Journal of Asian Economics 25 (2013) 1727

Contents lists available at SciVerse ScienceDirect

Journal of Asian Economics

Revisiting the Phillips curve for India and ination forecasting

Muneesh Kapur *
International Monetary Fund, 700 19th street, Washington, DC 20431, United States

A R T I C L E I N F O

A B S T R A C T

Article history:
Received 6 August 2012
Received in revised form 13 December 2012
Accepted 23 December 2012
Available online 4 January 2013

This paper focuses on modeling and forecasting ination in India using an augmented
Phillips curve framework. Both demand and supply factors are seen as drivers of ination.
Demand conditions are found to have a stronger impact on non-food manufactured
products (NFMP) ination vis-a-vis headline wholesale price ination; moreover, NFMP
ination is found to be more persistent than headline ination. Both these ndings support
the use of NFMP ination as a core measure of ination. But, the impact of global non-fuel
commodities on NFMP ination is found to be substantial. Ination in non-fuel
commodities is seen as a more important driver of domestic ination rather than fuel
ination. The exchange rate pass-through coefcient is found to be modest, but
nonetheless sharp depreciation in a short period of time can add to inationary pressures.
The estimated equations show a satisfactory in sample as well as out-of-sample
performance based on dynamic simulations. Nonetheless, forecasting challenges emanate
from volatility in international oil and other commodity prices and domestic food supply
dynamics.
2012 Elsevier Inc. All rights reserved.

JEL classication:
E31
E32
E52
E58
Keywords:
Exchange rate pass-through
India
Ination
Monetary policy
Phillips curve

1. Introduction
Low and stable ination is a key objective of monetary policy for central banks, whether ination targeting or otherwise.
However, achievement of low and stable ination is quite challenging. It is well-known that monetary policy affects output
and prices with lags, which are both long and variable. Accordingly, monetary policy has to be forward-looking, i.e.,
monetary policy needs to act today in anticipation of future growth and ination trajectory. Therefore, forecasts of growth
and ination play a critical role in the conduct and formulation of monetary policy and its ultimate success in achieving price
stability. This, in turn, depends upon success in modeling and forecasting ination and growth.
In India, low and stable ination remains a key objective of monetary policy along with growth and nancial stability.
Ination dynamics in emerging economies like India are, however, relatively more complex than advanced economies in
view of recurrent supply shocks and large weight of volatile components such as food items in the various price indices. This
makes ination modeling and forecasting more challenging in countries like India.
Ination in India has remained persistently high since early 2010, with headline WPI moving in a range of 910% until
October 2011, signicantly above its average of around 5% recorded during the 2000s. Non-food manufactured products
(NFMP) ination, a measure of underlying ination, also increased over the course of 2011 to a range of 78%. What explains
these ination dynamics? Can these be explained by standard models and approaches such as the Phillips curve framework?

The views expressed in the paper are those of the author and not necessarily of the institutions to which he belongs.
* Corresponding author. Tel.: +1 202 623 7443.
E-mail addresses: mkapur@imf.org, mkapur@rbi.org.in.

1049-0078/$ see front matter 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.asieco.2012.12.002

18

M. Kapur / Journal of Asian Economics 25 (2013) 1727

Empirical analysis undertaken in this paper indicates that the Phillips curve framework is able to explain ination dynamics
in the Indian context. Demand conditions are found to have a stronger impact on NFMP ination vis-a-vis headline WPI
ination. The impact of ination in global non-fuel commodities on NFMP ination is found to be substantial. Although the
exchange rate pass-through coefcient is modest, sharp depreciation can add to inationary pressures. While the Phillips
curve framework provides a satisfactory forecasting performance, forecasting challenges nonetheless emanate from
volatility in international commodity prices and domestic food supply dynamics.
Against this backdrop, Section 2 undertakes a review of the Phillips curve approach to modeling and forecasting ination and
discusses both the traditional Phillips curve and the New Keynesian Phillips Curve (NKPC) approaches. This section also assesses
the available cross-country empirical evidence for or against the Phillips curve. Section 3 attempts to model ination in India
based on the Phillips curve framework and assesses the forecasting performance of this approach. Section 4 concludes.
2. Modeling ination in the Phillips curve framework: theory and evidence
The Phillips curve framework provides one way of forecasting ination. Actual ination movements are inuenced not
only by demand side pressures but also by supply shocks. Ination also exhibits an inertia indicating that expectations are
largely adaptive. Lagged ination, therefore, remains an important determinant of ination and the lags could reect the
structure of the economy. Incorporating demand and supply factors as well as ination expectations leads to an augmented
Phillips Curve also termed as the triangle model of ination (Gordon, 1998). The phrase triangle stresses that ination
depends on a tripartite set of basic determinants: inertia (in ination), demand and supply shocks.

Pt aLPt1 bLDt cLZ t

(1)

where Pt, Dt and Zt denote ination, a measure of excess demand (unemployment gap or output gap) and supply shocks
(imported ination, exchange rate movements and weather-related shocks such as drought/excessive rainfall), respectively;
a(L), b(L) and c(L) are lag polynomials.
2.1. New Keynesian Phillips curve
As against the above traditional backward-looking ad hoc Phillips curve, in recent years, the Phillips curve has been
derived from micro-foundations, with optimal price setting by forward-looking monopolistically competitive rms. Such a
formulation leads to the NKPC, a purely forward-looking Phillips curve. In this specication, ination depends, inter alia,
upon expected future ination (EtPt+1); in contrast, ination depends on expected current ination (Et1Pt) in the traditional
expectations-augmented standard Phillips curve. The purely forward-looking Phillips curve, however, does not get much
empirical support. Lagged ination remains an important determinant of ination, and a purely backward-looking Phillips
curve seems to be preferred by the data, which has led to ad hoc hybrid Phillips curve with both forward- and backwardlooking ination components (Gali & Gertler, 1999; Gali, Gertler, & Lopez-Salido, 2005). Gali et al. (2005) nd that the
coefcient on expected ination is higher than that of lagged ination, which they argue as evidence in favor of the NKPC.
However, the NKPC and its empirical estimates are subject to serious identication issues as these specications do not
allow us to distinguish forward-looking models from backward-looking models; the higher weight on expected ination
may be due to misspecication resulting from omission of explanatory variables from the main equation. Typically, expected
ination is instrumented through lagged ination amongst the instrument set and this can bias the coefcient on expected
ination to be higher and the NKPC can yield large estimates of the coefcient on expected ination even when forwardlooking behavior is completely absent (Rudd & Whelan, 2007). Moreover, while many studies nd that the forward-looking
behavior dominates, the robust condence intervals are so wide that the results are consistent both with no backwardlooking dynamics as well as very substantial backward-looking behavior (Kleibergen & Mavroeidis, 2009).
2.2. Phillips curve forecasts: an assessment
While the Phillips curve framework remains the workhorse model for modeling ination and thinking about policy
issues, question marks have been raised over its forecasting abilities. According to Atkeson and Ohanian (2001), random walk
(naive) forecasts beat the (backward-looking) Phillips curve forecasts. Stock and Watson (2009) note that the Phillips curve
forecasts are better than other multivariate forecasts, but their performance is episodic, sometimes better than and
sometimes worse than a good univariate benchmark. Peach, Rich, and Cororation (2010) nd threshold effects in the Phillips
curve, i.e., if the output gap (or unemployment gap) is within a certain threshold, the relationship between ination and
activity is weak, but when the output gap is outside these thresholds, there is a signicant impact of economic activity on
ination. For the US, they estimate the threshold in terms of the unemployment gap to be 1.56%; thus, if the unemployment
gap is within 1.56%, there is no impact of unemployment on ination, and the effect on ination is signicant only when the
unemployment gap is outside this threshold of 1.56%. Meyer and Pasaogullari (2010) nd that no single specication
outperforms all others over all time periods; for example, for the US, they nd that the median and 16% trimmed-mean measures
outperform all other specications during the 1990s, and survey-based ination expectations seem to do better during volatile
periods. On the other hand, Fuhrer, Olivei, and Tootel (2009b) nd that the forecasting performance of the Phillips curve is better if
changing dynamics of ination, in particular the weakening impact of oil prices on ination, are taken into account.

