Determinants of inflation in India

Accepted Manuscript

Title: Determinants of inflation in India

Author: Deepak Mohanty Joice John

PII: S1049-0078(14)00067-0DOI: http://dx.doi.org/doi:10.1016/j.asieco.2014.08.002Reference: ASIECO 957

To appear in: ASIECO

Received date: 12-9-2013Revised date: 27-8-2014Accepted date: 30-8-2014

Please cite this article as: Mohanty, D., and John, J.,Determinants of inflation in India,Journal of Asian Economics (2014), http://dx.doi.org/10.1016/j.asieco.2014.08.002

This is a PDF file of an unedited manuscript that has been accepted for publication.As a service to our customers we are providing this early version of the manuscript.The manuscript will undergo copyediting, typesetting, and review of the resulting proofbefore it is published in its final form. Please note that during the production processerrors may be discovered which could affect the content, and all legal disclaimers thatapply to the journal pertain.

http://dx.doi.org/doi:10.1016/j.asieco.2014.08.002

http://dx.doi.org/10.1016/j.asieco.2014.08.002

of 29

Accep

ted

Man

uscr

ipt

Paper identifies the major determinants of inflation in India using a time varying

SVAR model for the period 1996-2014.

Influence of monetary policy on inflation remained almost steady during the study

period.

Output gap had an asymmetric impact on inflation with its influence having

weakened in the recent period.

Crude oil price was the predominant driver of inflation during 2009-11.

Fiscal deficit was a key determinant of inflation in 2011-12.

of 29

Accep

ted

Man

uscr

ipt

Title of the article: Determinants of inflation in India

Author names and affiliations: Deepak Mohanty Reserve Bank of India

Executive Director, 17th Floor, NCOB, Fort, Mumbai-400001, India, Tel.: +91-22-22633146, E-mail address: [email protected]

Joice John

Reserve Bank of India

Assistant Adviser, Department of Statistics and Information Management, C8/6th Floor, Bandra-Kurla Complex,

Mumbai-400051, India, Tel.: +91-22-26578315, E-mail address: [email protected]

Corresponding author: Joice John

Assistant Adviser, Reserve Bank of India, Department of Statistics and Information Management, C8/6th Floor, Bandra-Kurla

Complex, Mumbai-400051, India, Tel.: +91-22-26578315, E-mail address: [email protected]

http://ees.elsevier.com/asieco/viewRCResults.aspx?pdf=1&docID=1113&rev=2&fileID=18842&msid=2579504F-B852-4258-949B-A636D5240694

of 29

Accep

ted

Man

uscr

ipt

2

1. Introduction

Historically, inflation in India had remained moderate. Average annual inflation rate

measured by the headline Wholesale Price Index (WPI) for the 62- year period from

1950-51 to 2012-13 was around of 6.7 percent. In recent years, prior to the global

financial crisis, from 2000-01 to 2007-08, average annual inflation was even lower at

around 5.2 percent. However, headline inflation rose close to 10 percent during 2010-

11 and 2011-12 before showing some decline in 2012-13. This sudden surge in inflation

and its persistence in the face of significant negative output gap was puzzling.

While stylized facts attribute the rise in inflation to both global and domestic factors, in

addition to supply and demand factors, there is hardly any systematic empirical study on

this aspect. This could perhaps be because of rapid changes in the drivers of inflation

over a short period of time. Against this backdrop, the paper attempts to identify the

determinants of inflation in India using a structural vector auto regression (SVAR)

model. Further, in order to capture the temporal changes in inflation dynamics within

the structural framework, a time varying parameter SVAR model with stochastic volatility

is estimated. The time-varying SVAR model has been extensively used in the recent

literature for examining the changes in macroeconomic dynamics and provides a

flexible approach where the parameters in equations as well as the volatility are

permitted to change over time [1,2].

The paper is organized as follows. A brief narrative on the determinants of inflation

describing the stylized facts in the Indian context is given in section 2. Econometric

framework is presented in section 3. Section 4 presents the empirical results. Section 5

concludes.

2. Determinants of inflation

Canova, Gambetti, and Pappa [3] using a time-varying SVAR found that there are many

similarities in the structural behaviour of inflation and output across a number of

advanced countries: in the US, changes in demand shocks were found to be most

significant; in the euro area, changes in the monetary policy shocks and supply shocks

were the major determinants; and in the UK, demand shocks, supply shocks and the

monetary policy were important. For the emerging market economies (EME), Mohanty

of 29

Accep

ted

Man

uscr

ipt

3

and Klau [4], using the data for the 1980s and 1990s found that supply factors, including

large changes in the external factors and agricultural shocks, drove domestic inflation,

while traditional demand factors even though significant were relatively weak. Unsal and

Osorio [5] studying inflation dynamics in Asia showed that the contribution of monetary

and supply shocks to inflation declined and domestic demand played a major role in

driving inflation in the 2000s.

In this paper, global commodity prices, output gap, fiscal policy and monetary policy are

examined as the key elements in the determination of inflation in India.

2.1 Global commodity prices

With the gradual external liberalisation, the Indian economy has more opened up than

ever before (Fig.1). Hence the global commodity prices and the exchange rate are

playing increasingly important role in the determination of inflation. Currently, near about

80 percent of crude oil demand in India is met by imports. During 2009 to 2011, global

commodity prices had an adverse impact on domestic inflation. In 2012, the

depreciation of the Indian Rupee more than offset the beneficial impact of marginal

decline in global commodity prices on domestic inflation (Fig.2) [6]. Further, some

studies on exchange rate pass through to domestic prices in the Indian context suggest

that 100 basis points (bps) change in the exchange rate had around 10 bps impact on

inflation [7,8,9].

