Monetary Policy Rules in Practice: Evidence for Sri Lanka
Roshan Perera and Vishuddhi Jayawickrema
January 2014
Monetary Policy Rules in Practice: Evidence for Sri Lanka
Roshan Perera and Vishuddhi Jayawickrema
1
Abstract
The paper seeks to characterise the monetary policy decision making process for Sri Lanka
using standard Taylor-type monetary policy rules. Alternative monetary policy reaction
functions are estimated for Sri Lanka over the period 1996Q1 to 2013Q2. An open economy
reaction function is used in the analysis where the central bank is assumed to respond to
changes in inflation, the output gap and the exchange rate. A forward looking specification
of the reaction function is found to provide the most appropriate characterisation of policy
making at the Central Bank of Sri Lanka. The results indicate that the size of the coefficient
on the inflation gap has increased over time reflecting a greater focus on price stability.
However, the response of monetary policy to fluctuations in output has been greater than the
response to deviations in inflation reflecting the central bank’s preference and the lower
sensitivity of output to interest rate changes.
JEL Classification Numbers: C22, E43, E52
Keywords: Monetary Policy, Policy Interest Rates, Monetary Policy Reaction Function,
Taylor Rule
1
Roshan Perera is a Deputy Director and Vishuddhi Jayawickrema is an Economist in the Economic Research
Department, Central Bank of Sri Lanka. Corresponding author: [email protected]. The views expressed in this
paper are those of the authors and do not necessarily reflect those of the Central Bank of Sri Lanka.
3
I. INTRODUCTION
The goals of monetary policy determine what constitutes optimal monetary policy. In many
central banks the final goal of monetary policy is to maximise welfare by maintaining
inflation at a low and stable level and by reducing the deviation of actual output from its
potential. However, having a target or goal for monetary policy alone does not guarantee
that the target would be met. Further there are multiple instruments that could be used to
achieve a given target. Determining the optimal monetary policy is an issue that central
banks have to constantly address. Given the importance of the policy decisions facing
central banks a large literature has been developed which has tried to characterise the
relationship between the central bank’s monetary policy instrument and the central bank’s
objectives using monetary policy reaction functions. In the New Keynesian tradition, central
banks are characterised as conducting monetary policy to stabilise inflation and to reduce the
output gap
2
(Clarida, Gali and Gertler 1999, 2000; Svensson, 1999, 2002).
3
Empirical
studies have been carried out to determine the extent to which the rules proposed in
theoretical models such as the New Keynesian model actually reflect the conduct of
monetary policy in central banks. The ‘Taylor Rule’ which was proposed in Taylor (1993)
was considered the most appropriate characterision of monetary policy making by the
Federal Reserve Bank of the USA (Fed) during the period 1987 to 1992. According to the
Taylor rule, the Fed’s monetary policy instrument - the interest rate - was set in response to
deviations of actual inflation from a targeted level and actual output from its potential level.
Empirical estimates of the monetary policy rule for the Fed showed that the behavior of the
interest rate as determined by the monetary policy rule was closely related to the actual path
of the Federal Funds Rate during the period for which the estimation was carried out.
Estimates from monetary policy reaction functions provide valuable insights into how
central banks have conducted monetary policy in the past, while guiding central banks in the
setting of appropriate interest rates in various macroeconomic environments. With the shift
to inflation targeting frameworks an increasing number of central banks have adopted a
more rule based approach to monetary policy decision making. A key feature of inflation
targeting is that it leads to a more systematic response of central banks to inflation. When
central banks commit to following a rule they overcome the time inconsistency problem
associated with discretionary monetary policy making. Central banks that are guided by
policy rules are also able to better communicate their policy actions to the market. The
greater predictability in the behaviour of central banks has thus improved the transmission of
monetary policy. Rules based decision making also increases the accountability of central
2
The output gap is measured as the gap between actual output (y) and potential output (y*) expressed as a ratio
of potential output [(y-y*)/y*], where potential output is defined as a level of output that can be sustained over
a period of time without generating inflationary or deflationary pressures
3
In emerging market economies, central banks are also found to respond to movements in the exchange rate
(Mohanty and Klau, 2004).
4
banks and enhances the credibility of future policy actions. The breakdown of the direct
relationship between money supply growth and inflation has also led to more emphasis
being placed on Taylor type rules in the conduct of monetary policy (Blinder, 2006). The
shift towards more rule based monetary policy making was intended to mitigate the impact
on the economy of monetary and other shocks and thereby reduce the emergence of crises.
Although rules are a simplification of the monetary policy decision making process of
central banks it has been found to provide a fairly good approximation of the monetary
policy actions of central banks around the world. Empirically estimating monetary policy
reaction functions help describe the monetary policy decision making of central banks and
determine the extent to which they approximate rule-based behaviour. It enables an analysis
of how far the actual conduct of monetary policy deviates from that prescribed by rules.
In this paper we seek to estimate alternative monetary policy reaction functions for Sri
Lanka over the period 1996 to 2013:Q2. Although the objectives of monetary policy and
monetary operations have changed over this period, the short term interest rate has reflected
by and large the changes in the monetary policy stance of the Central Bank of Sri Lanka.
Section II provides an overview of the theoretical and empirical literature relating to
monetary policy rules and in section III the monetary policy framework in Sri Lanka and its
evolution over time are discussed. Section IV provides a discussion of the methodology
adopted and sets out the alternative specifications of the monetary policy reaction functions.
In section V the data used in the estimation are described and the results from the empirical
analysis are presented and discussed. The final section concludes and discusses some policy
implications and areas for further research.
II. LITERATURE REVIEW
In the optimsation problem, central banks are assumed to minimise a loss function by
stabilising inflation around an inflation target and stabilising output around potential
(Svennson, 1999).
where,
is inflation,
is the targeted level of inflation,
is actual output and
is the
potential output. The value placed on the coefficient indicates the relative preference of
the central bank towards stabilising inflation or output. Central banks that have greater
preference for inflation stabilisation would have a lower value for while central banks that
place greater weight on output and employment stabilisation would have a larger value for .
5
The rule proposed in Taylor (1993) was derived as a solution to the optimisation problem of
central banks. In the original version of the Taylor rule, the policy rate was set as a function
of a deviation of inflation from target and the actual output from potential or the output gap
4
.
where i
t
is the nominal policy interest rate, π
t
is the inflation rate, π* is the targeted or
desired rate of inflation, r* is the average equilibrium real interest rate, y
t
is actual y* is the
estimated potential output level.
The monetary policy rule provides a guide to how much the central bank should change its
policy interest rate in response to deviations of inflation and output from target or potential,
respectively. The weights assigned by central banks to the objectives of inflation and output
are reflected in the coefficients for the output gap and inflation gap. The relative weights
assigned by central banks to the inflation gap and the output gap would depend on the
preference of central banks as well as their legal mandates. The coefficient on the output
gap shows the trade-off between output and inflation with a higher coefficient on the output
gap indicating lower output variance. Clarida et al 2000 in an empirical analysis for the US
find that policymakers did not obey the Taylor principle during the 1970s and 1980s, while
their theoretical analysis suggests that the failure to obey the Taylor principle led to
indeterminacy of rational expectations equilibrium and possible sunset equilibria. Hence,
they conclude that the greater macroeconomic volatility experienced during that period was
due to poor policymaking. A fundamental requirement for stability in theoretical models is
the adherence to the ‘Taylor principle’, which is the rise in the nominal interest rate more
than one for one with inflation. This requires central banks to raise policy interest rates by
more than the increase in inflation, resulting in a rise in the real interest rate, which would
help dampen aggregate demand and bring inflation back to the targeted level
5
. Failure to
obey the Taylor principle leads to indeterminancy of rational expectations equilibrium and
possible sunset equilibria in theoretical models. Empirically it has been associated with
greater macroeconomic volatility. The estimated magnitude of the coefficient φ provides an
important yardstick for evaluating a central bank’s policy reaction function.
Patra and Kapur (2010) estimating alternative monetary policy rules for India find that the
dominant focus of monetary policy is inflation, which is accompanied by a strong
commitment to the stabilisation of output.