M. Kapur / Journal of Asian Economics 25 (2013) 1727

19

Between the backward-looking and the forward-looking specications, micro-founded versions of the Phillips curve can
be viewed as complementary to the standard backward-looking specications and there is little evidence suggesting that the
forward-looking Phillips curve specications provide more accurate ination forecasts than the standard backward-looking
specication (Fuhrer, Kodryzycki, Little, & Olivei, 2009a). According to Gordon (2011), the triangle model outperforms the
NKPC variant by orders of magnitude, not only in standard goodness-of-t statistics, but also in post-sample dynamic
simulations. Similarly, Ball and Mazumder (2011) nd the Great Recession of 200910 provides evidence against the NKPC.
As a result, the traditional backward-looking specication, augmented to account for supply shocks, continues to play a role
in shaping the ination outlook and the conduct of monetary policy.
2.3. Phillips curve studies for India
In the Indian context, previous attempts to model ination using the Phillips curve framework include Dholakia (1990),
RBI (2002, 2004), Kapur and Patra (2000), Srinivasan, Mahambare, and Ramachandra (2006), Dua and Gaur (2009), Paul
(2009), Patra and Ray (2010), Patra and Kapur (2012), Singh, Kanakaraj, and Sridevi (2011), and Mazumder (2011). RBI (2002,
2004) and Kapur and Patra (2000), using annual data, found evidence for the Phillips curve relationship in India, with role for
both excess demand conditions (output gap) and supply shocks (food ination and import prices). Srinivasan et al. (2006) did
not nd support for the Phillips curve with coefcients on the output gap terms being insignicant (although positive) with
headline ination as the dependent variable; the coefcient on the output gap was found to be negative with manufacturing
headline ination as the dependent variable. Their analysis is based on monthly data for the period April 1994March 2005
and industrial production as the activity variable. Using annual data, Paul (2009) is able to nd support in favor of the Phillips
curve only when industrial production is used as an indicator of economic activity (instead of overall GDP) and data are rearranged on a crop year basis (instead of scal year basis). Dua and Gaur (2009) investigated the Phillips curve relationship
for a number of Asian economies (both developing and developed) and found support for the Phillips curve in India (sample
period 19962005 using quarterly data) as well as other economies in their study; while they included imported ination as
an explanatory variable for the developed countries in their sample, it was not included in their developing countries
sample.
Patra and Ray (2010) employed the Phillips curve framework in the context of modeling ination expectations and found
support in favor of the relationship using monthly data for the period 19972008. Patra and Kapur (2012) estimated a range
of Phillips curves in the context of a new Keynesian model using quarterly data for 19962009. They estimated traditional
backward-looking Phillips curve as well as purely forward-looking NKPC and hybrid NKPC, while controlling for supply
shocks and using overall GDP as the activity variable. While Patra and Kapur (2012) found some support for the NKPC and the
hybrid version of the same, the estimated equations suffered from serial correlation. Moreover, in the hybrid NKPC, the
coefcient on lagged ination was found to be higher than that of the expected ination. The backward-looking Phillips
curve satised the various diagnostics. Singh et al. (2011) were able to nd the Phillips curve relationship in the latter part
(2004:Q22009:Q4) but not in the rst part (1997:Q42004:Q1) of their sample. Supply shocks were captured rather
crudely by looking at the outlier point in the contemporaneous relationship between ination and the output gap.
Mazumder (2011) found support for the Phillips curve relationship for India using quarterly data for the period 19702008
with economic activity proxied by industrial production and movements in oil prices as a control for supply shock; the
relationship was found to be stable across various monetary regimes proxied by the terms of various Governors. All the
above mentioned studies used the output gap estimates based on HodrickPrescott lter, except for Singh et al. (2011) who
used Kalman lter. Bhalla (2011) focused on the role of minimum support prices in driving ination dynamics in a simple bivariate framework, regressing headline ination on variation in minimum support prices.
A review of the existing studies in the Indian context shows a number of limitations. First, a number of studies use
industrial production as the activity variable to measure demand pressures. Overall ination in an economy reects
aggregate demand pressures and these are best captured by excess demand measures based on overall activity. Industrial
activity accounts for less than a fth of the Indian GDP and captures the overall demand pressures rather imperfectly.
Services sector now accounts for two-thirds of GDP and its exclusion may give a misleading picture. Second, a number of
studies are based on annual data, which reduces their relevance for policy purposes and moreover, such studies cannot
capture ination dynamics appropriately. Third, in almost all studies, the role of external supply shocks is limited to
international crude oil prices. In recent years, non-oil commodity prices have also witnessed a signicant jump amidst
elevated volatility. Moreover, while domestic fuel prices are administered in India as is the case in many other EMEs,
domestic non-oil prices are relatively freely determined. In this context, global non-oil commodity ination trends are
potentially important determinants of ination and their role needs to be assessed. Fourth, in the recent period, the Reserve
Bank has articulated NFMP ination as an indicator of demand-side pressures (RBI, 2011). According to Raj and Misra (2011),
who examine a host of core measures of ination, NFMP ination is the only measure which satises all the properties of a
core measure. None of the existing studies have modeled this component of ination. Fifth, the role of minimum support
prices is tested in a bivariate framework in Bhalla (2011), but not in a multivariate framework. Sixth, given the continued
dependence of agriculture on monsoon conditions, rainfall conditions remain a key determinant of domestic food and overall
ination. Finally, the sharp depreciation of the rupee during JulySeptember 2011 again brought into focus the role of
exchange rate pass-through. Existing studies have focused on movements in nominal exchange rate of the rupee vis-a`-vis the
US dollar, but sensitivity analysis to nominal effective exchange rate is missing. This issue is found to be important in the