2.2 Output gap

Traditional empirical work on the Phillips curve had tended to focus on the output gap

as a key indicator of inflationary pressures. The evidence on Phillips-curve relationship

in India is mixed. A number of studies covering data up to the 1990s or earlier did not

find the conventional Phillips-curve pattern [10,11,12,13,14,15]. However, Coe and

McDermott [16] found that the output gap is an important determinant of inflation in

almost all the Asian economies including India. Further, Paul [17] found that, in India,

after controlling for certain supply shocks, it is likely to have the short-run tradeoff

between inflation and industrial output.

of 29

Accep

ted

Man

uscr

ipt

4

2.3 Fiscal policy

Theoretical explanations of the fiscal impact of inflation are well postulated in the

macroeconomic theory. As argued by Sargent and Wallace [18], there is a dynamic

relationship between fiscal deficit and inflation. Since borrowing programs usually allow

governments to allocate ‘seigniorage’ over a period of time, the relationship between

fiscal deficit and inflation need not be contemporaneous. This might be a reason why

empirical research has had limited success in uncovering the relationship between fiscal

deficit and inflation [19,20]. However, Fischer, Sahay and Végh [21] using a panel data

of 94 countries, found that the fiscal deficits were one of the main drivers of high

inflation. Catao and Terrones [22] using a data set of 107 countries over the period of

1960–2001 found that the fiscal deficits were inflationary in most of the countries.

Further, this relationship was found to be especially strong for developing economies.

Lin and Chu [23] using a panel data of 91 countries for the period 1960 to 2006, found

that the fiscal deficit had a strong impact on inflation when inflation was high, and had a

weak impact when inflation was low. In a study on India covering the period 1953-2009,

Khundrakpam and Pattanaik [24] found that the fiscal deficit had a significant impact on

inflation. Traditionally the fiscal deficit in India remained relativity high. The fiscal

consolidation process during the period 2003-08 was reversed in 2008-09 mainly on

account of the financial crisis driven fiscal stimulus measures, which is postulated to

have had significant impact on inflation.

2.4 Monetary policy

Recent studies on monetary policy transmission in the Indian economy had found

increasing importance of the role of interest rates. It is also worth mentioning that from

May 2011 Reserve Bank of India has modified the operating procedure of the monetary

policy with a move to a single policy repo rate, with weighted average overnight call

money rate being explicitly recognized as the operating target of the monetary policy.

Anand, Peiris and Saxegaard [25] in a DSGE framework found that the peak impact of a

100 bps increase in the nominal policy rate was around 15 bps on inflation and it was

felt in the first quarter after the policy rate shock. Khundrakpam [26] observed that an

increase of 100 bps in the nominal policy rate was found to reduce bank credit by 2.2-

of 29

Accep

ted

Man

uscr

ipt

5

2.8 percent. Mohanty [27], using a quarterly SVAR model, found that the policy rate

increases had a negative effect on inflation; the peak effect on inflation was with a lag of

three quarters and the overall impact persisted through 8-10 quarters. Kapur and

Behera [9] found that 100 bps increase in the nominal policy rate lowered non-food

manufactured products’ inflation by 25 bps with a lag of 5 quarters.

3. Analytical framework

In order to assess the determinants of inflation a 5 variable SVAR is formulated, which

apart from inflation includes domestic crude oil prices, output gap, fiscal deficit and the

overnight weighted average call money rate. Inflation (πt) is measured as the

annualized growth rate of the de-seasonalised1 Wholesale Price Index (WPI). The WPI

based inflation was chosen for the study because during the study period (Q1: 1996-97

to Q3: 2013-14) the central bank was primarily focusing on WPI based inflation as the

main inflation measure, in absence of a nation-wide representative Consumer Price

Index (CPI). However, in this paper, for checking the robustness of the results, we also

used an alternate specification with inflation based on the annualized change in the de-

seasonalised1 Gross Domestic Product (GDP) deflator. Global commodity prices (gt) are

measured using the average crude oil price in Indian Rupees, which broadly represent

both international price and exchange rate pass-through to domestic inflation. This data

series is represented as the seasonally adjusted1 annualized growth rate. The output

gap (dt) is estimated by applying Hodrick-Prescott filter on de-seasonalised1 real Gross

Domestic Product (GDP). The ratio of seasonally adjusted1 gross fiscal deficit of the

central government to seasonally adjusted1 nominal GDP (at market prices) is used to

represent the fiscal policy (ft). The monetary policy (it) is represented using weighted

average call money rate, which is the operating target for monetary policy.

The structural identification restrictions for the SVAR estimation are as follows:

i) Crude oil price is considered to be the most exogenous to the framework.

ii) Inflation responds immediately to changes in crude oil price and to other

factors with a lag.

1Seasonal adjustment is done using X-12 ARIMA

of 29

Accep

ted

Man

uscr

ipt

6

iii) Output gap representing demand conditions is assumed to be sensitive to

international crude oil prices and domestic price movements.

iv) Fiscal deficit is sensitive to demand conditions and inflation.

v) Monetary policy is formulated considering the demand factors and price

situation besides being influenced by global factors and fiscal policy.

Let Yt denote a n ×1 vector gt, πt, dt, ft, it of 5 variables at time t. The structural

identifications described above can be incorporated in to a VAR framework by putting

the economic restrictions to draw inference about the structural relations. This can be

represented as:

AYt = F1Yt-1 +…+ FsYt-s + ut (1)

Where, ut is a (p=5) × 1 vector of structural shocks following N (0,Σ) where Σ = diag( σj).

For identifying this structural model, p(p + 1)/2 restrictions has to be imposed on A. Of

which p restrictions could be satisfied by normalizing the diagonal elements in A to

unity. The estimation in our case is with p=5, hence we have to impose 10 restrictions

on the contemporaneous correlations for identification of the five structural shocks. That

is straight away built into the system by the set of structural identification restrictions.