4
According to the reaction function formulated for the US Federal Reserve Bank in Taylor (1993) the Fed was
assumed to adjust the Federal Funds Rate according to the following rule:
,
where the inflation target was assumed to be 2 per cent and the constant real interest rate was also assumed to
be 2 per cent.
5
This requires that the long run coefficient on inflation be greater than one (φ> 1).
6
There are several issues that need to be addressed when estimating a monetary policy rule.
A choice needs to be made regarding the measure of inflation and the output gap
6
to be used
in the estimation. Orphanides (1999, 2001) find a high degree of incertainty surrounding
output gap estimates particularly when there are large deviations between real time data and
final revised data. A decision needs to be made regarding the timing of information flows
which will in turn determine whether contemporaneous, lagged or forward looking variables
are used in the analysis. In estimating forward looking specifications of the monetary policy
rule a choice needs to be made whether actual or forecasted values are to be used.
A choice also needs to be made whether interest rate smoothing behavior of central banks
should be taken into consideration. The monetary policy rule proposed by Taylor was
modified to take into consideration the interest rate smoothing behaviour of central banks
(Judd and Rudebusch, 1998; Clarida et al, 2000; Paez-Farrell, 2000)
7
. Central banks
typically change policy rates gradually to avoid sharp shocks to financial markets and to
reduce the possible risk of inaccurate policy actions which may then require a reversal of
policy actions which in turn could lead to a loss of credibility. The uncertainty of the
monetary policy transmission mechanism and the parameters linking the changes in the
policy instrument with the key variables in the economy as well as the uncertainty
surrounding the models used by central banks linking the variables of interest, have made
central banks cautious in their monetary policy making, to ensure they don’t create any
undue volatility by wrongly responding to macroeconomic developments. Patra and Kapur
(2010) find a high degree of interest rate smoothing in the case of India. Many empirical
studies estimating the monetary policy rule find that the lagged interest rate is highly
significant indicating that the policy adjusts gradually to changes in macroeconomic
conditions (English et al 2003).
Open economy policy functions including the response of policy makers to the exchange
rate have been developed to characeterise monetary policymaking in emerging market and
developing economies, particularly considering the importance of the exchange rate in these
economies. The exchange rate takes on greater importance for emerging market and
developing economies because the pass through from the exchange rate to domestic inflation
is high and the exchange rate is important for a country to maintain its external
competitiveness (Mohanty and Klau, 2004). Estimating a standard open economy monetary
policy reaction function for 13 emerging market economies, they find that in many of these
countries while monetary policy has increasingly focused on price stability, they also find a
strong response of interest rates to the exchange rate. In the case of India, Patra and Kapoor
(2010) find that in most specifications of the monetary policy rules estimated, the exchange
6
This includes making a choice about how potential output is to be measured.
7
Reviews of the interest rate smoothing behaviour of central banks can be found in Lowe and Willis (1997),
“The Smoothing of Official Interest Rates, Monetary Policy and Inflation Targeting”, Reserve Bank of
Australia, pp287-312 and Sack and Wieland (1999), “Interest Rate Smoothing and Optimal Monetary Policy:
A Review of Recent Empirical Evidence”, Finance and Economics Discussion Series, Federal Reserve Board,
Washington D.C.
7
rate is found to be insignificant which they conclude is a reflection of the Reserve Bank of
India’s approach to exchange rate management, wherein the policy rate is not used to target
a level or band of the exchange rate. McCauley (2006) who estimates a monetary policy rule
for Thailand find that the policy rate does not respond to changes in the exchange rate.
However, he concludes that this does not imply that the authorities are not concerned about
the exchange rate but rather that they have other instruments to deal with the exchange rate.
The time horizon adopted by the central bank as well as the view regarding the transmission
of monetary policy would determine whether the specification of the monetary policy rule
should be forward looking, backward looking or contemporaneous. Lags in the transmission
of monetary policy have made forward looking specifications of the monetary policy rule
more attractive (Batini and Haldane, 1999). A monetary policy reaction function
incorporating forward looking behaviour of agents and taking into consideration rational
expectations of agents was considered the most preferred specification for the US, Japan and
the UK by Clarida et al (1998). On the other hand, Taylor and William (2010) estimate
contemporaneous specifications of the monetary policy rule incorporating only information
about recent behaviour of inflation and output. Judd and Rudebusch (1998) estimate
backward looking specifications of the monetary policy rules and according to Rotemberg
and Woodford (1999) backward looking rules are quite good approximations of optimal
policy. The timing of information available to policy makers when they make policy
decisions would be another factor to be considered when determining whether
contemporaneous or lagged variables are used to estimate the policy reaction function.
In the estimation of the Taylor rule it is not without its critiques. According to McCallum
(1993) the Taylor rule is not ‘operational’ as it requires policymakers to have information
that is not necessarily available at the time monetary policy decisions are made. The
criticism relates to the timing of information on inflation and the output gap that is available
to policymakers at the time decision are made. Orphanides (1999, 2001) highlights the issue
with the measurement of the output gap which is an unobservable variable. He observes that
there is a high degree of uncertainty surrounding output gap estimates particularly when
there are large deviations between real time data and final revised data. In practice,
however, in the absence of real time data, ex-post data is commonly used to estimate
monetary policy rules.
In applying the Taylor rule there are issues that need to considered. According to Greenspan
(1997) monetary policy rules are at best “guideposts” to central banks, not inflexible rules
that eliminate discretion. The reason he gives for this is because the outcome of these rules
depend on some “key variables” primarily, the equilibrium real interest rate and the
estimation of potential output of the economy which are based on an analysis of historic
data. However, as he puts it “…history is not an infallible guide to the future, and the levels
of these two variables are currently under active debate.”
8
III. MONETARY POLICY FRAMEWORK IN SRI LANKA
The mandate of the Central Bank of Sri Lanka has evolved with the economic and financial
developments in Sri Lanka as well as the evolution of central banking around the world. In
the Monetary Law Act No. 58 of 1949 (MLA) under which the Central Bank of Sri Lanka
8
was established, the Bank was mandated with multiple objectives of stabilising the domestic
monetary value and the exchange rate of the Sri Lanka rupee vis-à-vis foreign currencies,
promoting a high level of production, employment and real income and encouraging and
promoting the full development of the productive resources of the country. In 2002, an
amendment to the MLA redefined the objectives of the Central Bank whereby the multiple
objectives of the Central Bank were replaced with two objectives: economic and price
stability and financial system stability.
Similarly the monetary policy framework, in which Sri Lanka has operated, as in the case of
most other countries, has evolved over time. From its inception to the early 1980s the
Central Bank adopted a more dirigiste approach to managing the economy by imposing
direct controls on credit and interest rates with a view to encouraging identified sectors in
the economy and imposing strict exchange controls. The focus during this period was
economic development even at the cost of high inflation. The liberalisation of the economy
in 1977 set the stage for the move away from direct instruments to more market oriented
monetary policy instruments. The ascendance of Monetarists economics led to an increasing
recognition of the long run relationship between monetary growth and inflation. In the
1980s the Central Bank formally adopted a monetary targeting policy framework. Under
this policy framework the Central Bank seeks to achieve its final objectives, by conducting
monetary policy so as to maintain reserve money, the Bank’s operating target, at a level that
is consistent with a desired growth of broad money, the Bank’s intermediate target. The
efficacy of this policy framework depends entirely on there being an identifiable relationship
between money supply growth and inflation which is econometrically determined by testing
for the stability of the money demand function. The development of the financial system and
financial innovations saw many central banks moving away from monetary targeting to
inflation targeting type policy frameworks. The Central Bank of Sri Lanka has also stated
that it is gradually refining its policy framework towards an inflation targeting type
monetary policy framework which does not depend on a strict relationship between money
and inflation. With the shift from a crawling band exchange rate regime to a floating
exchange rate system in January 2001, the role of the exchange rate for stabilisation has
reduced and reserve money became the nominal anchor of monetary policy.