M. Kapur / Journal of Asian Economics 25 (2013) 1727

20

context of modeling non-food manufactured products ination, as discussed in a later section. Overall, the various existing
studies touch upon some aspects of the determinants of ination, but a comprehensive assessment taking into account all
potential determinants is missing. This paper, therefore, attempts to overcome these limitations of the existing studies and
attempts to provide a comprehensive approach to ination dynamics in India.
3. Phillips curve modeling for India
3.1. Methodology and data
Given the empirical superiority of the traditional Phillips curve over the NKPC, this paper uses the former framework,
following Gordon (1998), to model ination in India. In India, wholesale price index (WPI) ination is the headline measure
of ination, although the Reserve Bank also takes into account trends in other ination indicators such as consumer prices
and GDP deator. As noted earlier, the Reserve Bank has recently emphasized NFMP ination as an indicator of demand-side
pressures; we, therefore, model both headline WPI ination and ination based on NFMP group within the WPI. Based on the
discussion in the preceding section, apart from the demand conditions, the potential explanatory variables are a host of
supply shocks such as global commodity ination, rainfall conditions, minimum support prices and the exchange rate. In the
context of high fuel ination, an issue of topical interest is the extent of spillover to NFMP ination. Drawing from the
discussion above, variants of the following general specication of Phillips curve are estimated:
WPIt a0 a1  YGAPt a2  Et1 WPIt a3  INFGLt a4  OILt a5  EXCHAt or; a5  NEERt a6  RAINt
a7  MSPt vt
NFMPt a0 a1  YNGAPt a2  Et1 NFMPt a3  INFGLt a4  OILt a5  EXCHAt or; a5  NEERt vt

(2)
(3)

The variables are dened as follows (all data are in percentage terms): WPI is ination (year-on-year, y-o-y) based on
headline wholesale price index; NFMP is ination (y-o-y) in WPIs non-food manufactured products group; INFGL
is variation (y-o-y) in global non-fuel commodity price index; OIL is variation (y-o-y) in international crude oil prices; MSP is
variation (y-o-y) in minimum support prices; EXCHA is variation (y-o-y) in the exchange rate of the Indian rupee vis-a`-vis
the US dollar; NEER is variation (y-o-y) in the 36-currency trade-weighted nominal effective exchange rate index of the
Indian rupee complied by the Reserve Bank of India; YGAP is output gap (actual real GDP less trend (HodrickPrescott (HP)
ltered) real GDP, using seasonally adjusted data);1 YNGAP is non-agricultural output gap (actual real GDP excluding
agriculture and non-community, personal and social services less its HP ltered trend, using seasonally adjusted data); RAIN
is deviation of actual rainfall during July from its normal level during July. Et1WPIt and Et1NFMPt denote expected ination
in period t  1 for the next period. Ination expectations are assumed to be adaptive and are captured through lags of
ination following Gordon (1998) and others.
Data on global non-fuel commodity prices and international crude oil prices are taken from the International Monetary Fund
(IMF); all other data are from the Reserve Bank of Indias Handbook of Statistics on the Indian Economy. All the data series are
stationary, with some ambiguity about MSP ination.2 Two lags of ination are included in the equations to ensure no residual
autocorrelation. The equations are estimated using quarterly data for the period 19962011 (AprilJune 1996 to January
March 2011); the sample period starts from 1996, since data on quarterly GDP are available from the quarter ended June 1996.
3.2. Empirical results
Results for headline WPI ination are set in Table 1 and those for non-food manufactured products ination are in Table 2.
The estimated coefcients are on the expected lines. Column 2 in both the tables reports results for the baseline Phillips
curve without supply shocks. Subsequent columns in both the tables estimate the augmented Phillips curve by adding
various supply shocks.
3.3. Headline WPI ination
Beginning with the specication for the headline ination, the key features of the estimates are: rst, excess demand
conditions have an upward pressure on ination, while deciency in demand pulls down ination. The estimates indicate
that if the output gap is 1% (i.e., actual output is 1% above the potential output level), then ination increases by 1925 basis
points with a lag of one quarter and the long-run impact3 is 4053 basis points (Table 1, Regressions 15). The demand
variable is signicant even in the baseline specication that does not incorporate supply shocks (Table 1, Regression 1). Thus,

Results are robust to the computation of the output gap using the BaxterKing band-pass lter.
Unit root tests (Augmented DickeyFuller tests, with lag selection based on BIC criteria) indicate that all data series are stationary except for MSP
ination. However, the KPSS test cannot reject the null of stationarity for MSP ination.
3
Long-run impact is the coefcient on the output gap (or other relevant variable of interest) divided by one minus the sum of the coefcients on the
lagged dependent variable.
2

M. Kapur / Journal of Asian Economics 25 (2013) 1727

21

Table 1
Phillips curve estimates for wholesale price ination.
Variable
1

Constant
YGAP(1)
WPI(1)
WPI(2)

Dependent variable: wholesale price ination (WPI)

2

Regression 1

Regression 2

Regression 3

Regression 4

Regression 5

Regression 6

AndrewsPloberger
statistic (p-values)

2.56
(4.3)
0.25
(2.2)
1.20
(16.1)
0.67
(4.9)

1.30
(3.4)
0.20
(2.3)
0.89
(13.9)
0.40
(4.7)

1.38
(3.3)
0.20
(2.1)
0.91
(13.2)
0.40
(4.4)

2.01
(4.8)
0.25
(2.5)
0.97
(13.1)
0.48
(5.3)

2.00
(4.7)
0.28
(2.5)
0.95
(14.1)
0.45
(4.8)

0.06

0.01

0.52

0.74

0.09

0.00

0.19

0.01

0.09
(8.4)
0.06
(3.3)

0.07
(6.3)

1.24
(2.7)
0.19
(2.1)
0.89
(13.9)
0.39
(4.1)
0.002
(0.3)
0.08
(8.5)
0.06
(3.6)

0.08
(4.7)
0.08
(3.4)