Thus A is given by a lower triangular matrix providing the 10 necessary restrictions for

identification.

Thus equation (1) could be rewritten as a reduced form VAR model.

Yt = B1Yt-1 +…+ BsYt-s + A-1Σεt (2)

Where εt follows N(0,I) and Bi = A-1Fi

Stacking the elements in the rows in Bi to form β and defining Xt = I ⊗(Yt-1'… Yt-s'), (2)

could be written as:

Yt = Xtβ + A-1Σεt (3)

The time varying dynamics of inflation determinants could be better understood if the

parameters of the model postulated in (3) are allowed to change over time. This is done

of 29

Accep

ted

Man

uscr

ipt

7

by analysing the structure using a Time-Varying Parameter Structural Vector Auto

Regression (TVP-VAR) framework, which is discussed in following section.

3.1 Time-varying parameter SVAR with stochastic volatility

In late 1990s TVP-VAR analysis became popular. Cogley and Sargent [28] used 3

variables with time-varying coefficients in a VAR model to study the persistence of

inflation in the US. Subsequently, Cogley and Sargent [29] introduced stochastic

volatility into the VAR model with time-varying coefficients but with a non-varying

structural shock. Primiceri [30] also used a three variable time varying VAR model,

which allowed all parameters to vary over time for the US economy. Cogley, Primiceri

and Sargent [1] used a time-varying VAR to identify persistence in inflation in the US

and defined a new measure of persistence. Benati and Surico [31] have also used

similar kind of approach to define the inflation persistence again for the US. Canova,

Gambetti and Pappa [3] used the time-varying SVAR to examine the dynamics of output

growth and inflation in the US, euro area and the UK. TVP-VAR was used for analyzing

time varying properties of the UK data by Benati [32]. Baumeister, Durinck and

Peersman [33] studied the effects of excess liquidity shocks on macroeconomic

variables in the euro area using a TVP-VAR model. TVP-VAR model for the Japanese

macroeconomic data was used by Nakajima, Kasuya and Watanabe [34] and Nakajima

[35]. Mumtaz and Plassmann [2] used TVP-VAR to study the time varying properties of

the real exchange rate for the UK, euro area and Canada.

All parameters in equation (3) are time-invariant. Following the methodology described

in Nakajima [36], this can be extended to TVP-VAR model by allowing the parameters

to change over time.

Yt = Xtβt + At-1Σtεt, t = s+1….n. (4)

Let at = (a21, a31, a32, a41, . . . , a54)' be a stacked vector of the lower-triangular elements

in At ; ht = (h1t, . . . , hkt)' with hjt = logσ2jt for j = 1, . . . , k, t = s+1, . . . , n and βt be the

vector of coefficients. Then following Primiceri [30] and Nakajima [36], it is assumed that

parameters in (4) follow a random walk process.

βt+1= βt +uβt

of 29

Accep

ted

Man

uscr

ipt

8

at+1= at +uat (5)

ht+1= ht +uht

with [εt, uβt, uat, uht]’ follows N(0, diag(I Σβ, Σa Σh)) for t = s+1, . .n.

Where , Σβ, Σa and Σh are the variance- covariance structure for the innovations of the

time-varying parameters. Following Cogley and Sargent [29] and Nakajima [36] Σβ, Σa,

and Σh are assumed to be diagonal. Further, following the related literature [1,30,36], a

tighter prior is set for Σβ and diffuse priors for Σa and Σh. The hyper parameters of Σβ are

simulated from an inverse Wishart distribution while Σa and Σh are drawn from an

inverse gamma distribution. Let Yt = ytt-1n; ω = (Σβ, Σa, Σh) and the prior density of ω be

π(ω). Given the data Y, the samples from the posterior distribution, π(β, a, h, ω|y) are

drawn following a Monte Carlo Marco Chain (MCMC) algorithm using the Matlab codes

developed by Nakajima [36]. The details of the estimation methodology are as available

in Nakajima [36].

4. Results

The 5 variable SVAR with the structural restrictions as imposed by the lower triangular

matrix A is first estimated in a time-invariant framework to understand the average

nature of dynamic relationship between the variables in the study period. In the

subsequent section the results of the TVP-VAR in the same framework are discussed.

Both estimations are done using the quarterly data from Q1:1996-97 to Q3:2013-14.

The choice of study period is on account of the availability of quarterly data on GDP and

to incorporate the post-liberalisation phase of the Indian economy, which assigned

greater role to market forces in determination of price, interest rate and exchange rate.

4.1 Empirical results from a time-invariant SVAR

The stationary properties of all the variables are tested and are found to be satisfactory

(Table 1). The lag length for the estimated SVAR is chosen to be 4 by the Schwarz

Bayesian Information Criterion (BIC). The regression diagnostics of no autocorrelation,

and homoscedasticity of residuals are found to be satisfactory. The impulse response

function (IRF) and forecast error variance decompositions (FEVD) are used to draw

conclusions on the determinants of inflation. The structural impulse response functions

of 29

Accep

ted

Man

uscr

ipt

9

with respect to one percentage point shock of different variables on inflation estimated

from the SVAR are presented in Fig.3. The 2-quarter, 4-quarter and 8-quarter ahead

structural FEVD are given in Table 2.

