policy affects
output growth, domestic prices and exchange rates. No doubt, the adoption of the
foating exchange rate system, the slogan of fnancial reforms, the trade liberalization
and relatively more independent central banks have enhanced the signifcance of
monetary policy. Therefore, both academics and policymakers are keen to understand
how, when and to what extent the economic aggregates respond to changes in monetary
policy.
In theory, the debate on monetary policy has evolved from policy ineffectiveness to
the identifcation of the long- and short-run impact of monetary policy on
macroeconomic performance. According to the monetarists, monetary policy is likely to
be effective in the short run but almost completely ineffective in the long run. While both
the New-Keynesian and classical schools of thought have similar views regarding the
neutrality of money in the long run, advocates of the former believe that monetary policy
may affect output and infation in the short run, as they presume that nominal wages are
rigid at least in the short run. On the other hand, rational expectation theory considers
expectations as crucially important for analyzing the effect of monetary policy (for
further theoretical debates, see Goodfriend (2005)).
Recent empirical studies such as Bernanke et al. (2005), Bernanke et al. (1998),
Eichenbaum et al. (1995) and Sims (1992) have signifcantly contributed to the
measurement of monetary policy and its innovations. These studies mainly utilize the
vector autoregressive methodology to measure the responsiveness of macroeconomic
aggregates to monetary policy shocks. Although the fndings of these studies provide
signifcant evidence on the response of macroeconomic variables, such as real economic
activity, national price levels and exchange rates, to changes in monetary policy, there
are number of measurement problems and various anomalies. These inconsistencies
generally include price, exchange rate and liquidity puzzles.
To overcome these issues, researchers have made numerous attempts to develop
advanced estimation methods and techniques measuring the shocks. Factor-augmented
vector autoregressive (FAVAR) approach developed by Bernanke et al. (2005) and
structural factor-augmented model proposed by Forni and Gambetti (2010) are
examples of these advancements[1].
In developing countries, monetary policy is more complicated and challenging due to
the lack of well-developed and deep fnancial markets, plus the weak channels of
transmission. Owing to the small fnancial markets and weak channels of transmission
in developing countries, the relationship between monetary measures and
macroeconomic aggregates tends to be weak and inconsistent. Above all, most of the
empirical studies have focused on developed countries. Thus, we know relatively less
how macroeconomic aggregates, such as output, national price levels and exchange
rates, respond to monetary policy shocks in developing countries. Yet, understanding
315
Monetary policy
shocks
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the effects of monetary policy on macroeconomic aggregates of developing countries is
of a great signifcance to policymakers and academics for enhancing macroeconomic
stability. Evidence on the effectiveness of monetary policy in developing countries
would enhance our overall understanding of using monetary policy as a tool of
macroeconomic stabilization.
Unlike most of the previous studies that have largely focused on developed countries,
this paper is focusing on a developing and relatively small open economy, namely,
Pakistan. Specifcally, the paper empirically investigates how shocks to monetary
policy measures – the short-termnominal interest rate and broad money growth – affect
economic aggregates: output growth of the economy, national price levels and the
nominal exchange rate. The study uses quarterly data covering the period from1980Q1
to 2009Q2. First, we study the effects of monetary policy on the long-run convergence of
these three macroeconomic aggregates. Second, we examine how these variables
respond to a one-standard deviation shock to the monetary measures.
To carry out our empirical investigation, we frst test the order of integration of the
variables by using the augmented Dickey and Fuller (1981) unit root tests. After
confrming the order of integration, we test for the long-run association among the
variables. Specifcally, we apply the Johansen’s (1995) cointegration process to testing
for the existence of the possible cointegration vectors. To examine the direction of the
short- and long-run causality, and to estimate the speed at which the variables converge
to its long-run equilibrium, we estimate the vector error correction model (hereafter
VECM). Finally, to investigate how macroeconomic aggregates respond to a
one-standard deviation shock to the monetary measures, we estimate impulse response
functions (hereafter IRFs) based on VECM. The empirical methodology starts with a
bivariate model of closed economy as a baseline model to which more explanatory
variables are gradually added. Finally, we extend our empirical model to an
open-economy model by incorporating bilateral nominal exchange rates and
international commodity prices. This approach enables us to examine howthe response
of underlying variable to monetary measures changes when we include more
information in the model. It also provides an interesting comparison between closed-
and open-economy models.
There is very little empirical literature for Pakistan on this topic. Qayyum (2008)
computes the monetary condition index (MCI) for Pakistan based on the estimated
weights of the measures of monetary policy, such as the interest rate and exchange rate.
Because the MCI index is more useful in the absence of supply shocks, and because the
supply shocks are dominant in Pakistan, the application of MCI for Pakistan is
questionable and unreliable. Therefore, the results of Qayyum’s study may not be
reliable[2]. Another study by Agha et al. (2005) uses six-month Treasury bill (T-bill)
rates as a measure of monetary policy and the VAR technique to examine the
effectiveness of monetary policy over a relatively short time span. Finally, recently,
Khan (2008) has made an attempt to investigate the impact of unanticipated changes in
monetary policy on output and infation by estimating structure VAR (SVAR). The
study uses nominal shocks in SVAR as a proxy for unexpected changes in monetary
policy. However, this measure suffers the problem of lack of theoretical rationales.
This paper contributes to the existing literature in three ways. First, we utilize
quarterly data rather than annual, and we cover a relatively recent sample period
spanning from 1980Q1 to 2009Q2. The use of quarterly data enables us not only to
JFEP
6,4
316
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
harvest the gain of higher degree of freedom but also allows us to use deeper lags of the
variables to identify a well-specifed model without losing the informational credibility
of the sample. Second, unlike most of the previous studies, we take into consideration the
time-series properties of the variables, such as non-stationary behavior. Finally, we use
VECMapproach, which improves the estimation over the SVARbecause the SVARdoes
not account for the long-run association. We also take into account the world oil prices
by including the world commodity price index as a control variable in our empirical
investigation.
The rest of the paper is organized as follows. Section 2 reviews the existing empirical
studies and highlights the strengths and weaknesses of their methodologies. Empirical
methods, data sources and the defnition of the variables are given in Section 3. Section 4
presents the empirical fndings. Finally, Section 5 concludes the study.
2. Literature survey
Forni and Gambetti (2010) examine the dynamic exogenous effect of monetary policy by
using a standard recursive scheme through a dynamic structural factor model for USA
covering the period 1973:3-2007:10. Their empirical analysis is based on the variables
which are used by Stock and Watson (1998). They argue that the factor analysis model
is superior to FAVAR proposed by Bernanke et al. (2005) because it helps in eliminating
the puzzles in monetary policy analysis. They fnd that a positive shock to Federal
Funds Rate (FFR) leads to an appreciation of real exchange rate. This confrms
overshooting hypothesis of Dornbusch (1976). Computing impulse response graphs,
they show the absence of price puzzle. Further, they argue that industrial production
falls, although temporary, to a large extent with a humped-shaped response.
Bjørnland (2008) examines the response of macroeconomic economic aggregates to
monetary policy by including the exchange rate in the model specifcation. He uses
quarterly data over the period 1993-2004. Further, he uses Cholesky ordering and the
Kim and Roubini’s (2000) identifcation procedures to determine the order of the
variables. Bjørnland (2008) shows that there is a temporary increase in the interest rate,
which normally takes four quarters to converge to its normal path. However, his
analysis does not provide any statistically signifcant evidence of the exchange rate
puzzle or price puzzle.
Ansari et al. (2007) explore the relationship between money income and domestic
prices by estimating VECM. They use both narrow and broad money as measures of
monetary policy. Using quarterly data, they document that for any divergence from
long-run equilibrium; output will increase by 6 per cent to adjust toward its long-run
equilibrium. Furthermore, they show that a positive shock to money policy leads to
adjustments in output after 5 quarters.
Bernanke et al. (2005) introduce a combination of VAR model and factor model to
capture large information set. They argue that a simple VAR analysis is unable to
incorporate such information. They use a diffusion index developed by Stock and
Watson (2002) to estimate the factors by utilizing a balanced panel of 120 monthly
macroeconomic series (1959:1-2001:8). Arecursive structure is assumed with identifying
assumption of no contemporaneous response of unobserved factors to monetary policy
shocks[3]. The comparison of 3-variable VAR model with two different FAVAR
specifcations reveals that the results fromthe standard VARmodel exhibit a signifcant
price puzzle and provide evidence of the inconsistent production response to a
317
Monetary policy
shocks
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
one-standard deviation monetary measure. However, the FAVAR approach improves
the results as price puzzle disappears after one year, real activity declines, monetary
aggregates fall and exchange rate appreciates for USA. As in Forni and Gambetti (2010),
as the study did not distinguish between the number of static factors and structural
shocks, a large number of economic restrictions are imposed to reach the identifcation.
Moreover, the restrictions are imposed on the IRFs of static factors instead of the IRF of
variables.
Jang and Ogaki (2004) examine the relationship between monetary policy shocks and
Dollar/Yen exchange rates, prices and output level for USA. The empirical analysis is
carried out, following the model of Jang and Ogaki (2004), through structural VECMand
VAR by employing long- and short-run restrictions on the model. They fnd that an
appreciation of exchange rate is the result of a contractionary monetary policy.
Furthermore, they fnd that output in domestic as well as in foreign country
signifcantly declines due to the long-run neutrality restrictions with an exception of
USAwhere a decline in output becomes negligible after four years. Finally, a fall in price
is observed as a result of tight monetary policy. While, estimating VECMand VARwith
short-run restrictions for variables in their levels, they fail to accept the UIP condition,
they fnd strong evidence in support of the existence of price puzzle.
Fullerton et al. (2001) apply an error correction model to study the behavior of the
exchange rate for Mexican peso over the period 1976-2000. The variables included in the
model are nominal exchange rates, consumer price index, liquid international reserves,
money supply and real gross domestic product (GDP) as non-policy variables while
one-and three-month T-bills rates as policy variables. Their fndings based on the
balance of payment framework and monetary model of exchange rate do not provide
any support to the established theory. However, balance of payment framework with
one-month T-bill rate is marginally better than the monetary model of exchange rate.
Wong (2000) empirically investigates the impact of monetary policy on
macroeconomic variables by applying a time-varying parameter model for USA over
the period 1959:1-1994:12. Output and prices are assumed to have lagged effect but FFR
and reserves are considered to have only contemporaneous effects. The rolling VARhas
been estimated with maximum three lags. The empirical results suggest that output
increases in response to a contractionary shock to monetary policy. The output is more
responsive to shocks during periods when the central bank adopts infation controlling
policy, whereas, it is less responsive when the central bank aims at promoting economic
growth. Overall, the plots of IRF provide the evidence of the presence of price puzzle.
Bernanke and Mihov (1998) develop a VAR-based methodology to measure and
assess the impact of monetary policy on macroeconomic variables. The measure of MP
is derived from an estimated model of Central Bank’s operating procedures and the
market for commercial bank reserves, which makes it more consistent than the
previously used instruments of monetary policy. The model has been estimated for
different time periods of post 1965-1996 for USA. The exogenous policy shocks are
computed through a standard VAR method by applying generalized methods of
moments in which the policy variables are placed last in variable ordering. The IRFs
indicate that there is an increase in output in response to an expansionary monetary
policy. Further, the plots provide evidence of a slower but a persistent rise in the prices.
Yet, their results considerably vary across different measures of monetary policy.
JFEP
6,4
318
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Although the study attempts to capture all the possible measures of monetary policy, it
fails to notify which of the measure is relatively more effective.
Eichenbaum et al. (1995) analyze the exchange rate transmission mechanism of
monetary policy for the period 1974:1-1990:5. They use three measures of monetary
policy commonly used in the literature. These measures are FFR, non-borrowed
reserves and the narrative measure of Romer and Romer (1989). They estimate a
multivariate VAR model by using the ordering of the variables based on the Wold
decomposition. The estimates on IRFs reveal that a contractionary monetary policy
leads to a signifcant and continual decline in US interest rate, a sharp and persistent
appreciation in US exchange rate, which is contradictory with the overshooting
hypothesis of exchange rate.
3. Empirical methodology, data and variable defnitions
3.1 Estimation methods
We use the Johansen’s (1995) procedure to test whether there is any long-run association
among the variables. The Johansen’s method is based on the relationship between the
rank of matrix and its characteristic roots (Enders, 2010). Specifcally, the model is
expressed as follows:
?X
t
? ?X
t?1
? ?
i?1
p?1
?
i
?X
t?i
? ?D
t
? ?
t
(1)
where X
t
is a vector of n endogenous variables and D
t
is a vector of m deterministic
variables. Furthermore, ?, ?
i
and ? are coeffcient matrices of dimension n ?n, n ?n
and n ?m, respectively. ?may have reduced rank, and hence, it can be informative to
decompose it with?? ??