0.08
(4.7)
0.08
(3.3)

OIL(4)
INFGL
EXCHA
NEER
RAIN_JULY(2)

0.03
(2.8)
0.03
(3.6)
0.10
(4.0)

RAIN_JULY(3)
MSP(1)

0.05
(2.3)
0.02
(2.1)
0.02
(2.4)
0.11
(4.9)

DUM2000Q4
DUM2003Q4

0.02
(2.6)
0.03
(3.7)
0.10
(4.4)

0.03
(2.6)
0.03
(3.6)

0.36

0.03
(2.6)
0.03
(3.5)

0.30
0.02

1.48
(4.0)
1.88
(10.7)
2.18
(14.4)

1.71
(4.8)
2.14
(13.5)
1.70
(7.8)

1.52
(3.6)
1.85
(9.7)
2.18
(15.2)

0.06
(1.6)
1.06
(3.7)
2.15
(13.1)
1.60
(9.7)

0.50
(10.0)
0.88
0.52
0.91
0.41

0.50
(10.1)
0.87
0.35
0.75
0.38

0.50
(9.5)
0.87
0.48
0.93
0.42

0.50
(7.5)
0.85
0.15
0.85
0.53

0.02
1.13
(3.9)
2.03
(10.3)
1.80
(11.5)

0.50
(7.5)
0.85
0.12
0.79
0.12

All coefcients
Sum of lagged ination
coefcients
R-bar2
LB-Q(4) test
JB test
White test

0.00

0.08

DMPSP(1)
DUM1998Q3

0.16
0.53
(4.5)
0.73
0.26
0.27
0.07

0.49

Source: Authors calculations based on data from International Monetary Fund and Reserve Bank of India.
Notes: Sample period for the estimation is 1996:22011:1. Figures in parentheses are t-statistics, with HAC standard errors corrected with NeweyWest/
Bartlett window and three lags. LB-Q test gives signicance level (p-value) of BoxPierceLjung Q-statistic for the null of no residual autocorrelation for 4
lags. JB test gives signicance level (p-value) for JarqueBera test for the null of normality of residuals. White test gives signicance level (p-value) for White
test for the null of homoskedasticity of residuals. Columns 8 and 9 give p-values for AndrewsPloberger statistics for stability of coefcients for Regressions
1 and 3, respectively.
Variables are dened as follows: WPI is headline wholesale price index ination (y-o-y); INFGL is variation (y-o-y) in global non-fuel commodity price
index; OIL is variation (y-o-y) in international crude oil prices; EXCHA is variation (y-o-y) in the exchange rate of the Indian rupee vis-a`-vis the US dollar;
NEER is variation (y-o-y) in the 36-currency trade-weighted nominal effective exchange rate index of the Indian rupee complied by the Reserve Bank of
India; YGAP is output gap (actual real GDP less its HP ltered trend, using seasonally adjusted data); RAIN_JULY is deviation of actual rainfall from its normal
level in July; MSP is variation (y-o-y) in minimum support prices for agricultural items announced by the government. DSMP is rst-difference of MSP. All
data are in percentage terms. DUM1998Q3, DUM2000Q4 and DUM2003Q4 are dummies for quarters 1998:3, 2002:4 and 2003:4, respectively.
Data on global non-fuel commodity prices and international crude oil prices are taken from the International Monetary Fund (IMF); all other data are from
the Reserve Bank of Indias Handbook of Statistics on the Indian Economy.

contrary to Paul (2009) and some other studies, the Phillips curve relation for India exists without the need to incorporate
supply shocks and other adjustments.
Second, the ination process is persistent, with the sum of lagged coefcients being around 0.5 and highly signicant.
Third, global commodity prices have a strong and quick pass-through (Table 1, Regressions 26). An increase of 10% in global
non-fuel commodity prices increases headline WPI ination by 7090 basis points in the same quarter, with the long-run
impact being double (140180 basis points); the coefcient needs to be juxtaposed with the large volatility in international
commodity prices witnessed in the recent years. Fourth, when we add international crude oil prices to the equation
(Regression 4), the coefcient is found to be positive but statistically insignicant. This could be reecting delayed and
incomplete pass-through of high international crude oil prices to domestic prices in view of the administered nature of
domestic fuel prices. The Government has also modulated the taxes and duties on petroleum products to smoothen the

M. Kapur / Journal of Asian Economics 25 (2013) 1727

22

Table 2
Phillips curve estimates for non-food manufactured products ination.
Variable

Dependent variable: non-food manufactured products ination (NFMP)

Regression 1

Regression 2

Regression 3

Regression 4

Regression 5

AndrewsPloberger
statistic (p-values)

1.18
(5.7)
0.30
(2.7)
1.33
(20.6)
0.64
(6.9)

0.97
(8.2)
0.25
(3.2)
1.11
(16.8)
0.44
(6.4)
0.05
(6.1)

1.00
(9.0)
0.26
(4.0)
1.10
(15.1)
0.44
(6.5)
0.05
(6.8)

1.06
(9.5)
0.29
(3.7)
1.10
(19.1)
0.48
(8.9)
0.05
(7.3)
0.01
(2.1)

0.84
(8.1)
0.23
(3.3)
1.06
(15.0)
0.41
(6.4)
0.05
(8.0)

0.03

0.11

0.21

0.07

0.12

0.23

0.22

0.29

Constant
YNGAP(1)
NFMP(1)
NFMP(2)
INFGL
OIL(4)

0.14

0.02
(2.2)

MINOIL
EXCHA

0.01
(0.4)

NEER
DUM1999Q1

1.87
(14.3)
1.90
(9.5)

DUM2005Q3

0.03
(2.3)
1.92
(13.4)
1.79
(11.2)

0.03
(2.0)
2.20
(11.9)
1.98
(13.1)

0.03
(2.8)
2.13
(15.1)
1.84
(14.4)

All coefcients
Sum of lagged ination
coefcients
R-bar2
LB-Q(4) test
JB test
White test