The impact of global factors represented by the domestic price of crude oil on inflation is

immediate and significant with the peak impact being in the first quarter. One

percentage point increase in the Indian basket crude oil price (in Rupees) leads to an

increase of 7 bps in domestic inflation accumulated over a period of 2 quarters. The

output gap is found to be insignificant at 95 per cent level of significance. The fiscal

deficit seems to impact WPI inflation with a longer lag of 5 quarters, with the peak

impact being estimated in the 5th quarter ahead. One percentage point increase in gross

fiscal deficit to GDP ratio is estimated to increase inflation by 106 bps over a period of 8

quarters. The peak impact of the policy rate on inflation is in the 4th quarter ahead. One

percentage point increase in the call money rate leads to 94 bps reduction in inflation

accumulated over a period of 8 quarters (Fig.3).

The Structural FEVD in Table 2 suggests that more than half the variation in inflation is

explained by its own shocks. Almost 22 percent of variation in WPI inflation 2 quarters

ahead is explained by the global factors: international crude oil price and exchange rate.

On an average around 3 percent of variation in inflation is explained by the output gap.

The fiscal deficit explains around 13 percent and the policy rate explains around 10

percent of variation in inflation in 8 quarters ahead.

4.2 Results with alternative measures of inflation

For checking the robustness of the above results we used an alternate specification with

inflation based on the GDP deflator. We did not use Consumer Price Index (CPI) based

measure of inflation because of the non-availability of a representative nation-wide CPI

series for India for the study period. The empirical results are more or less similar even

though the impact of various factors was found to be somewhat lesser in the deflator

based measure of inflation.

One percentage point increase in the Indian basket crude oil price (in Rupees) leads to

an increase of 4 bps in domestic inflation accumulated over a period of 2 quarters. The

output gap is not found to be significant at 95 per cent level of significance. The fiscal

of 29

Accep

ted

Man

uscr

ipt

10

deficit seems to significantly impact inflation in the 4th and 5th quarters. One percentage

point increase in gross fiscal deficit to GDP ratio is estimated to increase inflation by 76

bps over a period of 8 quarters. One percentage point increase in the call money rate

leads to 71 basis points reduction in inflation accumulated over a period of 8 quarters

with a significant impact in the 4th quarter (Fig.4).

As in the case with WPI inflation, the Structural FEVD suggests that over half the

variation in deflator-based inflation is explained by its own shocks. The output gap,

fiscal deficit and policy rate explain 4.8 per cent, 11.0 percent and 13.1 per cent of

variation respectively in the deflator based measure of inflation 8 quarters ahead (Table

3).

4.3 Empirical results from a time varying parameter SVAR with stochastic volatility

The TVP-VAR model with stochastic volatility described in equations (4) and (5) is used

to elucidate the time varying dynamics of various factors that explain inflation over the

study period. We followed the estimation procedure as available in Nakajima [36]. A

most commonly used method for setting the priors is laid down by Primiceri [30], where

priors are chosen based on the estimates of a time-invariant VAR model computed

using the pre-sample period. Even though it is reasonable to use this approach, due to

lack of sufficient data points in our sample we restricted the analysis by selecting a

reasonably flat prior for the initial state as we have not much information on the initial

state a priori [36]. The hyper parameters of Σβ, Σa and Σh are set as

Σβ~ Inverse Wishart (25, 0.01); (Σa)-2 ~ Gamma(4,0.02) and (Σh)

-2 ~ Gamma(4,0.02).

For the initial state of the parameters, flat priors are set with µβ0 = µa0 =µh0 = 0, and Σβ0

= Σa0 = Σh0 = 10 × I.

To compute2 the posterior estimates, 10,000 samples are drawn. However, the initial

1,000 samples are removed while calculating the posterior estimates. The sample paths

look stable and after the initial draws the sample autocorrelations are low. This indicates

that the sampling method produced the samples with low autocorrelation. The estimates

for the convergence diagnostics (CD) of Geweke [37], which are computed from the

2We used Matlab code developed by Nakajima [36] Available at (http://sites.google.com/site/jnakajimaweb/tvpvar.)

of 29

Accep

ted

Man

uscr

ipt

11

MCMC sample, shows that the convergence to the posterior distribution is not rejected

for the parameters. The inefficiency factors were found to be low indicating an efficient

sampling for the parameters.

Fig.5 provides the stochastic volatility of the five variables. Stochastic volatility of oil

prices showed a sharp increase in and around 2008-09 following the global financial

crisis. The volatility in Indian inflation exhibited a general downward trend after 2000.

The output gap showed some downward trend in volatility especially after 2003.

Volatility in call money rate remained at the same level, though with some temporal ups

and downs.

The time varying macroeconomic dynamics that determined Indian inflation during

1996-96 to 2012-13 are captured using the time varying IRFs and FEVDs of various

shocks on inflation in the TVP-VAR framework given in equations (4) and (5). The

accumulated impacts over 6 quarters ahead are presented in Fig. 6.

The impulse response functions reveal that the impact of global price shocks,

measured in terms of oil prices on domestic inflation was relatively high during 2009-10

and 2010-11. The accumulated impact of one percentage point positive shock in global

crude prices on inflation was around 12 bps during those periods. This came down to

about 5 bps in 2012-14. The output gap remained a notable determinant of inflation in

2007-08 and 2009-10. In 2007-08, the accumulated impact of one percentage point

positive shock to the output gap resulted in an increase of around 65 bps in inflation.

The impact got reduced in 2008-09, but went up to almost the earlier level in 2009-10.

Subsequently, this impact witnessed a general downward trend, with some cyclical ups

and downs. The fiscal shocks were relatively less adverse to inflation prior to 2005-06.