In the conduct of monetary policy there has been a move away from direct instruments to
more market oriented instruments with greater reliance placed on open market operations
(OMO) as the main instrument of monetary policy. Although initially OMO were ‘passive’
in that the Central Bank offered unlimited repurchase (repo) and reverse repurchase (reverse
8
It was known as the Central Bank of Ceylon until 31 December 1985.
9
repo) facilities to counterparties which they could avail at their discretion, to improve the
conduct of monetary policy the Central Bank moved to a system of more active open market
operations in March 2003. In this new system monetary policy is conducted to maintain
reserve money around a targeted level while ensuring that the short term interest rate is
maintained at a level which is compatible with the target of reserve money (Wijesinghe,
2006). A key element of this new system was the establishment of an interest rate corridor
formed by the lower bound of the overnight Repurchase (repo) rate and the upper bound by
the overnight Reverse Repurchase (reverse repo) rate. Monetary policy operations are
conducted to maintain the overnight interest rate (call market rate) at around the middle of
the corridor. With the move to more active open market operations the overnight call
market rate and consequently the interest rate channel took on a more important role in the
transmission of monetary policy. Although reserve money continues to be the operating
target of monetary policy, policy interest rates and specifically the policy interest rate
corridor is the main instrument used to signal the monetary policy stance of the Central Bank
of Sri Lanka.
Figure 1: Frequency of Policy Interest Rate and SRR Changes 2000 - 2013
Estimating a monetary policy rule for Sri Lanka is not straightforward given the changes in
the conduct of monetary policy. Although reserve money continues to be the target of
monetary policy there has been a shift towards the use of the interest rate corridor to signal
the stance of monetary policy. Further with the developments in financial markets there has
been an improvement in the transmission of policy rates to other interest rates further
justifying the use of the interest rate as the policy instrument. Difficulties arise in estimating
a monetary policy rule for Sri Lanka as it requires measuring potential output which is
unobserved and could change due to structural changes taking place in the economy.
0
1
2
3
4
5
6
7
8
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
Frequency
Policy Rate Changes SRR Changes
10
IV. METHODOLOGY
Several alternative specifications of the policy reaction function were estimated for Sri
Lanka. A contemporaneous specification of the form given in equation (1) was estimated.
The estimation results are given in Annex II. In the light of Judd and Rudebusch (1998) and
Rotemberg and Woodford (1999) finding that backward looking specifications of the Taylor
rule are relatively good approximations of optimal policy, a backward looking specification
of a monetary policy rule of the form set out in equation (2) was estimated.
(1)
where
is the short term interest rate or the policy rate of the central bank,
is the year on
year rate of inflation,
is the desired level of inflation,
is the output gap,
is the
change in the nominal exchange rate and
is a random disturbance term.
(2)
However, given the lags in the transmission of monetary policy, in practice central banks are
found to be more forward looking in their decision making. Hence, a forward looking
monetary specification of the policy reaction function of the form set out in equation (3) was
also estimated.
(3)
In the empirical literature (Clarida et al, 2000, Paez-Farrell, 2009) the interest rate
smoothing behaviour of central banks is taken into consideration by including a lagged value
of the interest rate.
Exchange rate smoothing has also been found to be an important consideration in the policy
reaction function of emerging economies (Mohanty and Klau, 2004). Given Sri Lanka is a
small open economy which is highly sensitive to movements in the exchange rate, a reaction
function including the exchange rate was chosen as the preferred specification.
V. DATA DESCRIPTION AND EMPIRICAL RESULTS
Quarterly data for the period 1996:Q1 to 2013:Q2 was used for the analysis. The choice of
the sample period was determined by the availability of quarterly data for GDP. Details of
the data series are given in Annex I and Table 1, while the stylised facts of the variables and
a summary of the descriptive statistics of the data series are found in Tables 2 and 3 of
Annex I. Empirical estimation of a Taylor rule requires a priori determination of three
parameters: the desired level of inflation, potential output and equilibrium real policy rate.
With regard to the choice of an inflation measure, there are several alternative measures of
inflation computed for Sri Lanka, such as the Colombo Consumer Price Index (CCPI), the
11
Wholesale Price Index (WPI) and the GDP deflator. Since the most widely accepted
measure of inflation is the CCPI, the year on year change in the CCPI, is the inflation
measure used in the analysis. The desired rate of inflation has been set at 5 per cent for the
entire period of analysis.
9
The year on year change in inflation as measured by CCPI had an average of 10.0 per cent
over the sample period 1996 to 2013:Q2. The inflation gap, which is the difference between
the actual inflation and the desired inflation, recorded a mean of 5.0 during the period under
consideration. The descriptive statistics of the data series were also analysed by dividing the
entire sample into two subsamples to evaluate the changes that have taken place since 2008.
Notably, the average inflation gap during the period 2008 to 2013:Q2 was 4.09 per cent,
which was lower than the average inflation gap of 5.44 per cent during the period 1996 to
2007.
The mean of the alternative measures of short term interest rates, namely the effective policy
rate, the average weighted call market rate and the 91-day Treasury bill rate were 11.3 per
cent, 11.6 per cent and 11.7 per cent respectively during the sample period 1996 to 2013:Q2.
Reflecting the movement in inflation, the average interest rates have also declined during the
period 2008 to 2013:Q2 from the rates prevailing during the period 1996 to. Meanwhile, the
average depreciation in the exchange rate was 5.1 per cent during the entire sample period,
and was also found to be lower in the latter subsample compared to the former.
Figure 2: Inflation (Spliced Index)
9
It is recognised that this rate may not have been the Central Bank of Sri Lanka’s target rate over the entire
sample period but it has been chosen as the target rate since the Central Bank of Sri Lanka has stated in policy
documents its desires to maintain inflation at mid-single digit levels over the medium term.
0
5
10
15
20
25
30
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
point-to-point change (per cent)
12
Measuring potential output, which cannot be observed, is one of the key issues that need to
be addressed when estimating a monetary policy rule. Correctly estimating the output gap is
crucial to obtaining reliable estimates of the monetary policy rule. There are several
alternative methods used to estimate potential output: Univariate filtering methods such as
the Hodrick Prescott filter (HP filter), Baxter King and Christiano-Fitzgerald band pass
filters as well as multivariate model based methods.
10
Alternative measures of the output
gap for Sri Lanka based on estimates of potential output using these various methods are
given in Figure 2. The figure shows a close correspondence between the alternative
measures of potential output. Since the potential output estimate using the HP filter measure
is available for the longest period, it is chosen for the empirical analysis.
Figure 3: Alternative Measures of the Output Gap
The central bank has operated under a monetary targeting framework since the 1980s.
However, since the commencement of open market operations, monetary policy has been
conducted to influence the short term interest rate. Hence, the short term interest rate was
10
Hodrick-Prescott (HP) filter: Potential output represents a filter that minimises the deviation of actual output
from the potential output, subject to a penalty on the maximum allowable change in potential growth between
the two periods. The standard practice is to use a smoothness parameter equal to 1,600 for quarterly data.
Baxter-King (BK) and Christiano-Fitzgerald (CF) band-pass filters: Accommodate business cycle dynamics
using a range of business cycle frequencies to separate the cyclical and trend components of output.
Multivariate (MV) filter: A model based approach to estimating potential output. The multivariate filter is used
to simulate the potential output by estimating the relationship between growth and other observable variables
including inflation, capacity utilisation and unemployment.
-8
-6
-4
-2
0
2
4
6
8
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
Per cent
HP filter BK filter CF filter MV filter
13
chosen to reflect the monetary policy stance of the Central Bank of Sri Lanka. Since there is
no one interest rate that appropriately reflects the stance of monetary policy over the entire
sample period, it was necessary to choose a short term interest rate that appropriately
reflected the monetary policy stance of the Central Bank. An ‘Effective Policy Rate’ was
constructed by choosing the policy interest rate that ‘best’ reflected the monetary policy
stance during each period under consideration. Until the commencement of open market
operations the Repo rate was considered as the effective policy interest rate. Thereafter
depending on macroeconomic conditions and liquidity conditions in the market either the
Repo rate or the Reverse Repo rate was chosen as the effective policy rate. To carry out
robustness checks, the monetary policy rule was also estimated out using the average
weighted call market rate (AWCMR) the 91-day Treasury bill rate. The correlation between
the AWCMR and the effective policy interest rate was found to be around 0.8, indicating the
close movement between the policy interest rate and the overnight market interest rate.