0.32

0.13
0.68
(10.6)
0.82
0.53
0.93
0.02

0.68
(21.6)
0.91
0.35
0.92
0.07

0.66
(20.0)
0.91
0.42
0.42
0.05

0.62
(16.2)
0.92
0.69
0.94
0.13

0.28

0.66
(23.6)
0.92
0.29
0.98
0.09

Source: Authors calculations based on data from International Monetary Fund and Reserve Bank of India.
Notes: Sample period for the estimation is 1996:22011:1. Figures in parentheses are t-statistics, with HAC standard errors corrected with NeweyWest/
Bartlett window and three lags. LB-Q test gives signicance level (p-value) of BoxPierceLjung Q-statistic for the null of no residual autocorrelation for 4
lags. JB test gives signicance level (p-value) for JarqueBera test for the null of normality of residuals; White test gives signicance level (p-value) for White
test for the null of homoskedasticity of residuals. Columns 7 and 8 give p-values for AndrewsPloberger statistics for stability of coefcients for Regressions
1 and 4, respectively.
Variables are dened as follows: NFMP is variation (y-o-y) in the non-food manufactured products component in the wholesale price index (WPI); INFGL is
variation (y-o-y) in global non-fuel commodity price index; OIL is variation (y-o-y) in international crude oil prices; MINOIL is variation (y-o-y) in mineral
oils sub-group in WPI; EXCHA is variation (y-o-y) in the exchange rate of the Indian rupee vis-a`-vis the US dollar; NEER is variation (y-o-y) in the 36currency trade-weighted nominal effective exchange rate index of the Indian rupee complied by the Reserve Bank of India; YNGAP is non-agricultural
output gap (actual real GDP excluding agriculture and non-community, personal and social services less its HP ltered trend, using seasonally adjusted
data); All data are in percentage terms. DUM1998Q1 and DUM2005Q3 are dummies for quarters 1998:1 and 2005:3, respectively.
Data on global non-fuel commodity prices and international crude oil prices are taken from the International Monetary Fund (IMF); all other data are from
the Reserve Bank of Indias Handbook of Statistics on the Indian Economy.

impact of volatility in international crude prices on domestic ination. These factors create a wedge between movements in
international crude oil prices and domestic fuel prices, which make it difcult to estimate the impact in the equation. Similar
ndings are reported by Dua and Gaur (2009) on the role of oil prices in the ination process for India as well as the other
three developing Asian countries (China, Philippines and Thailand) in their sample. Mohanty and Klau (2001) too report a
similar nding in only 5 out of 14 EMEs in their sample, oil prices are found to have a signicant impact on ination.
Fifth, the coefcient on the exchange rate indicates that the exchange rate pass-through is 0.06 in the short-run and 0.12 in
the long-run, i.e., 10% appreciation (depreciation) of rupee vis-a`-vis the US dollar reduces (increases) ination by 60 basis points
in the same quarter, while the long-run pass-through is 120 basis points. The results are broadly similar when we use the NEER
instead of the nominal exchange rate.4 The exchange rate pass-through coefcient is thus relatively low and is consistent with
other estimates (e.g., Khundrakpam, 2008; Patra & Kapur, 2012). The exchange rate pass-through for India is close to that of low
ination countries (0.16) (Choudhri & Hakura, 2001). The Indian rupee depreciated by around 10% vis-a-vis the US dollar during
JulySeptember 2011 and the pass-through estimates in this paper suggest that this depreciation could add almost 120 basis
points to headline ination in the long-run.

4
To maintain comparability between the results for the nominal exchange rate and the NEER, an increase (decrease) in both these variables is
depreciation (appreciation).

M. Kapur / Journal of Asian Economics 25 (2013) 1727

23

Sixth, given the importance of the south-west monsoon, rainfall shortage during the month of July the critical month for
kharif sowing is found to have an adverse impact on ination, with a lag of 23 quarters. A deciency of 10% in the rainfall in
July increases headline ination by 60 basis points with a lag of three quarters and the long-run impact turns out to be 120
basis points. Finally, minimum support prices (MSP) have a substantial impact: 10% increase in MSP increases headline WPI
ination by 100 basis points with lag of a quarter, and the long-run impact is 200 basis points. At the same time, MSP is also
found to respond to headline WPI ination with a lag. However, as noted earlier, the unit root tests provide conicting results
regarding MSP ination: while the ADF tests cannot reject the null hypothesis of unit root, the KPSS test accepts the null of
stationarity. Therefore, in order to check the robustness of the results, Table 1 (Regression 5) reports results when the rst
difference of MSP ination (DMSP) is used, while Regression 6 reports results when MSP ination is dropped from the
specication. In both cases, the results are broadly the same. When the rst difference of MSP ination is used, the coefcient
is found to be positive and somewhat lower, but it is signicant only at slightly below 10%. When the MSP ination variable is
dropped, the results are qualitatively similar as in the baseline (Regression 2); the only difference is that the coefcients on
the output gap and the exchange rate are somewhat higher.
Diagnostic tests are satisfactory and indicate residuals are normally distributed, free from serial correlation and are
homoskedastic (Table 1).5 The various equations have good explanatory power with adjusted R2 of 0.850.88. Formal stability
tests AndrewsPloberger tests indicate that the estimated specications are stable; Table 1 (columns 89) reports these
tests for the baseline specication (Regression 1) and the preferred specication (Regression 3). According to the tests, the null
of coefcient stability cannot be rejected for individual coefcients as well as all coefcients together for the baseline
specication (Regression 1). For the augmented and the preferred specication (Regression 3), there is some evidence of
instability in some of the individual coefcients; however, the null of stability for all coefcients jointly cannot be rejected.
The preferred equation (Regression 3) has a good explanatory power and captures the turning points in ination
relatively well, although there are periods of deviations (Chart 1, Panel A). These are partly due to movements in domestic
fuel and food prices, which are not explicitly modeled in the equation to avoid over-tting.
3.4. Non-food manufactured products ination
Turning to NFMP ination (Table 2), the results are qualitatively the same as in the headline ination case. There are,
however, a few notable differences from the estimates in Table 1. First, the impact of demand conditions on ination is now
higher than that in the headline ination case. A positive (negative) non-agricultural GDP gap of 1% increases (reduces) NFMP
ination by 2330 basis point with a lag of one quarter, and the long-run impact is 6894 basis points (Table 2, Regressions
15). The long-run impact of demand conditions on NFMP ination is more than that in the headline WPI ination case,
supporting the RBIs policy focus on NFMP ination as an indicator of demand pressures in the economy.
Second, NFMP ination is more persistent compared to headline ination, as may be seen from the sum of lagged ination
coefcients (0.620.68 in the case of NFMP and around 0.50 in the case of headline ination). The relatively stickier nature of
NFMP ination also extends support to it being used as a core measure of ination or an indicator of underlying ination
pressures. As Woodford (2003) has noted, it is the stickiness in prices or the persistence in ination that leads to deviation of
actual output from its natural (potential) level of output. As all goods prices are not sticky, central banks should target a
measure of core ination that places greater weight on those prices which are stickier.
Third, global commodity ination remains an important driver of NFMP ination: the coefcient estimates indicate that
an increase of 10% in global non-fuel prices increases NFMP ination by 50 basis points in the same quarter and 130160
basis points in the long-run. Both the immediate and the long-run impact of global non-fuel ination on NFMP ination (50
and 130160 basis points) is somewhat less than that on headline ination (7090 and 140180 basis points). Nonetheless,
the impact of global ination on NFMP ination is substantial this nding does not lend support to NFMP being a core
indicator of ination. Thus, whereas the ndings of strong demand impact and more persistent nature support the case of
NFMP as a core indicator of ination, the large impact of global commodity shocks on NFMP ination raises the question as to
whether NFMP ination is imported ination (Mohanty, 2011). We revert to this issue later.
Fourth, international crude oil prices are found to have a statistically signicant impact on NFMP ination, but the impact
is quite modest vis-a-vis non-fuel commodity prices. An increase of 10% in international crude oil prices increases NFMP
ination by only 10 basis points and that too with a lag of four quarters; the long-run impact is 30 basis points (Table 2,
Regression 4). The modest impact can be attributed to delayed and incomplete pass-through to domestic prices, as noted
earlier. When we use the WPI mineral group ination as an indicator of oil ination (in lieu of international crude oil
ination), the impact is somewhat stronger as well as quicker. An increase of 10% in WPI mineral oil group ination increases
NFMP ination by 20 basis points in the same quarter and 60 basis points in the long-run (Table 2, Regression 5).
Finally, the exchange rate coefcient surprisingly is not found to be signicant when we use the nominal exchange rate of
the rupee vis-a`-vis the US dollar (Table 2, Regression 3). However, when we use the movements in the NEER as an indicator of
the exchange rate, the coefcient is found to be statistically signicant. The exchange rate pass-through coefcient is 0.03
and 0.080.09 in the short-run and long-run, respectively (Table 2, Regressions 35). Thus, 10% depreciation of the NEER