This, however, showed some upward movement in the six years from 2005-06 to 2012-

13 with some cyclical ups and downs. There was a considerable increase in the

accumulated impact of one percentage point shock to fiscal factors on inflation during

2011-12. The impact of the policy shocks on inflation remained almost unchanged

during the period 2005-06 to 2013-14 barring a few quarters in 2010-11. In the recent

period, one percentage point positive shock in call money rate accumulates to around

120 bps reduction in inflation in 6 quarters.

of 29

Accep

ted

Man

uscr

ipt

12

In order to examine which of the identified shocks are important to explain the inflation

fluctuations we decompose the forecast error variance of inflation into contributions from

the five identified shocks (Fig. 7). The time-varying nature of the SVAR model implies

that this decomposition can be calculated at each point in time.

Global oil price shock was the dominant driver of inflation during 2009-10 and most part

of 2010-11 with contribution above 20 percent. This declined to below 20 percent during

2011-14. During 2006-08, the output gap was contributing to inflation dynamics in the

range of 15 to 25 percent, which reduced during the financial crisis period. Its impact

again rose briefly in the first half of 2009-10 but declined to under 10 percent

subsequently. The fiscal factors were contributing under 10 percent almost throughout

the study period. However, in 2011-12 their contribution rose to about 10 percent, which

reduced subsequently during 2012-14. The monetary policy represented by call money

rate was found to contribute appreciably to inflation only after 2005-06. Since then, its

contribution was hovering in and around 10 to 20 percent barring a few quarters during

2009-11.

It is a challenge to develop a consistent empirical explanation of the recent inflation

dynamics in India, not surprisingly though as the economy is undergoing rapid structural

changes and experienced supply and demand shocks in the post-crisis period. We

estimated a time varying SVAR model for explaining the inflation in India, which better

captures the time varying properties of the inflation process through its proximate

determinants, than a standard SVAR model with the same determinants. From the

supply side, crude oil prices and exchange rate were found to be playing an important

role in defining domestic prices as the bulk of the country’s petroleum requirements are

met by imports. From the demand side, even though the output gap was found to be

insignificant on an average, it had an asymmetric impact on inflation: largely

contributing to rise in inflation when the output gap in high and positive but having only

trivial contribution to lower inflation when the output gap is negative. This could be

because in a supply constrained economy, even as firms operate below capacity they

still protect their pricing power to some extent. The fiscal deficit was found to have

contributed significantly to inflation during the post-crisis period. While both global and

domestic factors have played a role in the recent inflation process in India, the role of

of 29

Accep

ted

Man

uscr

ipt

13

monetary and fiscal policy becomes important in the containment of inflation given their

significant impact.

5. Conclusion

The paper tries to examine the factors that may have contributed to inflation in India in

the one and a half decades from 1996-97 to 2013-14 using an econometric framework

by incorporating the standard determinants of inflation viz.: i) crude oil prices, ii) output

gap, iii) fiscal policy iv) monetary policy and v) intrinsic inflation persistence. Further it is

also attempted to analyze the temporal changes in inflation dynamics in India within a

structural framework using a time varying parameter SVAR model. The paper

documents that inflation process in India in recent years has been complex with the

drivers changing frequently. The global price shocks, measured by domestic price of

crude oil that was predominant during 2009-11 have moderated in 2012-13. The output

gap has an asymmetric effect on inflation with its impact having weakened since 2012-

13, as the economy registered negative output gap. The fiscal deficit emerged as one

of the key determinants of inflation in 2011-12. The monetary policy impact on inflation,

however, has remained broadly unchanged. The paper underscores the role of

monetary and fiscal policy in the containment of inflation irrespective of the nature of the

shock to the inflation process.

Acknowledgement

The views expressed in the paper are those of the authors and do not represent those

of the institution to which authors belong. The authors would like to acknowledge the

useful comments received from two anonymous referees. However remaining errors

and omissions, if any, are those of the authors.

References

[1] T. Cogley, G.E. Primiceri, T.J. Sargent, Inflation-Gap Persistence in the US, Am.

Econ. J. Macroecon. 2(1) (2010) 43-69.

of 29

Accep

ted

Man

uscr

ipt

14

[2] H. Mumtaz, L.Sunder‐Plassmann, Time‐Varying Dynamics of the Real Exchange

Rate: An Empirical Analysis, J. Appl. Econ. (2012) doi: 10.1002/jae.2270.

[3] F. Canova, L. Gambetti, E. Pappa, The Structural Dynamics of Output Growth and

Inflation: Some International Evidence, The Econ. J. 117(519) (2007) C167-C191.

[4] M.S Mohanty, M. Klau, What determines inflation in emerging market economies?,

Bank of International Settlements Papers 8 (2001), Available

athttp://www.bis.org/publ/bppdf/bispap08a.pdf, Accessed on February 08, 2013.

[5] D.F Unsal, C. Osorio, Inflation Dynamics in Asia: Causes Changes and Spillovers

from China, J. Asian. Eco. 24 (2013) 26–40.

[6] Reserve Bank of India, Macroeconomic and Monetary Developments First Quarter Review 2013-14 (2013), Available athttp://rbidocs.rbi.org.in/rdocs/Publications/PDFs/M290713FC66587B599.pdf ,Accessed on July 30, 2013.

[7] R. Bhattacharya, I. Patnaik, A. Shah, Exchange rate pass-through in India, National

Institute of Public Finance and Policy: New Delhi(2008), Available at

http://macrofinance.nipfp.org.in/PDF/BPS2008_erpt.pdf, Accessed on February 08,

2013.

[8] J.K. Khundrakpam, Have Economic Reforms Affected Exchange Rate Pass-Through

to Prices in India?, Econ. Polit. Wkly. (2008) 71-79.

[9] M. Kapur, H. Behera, Monetary Transmission Mechanism in India: A Quarterly

Model, Reserve Bank of India Working Paper 09 (2012), Available at

http://rbidocs.rbi.org.in/rdocs/Publications/PDFs/WP09250612FL.pdf, Accessed on

February 08, 2013.