/
, where both? and?are of dimensionn ?r andr is the rank
of ?.
As in Engle and Granger (1987), the dynamic behavior of a set of integrated variables
can be empirically analyzed through the VECM, which is the reduced formof the model.
The selected model is based on the backward looking behavior of output, domestic
prices and the exchange rate. The study employs a bivariate closed-economy model as in
Sims (1980) and Christiano et al. (1999), which is then frst extended to a multivariate and
fnally to an open-economy model to measure the relationship between macroeconomic
aggregates – output, domestic prices and the exchange rate – and monetary policy. In
matrix notation, the VECM can be written as follows:
?X
t
??
0
??
1
t ??X
t?1
?
?
i?1
p?1
?
i
?X
t?i
?
?
i?1
p?1
?
i
?Y
t
?e
t
(2)
where t ? 1, 2, …, T, X
t
? n ?1 vector of n endogenous I(I) variables included in the
VECM. These variables include industrial production, domestic prices and the exchange
rate. Y
t
?n ?1 vector of exogenous I(I) variables broad money supply, money market
lending rate and international commodity prices index. ?
0
?n ?1 vector of intercepts.
?
1
, ?
i
and ?
i
are n ?n matrices of coeffcients. e
t
? n ?1 vector of error terms
distributed as i.i.d. and fulflls the Gaussian properties of zero mean and constant
variances. ??matrix of parameters such that one element is non-zero. Moreover, ? is
the difference operator and all the variables are in log form except the interest rate.
319
Monetary policy
shocks
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The long-run behavior of the system depends on the rank of ?. Granger
representation theorem shows that if X
t
is integrated of order r then one can write
? ? ??
/
, where matrix ? contains matrix of r cointegrating vector. The matrix ?
known as the speed of adjustment, which measures how quickly ?X
t
reacts to a
deviation from equilibrium (Engle and Granger, 1987).
3.2 Data and defnition of variables
To examine the response of macroeconomic aggregates to monetary policy for Pakistan,
we use quarterly data covering the period 1980-2009. Pakistan has moved from fxed
exchange rate systemto managed exchange rate system; therefore, for the model where
the exchange rate is used, the time period starts form 1990Q1 to 2009Q2. All the data
except broad money are obtained fromthe International Financial Statistics database of
International Monetary Fund. Data on money supply are taken fromStatistical Bulletins
of Pakistan published by the State Bank of Pakistan (SBP, 2009). All the variables are in
log form except short-term interest rate. All the variables are on annual basis with
millions of Pak Rupee as unit of measurement. World commodity price index does not
have data for the full length of the sample period under investigation; therefore, few
missing values are interpolated by applying a two-quarter moving average formula.
Following prior studies, we use two alternative measures of monetary policy,
namely, broad money supply and the short-term interest rate. Bernanke and Blinder
(1992) and Sims (1998) argue that short-term interest rate is a superior measure of
monetary policy and it should be preferable over money supply. However, Berument
(2003) suggests that broad money is a better indicator of monetary policy in a small open
economy. In our empirical investigation, we utilize both measures with an aim to do a
comparison between both the said measures. Moreover, we use money supply as a
measure of monetary policy, as in Pakistan, it has been used to formulate the monetary
policy. However, recently, the monetary authority in Pakistan is giving relatively more
weightage to short-term interest rate.
4. Empirical fndings
We apply the ADF test to explore the order of integration of the underlying series. The
optimal lag structure for the ADF equations is selected based on AIC. The results are
given in Table I. The optimal lag length is market by “a”[4]. The ADF test results do not
provide any signifcant evidence to reject the null hypothesis of unit root for level series.
However, the frst difference of all the series appears stationary. These fndings are
robust across different lag lengths used in the estimation of the ADFequation. Thus, the
results suggest that the variables are integrated of order one.
To examine the long-run association, we apply the Johansen’s (1995) cointegration
test. Table II reports the results. As the results of the cointegration are very sensitive to
the lag length, we select the optimal lag length by applying the Bayesian Information
Criterion. Further, we also ensure that the estimated model has white noise
disturbances. We start by estimating the response of the underlying macroeconomic
aggregates to our both measures of monetary policy in bivariate framework, separately.
We then extend the bivariate model to the multivariate model by incorporating other
control variables. One should note that when we include the exchange rate in the
specifcations our model represents a case of an open economy[5].
JFEP
6,4
320
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The results reported in Table II provide evidence that there is a long-run relationship
between industrial production index and the short-terminterest rate, implying that both
variables have co-movement in the long run. This evidence is robust to the inclusion of
other variables in the specifcation. Our results suggest that there is a single long-run
stable relationship between industrial production index and monetary policy measures
regardless of whether we estimate model in the bivariate or multivariate framework.
The existence of the one cointegrating vector at different specifcations confrms the
validity of our results. When we turn to examine the long-run association between the
short-terminterest rate and our second response variable, namely, domestic price levels,
similar to the case of industrial production index, we fnd that there is only one
signifcant cointegrating vector. This observation holds true even when we include
industrial production index as an explanatory variable into the specifcation.
Next, we examine the long-run relationship by estimating the model for the case of an
open economy. Specifcally, we extend our closed-economy model to open economy
model by including, one-by-one, exchange rates and international commodity prices into
Table I.
Unit-root test results
Variables Model
Number of lags
Conclusion 8 6 4 2 1
Level series
LIPI Drift and trend ?2.742 ?2.743 ?1.176
a
?2.654 ?2.876 Non-stationary
Drift ?1.524 ?1.674 ?1.519
a
?1.765 ?1.543 Non-stationary
LCPI Drift and trend ?3.205 ?1.869 ?2.144
a
?1.234 ?1.234 Non-stationary
Drift 1.231 0.587
a
0.460 0.976 1.876 Non-stationary
LER Drift and trend ?1.345 ?1.302 ?1.654 ?1.854 ?0.679
a
Non-stationary
Drift ?0.657 ?0.674 -0.897 -0.964 ?0.987
a
Non-stationary
LCOMP Drift and trend ?2.077 ?1.567 ?1.974
a
?1.356 ?1.760 Non-stationary
Drift ?1.534 ?0.834 ?1.876 ?0.653
a
?0.365 Non-stationary
SR Drift and trend ?2.327 ?2.074 ?1.098 ?2.296
a
?2.645 Non-stationary
Drift ?2.730 ?2.198 ?1.832 ?2.543
a
?2.987 Non-stationary
LBMS Drift and trend ?2.089 ?2.010 ?3.198
a
?2.876 ?1.787 Non-stationary
Drift ?0.632 ?0.435 ?0.866
a
?0.643 ?0.465 Non-stationary
Firs-differenced series
LIPI Drift and trend ?2.435 ?2.830 ?3.807
a
?22.876 ?24.877 Stationary
Drift ?2.760 ?2.675 ?3.976
a
?22.643 ?23.159 Stationary
LCPI Drift and Trend ?1.856 ?1.980 ?3.680
a
?4.987 ?7.574 Stationary
Drift ?1.287 ?1.765 ?3.435
a
?3.754 ?7.334 Stationary
LER Drift and Trend ?2.763 ?4.123 ?4.965
a
?6.764 ?7.155 Stationary
Drift ?3.543 ?4.945 ?4.896
a
?6.210 ?7.765 Stationary
SR Drift and Trend ?3.987 ?4.987 ?5.265 ?9.976 ?11.876 Stationary
Drift ?3.734 ?4.976 ?5.109 ?10.145 ?11.643 Stationary
LCOMP Drift and Trend ?2.765 ?3.124 ?4.087 ?5.905
a
?7.111 Stationary
Drift ?2.845 ?3.976 ?3.900 ?5.687
a
?7.787 Stationary
LBMS Drift and Trend ?2.974 ?4.943 ?4.3743 ?7.365
a
?6.243 Stationary
Drift ?2.987 ?4.356 ?4.930 ?7.384
a
?6.296
Notes:
a
Represents optimum lag length selected by the Akaik Information Criterion (AIC); LIPI:
log(industrial production index); LCPI: log(consumer price index); LER: log(exchange rate); LCOMP:
log(world commodity price index); SR: sort-term interest rate; and LBMS: log(broad money supply)
321
Monetary policy
shocks
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the specifcation. The results are given in Table III. As we can see from the table, the
trace statistics signify the presence of one-cointegration relationship in both cases.
Third macroeconomic aggregate that we consider in our analysis is the bilateral
nominal exchange rate between PAK rupee and US dollar. Adopting the previous
strategy here too, we frst estimate a bivariate model, which is then extended to
multivariate model. The estimated eigenvalues showthat short-termrate of interest and
exchange rate appear to have a single linear combination in the long run, which is I(0).
We also fnd the evidence of the existence of the long-run relationship when we extend
the model by including domestic price levels and industrial production index.
As mentioned earlier, we also use an alternative measure of monetary policy, namely,
broad money supply. The results of the cointegration tests with this measure are given
in Table III. In general, the fndings are consistent with the results reported in Table II.
However, in two of the cases, we fnd two signifcant cointegrating vectors instead of
one. In case of more than one cointegrating vectors, the general practice is to select the
cointegrating vector, which has highest eigenvalue because it is most associated with
the stationary part of the model (Rashid, 2009). Following this, we select frst
cointegration vector as it has the highest value. The broad money supply appears to
have the long-run co-movement with the exchange rate regardless of whether the model
is estimated in bivariate or multivariate framework.
After confrming the presence of the long-run relationship, we estimate VECM for
each model to examine the speed of adjustment toward the long-run equilibrium. When
we regress the industrial production on short-term money market rate, we fnd that the
Table II.
Results for cointegration
test and VECM (Policy
variable: short-term
interest rate)
Model specifcation
No.
(Lags) Rank Q
r
Eigenvalue ECT
No.
(Lags)
Model-1: Dependent variable; LIPI
LIPI, SR 114 (4) 1 2.658 0.2094 0.0208*** (0.000) 114 (4)
LIPI, LCPI, SR 112 (6) 1 11.104 0.1836 ?0.0762*** (0.000) 114 (4)
LIPI, LCPI, LER, SR 115 (3) 1 28.062 0.1638 ?0.5622*** (0.000) 116 (2)
LIPI, LCPI, LER, LCOMP, SR 115 (3) 1 47.082 0.2452 ?0.6842*** (0.000) 116 (2)
Model-2: Dependent variable; LCPI
LCPI, SR 117 (1) 1 0.439 0.1601 0.0008* (0.052) 109 (9)
LCPI, LIPI, SR 114 (4) 1 9.934 0.1823 ?0.0199** (0.000) 111 (7)
LCPI, LIPI, LCOMP, SR 116 (2) 1 22.975 0.4584 ?0.0157*** (0.000) 116 (2)
LCPI, LIPI, LER, LCOMP, SR 115 (3) 1 47.082 0.2452 ?0.032*** (0.000) 111 (7)
Model-3: Dependent variable; LER
LER, SR 78 (7) 1 1.0381* 0.1752 ?0.0038** (0.017) 78 (7)
LER, LIPI, SR 78 (4) 1 12.7371* 0.2521 ?0.1682*** (0.001) 78 (6)
LER, LIPI, LCPI, SR 78 (4) 1 25.1796* 0.4194 ?0.2978*** (0.002) 78 (6)
Notes: This table displays the estimates for short-run interest rate as measure of monetary policy.
Lags in 2nd column refer to the number of lags for cointegrating vector, while lags in 7th column are for
estimation of vector error correction term; Qr is the LR trace statistic. ECT denotes error correction
term; LIPI: log(industrial production index); LCPI: log(consumer price index); LER: log(exchange rate);
LCOMP: log(world commodity price index); SR: sort-term interest rate; and LBMS: log(broad money
supply); ***, **, * denote the signifcance at the 1, 5 and 10 per cent levels, respectively
JFEP
6,4
322
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
error term is statistically negatively signifcant. The negative sign of the coeffcient
associated with error term is in line with the adjustment process, suggesting that
industrial production converges to its long-run equilibrium over the time. Specifcally,
the coeffcient of error term is 0.021, implying that any disequilibrium in industrial
production will be adjusted at the rate of 2.1 per cent in one quarter. However, the speed
of adjustment becomes 7.6 per cent per quarter when we add domestic prices to the
specifcation.