5
Unit root tests for residuals (not reported) both for headline ination (Table 1) and non-food manufactured products ination (Table 2) indicate that
these are stationary.

M. Kapur / Journal of Asian Economics 25 (2013) 1727

24

A: Wholesale Price Index (WPI) Inaon

12

Per cent

10
8
6
4
2

Actual

Dec-10

Jun-09

Mar-10

Sep-08

Dec-07

Jun-06

Mar-07

Sep-05

Dec-04

Jun-03

Mar-04

Sep-02

Dec-01

Jun-00

Mar-01

Sep-99

Dec-98

Jun-97

Mar-98

Sep-96

Fied (using specicaon Regression 3, Table 1)

B: WPI Non-Food Manufactured Products Inaon

10
8

Per cent

6
4
2

Actual

Dec-10

Mar-10

Jun-09

Sep-08

Dec-07

Mar-07

Jun-06

Sep-05

Dec-04

Mar-04

Jun-03

Sep-02

Dec-01

Mar-01

Jun-00

Sep-99

Dec-98

Mar-98

-4

Jun-97

-2

Sep-96

Fied (using specicaon Regression 4, Table 2)

Source: Authors Calculations, based on Tables 1 and 2.

Chart 1. Actual and tted ination.
Source: Authors calculations, based on Tables 1 and 2.

leads to an increase of 30 basis points in NFMP ination in the same quarter and 8090 basis points in the long-run. The
difference in the results using the nominal exchange rate and the NEER may be due to the fact that there have been periods
when the two measures indicate contradictory movements. This can occur due to cross-currency movements whereby the
rupee may depreciate vis-a`-vis the US dollar, but it may appreciate (or show a lower degree of depreciation) vis-a`-vis
currencies of other major trading partners.
Diagnostic tests indicate no residual serial correlation, and the residuals are normally distributed. The alternative
specications have high explanatory power (with adjusted R2 of 0.910.92). We chose Regression 4 as the preferred
specication and this has a relatively good t (Chart 1, Panel B); the null of homoskedastic errors cannot be rejected for this
specication (Table 2). AndrewsPloberger tests (reported in columns 7 and 8) for the baseline (Regression 1) and the
preferred (Regression 4) specications cannot reject the null of coefcient stability for the individual as well as all
coefcients taken together. Thus, the results are more robust vis-a-vis the headline ination case, where the stability tests
were unable to reject the null in the case of some individual variables.
3.5. NFMP as indicator of demand conditions
Empirical analysis in the preceding sections showed that the impact of demand conditions is stronger on NFMP ination
vis-a-vis headline ination; NFMP ination is more persistent than headline ination; and, the impact of commodity prices
on NFMP is almost the same as headline ination. The rst two ndings support the case of NFMP as a core indicator of
ination, but the third suggests otherwise.
In this context, it would be interesting to examine the impact of supply shocks from commodity ination on ination
dynamics in an advanced economy like the US. Accordingly, an exercise is undertaken to t the Phillips curve for the US
ination the exercise is attempted for both the CPI and the PPI ination in the US and for both headline as well as core for
these two measures of ination (Table 3).6 Core ination is measured as headline excluding food and energy. Estimation

6
The estimation period is the same as for India (i.e., AprilJune 2006 to JanuaryMarch 2011). Five lags of the respective ination variable are included in
each equation to remove residual serial correlation; however, ve lags (and even longer lags) were still unable to remove residual serial correlation in case of
headline PPI ination.

M. Kapur / Journal of Asian Economics 25 (2013) 1727

25

Table 3
Phillips curve estimates for ination in the US.
Dependent variable
US CPI ination

US core CPI ination

US PPI ination

US core PPI ination

Constant

0.81
(3.4)
0.19
(2.4)
0.55
(5.2)
0.01
(3.5)

0.16
(1.3)
0.05
(3.6)
0.90
(17.2)

0.61
(2.5)
0.21
(1.6)
0.51
(5.3)
0.02
(2.6)

0.19
(2.1)
0.06
(1.9)
0.83
(15.9)

YGAPSA_USA{1}
Lagged dependent variable@
OIL
OIL(1)

0.001
(0.9)
0.001
(1.8)

OIL(4)
INFGL

0.04
(2.3)

INFGL(1)
INFGL(4)

0.01
(2.5)
0.01
(1.9)

INFGL(5)
R-bar2
LB-Q(6)

0.003
(1.5)
0.84
0.27

0.89
0.28

0.86
0.00

0.90
0.71

Source: Authors calculations based on data from International Monetary Fund.