[10] S.S Bhalla, India's closed economy and world inflation, in: W.R. Cline and

Associates (Eds.), World Inflation and the Developing Countries, Brookings Institution,

Washington DC, 1981, pp.137.

[11] R.Chatterji, The behavior of industrial prices in India, Oxford University Press,

Delhi, 1989.

of 29

Accep

ted

Man

uscr

ipt

15

[12] S.K. Samanta, Price surprises and real output: The Indian evidence, Indian. Econ.

J. 34(2) (1986) 49–58.

[13] B.B. Bhattacharya, M. Lodh, Inflation in India: An analytical survey, ArthaVijnana

32(March) (1990) 25–68.

[14] R.H. Dholakia, Extended Phillips curve for the Indian economy, Indian.Econ.

J.38(1) (1990) 69-78.

[15] V. Virmani, Estimating Output Gap for the Indian Economy: Comparing Results

from Unobserved-Components Models and the Hodrick-Prescott Filter, Indian Institute

of Management Ahmadabad Research and Publication Department. No. WP2004-04-02

(2004), Available at http://www.iimahd.ernet.in/publications/data/2004-04-02vineet.pdf,

Accessed on February 08, 2013.

[16] D.T. Coe, C.J. McDermott, Does the gap model work in Asia?, Social Science

Research Network, Available at

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=882967, Accessed on February 08,

2013.

[17] B.P. Paul, In search of the Phillips curve for India, J. Asian. Econ. 20(4) (2009) 479-

488.

[18] T.J. Sargent, N. Wallace, Some unpleasant monetarist arithmetic, Fed. Reserve.

Bank. Minneapolis. Q. Rev. 5(3) (1981) 1-17.

[19] R.G. King, C.I. Plosser, Money, deficits, and inflation, in: Carnegie-Rochester

Conference Series on Public Policy, Vol. 22, No. 1, Elsevier, North-Holland, 1985, pp.

147-195.

[20] R.W. Click, Seigniorage in a cross-section of countries, J. Money. Credit. Bank.

30(2) (1998) 154-171.

[21] S. Fischer, R. Sahay, C.A. Végh, Modern hyper-and high inflations, J. Econ. Lit. 40

(2002) 837-880.

[22] L. A. Catao, M.E. Terrones, Fiscal deficits and inflation, J. Monet. Econ. 52(3)

(2005) 529-554.

of 29

Accep

ted

Man

uscr

ipt

16

[23] H.Y. Lin, H.P. Chu, Are fiscal deficits inflationary?, J. Int. Money. Financ. 32 (2013)

214–233.

[24] J.K Khundrakpam, S. Pattanaik, Fiscal Stimulus and Potential Inflationary Risks: An

Empirical Assessment of Fiscal Deficit and Inflation Relationship in India, J. Econ.

Integr. 25(4) (2010) 703-721.

[25] R. Anand, S.J. Peiris, M. Saxegaard, An estimated model with macro-financial

linkages for India, International Monetary Fund Working Paper WP/10/21 (2010),

Available at http://www.imf.org/external/pubs/ft/wp/2010/wp1021.pdf, Accessed on

February 08, 2013.

[26] J.K. Khundrakpam, Credit Channel of Monetary Transmission in India–How

Effective and Long is the Lag?, Reserve Bank of India Working Paper 20 (2011),

Available at http://rbidocs.rbi.org.in/rdocs/Publications/PDFs/CCMTIR51211.pdf,

Accessed on February 08, 2013.

[27] D. Mohanty, Evidence on Interest Rate Channel of Monetary Policy Transmission in

India. Reserve Bank of India Working Paper 06 (2012), Available at

http://rbidocs.rbi.org.in/rdocs/Publications/PDFs/WPS6180512FL.pdf, Accessed on

February 08, 2013.

[28] T. Cogley, T.J. Sargent, Evolving post-world war II US inflation dynamics, in: B.S.

Bernanke and K. Rogoff(Eds.), National Bureau of Economic Research

Macroeconomics Annual 2001 Volume 16, MIT press, 2002, pp. 331-388.

[29] T. Cogley, T.J. Sargent, Drifts and volatilities: monetary policies and outcomes in

the post WWII US, Rev. Econ. Dyn. 8(2) (2005) 262-302.

[30] G.E. Primiceri, Time varying structural vector auto-regressions and monetary

policy, Rev. Econ. Stud. 72(3) (2005) 821-852.

[31] L. Benati, P.Surico, Evolving US monetary policy and the decline of inflation

predictability, J. Eur. Econ. Assoc. 6(2‐3) (2008) 634-646.

[32] L. Benati, The “great moderation” in the United Kingdom, J. Money.

Credit.Bank.40(1) (2008) 121-147.

of 29

Accep

ted

Man

uscr

ipt

17

[33] C. Baumeister, E. Durinck, G. Peersman, (2008).Liquidity, inflation and asset

prices in a time-varying framework for the euro area, Social Science Research Network,

Available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1290842, Accessed on

February 08, 2013.

[34] J. Nakajima, M. Kasuya, T. Watanabe, Bayesian analysis of time-varying

parameter vector autoregressive model for the Japanese economy and monetary policy,

J. Jpn. Int. Econ. 25(3) (2011) 225-245.

[35] J. Nakajima, S. Shiratsuka, Y. Teranishi, The effects of monetary policy

commitment: Evidence from time-varying parameter VAR analysis, Bank of Japan

Institute for Monetary and Economic Studies Working Paper 10-E-06 (2010), Available

at http://www.imes.boj.or.jp/english/publication/edps/2010/10-E-06.pdf, Accessed on

February 08, 2013.