Figure 4: Central Bank Policy Interest Rates and Overnight Short Term Interest Rate
Sri Lanka being a small open economy is significantly affected by changes in its exchange
rate. Hence, the exchange rate has been included in the monetary policy reaction function of
the Central Bank. Following Patra and Kapur (2012) the annualised quarter on quarter
change in the nominal exchange rate was used in the analysis.
0
5
10
15
20
25
30
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
Per cent
AWCMR (Quarterly Average)
Repo
Reverse Repo
Penal Rate
Effective Policy Rate
14
Figure 5: Rs/US dollar Exchange Rate
A Chow breakpoint stability test was carried out on a selected contemporaneous monetary
policy reaction function.
11
Based on the results of the test, two statistically significant breaks
were detected in 2001 Q1 and 2008 Q1. The break in 2001 Q1 captures the shift to a free
floating exchange rate regime while the break in 2008 Q1 captures the impact of the global
financial crisis. A dummy variable has been included to take into account these
extraordinary events that had an undue impact on the volatility of the short term interest rate.
Contemporaneous and backward looking models were estimated using Ordinary Least
Squares (OLS). Forward looking specifications which include lead values of explanatory
variables were estimated using Generalised Method of Moments (GMM) to account for
possible endogeneity between variables (Clarida et al, 1998).
Table 1 presents a summary of the results from estimates of alternative contemporaneous
specifications of the monetary policy reaction function. Two alternative measures of short
term interest rate (AWCMR and the effective policy interest rate) were used in the
estimation. The baseline specifications (column 1 and 3 of Table 1) include
contemporaneous values for inflation and output gap, and the first lag of the short term
interest rate to take into account interest smoothing behaviour of the central bank. The
baseline specifications were augmented with the nominal exchange rate (column 2 and 4 of
Table 1). Similar estimations were also carried out using the 91-day Treasury bill rate in
order to perform robustness checks, and the results are given in Table 5 of Annex II.
11
Chow breakpoint stability test can only be done for single equations, and not for a system of equations.
Hence the break points detected are based on a contemporaneous specification of a monetary policy reaction
function.
50
60
70
80
90
100
110
120
130
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
Rs./USD
15
Estimates from contemporaneous monetary policy reaction function shows that the
coefficients on both inflation and output gap remain positive and significant for all
specifications. The coefficient on inflation is lower than the coefficient on the output gap
and the long-run coefficient on inflation is less than unity indicating that the Taylor principle
is not fulfilled. The use of a contemporaneous specification which ignores the lags in the
transmission of monetary policy could be the reason for this result. However, the coefficient
on the output gap is above unity. The exchange rate variable is found to be significant in one
of the augmented specification (column 2 of Table 1) although the coefficient is small.
Table 1: Estimates of the Contemporaneous Monetary Policy Reaction Function:
Sample Period: 1996 Q:1 – 2013 Q:2
Column number
1
2
3
4
Dependent variable
AWCMR
EFFECTIVE
Constant
2.26
2.63
2.34
2.44
(2.91)
***
(3.41)
***
(3.50)
***
(3.63)
***
INFCPIDEV
0.13
0.16
0.14
0.15
(2.61)
***
(3.18)
***
(3.05)
***
(3.25)
***
YGAPSA
0.36
0.37
0.52
0.53
(2.69)
***
(2.89)
***
(4.44)
***
(4.52)
***
AWCMR(-1)
0.75
0.68
(10.72)
***
(9.04)
***
EFFECTIVE(-1)
0.73
0.71
(11.66)***
(10.54)***
TBILL91(-1)
EXCH4(-1)
0.06
0.03
(2.24)**
(1.19)
Observations
69
69
69
69
Adjusted R-squared
0.74
0.75
0.80
0.80
F-statistic
65.05
53.07
89.63
68.00
S.E. of regression
1.96
1.90
1.73
1.73
Long-run coefficient on inflation
0.5
0.5
0.5
0.5
Long-run coefficient on output gap
1.4
1.1
2.0
1.8
Long-run coefficient on exchange rate
0.2
0.1
Neutral policy rate
8.9
8.8
Notes:
Absolute value of t-statistics is given in parentheses.
* significant at 10%; ** significant at 5%, *** significant at 1%
Estimation is by OLS methodology for the sample period 1996Q1 - 2013Q2
Output gap measure: Hodrick-Prescott filter
Table 2 presents a summary of the results from estimates of a backward looking
specification of a monetary policy reaction function. Two alternative measures of short term
16
interest rate (AWCMR and the effective policy interest rate) were used to estimate the
backward looking monetary policy reaction function. The baseline specifications (column 1
and 3 of Table 2) included lagged inflation and output gap, and the short term interest rate
with one lag. The baseline specifications were augmented with the nominal exchange rate
(column 2 and 4 of Table 2). Similar estimations were also carried out using the 91-day
Treasury bill rate in order to perform robustness checks, and the results are given in Table 6
of Annex II.
Table 2: Estimates of Monetary Policy Reaction Function - Backward Looking Specifications
Sample Period: 1996 Q:1 – 2013 Q:2
Column number
1
2
3
4
Dependent variable
AWCMR
EFFECTIVE
Constant
2.84
3.20
3.26
3.33
(3.58)***
(4.01)***
(4.55)***
(4.59)***
INFCPIDEV(-1)
0.11
0.14
0.10
0.11
(2.08)**
(2.62)***
(2.20)**
(2.30)**
YGAPSA(-1)
0.51
0.51
0.67
0.67
(3.80)***
(3.86)***
(5.38)***
(5.37)***
AWCMR(-1)
0.71
0.64
(9.64)***
(7.98)***
EFFECTIVE(-1)
0.67
0.65
(9.59)***
(8.68)***
EXCH4(-1)
0.06
0.02
(1.95)*
(0.69)
Observations
69
69
69
69
Adjusted R-squared
0.75
0.76
0.80
0.80
F-statistic
68.52
54.57
91.61
68.28
S.E. of regression
1.92
1.88
1.72
1.73
Long-run coefficient on inflation
0.4
0.4
0.3
0.3
Long-run coefficient on output gap
1.7
1.4
2.0
1.9
Long-run coefficient on exchange rate
0.2
0.0
Neutral policy rate
9.7
9.8
Notes:
Absolute value of t-statistics is given in parentheses.
* significant at 10%; ** significant at 5%, *** significant at 1%
Estimation is by OLS methodology for the sample period 1996Q1 - 2013Q2
Output gap measure: Hodrick-Prescott filter
Estimates from the backward looking monetary policy reaction function show that the
coefficients on both inflation and output gap remain positive and significant for all
specifications. However, the long-run coefficient on inflation is less than unity, whereas the
coefficient on the output gap is greater than one. This result is once again because the
backward looking specification does not take account of the lags in monetary transmission.
17
The exchange rate variable turns out to be significant in one of the augmented specifications
(column 2 of Table 2).
Table 3 presents a summary of the results from estimates of a forward looking specification
of a monetary policy reaction function. Similar to the contemporaneous and backward
looking specifications, the AWCMR and the effective policy interest rate were used as two
alternative measures of short term interest rate. One quarter ahead inflation and output gap
together with short term interest rate with one lag were used in the baseline specifications
(column 1 and 4 of Table 3). The baseline was augmented with a dummy variable to account
for the structural breaks identified in 2001 and 2008 (column 2 and 5 of Table 3). In
addition, the specifications were also augmented with movements in the nominal exchange
rate (column 3 and 6 of Table 3). Robustness checks were carried out using the 91-day
Treasury bill rate, and the results are given Table 7 of Annex II.