Notes: Sample period for the estimation is 1996:22011:1. Figures in parentheses are t-statistics, with standard errors corrected with NeweyWest/Bartlett
window and three lags. LB-Q gives signicance level (p-value) of BoxPierceLjung Q-statistic for the null of no residual autocorrelation for 6 lags.
@: sum of coefcients on the respective lagged dependent variable. Five lags of the lagged dependent variable were included in each equation to remove
residual serial correlation.
YGAPSA_USA = output gap for the US real GDP (based on seasonally adjusted series). Core CPI is headline CPI excluding food and energy. Headline PPI is PPI
for nished goods and core PPI is PPI for nished goods excluding food and energy. INFGL is variation (y-o-y) in global non-fuel commodity price index; OIL
is variation (y-o-y) in international crude oil prices.

results for the US show (i) stronger impact of demand conditions on core ination indicators vis-a-vis headline ination and
(ii) more persistence in core ination; both these results are in accordance with those noted for India. Turning to the impact
of commodity ination, both oil and non-oil commodity ination have a signicant impact on headline CPI as well as
headline PPI. However, in the case of core ination measures, the impact of oil and non-oil ination is muted. In the case of
core CPI ination, non-oil commodity ination is not found to have any statistically signicant impact. While oil prices
contemporaneously impact headline CPI ination, the impact on core CPI is with a lag of four quarters; moreover, the longrun impact is only a third of the impact on the headline ination. Moving to core PPI ination, oil ination is not found to have
any statistical signicant impact. On the other hand, non-oil commodity ination impacts core PPI ination with a lag of a
quarter (the impact on headline PPI was contemporaneous). Moreover, the long-run impact of non-oil commodity ination
on core PPI is lower than that on headline PPI. Overall, the currently used core measure of ination in India scores well on the
rst two of the three parameters of the core measure stronger impact of demand conditions, more persistence and lower
impact of supply shocks whereas the US core measures satisfy all the three.
3.6. Dynamic forecasting performance: in sample and out of sample
How good is the forecasting performance of the estimated Phillips curves in the Indian context? This section assesses both
in sample and out of sample forecasting performance.
As regards in sample dynamic forecasting, the results are based on the respective equations estimated over the full
sample period (i.e., up to the quarter ended March 2011). Various simulation statistics indicate a reasonably good
performance of the estimated equation (Table 4, Panel A). Dynamic forecasts over the full sample period (June 1997
March 2011) indicate that the root mean squared error (RMSE) of the estimated equation to be 66% of the random walk
model in the case of headline ination and 84% of the random walk model in the case of NFMP ination. The dynamic
forecasting evaluation over the full sample period is a relatively stringent test since it takes forecasted values of ination
at each iteration. This generates predictions of equations with the lagged dependent variable generated endogenously
rather than taking the actual values of lagged ination. If we focus on 4 and 8 quarters ahead forecast, the typical policy
focus horizon, the Phillips curve performs much better than a random walk. The Theils U falls to 0.290.41 for headline
ination, i.e., the RMSE of the estimated equations is only 2941% of a random walk model. For NFMP, the Theils U falls to
0.320.39.

M. Kapur / Journal of Asian Economics 25 (2013) 1727

26
Table 4
Forecasting performance of Phillips curve.
Item

Dynamic forecasts over

the period 1997:22011:1

1-Step ahead
forecast

4-Step ahead
forecast

8-Step ahead
forecast

0.03
0.59
0.74
0.44

0.11
0.92
1.12
0.29

0.28
0.83
0.99
0.41

0.02
0.46
0.58
0.44

0.01
0.92
1.10
0.32

0.03
0.99
1.17
0.39

0.02
1.18
1.47
0.57

0.01
2.61
2.99
0.48

0.42
3.24
3.48
1.07

0.04
0.78
0.91
0.48

0.21
1.72
2.09
0.40

1.03
2.01
2.39
0.57

(A) In sample forecasting performance

WPI ination equation (Table 1, Regression 3)
Mean error
0.07
Mean absolute error
0.91
1.11
Root mean square error
Theils U
0.66
Non-food manufactured products ination equation (Table 2, Regression 4)
Mean error
0.03
Mean absolute error
0.91
Root mean square error
1.11
Theils U
0.84
(B) Out of sample forecasting performance
WPI ination equation (Table 1, Regression 3)
Mean error
Mean absolute error
Root mean square error
Theils U
Non-food manufactured products ination equation (Table 2, Regression 4)
Mean error
Mean absolute error
Root mean square error
Theils U

Source: Authors calculations based on data from International Monetary Fund and Reserve Bank of India.
Notes: Out of sample forecasts in Panel B are generated by re-estimating the preferred specications for both headline WPI ination and NFMP ination for
the period up to the quarter ended March 2007 and generating dynamic forecasts up to March 2011.

Turning to out of sample forecasts, the preferred specications of Tables 1 and 2 are re-estimated for the period just prior
to the onset of the global nancial crisis, i.e., for the period up to the quarter ended March 2007 and then dynamic forecasts
are generated for various horizons up to the quarter ended March 2011. The results indicate that the Phillips curve
specications outperform the random walk model for all horizons in the case of NFMP ination and for up to four quarters in
the case of headline WPI ination (Table 4, Panel B). Unlike Atkeson and Ohanian (2001) nding in the US context, the Phillips
curve forecasts are found to outperform a random walk model over the sample period.
The better performance of the estimated Phillips curve vis-a-vis the random walk, however, benets from the fact that
this exercise uses actual realized values of other explanatory variables like output growth, global commodity ination and
movements in the exchange rate. Global commodity ination exhibits a signicant quarter-to-quarter volatility, relatively
difcult to forecast and often the source of actual ination deviating from the forecasted ination on a real time basis. In
addition, food prices exhibit signicant volatility depending upon weather conditions and these add further volatility to
headline ination. Moreover, the weight of food items in the price indices in India is signicantly higher than advanced
economies and even many other emerging market economies. Higher food weights coupled with more volatility in food
ination, therefore, lead to more volatility in headline ination. With half the basket on account of food items, core measures
of ination have limitations. Overall, while the Phillips curve framework provides a useful way to forecast ination, the
volatility in global commodity prices and domestic agricultural shocks makes accurate forecasts a challenging job.
4. Conclusion
In view of long and variable transmission lags, it is important for monetary policy to respond to expected ination and
output dynamics. Reliable forecasts of growth and ination are, therefore, important for effective monetary management.
This paper focused on modeling and forecasting ination in India using an augmented Phillips curve framework. Both
demand and supply factors are seen as drivers of ination. Demand conditions are found to have a stronger impact on nonfood manufactured products (NFMP) ination vis-a-vis headline WPI ination; moreover, NFMP ination is found to be more
persistent than headline ination. Both these ndings support the use of NFMP ination as a core measure of ination. But,
the impact of movements in global prices of non-fuel commodities on NFMP ination is found to be substantial, which
tempers the case for using NFMP ination as a core measure. Ination in non-fuel commodities is seen as a more important
driver of domestic ination rather than fuel ination, although most of the focus is typically on fuel prices.
The exchange rate pass-through coefcient is found to be modest, but nonetheless sharp depreciation in a short period of
time can add to inationary pressures. The estimated equations show a satisfactory in-sample performance based on
dynamic simulations. Nonetheless, forecasting challenges emanate from volatility in international oil and other commodity
prices and domestic food supply dynamics. These supply side factors have exhibited signicant volatility in the recent years
and add complexity to ination dynamics and its forecasting. Finally, structural food ination emanating from protein-rich