[36] J. Nakajima, Time-varying parameter VAR model with stochastic volatility: An

overview of methodology and empirical applications, Bank of Japan Institute for

Monetary and Economic Studies Working Paper 11-E-09 (2011), Available at

http://www.imes.boj.or.jp/research/papers/english/me29-6.pdf, Accessed on February

08, 2013.

[37] J. Geweke, Evaluating the accuracy of sampling-based approaches to the

calculation of posterior moments, in: J. M. Bernardo, J. O. Berger, A. P. Dawid, and A.

F. M. Smith (Eds.), Bayesian Statistics Volume 4, Oxford University Press, New York,

1992, pp. 169–188.

of 29

Accep

ted

Man

uscr

ipt

18

Figure Captions

Fig. 1: Total trade to GDP ratio

Fig. 2: Annual percent change in crude prices and exchange rate (India Rupee per US

Dollar)

Fig. 3: Response of inflation (WPI) to one percentage point shock in different variables

Fig. 4: Response of inflation (GDP deflator) to one percentage point shock in different

variables

Fig. 5: Stochastic volatility in different variables

Fig.6: Time-varying accumulated impulse response function – Response of inflation to

one percentage point shock in other variables

Fig. 7: Time varying forecast error variance decomposition (over 6 quarters) of inflation

of 29

Accep

ted

Man

uscr

ipt

Based on authors’ calculations using data on ‘India’s foreign trade’ and ‘GDP at current market prices’ (Source: Database on Indian

Economy, Reserve Bank of India)

0

10

20

30

40

50

1970-7

1

1972-7

3

1974-7

5

1976-7

7

1978-7

9

1980-8

1

1982-8

3

1984-8

5

1986-8

7

1988-8

9

1990-9

1

1992-9

3

1994-9

5

1996-9

7

1998-9

9

2000-0

1

2002-0

3

2004-0

5

2006-0

7

2008-0

9

2010-1

1

2012-1

3

%

Total trade to GDP ratio

Fig. 1: Total trade to GDP ratio

of 29

Accep

ted

Man

uscr

ipt

Based on authors’ calculations using data on ‘Average forex rates’ (Source: Database on Indian Economy, Reserve Bank of India)

and ‘IMF Primary Commodity Prices’ (Source: International Monetary Fund)

-80

-60

-40

-20

0

20

40

60

80

Mar-

09

Jun-0

9

Sep-0

9

Dec-0

9

Mar-

10

Jun-1

0

Sep-1

0

Dec-1

0

Mar-

11

Jun-1

1

Sep-1

1

Dec-1

1

Mar-

12

Jun-1

2

Sep-1

2

Dec-1

2

Mar-

13

Jun-1

3

Sep-1

3

Dec-1

3

%

Crude oil price Exchange rate (Rs./$)

Fig. 2: Annual percent change in crude prices and exchange rate

of 29

Accep

ted

Man

uscr

ipt

(a) Crude oil price shock

(b) Output gap shock

(c) Fiscal shock

(d) Monetary policy shock

Based on authors’ calculations using data on ‘Wholesale Price Index’, ‘GDP at constant factor cost’, ‘Fiscal deficit of central

government’ , Weighted average call money rate’ and ‘Average forex rates’ (Source: Database on Indian Economy, Reserve Bank

of India) and ‘IMF Primary Commodity Prices’ (Source: International Monetary Fund).

-0.06

-0.04

-0.02

0

0.02

0.04

0.06

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

95 % CI

-1.5

-1

-0.5

0

0.5

1

1.5

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

95 % CI

-1

-0.5

0

0.5

1

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

95 % CI

-1.5

-1

-0.5

0

0.5

1

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

95 % CI

Fig. 3: Response of inflation (WPI) to 1 percentage point shock

of 29

Accep

ted

Man

uscr

ipt



(c) Fiscal shock


Based on authors’ calculations using data on ‘GDP at current factor cost’ ,‘GDP at constant factor cost’, ‘Fiscal deficit of central



-0.05

-0.03

-0.01

0.01

0.03

0.05

0.07

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

95% CI

-1.5

-1

-0.5

0

0.5

1

1.5

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

95% CI

-0.8

-0.4

0

0.4

0.8

1.2

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

95% CI

-1.5

-1

-0.5

0

0.5

1

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

95% CI

Fig.4: Response of inflation (GDP) to 1 percentage point shock

of 29

Accep

ted

Man

uscr

ipt

Crude oil price

Inflation

Output gap

Fiscal deficit

Call money rate

Based on authors’ calculations using data on ‘Wholesale Price Index’, ‘GDP at constant factor cost’, ‘Fiscal deficit of central government’ ,

Weighted average call money rate’ and ‘Average forex rates’ (Source: Database on Indian Economy, Reserve Bank of India) and ‘IMF

Primary Commodity Prices’ (Source: International Monetary Fund).