The estimates on the speed of adjustment have considerably been improved in the
case of the open-economy model. For instance, the disequilibrium in industrial
production is now corrected by the rate of 56.2 per cent in one quarter, while the
inclusion of international commodity prices into the model further increases the
adjustment speed to 68.42 per cent per quarter. This suggests that the adjustment
process of industrial production to disequilibrium is relatively fast in an open economy
framework.
When we estimate the model of industrial production using an alternative measure of
monetary policy, namely, broad money supply, the speed of adjustment is remarkably
high for the baseline model, suggesting that the disequilibrium is corrected by the rate
of 76 per cent in one quarter. Contrary to the frst measure, the rate of convergence
declines with the addition of more information into the model. This fnding suggests
that when the monetary policy is measured by the short-term interest rate it is more
effective for an open economy, whereas the measures of money supply has signifcant
role to play in a closed economy. These fndings make sense as theoretically the interest
Table III.
Results for cointegration
test and VECM (Policy
variable: broad money
supply)
Model specifcation
No.
(Lags) Rank Qr
Eigen
value ECT
No.
(Lags)
Model-1: Dependent variable; LIPI
LIPI, LBMS 116 (2) 1 0.0004 0.3743 ?0.7595*** (0.000) 116 (2)
LIPI, LCPI, LBMS 113 (5) 1 9.033 0.2365 ?0.3795*** (0.000) 113 (5)
LIPI, LCPI, LER, LBMS 116 (2) 1 28.742 0.5715 ?0.3571*** (0.000) 113 (5)
LIPI, LCPI, LER, LCOMP, LBMS 115 (3) 2 27.761 0.2286 ?0.4549*** (0.000) 113 (5)
Model-2: Dependent variable: LCPI
LCPI, LBMS 109 (9) 1 0.4011 0.1343 ?0.0573*** (0.002) 113 (5)
LCPI, LIPI, LBMS 113 (5) 1 9.0338 0.2365 ?0.1524*** (0.003) 110 (8)
LCPI, LIPI, LCOMP, LBMS 114 (4) 1 20.1580 0.3805 ?0.1441 (0.033) 112 (6)
LCPI, LIPI, LER, LCOMP, LBMS 115 (3) 2 27.7615 0.2286 ?0.2207** (0.007) 112 (6)
Model-3: Dependent variable; LER
LER, LBMS 78 (11) 1 2.526 0.1578 ?0.0804** (0.019) 78 (4)
LER, LIPI, LBMS 78 (2) 1 9.806 0.2813 ?0.2486*** (0.000) 78 (5)
LER, LIPI, LCPI, LBMS 78 (4) 1 26.798 0.3336 ?0.3298*** (0.000) 78 (6)
Notes: This table displays the estimates for money supply (M2) as measure of monetary policy; Lags
in 2nd column refer to the number of lags for cointegrating vector while lags in 7th column are for
estimation of vector error correction term; Qr is the LRtrace statistic; ECTdenotes error correction term;
LIPI: log(industrial production index); LCPI: log(consumer price index); LER: log(exchange rate);
LCOMP: log(world commodity price index); SR: sort-term interest rate; and LBMS: log(broad money
supply); ***, **denote the signifcance at the 1 and 5 and per cent level, respectively
323
Monetary policy
shocks
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
rate plays a signifcant role in external capital fows along with its internal effectiveness,
whereas money supply is relatively more effective to make adjustments in domestic
accounts.
Regarding the long-run relationship between domestic prices and the short-term
nominal interest rate, we fnd that prices diverge fromtheir long-run equilibriumby 0.08
per cent per quarter. It is interesting to note that the inclusion of industrial production to
the model yields statistically negatively signifcant estimates on error correction term.
This implies that the short-run interest rate may affect price levels through its effects on
industrial output. Next, we include the exchange rate into the model to represent the case
of an open economy. Based upon one cointegration vector, the long-run equilibrium
relationship appears to be statistically negatively signifcant. A 1.6 per cent pace is
identifed by the error correction termwith which the disequilibriumwill be corrected by
domestic prices. However, the extension of the model by including industrial production
further improves the speed of adjustment to 3.2 per cent. Thus, there is evidence that
prices respond more quickly, and the pace of movement to the equilibrium point has
been increased by moving from a closed-economy to an open-economy model.
The model of domestic prices is empirically tested for long-run equilibrium
relationship by using the alternative measure of monetary policy as well. In a bivariate
closed-economy situation, there is a negative and statistically signifcant long-run
relationship as identifed by the error correction term. The system converges to its
long-run equilibriumposition at the rate of 5.7 per cent per quarter. In the open-economy
framework, adding international commodity prices once again increases the speed of
adjustment toward the long-run equilibrium.
For the third dependent variable, namely, the nominal exchange rate, the error
correction term appears negative and statistically signifcant. However, the magnitude
of the coeffcient indicates that the system converges to its long-run equilibrium by a
marginal rate. However, including industrial production to the model, we fnd that the
adjustment speed has been increased to 17 per cent per quarter, while the inclusion of
price levels into the specifcation further enhances the process of adjustments (at the rate
of 30 per cent per quarter) toward their long-run equilibrium position. These results
suggest that the both industrial production and prices have a signifcant role to play in
adjustment mechanism.
Regarding the money supply as a measure of monetary policy, we fnd that the speed
of adjustment is 8.04 per cent for exchange rate – money supply model. In addition, the
inclusion of domestic prices in the exchange rate model increases the speed of
adjustment to 25.0 per cent in one quarter. Further, when we add industrial production
to the specifcation, the estimates provide evidence that any deviation from the
equilibrium is adjusted to the long-run equilibrium position with the rate of 33 per cent
per quarter.
Finally, we compute the IRFs to examine the response of industrial production,
domestic prices and the exchange rate to a one-standard deviation shock to our
monetary measures. Similar to the case of cointegration and VECMestimation, we start
from a model of closed economy (a bivariate case), and we then extend the model to an
open economy by incorporating the exchange rate and international commodity prices
index into our specifcation. One of the major problems in IRFs is the sensitivity with
respect to variables ordering in the system. Therefore, following the empirical literature,
such as Forni and Gambetti (2010), Bernanke et al. (2005), Bjørnland (2008) and
JFEP
6,4
324
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Holtemoller (2004), this study assumes a recursive structure of ordering for the closed
economy in which policy variables are ordered at last. This implies that macroeconomic
aggregates do not respond contemporaneously to monetary policy innovations but
monetary policy might react toward any news frommacro aggregates within the period.
This is consistent with the transmission mechanism of monetary policy as highlighted
by several empirical studies, such as Svensson (1997). In the case of closed economy, we
order the variables as LIPI
t
, LCPI
t
, and SR
t
.
In an open-economy model, the exchange rate is placed last in the order of variables
as suggested by Eichenbaum et al. (1995). It ensures a lagged response of monetary
policy due to any exchange rate led shock but this identifcation results in a delayed
exchange rate response to monetary policy (Bjørnland, 2008). Kim and Roubini (2000)
propose a contemporaneous interaction between monetary policy and the exchange rate
to solve the problem of exchange rate puzzle. We employ Kim and Roubini (2000)
methodology and introduce the exchange rate and monetary policy interaction of
contemporaneous impact (by reversing the place in variable ordering for IRF). As
Pakistan is a small open economy, international prices are assumed exogenous for the
economy. In other words, central bank does not have international prices in its
information set (Juang et al., 2003). Therefore, international commodity prices are placed
after the exchange rate. Specifcally, in our case, the variable ordering for the
open-economy model is LIPI
t
, LCPI
t
, SR
t
, LER
t
, and LCOMP
t
.
The IRFs for the closed economy by using money market rate and broad money as
measures of monetary policy are presented in Figures 1 and 2, respectively. In the
bilateral model, a one-standard deviation shock to the short-terminterest rate appears to
have a negative impact on industrial production with a margin positive start. The
overall impact is composed of downward and upward fuctuations of industrial
production curve with a declining trend. It is interesting to note that the inclusion of
other variables in the baseline model changes neither the initial nor the long-run
response of industrial production to a one-standard deviation shock to money market
rate. These fndings are consistent with the results of prior empirical studies, such as
Forni and Gambetti (2010) and Bjørnland (2008), who also report a negative response of
output to monetary policy shocks. The response of industrial production remains
negative with respect to short-term interest rate shocks even for a multivariate
closed-economy model.
Domestic prices respond positively, providing the evidence of the existence of price
puzzle. Sims (1992) presents similar evidence. However, our fndings are in contrast to
Ogaki et al. (2003), who fnd that there is no price puzzle[6]. The inclusion of exchange
rate and international commodity prices improves the response of industrial production
to positive monetary policy innovations. A positive shock to money market rate
decreases the industrial production at a sharp rate before it starts increasing. This fact
is in line with the fndings of Bernanke et al. (2005) and Forni and Gambetti (2010) that
there is a decline in output after a positive shock to FFR.
On the other hand, when we consider the case of an open economy, a sharp and
abrupt increase in price level is observed after a shock to monetary policy. This point
outs price puzzles (Leeper et al., 1992; Bernanke and Blinder, 1992). Sims (1992) have
suggested that price puzzle can be tackled by including the international commodity
prices but in our case, the inclusion of international prices does not provide any
signifcant help in eliminating the puzzle. Yet, our fnding is consistent with
325
Monetary policy
shocks
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Eichenbaumet al. (1995), who argue that the inclusion of international commodity prices
does not improve the responsiveness of prices toward monetary policy innovations.
As one can see from the fgure, the initial response of exchange rate to monetary
policy innovations is almost zero. However, later on it appreciates followed by a
considerable depreciation, confrming the overshooting hypothesis of Dornbusch (1976).
This suggests that a contractionary monetary policy would lead to an appreciation in
the nominal exchange rate before it depreciates in the long run. However, we fnd the
evidence of exchange rate puzzle when we include international commodity price in the
model.
The results for the alternative measure of monetary policy are consistent with the
baseline model as depicted by IRFs. In the bivariate closed-economy models, the
LIPI LCPI LER
?0.06
?0.04
?0.02
0
0 50
VEC5, R, LIPI
step
Graphs by irfname, impulse variable, and response variable
0
0.005
0.01
0.015
0 50
VEC6, R, LCPI
step
Graphs by irfname, impulse variable, and response variable
?0.04
?0.02
0
0.02
0
VEC6, R, LIPI
step
Graphs by irfname, impulse variable, and response vari
0
.005
.01
.015
0 50
VEC6, R, LCPI
step
Graphs by irfname, impulse variable, and response variable
?0.05
0
0 50
VEC7, R, LIPI
step
Graphs by irfname, impulse variable, and response variable
0
0.002
0.004
0.006
0 50
VEC7, R, LCPI
step
Graphs by irfname, impulse variable, and response variable
?0.003
?0.002
?0.001
0
0 50
VEC7, R, LER
step
Graphs by irfname, impulse variable, and response variable
?0.06
?0.04
?0.02
0
0 50
VEC8, R, LIPI
step
Graphs by irfname, impulse variable, and response variable
0
0.001
0.002
0.003
0.004
0 50
VEC8, R, LCPI
step
Graphs by irfname, impulse variable, and response variable
?0.002
?0.001
0
?0.001
0 50
VEC8, R, LER
step
Graphs by irfname, impulse variable, and response variable
Figure 1.
IRFs: effects of shocks to
short-term interest rate
JFEP
6,4
326
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
industrial production responds to broad money supply-led shocks non-monotonically.