M. Kapur / Journal of Asian Economics 25 (2013) 1727

27

items and fruits and vegetables has emerged as a key driver of domestic ination (Gokarn, 2010; Subbarao, 2011). This factor
has not been considered in this paper and it would be useful to incorporate it in the modeling framework for a better
understanding of ination dynamics.
References
Atkeson, A., & Ohanian, L. (2001). Are Phillips curves useful for forecasting ination? Federal Reserve Bank of Minneapolis Quarterly Review25(1) Winter, 211.
Ball, L., & Mazumder, S. (2011). Ination dynamics and the great recession. Working Paper 17044, National Bureau of Economic Research.
Bhalla, S. (2011). Populism, politics and procurement. Financial Express, September 13.
Choudhri, E., & Hakura, D. (2001). Exchange rate pass-through to domestic prices: Does the inationary environment matter? Working Paper 194, International
Monetary Fund.
Dholakia, R. (1990). Extended Phillips curve for the Indian economy. Indian Economic Journal, 38(1), 6978.
Dua, P., & Gaur, U. (2009). Determination of ination in an open economy Phillips curve framework: The case for developed and developing economies. Working Paper
178, Centre for Development Economics, Delhi School of Economics.
Fuhrer, J., Kodryzycki, Y., Little, J., & Olivei, G. (2009). The Phillips curve in historical context. In J. Fuhrer, Y. Kodryzycki, J. Little, & G. Olivei (Eds.), Understanding
ination and the implications for monetary policy: A Phillips curve retrospective, (pp. 368). Cambridge, MA/London: MIT Press.
Fuhrer, J., Olivei, G., & Tootel, G. (2009b). Empirical estimates of changing ination dynamics. Working Paper 09-4, Federal Reserve Bank of Boston.
Gali, J., & Gertler, M. (1999). Ination dynamics: A structural econometric analysis. Journal of Monetary Economics, 44, 195222.
Gali, J., Gertler, M., & Lopez-Salido, D. (2005). Robustness of the estimates of the hybrid New Keynesian Phillips curve. Journal of Monetary Economics, 52(6), 1107
1118.
Gokarn, S. (2010). The price of protein. Reserve Bank of India Bulletin, November, available at http://www.rbi.org.in/scripts/publications.aspx?publication=
Monthly#.
Gordon, R. (1998). Foundations of the goldilocks economy: Supply shocks and the time-varying NAIRU. Brookings Papers on Economic Activity, 29, 297333.
Gordon, R. (2011). The history of the Phillips curve: Consensus and bifurcation. Economica, 78(January), 1050.
Kapur, M., & Patra, M. (2000). The price of low ination. Occasional Papers, 21(23), 191233, Reserve Bank of India.
Khundrakpam, J. (2008). Have economic reforms affected exchange rate pass-through to prices in India. Economic and Political Weekly, 43(16), 7179, April 19.
Kleibergen, F., & Mavroeidis, S. (2009). Weak instrument robust tests in GMM and the New Keynesian Phillips curve. Journal of Business and Economic Statistics,
27(July (3)), 293311.
Mazumder, S. (2011). The stability of the Phillips curve in India: Does the Lucas critique apply? Journal of Asian Economics, 22(6), 528539.
Meyer, B., & Pasaogullari, M. (2010). Simple ways to forecast ination: What works best? Economic Commentary, available at http://www.clevelandfed.org/
research/commentary/2010/2010-17.cfm.
Mohanty, D. (2011). Changing ination dynamics in India. Reserve Bank of India Bulletin, September, available at http://www.rbi.org.in/scripts/publications.
aspx?publication=Monthly#.
Mohanty, M., & Klau, M. (2001). What determines ination in emerging market countries? Modeling aspects of the ination process and monetary transmission
mechanism in emerging market countries.
Patra, M., & Kapur, M. (2012). A monetary policy model for India. Macroeconomics and Finance in Emerging Market Economies, 5(1), 1639.
Patra, M., & Ray, P. (2010). Ination expectations and monetary policy in India: An empirical exploration. Working Paper WP/10/84, International Monetary Fund.
Paul, B. (2009). In search of the Phillips curve for India. Journal of Asian Economics, 20, 479488.
Peach, R., Rich, R., & Cororation, A. (2010). How does slack inuence ination? Current Issues in Economic and Finance17(3) Federal Reserve Bank of New York.
Raj, J., & Misra, S. (2011). Measures of core ination in India an empirical evaluation. Working Paper (DEPR) 16/2011, Reserve Bank of India.
Reserve Bank of India (2002). Report on currency and nance 200001, available at http://www.rbi.org.in/scripts/publications.aspx?publication=Annual.
Reserve Bank of India (2004), Report on currency and nance 200304, available at http://www.rbi.org.in/scripts/publications.aspx?publication=Annual.
Reserve Bank of India (2011). Monetary policy statement 201112, available at http://www.rbi.org.in/scripts/publications.aspx?publication=Annual.
Rudd, J., & Whelan, K. (2007). Modeling ination dynamics: A critical review of recent research. Journal of Money, Credit and Banking, 39(1), 155170.
Singh, B., Kanakaraj, A., & Sridevi, T. (2011). Revisiting the empirical existence of the Phillips curve for India. Journal of Asian Economics, 22, 247258.
Srinivasan, N., Mahambare, V., & Ramachandra, M. (2006). Modeling ination in India: Critique of the structuralist approach. Journal of Quantitative Economics,
New Series, 4(2), 4558.
Stock, J., & Watson, M. (2009). Phillips curve ination forecasts. In J. Fuhrer, Y. Kodryzycki, J. Little, & G. Olivei (Eds.), Understanding ination and the implications for
monetary policy: A Phillips curve retrospective (pp. 99184). Cambridge, MA/London: MIT Press.
Subbarao, D. (2011). Monetary policy dilemmas: Some RBI perspectives. Reserve Bank of India Bulletin, October, available at http://www.rbi.org.in/scripts/
publications.aspx?publication=Monthly#.
Woodford, M. (2003). Interest and prices: Foundations of a theory of monetary policy. Princeton: Princeton University Press.