The coefficients on inflation and the output gap are statistically significant and have the right
sign. However, the coefficient on the output gap is larger than the coefficient on inflation.
The coefficient on lagged interest rate is large and significant implying a relatively high
degree of interest rate smoothing. This indicates that the central bank generally changes its
policy interest rate in small steps in response to macroeconomic developments. Increasing
uncertainty of the macroeconomic environment and the monetary transmission mechanism
have made central banks more cautious in the conduct of monetary policy. The long run
coefficient
12
on inflation is greater than 1 in all forward looking specifications of the
monetary policy reaction function indicating that in the forward looking specification of the
monetary policy rule the Taylor principle
13
is satisfied. The long run coefficient on the
output gap is larger than the coefficient inflation indicating that monetary policy seems to
react more strongly to fluctuations in output than to deviations in inflation. The higher
coefficient on the output gap may reflect a higher preference towards output stabilisation but
it could also reflect a lower sensitivity of output to the interest rate and hence the need for a
stronger response of monetary policy towards output stabilisation (Hayo and Hoffman,
2005). In emerging markets and developing countries, monetary policy is supposed to react
more strongly to movements in the exchange rate. In the forward looking specification the
coefficient on the exchange rate is of the right sign, indicating that monetary policy is
tightened in response to depreciation in the exchange rate but the coefficient is very small
and not significant. According to Mohanty and Klau (2004), the strength of the monetary
12
13
As discussed in Taylor (1999) if the coefficient on inflation is less than one, the real interest would decline
with a rise in inflation, leading to higher inflation in the future.
18
policy response to the exchange rate depends on whether a central bank is able to use other
instruments such as intervention in the foreign exchange market, temporary capital controls,
swaps and other derivative instruments to stabilise the exchange rate.
Table 3: Estimates of Monetary Policy Reaction Function - Forward Looking Specifications
Sample Period: 1996 Q:1 – 2013 Q:2
Column number
1
2
3
4
5
6
Dependent variable
AWCMR
EFFECTIVE
Constant
1.87
-0.91
-1.05
1.68
-0.01
-0.05
(2.01)
**
(-0.25)
(-0.28)
(1.88)
*
(0.00)
(-0.01)
INFCPIDEV(+1)
0.29
0.36
0.42
0.28
0.29
0.31
(1.96)
**
(1.79)
*
(1.96)
**
(2.71)
***
(2.02)
**
(2.11)
**
YGAPSA(+1)
0.65
0.69
0.59
0.95
0.97
0.94
(1.13)
(1.21)
(1.03)
(1.96)
**
(1.90)
*
(1.94)
*
AWCMR(-1)
0.72
0.77
0.70
(7.68)
***
(6.89)
***
(5.60)
***
EFFECTIVE(-1)
0.73
0.78
0.76
(9.12)
***
(4.86)
***
(4.12)
***
TBILL91(-1)
EXCH4(-1)
0.06
0.02
(1.41)
(0.58)
DUMMY
2.02
2.30
1.12
1.16
(0.71)
(0.81)
(0.35)
(0.36)
Observations
68
68
68
68
68
68
Adjusted R-squared
0.62
0.57
0.57
0.61
0.60
0.60
J-statistic
0.00
0.00
0.00
0.00
0.00
0.00
S.E. of regression
2.35
2.50
2.52
2.40
2.45
2.44
Long-run coefficient on inflation
1.0
1.5
1.4
1.0
1.4
1.3
Long-run coefficient on output gap
2.3
3.0
2.0
3.6
4.5
4.0
Long-run coefficient on exchange rate
0.2
0.1
Neutral policy rate
6.6
4.8
6.3
5.1
Notes:
Absolute value of t-statistics is given in parentheses
* significant at 10%; ** significant at 5%, *** significant at 1%
Estimation is by GMM methodology for the sample period 1996Q1 - 2013Q2
Instruments used for GMM estimation: INFCPIDEV(-1), YGAPSA(-1), AWCMR(-1), EFFECTIVE(-1),
EXCH4(-1) and DUMMY
Output gap measure: Hodrick-Prescott filter
Figure 5 plots the actual effective rate and the policy interest rate based on the estimates of
the monetary policy rule from the preferred specification, i.e., the forward looking monetary
19
policy reaction function augmented with a dummy variable (column 5 of Table 3).
14
The
policy interest rate estimated from the model appears to closely track the effective rate
reasonably well, although there are was some deviation during 2000-2001 and 2006-2007.
Figure 6: Actual Versus Estimated Effective Policy Rate
Based on the Forward Looking Specification of the Monetary Policy Rule
Determining the neutral policy rate
15
is vital for the estimation of a monetary policy rule. It
is possible to estimate this value using a general equilibrium model of the economy.
However, a crude estimate can be obtained from the estimation of the Taylor rule itself. The
estimates of the neutral policy interest rate in the contemporaneous specification is around
8.9 per cent while for the backward looking specification it is around 9.8 per cent indicating
the need for high interest rates if the lags in transmission of monetary policy are not taken
into consideration by policy makers. According to estimates of forward looking
specifications of the monetary policy reaction function, the neutral policy rate is around 6.2 -
6.6 per cent.
16
Assuming a higher level of desired inflation during the first sub sample
period results in a neutral policy rate estimate of between 10-11 per cent.
14
The selected specification has a Durbin-Watson statistic of 1.77, which is approximately close to the desired
level of 2, the level of serial correlation is not significant. According to the Ljung-Box Q-statistic the null
hypothesis that there is no autocorrelation can be accepted.
15
The neutral policy rate is the policy rate at which the economy is assumed to be growing at its potential level
and inflation is maintained at the desired level.
16
However, the estimates of the neutral policy rate should be treated as indicative and within wide confidence
intervals, as the assessment of the neutral rate is conditional upon the view on the rate of potential output
growth (Patra and Kapur, 2010).
(continued…)
-10
-5
0
5
10
15
20
25
30
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
Effective policy rate
Fitted
Actual
Residual
20
Figure 7: Interest Rate Gap and Inflation Gap
Figure 7 plots the interest rate gap which is the difference between the effective policy rate
and the interest rate implied by the estimated monetary policy rule and the inflation gap
which is the difference between the inflation rate and targeted inflation. The figure shows an
inverse relationship between the interest rate gap and the inflation gap. Periods during which
the actual interest rate was close to the implied rate implied by the estimated Taylor rule,
actual inflation is closer to the desired/targeted rate of inflation. In periods where there is a
deviation of the effective policy rate from the policy rate implied by the Taylor rule, the
larger the gap between actual inflation and the desired/targeted rate of inflation. A widening
gap is observed during the period 2007-2008, coinciding with the Global Financial Crisis.
A recursive regression was carried out for the backward looking specification in column 3 of
Table 2 to assess the evolution of the coefficients on the inflation gap and output gap over
time. The results are presented in Figure 7. According to the estimates the response of
monetary policy to deviations of inflation from the desired level and the output gap has
strengthened since 2007, reaching a peak in 2009. The response of monetary policy to
inflation has stabilised thereafter, while the response to the output gap appears to have
gradually declined reflecting the improvement in the transmission of monetary policy. The
coefficient on the output gap has been consistently higher than the coefficient on the
inflation gap reflecting the lower sensitivity of output to interest rates.
Figure 8: Recursive Estimates of Coefficients of Inflation Gap and Output Gap
-10
-5
0
5
10
15
20
25
-10
-8
-6
-4
-2
0
2
4
6
8
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
Inflation
Interest rate
Interest Rate gap (LHS) Inflation gap (RHS)
21
a. Inflation Gap
b. Output Gap
Since there appears to be a definite shift in the coefficients on inflation gap and output gap
after 2007, the monetary policy reaction function was estimated over two sub sample
periods. The first sample period covered the period 1996 Q:1 to 2007 Q:4, while the second
sample period was from 2008 Q:1 to 2013 Q:2. Due to insufficient number of observations
in the second sample period the results from that period are not reported. However,
comparing the results from the first sample period and the entire sample provide some
important insights into the changes that have taken place in the conduct of monetary policy.