Specifcally, a positive shock to broad money supply appears to increase the industrial
production at frst, then it results in a decline in the production, and fnally, it increases
the production. When the model is extended to an open economy by incorporating the
exchange rate and international commodity prices, the response of industrial production
remains similar to our earlier results. Overall, our results are in line with Berument
(2003) and Ansari et al. (2007), who also fnd a positive response of GDP to a positive
shock to money supply. We did not fnd any evidence of price puzzle when broad money
supply is used as a measure of monetary policy. This fnding is contrast to the case when
we use money market rates as a measure of monetary policy. These fndings are similar
to Juang et al. (2003), Forni and Gambetti (2010), and Bernanke et al. (2005). Similarly, the
exchange rate appears to appreciate in response to a positive shock to monetary policy
LIPI LCPI LER
0
0.005
0.01
0.015
0 50
VEC1, LM, LIPI
step
Graphs by irfname, impulse variable, and response variable
0
0.005
0.01
0.015
0 50
VEC2, LM, LCPI
step
Graphs by irfname, impulse variable, and response variable
0
0.01
0.02
0.03
0 50
VEC2, LM, LIPI
step
Graphs by irfname, impulse variable, and response variable
0
0.005
0.01
0.015
0 50
VEC2, LM, LCPI
step
Graphs by irfname, impulse variable, and response variable
?0.01
0
0.01
0.02
0.03
0 50
VEC3, LM, LIPI
step
Graphs by irfname, impulse variable, and response variable
0
0.005
0.01
?0.015
0 50
VEC3, LM, LCPI
step
Graphs by irfname, impulse variable, and response variable
?0.01
?0.005
0
0.005
0 50
VEC3, Lm, LER
step
Graphs by irfname, impulse variable, and response variable
?0.01
0
0.01
?0.02
0 50
VEC4, LM, LIPI
step
Graphs by irfname, impulse variable, and response variable
0
0.005
0.01
0.015
0 50
VEC4, LM LCPI
step
Graphs by irfname, impulse variable, and response variable
0
0.005
0.01
0 50
VEC4, LM, LER
step
Graphs by irfname, impulse variable, and response variable
Figure 2.
IRFs: effects of shocks to
broad money supply
327
Monetary policy
shocks
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and then depreciates. This confrms the overshooting hypothesis of Dornbusch (1976)
and provides no evidence of any exchange rate puzzle[7].
Collectively, our results suggest that there is a signifcant interaction between the
real and monetary side of the economy in Pakistan. The statistical signifcance of the
error correction term clearly refects the convergence to the long-run equilibrium of
industrial production, prices and the exchange rate. This holds for both measures of
monetary policy we used in our analysis. In other words, our results suggest that that
there is a signifcant response of the macroeconomic aggregates to shocks to monetary
policy.
5. Conclusions and policy implications
The study empirically analyzes the response of macroeconomic aggregates, such as real
economic activity proxied by industrial production, national price levels and the
exchange rate, to shocks to monetary policy for Pakistan during the period 1980:
Q1-2009:Q2. The fndings are summarized as follows. First, we fnd that there is a
signifcant co-movement between macroeconomic aggregates and our monetary policy
measures, namely, the short-term nominal interest rate and broad money supply.
Second, the estimates of error correction term provide evidence that the industrial
production adjusts at faster speed relative to prices and the exchange rate over the
examined period. Finally, we show that the short-term interest rate has relatively
stronger effects on output as compared to broad money supply, whereas prices and
exchange rates adjust more quickly to their long-run equilibriumwhen money supply is
used as a measure of monetary policy.
The graphs of IRF provide the evidence of price and exchange rate puzzles when
money market rate is used as monetary policy. In contrast, when broad money supply is
used as a measure of monetary policy, no evidence of exchange rate puzzle is witnessed.
This fnding is in line with the Dornbusch (1976) overshooting hypothesis. The fndings
of the analysis suggest that the interest rate-oriented monetary policy is more effective
when the monetary authorities have objectives to enhance the output growth of the
economy. However, if the objective is to control the infation rate, then the broad money
supply seems more appropriate instrument. Furthermore, our fndings suggest that the
monetary policy has a signifcant role in stabilizing both real and nominal sectors of the
economy.
Notes
1. Despite a large body of empirical work using intensifed methodological applications, the
fndings are still inconclusive at best (for further details, see Bjørnland (2009)).
2. Although Khan and Qayyum (2004) provide empirical evidence of the superiority of the MCI
over Bernanke and Mahiv (1995) measure of monetary policy while measuring the
macroeconomic impact of monetary policy for Pakistan, Bernanke and Mihov’s measure of
monetary policy is better theoretically, as it uses more fnancial variables, which play an
important role in monetary policy formulation.
3. Monetary policy variable is placed last in the variable ordering.
4. However, we also estimate the ADF equations with other lag lengths to check the robustness
of the ADF results at different lags.
JFEP
6,4
328
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
5. Althoughwe use response variables as control variables as well, we estimate a separate model
for each response variable to test for cointegration.
6. However, he used different technique namely SVAR.
7. Bjørnland (2008) and Forni and Gambetti (2010) also fnd an appreciation of ERin response to
a positive shock to monetary policy.
References
Agha, A., Ahmed, I., Mubarik, N., Ali, Y. and Shah, H. (2005), “Transmission mechanism of
monetary policy in Pakistan”, SBP-Research Bulletin, Vol. 1 No. 1, pp. 1-23.
Bernanke, B. and Blinder, A. (1992), “The federal funds rate and the channels of monetary
transmission”, The American Economic Review, Vol. 82 No. 4, pp. 901-921.
Bernanke, B. and Mihov, I. (1998), “Measuring monetary policy”, Quarterly Journal of Economics,
Vol. 113 No. 3, pp. 869-902.
Bernanke, B., Boivin, J. and Eliasz, P. (2005), “Measuring the effects of monetary policy: a
factor-augmented vector autoregressive (FAVAR) approach”, Quarterly Journal of
Economics, Vol. 120 No. 1, pp. 387-422.
Bjørnland, H. (2008), “Monetary policy and exchange rate interactions in a small open economy”,
Scandinavian Journal of Economics, Vol. 110 No. 1, pp. 197-221.
Bjørnland, H. (2009), “Monetary policy and exchange rate overshooting: Dornbusch was right
after all”, Journal of International Economics, Vol. 79 No. 1, pp. 64-77.
Christiano, L., Eichenbaum, M. and Evans, C. (1999), “Monetary policy shocks: what have we
learned and to what end?”, Handbook of Macroeconomics, Vol. 1, pp. 65-148.
Dickey, D. and Fuller, W. (1981), “Likelihood ratio statistics for autoregressive time series with a
unit root”, Econometrica, Vol. 49 No. 4, pp. 1057-1072.
Dornbusch, R. (1976), “Expectations and exchange rate dynamics”, The Journal of Political
Economy, Vol. 84 No. 6, pp. 1161-1176.
Eichenbaum, M. and Evans, C. (1995), “Some empirical evidence on the effects of shocks to
monetary policy on exchange rates”, Quarterly Journal of Economics, Vol. 110 No. 4,
pp. 975-1009.
Enders, W. (2010), Applied Econometric Time Series, 3rd ed.
Engle, R. and Granger, C. (1987), “Cointegration and error correction: representation, estimation,
and testing”, Econometrica, Vol. 55 No. 2, pp. 251-276.
Forni, M. and Gambetti, L. (2010), “The dynamic effects of monetary policy: a structural factor
model approach”, Journal of Monetary Economics, Vol. 57 No. 2, pp. 203-216.
Friedman, M. (1968), “Money and business cycle”, American Economic Review, Vol. 58 No. 1,
pp. 1-17.
Fullerton, T., Hattori, M. and Calderón, C. (2001), “Error correction exchange rate modeling:
evidence for Mexico”, Journal of Economics and Finance, Vol. 25 No. 3, pp. 358-368.
Goodfriend, M. (2005), “The monetary policy debate since October 1979: lessons for theory and
practice”, Federal Reserve Bank of St. Louis Review, Vol. 87 No. 2, pp. 243-262.
Jang, K. and Ogaki, M. (2004), “The effects of monetary policy shocks on exchange rates: a
structural vector error correction model approach”, Journal of the Japanese and
International Economies, Vol. 18 No. 1, pp. 99-114.
Johansen, S. (1995), Likelihood-Based Inference in Cointegrated Vector Autoregressive Models,
Oxford University Press, Oxford.
329
Monetary policy
shocks
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Khan, M. (2008), “Short-run effects of unanticipated changes in monetary policy: interpreting
macroeconomic dynamics in Pakistan”, State Bank of Pakistan Working Paper, 22.
Khan, S. and Qayyum, A. (2007), “Measures of monetary policy stance: the case of Pakistan”,
Pakistan Institute of Development Economics, Working Papers 2007 (39), pp. 1-24.
Kim, S. and Roubini, N. (2000), “Exchange rate anomalies in the industrial countries: a solution
with a structural VAR approach”, Journal of Monetary Economics, Vol. 45 No. 3,
pp. 561-586.
Leeper, E. and Gordon, D. (1992), “In search of the liquidity effect”, Journal of Monetary
Economics, Vol. 29 No. 3, pp. 341-369.
Qayyum, A. (2008), “Does monetary policy play effective role in controlling infation in Pakistan”.
Rashid, A. (2009), “Testing the modifed-combined Ppp and Uip hypothesis in South Asian
economies”, Applied Econometrics and International Development, Vol. 9 No. 1, pp. 199-218.
SBP (2009), “Statistical bulletins of State Bank of Pakistan”.
Sims, C. (1980), “Macroeconomics and reality”, Econometrica, Vol. 48 No. 1, pp. 1-48.
Sims, C. (1992), “Interpreting the macroeconomic time series facts: the effects of monetary policy”,
European Economic Review, Vol. 36 No. 5, pp. 975-1000.
Sims, C. (1998), “Comment on Glenn Rudebusch’s ‘Do measures of monetary policy in a VARmake
sense?’”, International Economic Review, Vol. 39 No. 4, pp. 933-941.
Wong, K. (2000), “Variability in the effects of monetary policy on economic activity”, Journal of
Money, Credit and Banking, Vol. 32 No. 2, pp. 179-198.
Further reading
Ahmed, S. and Ansari, M. (2008), “Does money matter? Evidence from vector error-correction for
Mexico”, The Journal of Developing Areas, Vol. 41 No. 1, pp. 185-202.
Bernanke, B. and Gertler, M. (1995), “Inside the black box: the credit channel of monetary policy
transmission”, The Journal of Economic Perspectives, Vol. 9 No. 4, pp. 27-48.
Campbell, J. and Perron, P. (1991), “Pitfalls and opportunities: what macroeconomists should
know about unit roots”, NBER Macroeconomics Annual, Vol. 6, pp. 141-201.
Clarida, R., Gali, J. and Gertler, M. (1999), “The science of monetary policy: a new Keynesian
perspective”, Journal of Economic Literature, Vol. 37 No. 4, pp. 1661-1707.
Corresponding author
Abdul Rashid can be contacted at: [email protected]
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
JFEP
6,4
330
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_937093837.pdf
This paper aims to empirically examine how shocks to monetary policy measures (the
short-term nominal interest rate and broad money supply) affect macroeconomic aggregates, namely,
output growth of the economy, national price levels and the nominal exchange rate.
Journal of Financial Economic Policy
The response of macroeconomic aggregates to monetary policy shocks in Pakistan
Abdul Rashid Zainab J ehan
Article information:
To cite this document:
Abdul Rashid Zainab J ehan , (2014),"The response of macroeconomic aggregates to monetary policy
shocks in Pakistan", J ournal of Financial Economic Policy, Vol. 6 Iss 4 pp. 314 - 330
Permanent link to this document:http://dx.doi.org/10.1108/J FEP-04-2013-0016
Downloaded on: 24 January 2016, At: 21:50 (PT)
References: this document contains references to 33 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 228 times since 2014*
Users who downloaded this article also downloaded:
Dogga Satyanarayana Murthy, Suresh Kumar Patra, Amaresh Samantaraya, (2014),"Trade Openness,
Financial Development Index and Economic Growth: Evidence from India (1971-2012)", J ournal of Financial
Economic Policy, Vol. 6 Iss 4 pp. 362-375http://dx.doi.org/10.1108/J FEP-10-2013-0056
Tess DeLean, J oseph P. J oyce, (2014),"Stock markets and the costs of banking crises", J ournal of Financial
Economic Policy, Vol. 6 Iss 4 pp. 342-361http://dx.doi.org/10.1108/J FEP-01-2014-0003
David Walker, (2014),"Market vs. administered Federal Reserve policy rates", J ournal of Financial Economic
Policy, Vol. 6 Iss 4 pp. 331-341http://dx.doi.org/10.1108/J FEP-05-2014-0032
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The response of macroeconomic
aggregates to monetary policy
shocks in Pakistan
Abdul Rashid
International Institute of Islamic Economics (IIIE), International Islamic
University, Islamabad, Pakistan, and
Zainab Jehan
Department of Economics, Fatima Jinnah Women University,
Rawalpindi, Pakistan
Abstract
Purpose – This paper aims to empirically examine how shocks to monetary policy measures (the
short-term nominal interest rate and broad money supply) affect macroeconomic aggregates, namely,
output growth of the economy, national price levels and the nominal exchange rate.
Design/methodology/approach – Johansen’s (1995) cointegration technique and error correction
models are used to explore the long-run relationship among variables. To investigate how
macroeconomic aggregates respond to a one-standard deviation shock to the underlying monetary
measures, the authors estimate impulse response functions based on error correction models. The
study uses quarterly data covering the period 1980-2009.
Findings – The results provide evidence that there is a long-run stable relationship between the
authors’ monetary measures and the underlying macroeconomic aggregates. They also fnd that
the industrial production adjusts at a faster speed relative to commodity prices and the exchange
rate over the examined period. Further, they show that the short-term interest rate has relatively
stronger effects on output as compared to broad money supply, whereas prices and exchange rates
adjust more quickly to their long-run equilibrium when money supply is used as a measure of
monetary policy. Finally, the authors fnd signifcant evidence of a price puzzle regardless of
whether they consider a closed or an open economy case. However, an initial appreciation
of exchange rate is observed in response to a one-standard deviation shock to money supply,
indicating the overshooting hypothesis phenomenon.
Practical implications – The fndings of the analysis suggest that the interest rate-oriented
monetary policy is more effective when the monetary authorities’ objective is to enhance the output
growth of the economy. However, in case of infation targeting, the broad money supply seems a
more appropriate instrument. Our fndings also suggest that the monetary policy has a signifcant
role in stabilizing both real and nominal sectors of the economy.
Originality/value – The main value of this paper is to examine the signifcance of monetary policy
for a developing and relatively small open economy, namely, Pakistan. The authors use the error
correction model, which improves the estimation by accounting for the long-run association. They
also take into account the world oil prices by including the world commodity price index as a
control variable in their empirical investigation. Finally, they utilize quarterly data rather than
annual, and they cover a relatively recent sample period.
JEL classifcation – C3, E4, E5
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1757-6385.htm
JFEP
6,4
314
Journal of Financial Economic Policy
Vol. 6 No. 4, 2014
pp. 314-330
©Emerald Group Publishing Limited
1757-6385
DOI 10.1108/JFEP-04-2013-0016
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Keywords Monetary policy, Convergence, Error correction model, Impulse response function,
Macroeconomic aggregates, Long-run relationship
Paper type Research paper
1. Introduction
Since the seminal work by Friedman (1968), the role of monetary policy in
macroeconomic stabilization is an inconclusive issue. Besides the development on
theoretical grounds, a substantial body of empirical literature has contributed to the
ongoing debate by providing signifcant evidence on how monetary policy affects
output growth, domestic prices and exchange rates. No doubt, the adoption of the
foating exchange rate system, the slogan of fnancial reforms, the trade liberalization
and relatively more independent central banks have enhanced the signifcance of
monetary policy. Therefore, both academics and policymakers are keen to understand
how, when and to what extent the economic aggregates respond to changes in monetary
policy.
In theory, the debate on monetary policy has evolved from policy ineffectiveness to
the identifcation of the long- and short-run impact of monetary policy on
macroeconomic performance. According to the monetarists, monetary policy is likely to
be effective in the short run but almost completely ineffective in the long run. While both
the New-Keynesian and classical schools of thought have similar views regarding the
neutrality of money in the long run, advocates of the former believe that monetary policy
may affect output and infation in the short run, as they presume that nominal wages are
rigid at least in the short run. On the other hand, rational expectation theory considers
expectations as crucially important for analyzing the effect of monetary policy (for
further theoretical debates, see Goodfriend (2005)).
Recent empirical studies such as Bernanke et al. (2005), Bernanke et al. (1998),
Eichenbaum et al. (1995) and Sims (1992) have signifcantly contributed to the
measurement of monetary policy and its innovations. These studies mainly utilize the
vector autoregressive methodology to measure the responsiveness of macroeconomic
aggregates to monetary policy shocks. Although the fndings of these studies provide
signifcant evidence on the response of macroeconomic variables, such as real economic
activity, national price levels and exchange rates, to changes in monetary policy, there
are number of measurement problems and various anomalies. These inconsistencies
generally include price, exchange rate and liquidity puzzles.
To overcome these issues, researchers have made numerous attempts to develop
advanced estimation methods and techniques measuring the shocks. Factor-augmented
vector autoregressive (FAVAR) approach developed by Bernanke et al. (2005) and
structural factor-augmented model proposed by Forni and Gambetti (2010) are
examples of these advancements[1].
In developing countries, monetary policy is more complicated and challenging due to
the lack of well-developed and deep fnancial markets, plus the weak channels of
transmission. Owing to the small fnancial markets and weak channels of transmission
in developing countries, the relationship between monetary measures and
macroeconomic aggregates tends to be weak and inconsistent. Above all, most of the
empirical studies have focused on developed countries. Thus, we know relatively less
how macroeconomic aggregates, such as output, national price levels and exchange
rates, respond to monetary policy shocks in developing countries. Yet, understanding
315
Monetary policy
shocks
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the effects of monetary policy on macroeconomic aggregates of developing countries is
of a great signifcance to policymakers and academics for enhancing macroeconomic
stability. Evidence on the effectiveness of monetary policy in developing countries
would enhance our overall understanding of using monetary policy as a tool of
macroeconomic stabilization.
Unlike most of the previous studies that have largely focused on developed countries,
this paper is focusing on a developing and relatively small open economy, namely,
Pakistan. Specifcally, the paper empirically investigates how shocks to monetary
policy measures – the short-termnominal interest rate and broad money growth – affect
economic aggregates: output growth of the economy, national price levels and the
nominal exchange rate. The study uses quarterly data covering the period from1980Q1
to 2009Q2. First, we study the effects of monetary policy on the long-run convergence of
these three macroeconomic aggregates. Second, we examine how these variables
respond to a one-standard deviation shock to the monetary measures.
To carry out our empirical investigation, we frst test the order of integration of the
variables by using the augmented Dickey and Fuller (1981) unit root tests. After
confrming the order of integration, we test for the long-run association among the
variables. Specifcally, we apply the Johansen’s (1995) cointegration process to testing
for the existence of the possible cointegration vectors. To examine the direction of the
short- and long-run causality, and to estimate the speed at which the variables converge
to its long-run equilibrium, we estimate the vector error correction model (hereafter
VECM). Finally, to investigate how macroeconomic aggregates respond to a
one-standard deviation shock to the monetary measures, we estimate impulse response
functions (hereafter IRFs) based on VECM. The empirical methodology starts with a
bivariate model of closed economy as a baseline model to which more explanatory
variables are gradually added. Finally, we extend our empirical model to an
open-economy model by incorporating bilateral nominal exchange rates and
international commodity prices. This approach enables us to examine howthe response
of underlying variable to monetary measures changes when we include more
information in the model. It also provides an interesting comparison between closed-
and open-economy models.
There is very little empirical literature for Pakistan on this topic. Qayyum (2008)
computes the monetary condition index (MCI) for Pakistan based on the estimated
weights of the measures of monetary policy, such as the interest rate and exchange rate.
Because the MCI index is more useful in the absence of supply shocks, and because the
supply shocks are dominant in Pakistan, the application of MCI for Pakistan is
questionable and unreliable. Therefore, the results of Qayyum’s study may not be
reliable[2]. Another study by Agha et al. (2005) uses six-month Treasury bill (T-bill)
rates as a measure of monetary policy and the VAR technique to examine the
effectiveness of monetary policy over a relatively short time span. Finally, recently,
Khan (2008) has made an attempt to investigate the impact of unanticipated changes in
monetary policy on output and infation by estimating structure VAR (SVAR). The
study uses nominal shocks in SVAR as a proxy for unexpected changes in monetary
policy. However, this measure suffers the problem of lack of theoretical rationales.
This paper contributes to the existing literature in three ways. First, we utilize
quarterly data rather than annual, and we cover a relatively recent sample period
spanning from 1980Q1 to 2009Q2. The use of quarterly data enables us not only to
JFEP
6,4
316
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
harvest the gain of higher degree of freedom but also allows us to use deeper lags of the
variables to identify a well-specifed model without losing the informational credibility
of the sample. Second, unlike most of the previous studies, we take into consideration the
time-series properties of the variables, such as non-stationary behavior. Finally, we use
VECMapproach, which improves the estimation over the SVARbecause the SVARdoes
not account for the long-run association. We also take into account the world oil prices
by including the world commodity price index as a control variable in our empirical
investigation.
The rest of the paper is organized as follows. Section 2 reviews the existing empirical
studies and highlights the strengths and weaknesses of their methodologies. Empirical
methods, data sources and the defnition of the variables are given in Section 3. Section 4
presents the empirical fndings. Finally, Section 5 concludes the study.
2. Literature survey
Forni and Gambetti (2010) examine the dynamic exogenous effect of monetary policy by
using a standard recursive scheme through a dynamic structural factor model for USA
covering the period 1973:3-2007:10. Their empirical analysis is based on the variables
which are used by Stock and Watson (1998). They argue that the factor analysis model
is superior to FAVAR proposed by Bernanke et al. (2005) because it helps in eliminating
the puzzles in monetary policy analysis. They fnd that a positive shock to Federal
Funds Rate (FFR) leads to an appreciation of real exchange rate. This confrms
overshooting hypothesis of Dornbusch (1976). Computing impulse response graphs,
they show the absence of price puzzle. Further, they argue that industrial production
falls, although temporary, to a large extent with a humped-shaped response.
Bjørnland (2008) examines the response of macroeconomic economic aggregates to
monetary policy by including the exchange rate in the model specifcation. He uses
quarterly data over the period 1993-2004. Further, he uses Cholesky ordering and the
Kim and Roubini’s (2000) identifcation procedures to determine the order of the
variables. Bjørnland (2008) shows that there is a temporary increase in the interest rate,
which normally takes four quarters to converge to its normal path. However, his
analysis does not provide any statistically signifcant evidence of the exchange rate
puzzle or price puzzle.
Ansari et al. (2007) explore the relationship between money income and domestic
prices by estimating VECM. They use both narrow and broad money as measures of
monetary policy. Using quarterly data, they document that for any divergence from
long-run equilibrium; output will increase by 6 per cent to adjust toward its long-run
equilibrium. Furthermore, they show that a positive shock to money policy leads to
adjustments in output after 5 quarters.
Bernanke et al. (2005) introduce a combination of VAR model and factor model to
capture large information set. They argue that a simple VAR analysis is unable to
incorporate such information. They use a diffusion index developed by Stock and
Watson (2002) to estimate the factors by utilizing a balanced panel of 120 monthly
macroeconomic series (1959:1-2001:8). Arecursive structure is assumed with identifying
assumption of no contemporaneous response of unobserved factors to monetary policy
shocks[3]. The comparison of 3-variable VAR model with two different FAVAR
specifcations reveals that the results fromthe standard VARmodel exhibit a signifcant
price puzzle and provide evidence of the inconsistent production response to a
317
Monetary policy
shocks
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
one-standard deviation monetary measure. However, the FAVAR approach improves
the results as price puzzle disappears after one year, real activity declines, monetary
aggregates fall and exchange rate appreciates for USA. As in Forni and Gambetti (2010),
as the study did not distinguish between the number of static factors and structural
shocks, a large number of economic restrictions are imposed to reach the identifcation.
Moreover, the restrictions are imposed on the IRFs of static factors instead of the IRF of
variables.
Jang and Ogaki (2004) examine the relationship between monetary policy shocks and
Dollar/Yen exchange rates, prices and output level for USA. The empirical analysis is
carried out, following the model of Jang and Ogaki (2004), through structural VECMand
VAR by employing long- and short-run restrictions on the model. They fnd that an
appreciation of exchange rate is the result of a contractionary monetary policy.
Furthermore, they fnd that output in domestic as well as in foreign country
signifcantly declines due to the long-run neutrality restrictions with an exception of
USAwhere a decline in output becomes negligible after four years. Finally, a fall in price
is observed as a result of tight monetary policy. While, estimating VECMand VARwith
short-run restrictions for variables in their levels, they fail to accept the UIP condition,
they fnd strong evidence in support of the existence of price puzzle.
Fullerton et al. (2001) apply an error correction model to study the behavior of the
exchange rate for Mexican peso over the period 1976-2000. The variables included in the
model are nominal exchange rates, consumer price index, liquid international reserves,
money supply and real gross domestic product (GDP) as non-policy variables while
one-and three-month T-bills rates as policy variables. Their fndings based on the
balance of payment framework and monetary model of exchange rate do not provide
any support to the established theory. However, balance of payment framework with
one-month T-bill rate is marginally better than the monetary model of exchange rate.