The long run coefficient on inflation was less than one which was below the threshold
prescribed by the Taylor principle, implying that during this period monetary policy has
reacted less than proportional to changes in inflation. On the other hand, for the entire
sample period, the long run coefficient was above 1, indicating that during the second
sample period the Taylor principle was met. With monetary policy reacting more than
proportionately to the inflation gap, there is an increase in the real interest rate leading to
0.00
0.02
0.04
0.06
0.08
0.10
0.12
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
Coefficient of inflation gap
0.55
0.60
0.65
0.70
0.75
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
Coefficient of output gap
22
lower inflation. Further the long run coefficient on output gap is higher during the second
period indicating a higher weight on output stabilisation.
Table 4: Estimates of Monetary Policy Reaction Function - Forward Looking Specifications
Sample Periods: 1996:Q1 to 2007:Q4 and 1996 Q:1 to 2013 Q:2
Column number
1
2
3
4
5
6
Sample period
1996Q1 2007Q4
1996Q1 2013Q2
Constant
2.28
9.52
9.53
1.68
-0.01
-0.05
(1.43)
(1.55)
(1.52)
(1.88)
*
(0.00)
(-0.01)
INFCPIDEV(+1)
0.28
0.26
0.26
0.28
0.29
0.31
(1.46)
(1.59)
(1.63)
(2.71)
***
(2.02)
**
(2.11)
**
YGAPSA(+1)
1.18
1.17
1.15
0.95
0.97
0.94
(1.86)
*
(1.96)
*
(1.98)
**
(1.96)
**
(1.90)
*
(1.94)
*
EFFECTIVE(-1)
0.68
0.42
0.36
0.73
0.78
0.76
(6.29)
***
(1.65)
(1.26)
(9.12)
***
(4.86)
***
(4.12)
***
EXCH4(-1)
0.06
0.02
(1.25)
(0.58)
DUMMY
-4.50
-4.20
1.12
1.16
(-1.07)
(-1.00)
(0.35)
(0.36)
Observations
47
47
47
68
68
68
Adjusted R-squared
0.32
0.36
0.37
0.61
0.60
0.60
J-statistic
0.00
0.00
0.00
0.00
0.00
0.00
S.E. of regression
2.86
2.79
2.76
2.40
2.45
2.44
Long-run coefficient on inflation
0.9
0.4
0.4
1.0
1.4
1.3
Long-run coefficient on output gap
3.6
2.0
1.8
3.6
4.5
4.0
Long-run coefficient on exchange rate
0.1
0.1
Neutral policy rate
7.0
8.7
6.3
5.1
Notes:
Absolute value of t-statistics is given in parentheses
* significant at 10%; ** significant at 5%, *** significant at 1%
Estimation is by GMM methodology for the effective interest rate
Instruments used for GMM estimation: INFCPIDEV(-1), YGAPSA(-1), AWCMR(-1), EFFECTIVE(-1),
EXCH4(-1) and DUMMY
Output gap measure: Hodrick-Prescott filter
VI. CONCLUSION
The paper provides an empirical characterisation of the monetary policy reaction function of
the Central Bank of Sri Lanka over the period 1996Q1 to 2013Q2. . Although the
objectives of the Central Bank have changed during this period as have its operating
methods, the monetary policy stance of the Central Bank appears to have been well
characterised by movements in the short term interest rate. Estimates of the Central Bank’s
monetary policy reaction function provide an understanding of the relative weights placed
23
by the Central Bank on inflation and output during the period for which the analysis was
conducted.
The estimates provide evidence of a change in the coefficients for the inflation gap and the
output gap during the period of analysis, in particular with a stronger response of monetary
policy to the inflation gap and the output gap being observed since 2007. There is also
evidence of a greater weight being placed on output stabilisation, which could reflect both
the preference of the central bank and structural issues relating to the slower transmission of
monetary policy. A relatively strong response to the output gap may be attributed to a lower
sensitivity of output to the interest rate. However, there appears to be a shift in monetary
policy from greater responsiveness to the output gap to more focus on inflation. The
empirical analysis however, does not provide any evidence that monetary policy is
responsive to the exchange rate.
The challenges discussed previously in the estimation of monetary policy reaction functions
were encountered in the case of Sri Lanka as well. Generating reliable estimates of potential
output is particularly challenging given the structural changes that have taken place in the
economy. Difficulties also arise in determining the equilibrium real interest rate which may
therefore give rise to policy error. Further identifying a single targeted or desired inflation
rate for the entire period of the analysis is challenging.
Taylor type monetary policy rules provide a simple and transparent framework for
conducting monetary policy (Taylor, 2008). However, mechanically following rule based
monetary policy formulation is not what is recommended. Monetary policy rules provide
only a guide to policy makers in their decision making process. Judgment is required when
evaluating macroeconomic developments in the decision making process. Unexpected
events may also require changes in the interest rate to smoothen the volatility in financial
markets. While monetary policy rules are useful to guide monetary policy formulation
slavishly following simple policy rules may not always lead to optimal monetary policy
making (Svensson, 2003; Woodford, 2001).
References
Ball, Laurence, 1999. “Policy rules for open economies”, in John B Taylor (ed), Monetary
policy rules, NBER, 127-56.
Batini, Nicoletta and Andrew G Haldane, 1999. “Forward-Looking Rules for Monetary
Policy,” In Taylor, J.B. (ed.), Monetary Policy Rules. University of Chicago Press: Chicago.
Beck, Guenter and Wieland, Volker, 2008. "Central bank misperceptions and the role of
money in interest rate rules," Working Paper Research 147, National Bank of Belgium.
Blinder, Alan, 2006. "Monetary Policy Today: Sixteen Questions and about Twelve
Answers," Working Papers 73, Princeton University, Department of Economics, Center for
Economic Policy Studies.
Clarida, Richard, Jordi Gali and Mark Gerlter. 1998. “Monetary Policy Rules in Practice:
Some International Evidence”, European Economic Review, 42, 1033-1067.
Clarida, Richard, Jordi Gali and Mark Gerlter. 1999 “The Science of Monetary Policy: A
New Keynesian Perspective”, Journal of Economic Literature, Vol. 37 (4), 1661-1707.
Clarida, Richard, Jordi Gali and Mark Gerlter. 2000. “Monetary Policy Rules and
Macroeconomic Stability”, Quarterly Journal of Economics, 115, 147-180.
Dattels, Peter and others. 2001. Sri Lanka: “A Strategy for Strengthening the Framework and
Implemenation of Monetary and Foreign Exchange Operations.” Technical Assistance
Report, International Monetary Fund, Washington D.C.
Debabrata, Michael Patra and Muneesh Kapur. 2010, “A Monetary Policy Model without
Money for India” IMF Working Paper WP/10/183.
English, William, William Nelson and Brian Sack. 2003, “Interpreting the Significance of the
Lagged Interest Rate in Estimated Monetary Policy Rules” Contributions to Macroeconomics
Vol 3(1).
Gali, Jordi and Mark Gertler, 2007. “Macroeconomic Modelling for Monetary Policy
Evaluation”, Journal of EconomicmPerspectives, 21(4), 25-45.
Judd, John P and Glenn D. Rudebusch, 1998. “Taylor’s Rule and the Fed: 1970-1997,”
Federal Reserve Bank of San Francisco Economic Review, no.3, 3-16.
Kydland, Finn and Edward Prescott (1977). “ Rules Rather than Discretion: The
Inconsistency of Optimal Plans”, Journal of Political Economy, 85(3), 473-492.
25
Lowe, Philip and Luci Ellis, 1997. “The smoothing of official interest rates”, Monetary
policy and inflation targeting, Reserve Bank of Australia, 287-312.
McCallum, B.T., 1999. “Recent Developments in the Analysis of Monetary Policy Rules”,
Homer Jones memorial Lecture, March 11, 1999, 11.
McCallum, B.T., 2000. “Alternative Monetary Policy Rules: A Comparison with Historical
Settings for the United States, the United Kingdom, and Japan," Working Paper 7725,
National Bureau of Economic Research.