Wong (2000) empirically investigates the impact of monetary policy on
macroeconomic variables by applying a time-varying parameter model for USA over
the period 1959:1-1994:12. Output and prices are assumed to have lagged effect but FFR
and reserves are considered to have only contemporaneous effects. The rolling VARhas
been estimated with maximum three lags. The empirical results suggest that output
increases in response to a contractionary shock to monetary policy. The output is more
responsive to shocks during periods when the central bank adopts infation controlling
policy, whereas, it is less responsive when the central bank aims at promoting economic
growth. Overall, the plots of IRF provide the evidence of the presence of price puzzle.
Bernanke and Mihov (1998) develop a VAR-based methodology to measure and
assess the impact of monetary policy on macroeconomic variables. The measure of MP
is derived from an estimated model of Central Bank’s operating procedures and the
market for commercial bank reserves, which makes it more consistent than the
previously used instruments of monetary policy. The model has been estimated for
different time periods of post 1965-1996 for USA. The exogenous policy shocks are
computed through a standard VAR method by applying generalized methods of
moments in which the policy variables are placed last in variable ordering. The IRFs
indicate that there is an increase in output in response to an expansionary monetary
policy. Further, the plots provide evidence of a slower but a persistent rise in the prices.
Yet, their results considerably vary across different measures of monetary policy.
JFEP
6,4
318
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Although the study attempts to capture all the possible measures of monetary policy, it
fails to notify which of the measure is relatively more effective.
Eichenbaum et al. (1995) analyze the exchange rate transmission mechanism of
monetary policy for the period 1974:1-1990:5. They use three measures of monetary
policy commonly used in the literature. These measures are FFR, non-borrowed
reserves and the narrative measure of Romer and Romer (1989). They estimate a
multivariate VAR model by using the ordering of the variables based on the Wold
decomposition. The estimates on IRFs reveal that a contractionary monetary policy
leads to a signifcant and continual decline in US interest rate, a sharp and persistent
appreciation in US exchange rate, which is contradictory with the overshooting
hypothesis of exchange rate.
3. Empirical methodology, data and variable defnitions
3.1 Estimation methods
We use the Johansen’s (1995) procedure to test whether there is any long-run association
among the variables. The Johansen’s method is based on the relationship between the
rank of matrix and its characteristic roots (Enders, 2010). Specifcally, the model is
expressed as follows:
?X
t
? ?X
t?1
? ?
i?1
p?1
?
i
?X
t?i
? ?D
t
? ?
t
(1)
where X
t
is a vector of n endogenous variables and D
t
is a vector of m deterministic
variables. Furthermore, ?, ?
i
and ? are coeffcient matrices of dimension n ?n, n ?n
and n ?m, respectively. ?may have reduced rank, and hence, it can be informative to
decompose it with?? ??
/
, where both? and?are of dimensionn ?r andr is the rank
of ?.
As in Engle and Granger (1987), the dynamic behavior of a set of integrated variables
can be empirically analyzed through the VECM, which is the reduced formof the model.
The selected model is based on the backward looking behavior of output, domestic
prices and the exchange rate. The study employs a bivariate closed-economy model as in
Sims (1980) and Christiano et al. (1999), which is then frst extended to a multivariate and
fnally to an open-economy model to measure the relationship between macroeconomic
aggregates – output, domestic prices and the exchange rate – and monetary policy. In
matrix notation, the VECM can be written as follows:
?X
t
??
0
??
1
t ??X
t?1
?
?
i?1
p?1
?
i
?X
t?i
?
?
i?1
p?1
?
i
?Y
t
?e
t
(2)
where t ? 1, 2, …, T, X
t
? n ?1 vector of n endogenous I(I) variables included in the
VECM. These variables include industrial production, domestic prices and the exchange
rate. Y
t
?n ?1 vector of exogenous I(I) variables broad money supply, money market
lending rate and international commodity prices index. ?
0
?n ?1 vector of intercepts.
?
1
, ?
i
and ?
i
are n ?n matrices of coeffcients. e
t
? n ?1 vector of error terms
distributed as i.i.d. and fulflls the Gaussian properties of zero mean and constant
variances. ??matrix of parameters such that one element is non-zero. Moreover, ? is
the difference operator and all the variables are in log form except the interest rate.
319
Monetary policy
shocks
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The long-run behavior of the system depends on the rank of ?. Granger
representation theorem shows that if X
t
is integrated of order r then one can write
? ? ??
/
, where matrix ? contains matrix of r cointegrating vector. The matrix ?
known as the speed of adjustment, which measures how quickly ?X
t
reacts to a
deviation from equilibrium (Engle and Granger, 1987).
3.2 Data and defnition of variables
To examine the response of macroeconomic aggregates to monetary policy for Pakistan,
we use quarterly data covering the period 1980-2009. Pakistan has moved from fxed
exchange rate systemto managed exchange rate system; therefore, for the model where
the exchange rate is used, the time period starts form 1990Q1 to 2009Q2. All the data
except broad money are obtained fromthe International Financial Statistics database of
International Monetary Fund. Data on money supply are taken fromStatistical Bulletins
of Pakistan published by the State Bank of Pakistan (SBP, 2009). All the variables are in
log form except short-term interest rate. All the variables are on annual basis with
millions of Pak Rupee as unit of measurement. World commodity price index does not
have data for the full length of the sample period under investigation; therefore, few
missing values are interpolated by applying a two-quarter moving average formula.
Following prior studies, we use two alternative measures of monetary policy,
namely, broad money supply and the short-term interest rate. Bernanke and Blinder
(1992) and Sims (1998) argue that short-term interest rate is a superior measure of
monetary policy and it should be preferable over money supply. However, Berument
(2003) suggests that broad money is a better indicator of monetary policy in a small open
economy. In our empirical investigation, we utilize both measures with an aim to do a
comparison between both the said measures. Moreover, we use money supply as a
measure of monetary policy, as in Pakistan, it has been used to formulate the monetary
policy. However, recently, the monetary authority in Pakistan is giving relatively more
weightage to short-term interest rate.
4. Empirical fndings
We apply the ADF test to explore the order of integration of the underlying series. The
optimal lag structure for the ADF equations is selected based on AIC. The results are
given in Table I. The optimal lag length is market by “a”[4]. The ADF test results do not
provide any signifcant evidence to reject the null hypothesis of unit root for level series.
However, the frst difference of all the series appears stationary. These fndings are
robust across different lag lengths used in the estimation of the ADFequation. Thus, the
results suggest that the variables are integrated of order one.
To examine the long-run association, we apply the Johansen’s (1995) cointegration
test. Table II reports the results. As the results of the cointegration are very sensitive to
the lag length, we select the optimal lag length by applying the Bayesian Information
Criterion. Further, we also ensure that the estimated model has white noise
disturbances. We start by estimating the response of the underlying macroeconomic
aggregates to our both measures of monetary policy in bivariate framework, separately.
We then extend the bivariate model to the multivariate model by incorporating other
control variables. One should note that when we include the exchange rate in the
specifcations our model represents a case of an open economy[5].
JFEP
6,4
320
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The results reported in Table II provide evidence that there is a long-run relationship
between industrial production index and the short-terminterest rate, implying that both
variables have co-movement in the long run. This evidence is robust to the inclusion of
other variables in the specifcation. Our results suggest that there is a single long-run
stable relationship between industrial production index and monetary policy measures
regardless of whether we estimate model in the bivariate or multivariate framework.
The existence of the one cointegrating vector at different specifcations confrms the
validity of our results. When we turn to examine the long-run association between the
short-terminterest rate and our second response variable, namely, domestic price levels,
similar to the case of industrial production index, we fnd that there is only one
signifcant cointegrating vector. This observation holds true even when we include
industrial production index as an explanatory variable into the specifcation.
Next, we examine the long-run relationship by estimating the model for the case of an
open economy. Specifcally, we extend our closed-economy model to open economy
model by including, one-by-one, exchange rates and international commodity prices into
Table I.
Unit-root test results
Variables Model
Number of lags
Conclusion 8 6 4 2 1
Level series
LIPI Drift and trend ?2.742 ?2.743 ?1.176
a
?2.654 ?2.876 Non-stationary
Drift ?1.524 ?1.674 ?1.519
a
?1.765 ?1.543 Non-stationary
LCPI Drift and trend ?3.205 ?1.869 ?2.144
a
?1.234 ?1.234 Non-stationary
Drift 1.231 0.587
a
0.460 0.976 1.876 Non-stationary
LER Drift and trend ?1.345 ?1.302 ?1.654 ?1.854 ?0.679
a
Non-stationary
Drift ?0.657 ?0.674 -0.897 -0.964 ?0.987
a
Non-stationary
LCOMP Drift and trend ?2.077 ?1.567 ?1.974
a
?1.356 ?1.760 Non-stationary
Drift ?1.534 ?0.834 ?1.876 ?0.653
a
?0.365 Non-stationary
SR Drift and trend ?2.327 ?2.074 ?1.098 ?2.296
a
?2.645 Non-stationary
Drift ?2.730 ?2.198 ?1.832 ?2.543
a
?2.987 Non-stationary
LBMS Drift and trend ?2.089 ?2.010 ?3.198
a
?2.876 ?1.787 Non-stationary
Drift ?0.632 ?0.435 ?0.866
a
?0.643 ?0.465 Non-stationary
Firs-differenced series
LIPI Drift and trend ?2.435 ?2.830 ?3.807
a
?22.876 ?24.877 Stationary
Drift ?2.760 ?2.675 ?3.976
a
?22.643 ?23.159 Stationary
LCPI Drift and Trend ?1.856 ?1.980 ?3.680
a
?4.987 ?7.574 Stationary
Drift ?1.287 ?1.765 ?3.435
a
?3.754 ?7.334 Stationary
LER Drift and Trend ?2.763 ?4.123 ?4.965
a
?6.764 ?7.155 Stationary
Drift ?3.543 ?4.945 ?4.896
a
?6.210 ?7.765 Stationary
SR Drift and Trend ?3.987 ?4.987 ?5.265 ?9.976 ?11.876 Stationary
Drift ?3.734 ?4.976 ?5.109 ?10.145 ?11.643 Stationary
LCOMP Drift and Trend ?2.765 ?3.124 ?4.087 ?5.905
a
?7.111 Stationary
Drift ?2.845 ?3.976 ?3.900 ?5.687
a
?7.787 Stationary
LBMS Drift and Trend ?2.974 ?4.943 ?4.3743 ?7.365
a
?6.243 Stationary
Drift ?2.987 ?4.356 ?4.930 ?7.384
a
?6.296
Notes:
a
Represents optimum lag length selected by the Akaik Information Criterion (AIC); LIPI:
log(industrial production index); LCPI: log(consumer price index); LER: log(exchange rate); LCOMP:
log(world commodity price index); SR: sort-term interest rate; and LBMS: log(broad money supply)
321
Monetary policy
shocks
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the specifcation. The results are given in Table III. As we can see from the table, the
trace statistics signify the presence of one-cointegration relationship in both cases.
Third macroeconomic aggregate that we consider in our analysis is the bilateral
nominal exchange rate between PAK rupee and US dollar. Adopting the previous
strategy here too, we frst estimate a bivariate model, which is then extended to
multivariate model. The estimated eigenvalues showthat short-termrate of interest and
exchange rate appear to have a single linear combination in the long run, which is I(0).
We also fnd the evidence of the existence of the long-run relationship when we extend
the model by including domestic price levels and industrial production index.
As mentioned earlier, we also use an alternative measure of monetary policy, namely,
broad money supply. The results of the cointegration tests with this measure are given
in Table III. In general, the fndings are consistent with the results reported in Table II.
However, in two of the cases, we fnd two signifcant cointegrating vectors instead of
one. In case of more than one cointegrating vectors, the general practice is to select the
cointegrating vector, which has highest eigenvalue because it is most associated with
the stationary part of the model (Rashid, 2009). Following this, we select frst
cointegration vector as it has the highest value. The broad money supply appears to
have the long-run co-movement with the exchange rate regardless of whether the model
is estimated in bivariate or multivariate framework.
After confrming the presence of the long-run relationship, we estimate VECM for
each model to examine the speed of adjustment toward the long-run equilibrium. When
we regress the industrial production on short-term money market rate, we fnd that the
Table II.
Results for cointegration
test and VECM (Policy
variable: short-term
interest rate)
Model specifcation
No.
(Lags) Rank Q
r
Eigenvalue ECT
No.
(Lags)
Model-1: Dependent variable; LIPI
LIPI, SR 114 (4) 1 2.658 0.2094 0.0208*** (0.000) 114 (4)
LIPI, LCPI, SR 112 (6) 1 11.104 0.1836 ?0.0762*** (0.000) 114 (4)
LIPI, LCPI, LER, SR 115 (3) 1 28.062 0.1638 ?0.5622*** (0.000) 116 (2)
LIPI, LCPI, LER, LCOMP, SR 115 (3) 1 47.082 0.2452 ?0.6842*** (0.000) 116 (2)
Model-2: Dependent variable; LCPI
LCPI, SR 117 (1) 1 0.439 0.1601 0.0008* (0.052) 109 (9)
LCPI, LIPI, SR 114 (4) 1 9.934 0.1823 ?0.0199** (0.000) 111 (7)
LCPI, LIPI, LCOMP, SR 116 (2) 1 22.975 0.4584 ?0.0157*** (0.000) 116 (2)
LCPI, LIPI, LER, LCOMP, SR 115 (3) 1 47.082 0.2452 ?0.032*** (0.000) 111 (7)
Model-3: Dependent variable; LER
LER, SR 78 (7) 1 1.0381* 0.1752 ?0.0038** (0.017) 78 (7)
LER, LIPI, SR 78 (4) 1 12.7371* 0.2521 ?0.1682*** (0.001) 78 (6)
LER, LIPI, LCPI, SR 78 (4) 1 25.1796* 0.4194 ?0.2978*** (0.002) 78 (6)
Notes: This table displays the estimates for short-run interest rate as measure of monetary policy.
Lags in 2nd column refer to the number of lags for cointegrating vector, while lags in 7th column are for
estimation of vector error correction term; Qr is the LR trace statistic. ECT denotes error correction
term; LIPI: log(industrial production index); LCPI: log(consumer price index); LER: log(exchange rate);
LCOMP: log(world commodity price index); SR: sort-term interest rate; and LBMS: log(broad money
supply); ***, **, * denote the signifcance at the 1, 5 and 10 per cent levels, respectively
JFEP
6,4
322
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
error term is statistically negatively signifcant. The negative sign of the coeffcient
associated with error term is in line with the adjustment process, suggesting that
industrial production converges to its long-run equilibrium over the time. Specifcally,
the coeffcient of error term is 0.021, implying that any disequilibrium in industrial
production will be adjusted at the rate of 2.1 per cent in one quarter. However, the speed
of adjustment becomes 7.6 per cent per quarter when we add domestic prices to the
specifcation.
The estimates on the speed of adjustment have considerably been improved in the
case of the open-economy model. For instance, the disequilibrium in industrial
production is now corrected by the rate of 56.2 per cent in one quarter, while the
inclusion of international commodity prices into the model further increases the
adjustment speed to 68.42 per cent per quarter. This suggests that the adjustment
process of industrial production to disequilibrium is relatively fast in an open economy
framework.
When we estimate the model of industrial production using an alternative measure of
monetary policy, namely, broad money supply, the speed of adjustment is remarkably
high for the baseline model, suggesting that the disequilibrium is corrected by the rate
of 76 per cent in one quarter. Contrary to the frst measure, the rate of convergence
declines with the addition of more information into the model. This fnding suggests
that when the monetary policy is measured by the short-term interest rate it is more
effective for an open economy, whereas the measures of money supply has signifcant
role to play in a closed economy. These fndings make sense as theoretically the interest
Table III.
Results for cointegration
test and VECM (Policy
variable: broad money
supply)
Model specifcation
No.
(Lags) Rank Qr
Eigen
value ECT
No.
(Lags)
Model-1: Dependent variable; LIPI
LIPI, LBMS 116 (2) 1 0.0004 0.3743 ?0.7595*** (0.000) 116 (2)
LIPI, LCPI, LBMS 113 (5) 1 9.033 0.2365 ?0.3795*** (0.000) 113 (5)
LIPI, LCPI, LER, LBMS 116 (2) 1 28.742 0.5715 ?0.3571*** (0.000) 113 (5)
LIPI, LCPI, LER, LCOMP, LBMS 115 (3) 2 27.761 0.2286 ?0.4549*** (0.000) 113 (5)
Model-2: Dependent variable: LCPI
LCPI, LBMS 109 (9) 1 0.4011 0.1343 ?0.0573*** (0.002) 113 (5)
LCPI, LIPI, LBMS 113 (5) 1 9.0338 0.2365 ?0.1524*** (0.003) 110 (8)
LCPI, LIPI, LCOMP, LBMS 114 (4) 1 20.1580 0.3805 ?0.1441 (0.033) 112 (6)
LCPI, LIPI, LER, LCOMP, LBMS 115 (3) 2 27.7615 0.2286 ?0.2207** (0.007) 112 (6)
Model-3: Dependent variable; LER
LER, LBMS 78 (11) 1 2.526 0.1578 ?0.0804** (0.019) 78 (4)
LER, LIPI, LBMS 78 (2) 1 9.806 0.2813 ?0.2486*** (0.000) 78 (5)
LER, LIPI, LCPI, LBMS 78 (4) 1 26.798 0.3336 ?0.3298*** (0.000) 78 (6)
Notes: This table displays the estimates for money supply (M2) as measure of monetary policy; Lags
in 2nd column refer to the number of lags for cointegrating vector while lags in 7th column are for
estimation of vector error correction term; Qr is the LRtrace statistic; ECTdenotes error correction term;
LIPI: log(industrial production index); LCPI: log(consumer price index); LER: log(exchange rate);
LCOMP: log(world commodity price index); SR: sort-term interest rate; and LBMS: log(broad money
supply); ***, **denote the signifcance at the 1 and 5 and per cent level, respectively
323
Monetary policy
shocks
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
rate plays a signifcant role in external capital fows along with its internal effectiveness,
whereas money supply is relatively more effective to make adjustments in domestic
accounts.
Regarding the long-run relationship between domestic prices and the short-term
nominal interest rate, we fnd that prices diverge fromtheir long-run equilibriumby 0.08
per cent per quarter. It is interesting to note that the inclusion of industrial production to
the model yields statistically negatively signifcant estimates on error correction term.
This implies that the short-run interest rate may affect price levels through its effects on
industrial output. Next, we include the exchange rate into the model to represent the case
of an open economy. Based upon one cointegration vector, the long-run equilibrium
relationship appears to be statistically negatively signifcant. A 1.6 per cent pace is
identifed by the error correction termwith which the disequilibriumwill be corrected by
domestic prices. However, the extension of the model by including industrial production
further improves the speed of adjustment to 3.2 per cent. Thus, there is evidence that
prices respond more quickly, and the pace of movement to the equilibrium point has
been increased by moving from a closed-economy to an open-economy model.
The model of domestic prices is empirically tested for long-run equilibrium
relationship by using the alternative measure of monetary policy as well. In a bivariate
closed-economy situation, there is a negative and statistically signifcant long-run
relationship as identifed by the error correction term. The system converges to its
long-run equilibriumposition at the rate of 5.7 per cent per quarter. In the open-economy
framework, adding international commodity prices once again increases the speed of
adjustment toward the long-run equilibrium.
For the third dependent variable, namely, the nominal exchange rate, the error
correction term appears negative and statistically signifcant. However, the magnitude
of the coeffcient indicates that the system converges to its long-run equilibrium by a
marginal rate. However, including industrial production to the model, we fnd that the
adjustment speed has been increased to 17 per cent per quarter, while the inclusion of
price levels into the specifcation further enhances the process of adjustments (at the rate
of 30 per cent per quarter) toward their long-run equilibrium position. These results
suggest that the both industrial production and prices have a signifcant role to play in
adjustment mechanism.
Regarding the money supply as a measure of monetary policy, we fnd that the speed
of adjustment is 8.04 per cent for exchange rate – money supply model. In addition, the
inclusion of domestic prices in the exchange rate model increases the speed of
adjustment to 25.0 per cent in one quarter. Further, when we add industrial production
to the specifcation, the estimates provide evidence that any deviation from the
equilibrium is adjusted to the long-run equilibrium position with the rate of 33 per cent
per quarter.
Finally, we compute the IRFs to examine the response of industrial production,
domestic prices and the exchange rate to a one-standard deviation shock to our
monetary measures. Similar to the case of cointegration and VECMestimation, we start
from a model of closed economy (a bivariate case), and we then extend the model to an
open economy by incorporating the exchange rate and international commodity prices
index into our specifcation. One of the major problems in IRFs is the sensitivity with
respect to variables ordering in the system. Therefore, following the empirical literature,
such as Forni and Gambetti (2010), Bernanke et al. (2005), Bjørnland (2008) and
JFEP
6,4
324
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Holtemoller (2004), this study assumes a recursive structure of ordering for the closed
economy in which policy variables are ordered at last. This implies that macroeconomic
aggregates do not respond contemporaneously to monetary policy innovations but
monetary policy might react toward any news frommacro aggregates within the period.
This is consistent with the transmission mechanism of monetary policy as highlighted
by several empirical studies, such as Svensson (1997). In the case of closed economy, we
order the variables as LIPI
t
, LCPI
t
, and SR
t
.
In an open-economy model, the exchange rate is placed last in the order of variables
as suggested by Eichenbaum et al. (1995). It ensures a lagged response of monetary
policy due to any exchange rate led shock but this identifcation results in a delayed
exchange rate response to monetary policy (Bjørnland, 2008). Kim and Roubini (2000)
propose a contemporaneous interaction between monetary policy and the exchange rate
to solve the problem of exchange rate puzzle. We employ Kim and Roubini (2000)
methodology and introduce the exchange rate and monetary policy interaction of
contemporaneous impact (by reversing the place in variable ordering for IRF). As
Pakistan is a small open economy, international prices are assumed exogenous for the
economy. In other words, central bank does not have international prices in its
information set (Juang et al., 2003). Therefore, international commodity prices are placed
after the exchange rate. Specifcally, in our case, the variable ordering for the
open-economy model is LIPI
t
, LCPI
t
, SR
t
, LER
t
, and LCOMP
t
.
The IRFs for the closed economy by using money market rate and broad money as
measures of monetary policy are presented in Figures 1 and 2, respectively. In the
bilateral model, a one-standard deviation shock to the short-terminterest rate appears to
have a negative impact on industrial production with a margin positive start. The
overall impact is composed of downward and upward fuctuations of industrial
production curve with a declining trend. It is interesting to note that the inclusion of
other variables in the baseline model changes neither the initial nor the long-run
response of industrial production to a one-standard deviation shock to money market
rate. These fndings are consistent with the results of prior empirical studies, such as
Forni and Gambetti (2010) and Bjørnland (2008), who also report a negative response of
output to monetary policy shocks. The response of industrial production remains
negative with respect to short-term interest rate shocks even for a multivariate
closed-economy model.
Domestic prices respond positively, providing the evidence of the existence of price
puzzle. Sims (1992) presents similar evidence. However, our fndings are in contrast to
Ogaki et al. (2003), who fnd that there is no price puzzle[6]. The inclusion of exchange
rate and international commodity prices improves the response of industrial production
to positive monetary policy innovations. A positive shock to money market rate
decreases the industrial production at a sharp rate before it starts increasing. This fact
is in line with the fndings of Bernanke et al. (2005) and Forni and Gambetti (2010) that
there is a decline in output after a positive shock to FFR.
On the other hand, when we consider the case of an open economy, a sharp and
abrupt increase in price level is observed after a shock to monetary policy. This point
outs price puzzles (Leeper et al., 1992; Bernanke and Blinder, 1992). Sims (1992) have
suggested that price puzzle can be tackled by including the international commodity
prices but in our case, the inclusion of international prices does not provide any
signifcant help in eliminating the puzzle. Yet, our fnding is consistent with
325
Monetary policy
shocks
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Eichenbaumet al. (1995), who argue that the inclusion of international commodity prices
does not improve the responsiveness of prices toward monetary policy innovations.
As one can see from the fgure, the initial response of exchange rate to monetary
policy innovations is almost zero. However, later on it appreciates followed by a
considerable depreciation, confrming the overshooting hypothesis of Dornbusch (1976).
This suggests that a contractionary monetary policy would lead to an appreciation in
the nominal exchange rate before it depreciates in the long run. However, we fnd the
evidence of exchange rate puzzle when we include international commodity price in the
model.