Mohanty, M.S. and Marc Klau, 2004. “Monetary Policy Rules in Emerging Market
Economies: Issues and Evidence”, Bank of International Settlements Working Paper No.
149.
Orphanides, A. and S. van Norden, 1999. The Reliability of Output Gap Estimates in Real-
Time, Board of Governors of the Federal Reserve System Finance and Economics
Discussion Series No. 38
Orphanides, Athanasios, 2001. “Monetary policy rules, macroeconomic stability and
inflation: a view from the trenches,” Finance and Economics Discussion Series 2001-62,
Board of Governors of the Federal Reserve System.
Orphanides, Athanosios, 2003. “Historical Monetary Policy Analysis and the Taylor Rule”,
Journal of Monetary Economics, 50(5), 983-1022.
Orphanides, Athanosios, 2007. “Taylor Rules”, Working Paper FEDS 2007-18, Board of
Governors of the Federal Reserve System, January.
Paez-Farrell, Juan, 2009. “Monetary Policy Rules in Theory and Practice: Evidence from the
UK and US”, Applied Economics, 4(16), 2037-2046.
Rotemberg, Julio J., and Michael Woodford, 1999. “Interest Rules in an Estimated Sticky
Price Model”, in John B. Taylor (ed.), 1999, Monetary Policy Rules, University of Chicago
Press, Chicago, 57-126.
Sack, Brian and Volker Wieland, 1999. “Interest-Rate Smoothing and Optimal Monetary
Policy: A Review of Recent Empirical Evidence,” Finance and Economics Discussion Series,
1999-39, Federal Reserve Board, (August).
Svensson Lars E O, 1999. “Inflation targeting as a monetary policy rule”, Journal of
Monetary Economics, 43, 607-54.
26
Svensson Lars E O, 2002. “Inflation targeting: should it be modelled as an instrument rule or
a target rule”. European Economic Review, 46, 771-80.
Svensson Lars E O, 2003. “Escaping from a Liquidity Trap: The Foolproof Way and
Others”, Journal of Economic perspectives, Vol. 17(4), 145-66.
Taylor, John B., 1993. “Discretion versus Policy Rules in Practice”, Carnegie-Rochester
Conference Series on Public Policy, 39, 195-214.
Taylor, John B., 1999. “A Historical Analysis of Monetary Policy Rules” in John B. Tayloe
(ed.), “Monetary Policy Rules”, University of Chicago Press, Chicago, 319-348.
Taylor, John B., 2002. “Using Monetary Policy Rules in Emerging Market Economies”
Stabilisation and Monetary Policy, Banco de Mexico, 441-57.
Taylor, John B. and John C. Williams, 2010. “Simple and Robust Rules for Monetary
Policy”, National Bureau of Economic Research Working Paper No. 15908.
Wijesinghe, D.S. 2006. “Active Open Market Operations”, Staff Studies Central Bank of Sri
Lanka , 36(1 & 2), 15-35.
Woodford, Michael, 2001. "Imperfect Common Knowledge and the Effects of Monetary
Policy," NBER Working Papers 8673, National Bureau of Economic Research, Inc.
27
Annex I
Table 1: Data Description
Variable Name
Definition
Period
Source
AWCMR
Average weighted call money rate
(quarterly average)
1996 – 2013:Q2
CBSL
1
EFFECTIVE
Effective policy rate
1996 - 2012:Q2
CBSL
TBILL91
91 day Treasury bill rate
1996 - 2012:Q2
CBSL
EXCH4
Annualised quarter-on-quarter variation in
the monthly average exchange rate
1996 - 2012:Q2
CBSL
INFCPIDEV
Deviation of actual inflation (change of
the Colombo Consumers’ Price Index
(CCPI)) from the indicative inflation
projection of 5 per cent
1996 - 2012:Q2
DCS
2
Author’s
estimates
YGAPSA
Output gap measure (computed using
seasonally adjusted GDP)
1996 - 2012:Q2
Author’s
estimates
DUMMY
2001:Q1-Q3 and 2008:Q1-Q3 are set to 0
-
-
1/ CBSL – Central Bank of Sri Lanka
2/ DCS – Department of Census and Statistics, Sri Lanka
Table 2: Sylised Facts (Averge for Period)
Per cent
Period
Headline
Inflation
(CCPI)
Real GDP
Growth Rate
(YRATE)
Output Gap
(YGAPSA
Average
Weighted
Call Market
Rate
(AWCMR)
Depreciation
of Rs/US $
Exchange
Rate
(EXCH4)
1996-2013Q2
10.0
5.2
0.0
11.6
5.1
1996-2001
10.1
4.0
0.7
14.5
9.3
2002-2007
10.8
5.4
-0.6
10.8
3.1
2008-2013Q2
9.1
6.4
-0.1
9.5
2.6
Sources: Central Bank of Sri Lanka, Department of Census and Statistics, Author’s calculations
28
Table 3: Descriptive Statistics
Table 3.1: Descriptive Statistics: 1996 – 2013
AWCMR
EFFECTIVE
TBILL91
EXCH4
INFCPIDEV
YGAPSA
Observations
70
70
70
70
70
70
Mean
11.64
11.31
11.69
5.07
5.02
-0.01
Median
11.05
10.50
11.29
5.15
4.20
-0.20
Maximum
23.83
22.00
21.30
37.50
23.40
6.68
Minimum
7.02
7.00
6.98
-18.90
-4.30
-5.57
Std. Dev.
3.80
3.82
3.67
8.98
5.34
1.78
Skewness
1.24
1.11
0.72
1.00
1.05
0.41
Kurtosis
4.57
3.62
2.67
6.21
4.42
5.59
Jarque-Bera
25.00
15.46
6.43
41.83
18.76
21.57
Probability
0.00
0.00
0.04
0.00
0.00
0.00
Sum
814.47
791.46
818.13
355.10
351.10
-0.51
Sum Sq. Dev.
994.44
1007.50
928.55
5566.28
1970.85
219.31
29
Table 3.2 Descriptive Statistics: 1996 – 2007
AWCMR
EFFECTIVE
TBILL91
EXCH4
INFCPIDEV
YGAPSA
Observations
48
48
48
48
48
48
Mean
12.63
11.72
12.13
6.21
5.44
0.05
Median
12.27
11.50
11.79
5.65
5.75
-0.20
Maximum
23.83
22.00
21.30
35.40
16.50
6.68
Minimum
7.48
7.00
7.00
-18.90
-1.60
-5.57
Std. Dev.
3.88
3.45
3.56
7.54
4.29
1.98
Skewness
1.18
1.25
0.72
0.54
0.23
0.44
Kurtosis
4.27
4.75
2.92
8.24
2.52
5.19
Jarque-Bera
14.39
18.70
4.11
57.35
0.88
11.14
Probability
0.00
0.00
0.13
0.00
0.65
0.00
Sum
606.03
562.71
582.34
298.20
261.10
2.32
Sum Sq. Dev.
706.11
559.17
595.92
2674.85
863.91
184.57
Table 3.3: Descriptive Statistics: 2008 – 2013
AWCMR
EFFECTIVE
TBILL91
EXCH4
INFCPIDEV
YGAPSA
Observations
22
22
22
22
22
22
Mean
9.47
10.40
10.72
2.59
4.09
-0.13
Median
8.48
8.75
9.48
-0.60
1.95
0.04
Maximum
15.09
19.00
18.39
37.50
23.40
1.72
Minimum
7.02
7.00
6.98
-10.60
-4.30
-2.49
Std. Dev.
2.57
4.48
3.80
11.32
7.17
1.28
Skewness
1.17
1.25
0.93
1.68
1.60
-0.40
Kurtosis
2.86
2.88
2.48
5.61
4.58
2.02
Jarque-Bera
5.02
5.74
3.45
16.63
11.64
1.46
Probability
0.08
0.06
0.18
0.00
0.00
0.48
Sum
208.44
228.75
235.79
56.90
90.00
-2.83
Sum Sq. Dev.