The results for the alternative measure of monetary policy are consistent with the
baseline model as depicted by IRFs. In the bivariate closed-economy models, the
LIPI LCPI LER
?0.06
?0.04
?0.02
0
0 50
VEC5, R, LIPI
step
Graphs by irfname, impulse variable, and response variable
0
0.005
0.01
0.015
0 50
VEC6, R, LCPI
step
Graphs by irfname, impulse variable, and response variable
?0.04
?0.02
0
0.02
0
VEC6, R, LIPI
step
Graphs by irfname, impulse variable, and response vari
0
.005
.01
.015
0 50
VEC6, R, LCPI
step
Graphs by irfname, impulse variable, and response variable
?0.05
0
0 50
VEC7, R, LIPI
step
Graphs by irfname, impulse variable, and response variable
0
0.002
0.004
0.006
0 50
VEC7, R, LCPI
step
Graphs by irfname, impulse variable, and response variable
?0.003
?0.002
?0.001
0
0 50
VEC7, R, LER
step
Graphs by irfname, impulse variable, and response variable
?0.06
?0.04
?0.02
0
0 50
VEC8, R, LIPI
step
Graphs by irfname, impulse variable, and response variable
0
0.001
0.002
0.003
0.004
0 50
VEC8, R, LCPI
step
Graphs by irfname, impulse variable, and response variable
?0.002
?0.001
0
?0.001
0 50
VEC8, R, LER
step
Graphs by irfname, impulse variable, and response variable
Figure 1.
IRFs: effects of shocks to
short-term interest rate
JFEP
6,4
326
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
industrial production responds to broad money supply-led shocks non-monotonically.
Specifcally, a positive shock to broad money supply appears to increase the industrial
production at frst, then it results in a decline in the production, and fnally, it increases
the production. When the model is extended to an open economy by incorporating the
exchange rate and international commodity prices, the response of industrial production
remains similar to our earlier results. Overall, our results are in line with Berument
(2003) and Ansari et al. (2007), who also fnd a positive response of GDP to a positive
shock to money supply. We did not fnd any evidence of price puzzle when broad money
supply is used as a measure of monetary policy. This fnding is contrast to the case when
we use money market rates as a measure of monetary policy. These fndings are similar
to Juang et al. (2003), Forni and Gambetti (2010), and Bernanke et al. (2005). Similarly, the
exchange rate appears to appreciate in response to a positive shock to monetary policy
LIPI LCPI LER
0
0.005
0.01
0.015
0 50
VEC1, LM, LIPI
step
Graphs by irfname, impulse variable, and response variable
0
0.005
0.01
0.015
0 50
VEC2, LM, LCPI
step
Graphs by irfname, impulse variable, and response variable
0
0.01
0.02
0.03
0 50
VEC2, LM, LIPI
step
Graphs by irfname, impulse variable, and response variable
0
0.005
0.01
0.015
0 50
VEC2, LM, LCPI
step
Graphs by irfname, impulse variable, and response variable
?0.01
0
0.01
0.02
0.03
0 50
VEC3, LM, LIPI
step
Graphs by irfname, impulse variable, and response variable
0
0.005
0.01
?0.015
0 50
VEC3, LM, LCPI
step
Graphs by irfname, impulse variable, and response variable
?0.01
?0.005
0
0.005
0 50
VEC3, Lm, LER
step
Graphs by irfname, impulse variable, and response variable
?0.01
0
0.01
?0.02
0 50
VEC4, LM, LIPI
step
Graphs by irfname, impulse variable, and response variable
0
0.005
0.01
0.015
0 50
VEC4, LM LCPI
step
Graphs by irfname, impulse variable, and response variable
0
0.005
0.01
0 50
VEC4, LM, LER
step
Graphs by irfname, impulse variable, and response variable
Figure 2.
IRFs: effects of shocks to
broad money supply
327
Monetary policy
shocks
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and then depreciates. This confrms the overshooting hypothesis of Dornbusch (1976)
and provides no evidence of any exchange rate puzzle[7].
Collectively, our results suggest that there is a signifcant interaction between the
real and monetary side of the economy in Pakistan. The statistical signifcance of the
error correction term clearly refects the convergence to the long-run equilibrium of
industrial production, prices and the exchange rate. This holds for both measures of
monetary policy we used in our analysis. In other words, our results suggest that that
there is a signifcant response of the macroeconomic aggregates to shocks to monetary
policy.
5. Conclusions and policy implications
The study empirically analyzes the response of macroeconomic aggregates, such as real
economic activity proxied by industrial production, national price levels and the
exchange rate, to shocks to monetary policy for Pakistan during the period 1980:
Q1-2009:Q2. The fndings are summarized as follows. First, we fnd that there is a
signifcant co-movement between macroeconomic aggregates and our monetary policy
measures, namely, the short-term nominal interest rate and broad money supply.
Second, the estimates of error correction term provide evidence that the industrial
production adjusts at faster speed relative to prices and the exchange rate over the
examined period. Finally, we show that the short-term interest rate has relatively
stronger effects on output as compared to broad money supply, whereas prices and
exchange rates adjust more quickly to their long-run equilibriumwhen money supply is
used as a measure of monetary policy.
The graphs of IRF provide the evidence of price and exchange rate puzzles when
money market rate is used as monetary policy. In contrast, when broad money supply is
used as a measure of monetary policy, no evidence of exchange rate puzzle is witnessed.
This fnding is in line with the Dornbusch (1976) overshooting hypothesis. The fndings
of the analysis suggest that the interest rate-oriented monetary policy is more effective
when the monetary authorities have objectives to enhance the output growth of the
economy. However, if the objective is to control the infation rate, then the broad money
supply seems more appropriate instrument. Furthermore, our fndings suggest that the
monetary policy has a signifcant role in stabilizing both real and nominal sectors of the
economy.
Notes
1. Despite a large body of empirical work using intensifed methodological applications, the
fndings are still inconclusive at best (for further details, see Bjørnland (2009)).
2. Although Khan and Qayyum (2004) provide empirical evidence of the superiority of the MCI
over Bernanke and Mahiv (1995) measure of monetary policy while measuring the
macroeconomic impact of monetary policy for Pakistan, Bernanke and Mihov’s measure of
monetary policy is better theoretically, as it uses more fnancial variables, which play an
important role in monetary policy formulation.
3. Monetary policy variable is placed last in the variable ordering.
4. However, we also estimate the ADF equations with other lag lengths to check the robustness
of the ADF results at different lags.
JFEP
6,4
328
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
5. Althoughwe use response variables as control variables as well, we estimate a separate model
for each response variable to test for cointegration.
6. However, he used different technique namely SVAR.
7. Bjørnland (2008) and Forni and Gambetti (2010) also fnd an appreciation of ERin response to
a positive shock to monetary policy.
References
Agha, A., Ahmed, I., Mubarik, N., Ali, Y. and Shah, H. (2005), “Transmission mechanism of
monetary policy in Pakistan”, SBP-Research Bulletin, Vol. 1 No. 1, pp. 1-23.
Bernanke, B. and Blinder, A. (1992), “The federal funds rate and the channels of monetary
transmission”, The American Economic Review, Vol. 82 No. 4, pp. 901-921.
Bernanke, B. and Mihov, I. (1998), “Measuring monetary policy”, Quarterly Journal of Economics,
Vol. 113 No. 3, pp. 869-902.
Bernanke, B., Boivin, J. and Eliasz, P. (2005), “Measuring the effects of monetary policy: a
factor-augmented vector autoregressive (FAVAR) approach”, Quarterly Journal of
Economics, Vol. 120 No. 1, pp. 387-422.
Bjørnland, H. (2008), “Monetary policy and exchange rate interactions in a small open economy”,
Scandinavian Journal of Economics, Vol. 110 No. 1, pp. 197-221.
Bjørnland, H. (2009), “Monetary policy and exchange rate overshooting: Dornbusch was right
after all”, Journal of International Economics, Vol. 79 No. 1, pp. 64-77.
Christiano, L., Eichenbaum, M. and Evans, C. (1999), “Monetary policy shocks: what have we
learned and to what end?”, Handbook of Macroeconomics, Vol. 1, pp. 65-148.
Dickey, D. and Fuller, W. (1981), “Likelihood ratio statistics for autoregressive time series with a
unit root”, Econometrica, Vol. 49 No. 4, pp. 1057-1072.
Dornbusch, R. (1976), “Expectations and exchange rate dynamics”, The Journal of Political
Economy, Vol. 84 No. 6, pp. 1161-1176.
Eichenbaum, M. and Evans, C. (1995), “Some empirical evidence on the effects of shocks to
monetary policy on exchange rates”, Quarterly Journal of Economics, Vol. 110 No. 4,
pp. 975-1009.
Enders, W. (2010), Applied Econometric Time Series, 3rd ed.
Engle, R. and Granger, C. (1987), “Cointegration and error correction: representation, estimation,
and testing”, Econometrica, Vol. 55 No. 2, pp. 251-276.
Forni, M. and Gambetti, L. (2010), “The dynamic effects of monetary policy: a structural factor
model approach”, Journal of Monetary Economics, Vol. 57 No. 2, pp. 203-216.
Friedman, M. (1968), “Money and business cycle”, American Economic Review, Vol. 58 No. 1,
pp. 1-17.
Fullerton, T., Hattori, M. and Calderón, C. (2001), “Error correction exchange rate modeling:
evidence for Mexico”, Journal of Economics and Finance, Vol. 25 No. 3, pp. 358-368.
Goodfriend, M. (2005), “The monetary policy debate since October 1979: lessons for theory and
practice”, Federal Reserve Bank of St. Louis Review, Vol. 87 No. 2, pp. 243-262.
Jang, K. and Ogaki, M. (2004), “The effects of monetary policy shocks on exchange rates: a
structural vector error correction model approach”, Journal of the Japanese and
International Economies, Vol. 18 No. 1, pp. 99-114.
Johansen, S. (1995), Likelihood-Based Inference in Cointegrated Vector Autoregressive Models,
Oxford University Press, Oxford.
329
Monetary policy
shocks
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Khan, M. (2008), “Short-run effects of unanticipated changes in monetary policy: interpreting
macroeconomic dynamics in Pakistan”, State Bank of Pakistan Working Paper, 22.
Khan, S. and Qayyum, A. (2007), “Measures of monetary policy stance: the case of Pakistan”,
Pakistan Institute of Development Economics, Working Papers 2007 (39), pp. 1-24.
Kim, S. and Roubini, N. (2000), “Exchange rate anomalies in the industrial countries: a solution
with a structural VAR approach”, Journal of Monetary Economics, Vol. 45 No. 3,
pp. 561-586.
Leeper, E. and Gordon, D. (1992), “In search of the liquidity effect”, Journal of Monetary
Economics, Vol. 29 No. 3, pp. 341-369.
Qayyum, A. (2008), “Does monetary policy play effective role in controlling infation in Pakistan”.
Rashid, A. (2009), “Testing the modifed-combined Ppp and Uip hypothesis in South Asian
economies”, Applied Econometrics and International Development, Vol. 9 No. 1, pp. 199-218.
SBP (2009), “Statistical bulletins of State Bank of Pakistan”.
Sims, C. (1980), “Macroeconomics and reality”, Econometrica, Vol. 48 No. 1, pp. 1-48.
Sims, C. (1992), “Interpreting the macroeconomic time series facts: the effects of monetary policy”,
European Economic Review, Vol. 36 No. 5, pp. 975-1000.
Sims, C. (1998), “Comment on Glenn Rudebusch’s ‘Do measures of monetary policy in a VARmake
sense?’”, International Economic Review, Vol. 39 No. 4, pp. 933-941.
Wong, K. (2000), “Variability in the effects of monetary policy on economic activity”, Journal of
Money, Credit and Banking, Vol. 32 No. 2, pp. 179-198.
Further reading
Ahmed, S. and Ansari, M. (2008), “Does money matter? Evidence from vector error-correction for
Mexico”, The Journal of Developing Areas, Vol. 41 No. 1, pp. 185-202.
Bernanke, B. and Gertler, M. (1995), “Inside the black box: the credit channel of monetary policy
transmission”, The Journal of Economic Perspectives, Vol. 9 No. 4, pp. 27-48.
Campbell, J. and Perron, P. (1991), “Pitfalls and opportunities: what macroeconomists should
know about unit roots”, NBER Macroeconomics Annual, Vol. 6, pp. 141-201.
Clarida, R., Gali, J. and Gertler, M. (1999), “The science of monetary policy: a new Keynesian
perspective”, Journal of Economic Literature, Vol. 37 No. 4, pp. 1661-1707.
Corresponding author
Abdul Rashid can be contacted at: [email protected]
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
JFEP
6,4
330
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
5
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_937093837.pdf