138.53
421.83
302.46
2693.07
1079.50
34.27
30
Table 4: Unit Root Tests
Variable
Augmented Dickey-Fuller test statistic
t-Statistic
Probability
AWCMR
-2.88983
0.0518
EFFECTIVE
-2.33379
0.1645
TBILL91
-2.64335
0.0895
EXCH4
-5.06895
0.0001
INFCPIDEV
-3.92929
0.0031
YGAPSA
-3.65812
0.0069
31
Annex II
Table 5: Estimates of the Contemporaneous Monetary Policy Reaction Function
Column number
1
2
3
4
5
6
Dependent variable
AWCMR
EFFECTIVE
TBILL91
Constant
2.26
2.63
2.34
2.44
1.74
1.98
(2.91)
***
(3.41)
***
(3.50)
***
(3.63)
***
(2.93)
***
(3.34)
***
INFCPIDEV
0.13
0.16
0.14
0.15
0.11
0.13
(2.61)
***
(3.18)
***
(3.05)
***
(3.25)
***
(2.79)
***
(3.30)
***
YGAPSA
0.36
0.37
0.52
0.53
0.41
0.42
(2.69)
***
(2.89)
***
(4.44)
***
(4.52)
***
(4.38)
***
(4.57)
***
AWCMR(-1)
0.75
0.68
(10.72)
***
(9.04)
***
EFFECTIVE(-1)
0.73
0.71
(11.66)
***
(10.54)
***
TBILL91(-1)
0.80
0.75
(14.28)
***
(12.59)
***
EXCH4(-1)
0.06
0.03
0.04
(2.24)
**
(1.19)
(1.98)
**
Observations
69
69
69
69
69
69
Adjusted R-squared
0.74
0.75
0.80
0.80
0.86
0.86
F-statistic
65.05
53.07
89.63
68.00
137.94
109.07
S.E. of regression
1.96
1.90
1.73
1.73
1.39
1.36
Long-run coefficient on
inflation
0.5
0.5
0.5
0.5
0.5
0.5
Long-run coefficient on
output gap
1.4
1.1
2.0
1.8
2.1
1.7
Long-run coefficient on
exchange rate
0.2
0.1
0.2
Neutral policy rate
8.9
8.8
8.8
Notes:
Absolute value of t-statistics is given in parentheses
* significant at 10%; ** significant at 5%, *** significant at 1%
Estimation method: OLS
Sample period: 1996Q1 - 2013Q2
Output gap measure: HP filter
32
Table 6: Estimates of the Backward Looking Monetary Policy Reaction Function
Column number
1
2
3
4
5
6
Dependent variable
AWCMR
EFFECTIVE
TBILL91
Constant
2.84
3.20
3.26
3.33
2.44
2.68
(3.58)
***
(4.01)
***
(4.55)
***
(4.59)
***
(3.91)
***
(4.24)
***
INFCPIDEV(-1)
0.11
0.14
0.10
0.11
0.11
0.13
(2.08)
**
(2.62)
***
(2.20)
**
(2.30)
**
(2.79)
***
(3.23)
***
YGAPSA(-1)
0.51
0.51
0.67
0.67
0.49
0.49
(3.80)
***
(3.86)
***
(5.38)
***
(5.37)
***
(5.04)
***
(5.13)
***
AWCMR(-1)
0.71
0.64
(9.64)
***
(7.98)
***
EFFECTIVE(-1)
0.67
0.65
(9.59)
***
(8.68)
***
TBILL91(-1)
0.74
0.69
(12.22)
***
(10.60)
***
EXCH4(-1)
0.06
0.02
0.03
(1.95)
*
(0.69)
(1.69)
*
Observations
69
69
69
69
69
69
Adjusted R-squared
0.75
0.76
0.80
0.80
0.86
0.87
F-statistic
68.52
54.57
91.61
68.28
141.91
110.20
S.E. of regression
1.92
1.88
1.72
1.73
1.37
1.36
Long-run coefficient on
inflation
0.4
0.4
0.3
0.3
0.4
0.4
Long-run coefficient on
output gap
1.7
1.4
2.0
1.9
1.9
1.6
Long-run coefficient on
exchange rate
0.2
0.0
0.1
Neutral policy rate
9.7
9.8
9.4
Notes:
Absolute value of t-statistics is given in parentheses
* significant at 10%; ** significant at 5%, *** significant at 1%
Estimation method: OLS
Sample period: 1996Q1 - 2013Q2
Output gap measure: HP filter
33
Table 7: Estimates of the Forward Looking Monetary Policy Reaction Function
Column number
1
2
3
4
5
6
7
8
9
10
11
12
13
Dependent variable
AWCMR
EFFECTIVE
TBILL91
Constant
1.87
-0.91
2.12
-1.05
1.68
-0.01
1.71
1.88
-0.05
1.96
-2.40
2.35
-2.34
(2.01)
**
(-0.25)
(2.20)
**
(-0.28)
(1.88)
*
(0.00)
(1.88)
*
(2.10)
**
(-0.01)
(1.63)
(-0.72)
(1.46)
(-0.65)
INFCPIDEV(+1)
0.29
0.36
0.34
0.42
0.28
0.29
0.30
0.25
0.31
0.34
0.44
0.42
0.52
(1.96)
**
(1.79)
*
(2.44)
**
(1.96)
**
(2.71)
***
(2.02)
**
(2.72)
***
(3.41)
***
(2.11)
**
(3.45)
***
(2.34)
**
(2.81)
***
(2.02)
**
YGAPSA(+1)
0.65
0.69
0.52
0.59
0.95
0.97
0.92
1.01
0.94
0.57
0.66
0.45
0.55
(1.13)
(1.21)
(0.93)
(1.03)
(1.96)
**
(1.90)
*
(1.98)
**
(1.94)
*
(1.94)
*
(1.65)
*
(1.61)
(1.20)
(1.30)
AWCMR(-1)
0.72
0.77
0.64
0.70
(7.68)
***
(6.89)
***
(5.93)
***
(5.60)
***
EFFECTIVE(-1)
0.73
0.78
0.71
0.72
0.76
(9.12)
***
(4.86)
***
(7.72)
***
(8.21)
***
(4.12)
***
TBILL91(-1)
0.68
0.77
0.60
0.70
(5.54)
***
(5.39)
***
(2.89)
***
(3.09)
***
EXCH4(-1)
0.06
0.06
0.03
0.02
0.05
0.05
(1.52)
(1.41)
(0.77)
(0.58)
(1.10)
(0.95)
REER(-1)
0.03
(0.48)
DUMMY
2.02
2.30
1.12
1.16
3.14
3.36
(0.71)
(0.81)
(0.35)
(0.36)
(1.16)
(1.23)
Observations
68
68
68
68
68
68
68
68
68
68
68
68
68
Adjusted R-squared
0.62
0.57
0.63
0.57
0.61
0.60
0.62
0.60
0.60
0.72
0.65
0.69
0.60
J-statistic
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
S.E. of regression
2.35
2.50
2.33
2.52
2.40
2.45
2.39
2.45
2.44
1.96
2.20
2.07
2.35
Long-run coefficient
on inflation
1.0
1.5
1.0
1.4
1.0
1.4
1.0
0.9
1.3
1.1
1.9
1.0
1.7
Long-run coefficient
on output gap
2.3
3.0
1.5
2.0
3.6
4.5
3.1
3.6
4.0
1.8
2.9
1.1
1.8
Long-run coefficient
on exchange rate
0.2
0.2
0.1
0.1
0.1
0.1
0.2
Neutral policy rate
6.6
6.3
6.2
Notes:
Absolute value of t-statistics is given in parentheses
* significant at 10%; ** significant at 5%, *** significant at 1%
Estimation method: GMM
Instruments used for GMM estimation: INFCPIDEV(-1), YGAPSA(-1), AWCMR(-1), EFFECTIVE(-1), TBILL91(-1),
EXCH4(-1) and DUMMY
Sample period: 1996Q1 - 2013Q2
Output gap measure: HP filter