0.0

0.1

0.2

0.3

0.4

0.5

Q1:9

7-9

8

Q3:9

8-9

9

Q1:0

0-0

1

Q3:0

1-0

2

Q1:0

3-0

4

Q3:0

4-0

5

Q1:0

6-0

7

Q3:0

7-0

8

Q1:0

9-1

0

Q3:1

0-1

1

Q1:1

2-1

3

Q3:1

3-1

4 2.E-04

2.E-04

3.E-04

3.E-04

Q1:9

7-9

8

Q3:9

8-9

9

Q1:0

0-0

1

Q3:0

1-0

2

Q1:0

3-0

4

Q3:0

4-0

5

Q1:0

6-0

7

Q3:0

7-0

8

Q1:0

9-1

0

Q3:1

0-1

1

Q1:1

2-1

3

Q3:1

3-1

4 4.E-06

4.E-06

5.E-06

5.E-06

Q1:9

7-9

8

Q3:9

8-9

9

Q1:0

0-0

1

Q3:0

1-0

2

Q1:0

3-0

4

Q3:0

4-0

5

Q1:0

6-0

7

Q3:0

7-0

8

Q1:0

9-1

0

Q3:1

0-1

1

Q1:1

2-1

3

Q3:1

3-1

4

9.0E-06

9.5E-06

1.0E-05

Q1:9

7-9

8

Q3:9

8-9

9

Q1:0

0-0

1

Q3:0

1-0

2

Q1:0

3-0

4

Q3:0

4-0

5

Q1:0

6-0

7

Q3:0

7-0

8

Q1:0

9-1

0

Q3:1

0-1

1

Q1:1

2-1

3

Q3:1

3-1

4

1.9E-07

2.0E-07

2.0E-07

2.1E-07

2.1E-07

Q1:9

7-9

8

Q3:9

8-9

9

Q1:0

0-0

1

Q3:0

1-0

2

Q1:0

3-0

4

Q3:0

4-0

5

Q1:0

6-0

7

Q3:0

7-0

8

Q1:0

9-1

0

Q3:1

0-1

1

Q1:1

2-1

3

Q3:1

3-1

4

Fig. 5: Stochastic volatility in different variables

of 29

Accep

ted

Man

uscr

ipt



(c) Fiscal shock





-0.10

0.00

0.10

0.20

0.30

Q1:0

6-0

7

Q4:0

6-0

7

Q3:0

7-0

8

Q2:0

8-0

9

Q1:0

9-1

0

Q4:0

9-1

0

Q3:1

0-1

1

Q2:1

1-1

2

Q1:1

2-1

3

Q4:1

2-1

3

Q3:1

3-1

4

%

0.0

0.2

0.4

0.6

0.8

1.0

Q1:0

6-0

7

Q4:0

6-0

7

Q3:0

7-0

8

Q2:0

8-0

9

Q1:0

9-1

0

Q4:0

9-1

0

Q3:1

0-1

1

Q2:1

1-1

2

Q1:1

2-1

3

Q4:1

2-1

3

Q3:1

3-1

4

%

0.0

0.2

0.4

0.6

0.8

1.0

Q1:0

6-0

7

Q4:0

6-0

7

Q3:0

7-0

8

Q2:0

8-0

9

Q1:0

9-1

0

Q4:0

9-1

0

Q3:1

0-1

1

Q2:1

1-1

2

Q1:1

2-1

3

Q4:1

2-1

3

Q3:1

3-1

4

%

-1.6

-1.2

-0.8

-0.4

0.0

Q1:0

6-0

7

Q4:0

6-0

7

Q3:0

7-0

8

Q2:0

8-0

9

Q1:0

9-1

0

Q4:0

9-1

0

Q3:1

0-1

1

Q2:1

1-1

2

Q1:1

2-1

3

Q4:1

2-1

3

Q3:1

3-1

4

%

Fig.6: Time-varying accumulated impulse response function

of 29

Accep

ted

Man

uscr

ipt




0.0

20.0

40.0

60.0

80.0

100.0

Q1:0

6-0

7

Q2:0

6-0

7

Q3:0

6-0

7

Q4:0

6-0

7

Q1:0

7-0

8

Q2:0

7-0

8

Q3:0

7-0

8

Q4:0

7-0

8

Q1:0

8-0

9

Q2:0

8-0

9

Q3:0

8-0

9

Q4:0

8-0

9

Q1:0

9-1

0

Q2:0

9-1

0

Q3:0

9-1

0

Q4:0

9-1

0

Q1:1

0-1

1

Q2:1

0-1

1

Q3:1

0-1

1

Q4:1

0-1

1

Q1:1

1-1

2

Q2:1

1-1

2

Q3:1

1-1

2

Q4:1

1-1

2

Q1:1

2-1

3

Q2:1

2-1

3

Q3:1

2-1

3

Q4:1

2-1

3

Q1:1

3-1

4

Q2:1

3-1

4

Q3:1

3-1

4

%

Inflation Crude oil price Output gap Fiscal deficit Call money rate

Fig. 7: Time varying forecast error variance decomposition

of 29

Accep

ted

Man

uscr

ipt

Table 1: Phillips-Perron test for unit root

Variables Test statistic p-value*

Inflation^ -WPI -5.468 0.000 Inflation^ - GDP Deflector -5.872 0.000 Crude oil price change** -5.849 0.000 Output gap -2.724 0.0699 Fiscal deficit -6.806 0.000 Policy rate -4.112 0.001

*MacKinnon approximate p-value

^ Measured as annualized quarterly percent change in seasonally adjusted data.

** Measured as annualized quarterly percent change in seasonally adjusted domestic prices.

For all variables expect output gap, unit root rejected at 1% level of significance. In case of output gap unit root is

rejected at 7% i.e. Stationary at 7% level of significance .

Based on authors’ calculations using data on ‘Wholesale Price Index’, ‘GDP at current factor cost’, ‘GDP at constant

factor cost’, ‘Fiscal deficit of central government’ , Weighted average call money rate’ and ‘Average forex rates’

(Source: Database on Indian Economy, Reserve Bank of India) and ‘IMF Primary Commodity Prices’ (Source:

International Monetary Fund).

Table 1: Phillips-Perron test for unit root

http://ees.elsevier.com/asieco/download.aspx?id=18837&guid=4737772d-65c9-4a25-8a57-64602f8458b4&scheme=1

of 29

Accep

ted

Man

uscr

ipt

Table 2: Forecast error variance decomposition – WPI inflation

Variables Variation in inflation explained- 2 quarters ahead (%)

Variation in inflation explained- 4 quarters ahead (%)


Inflation (WPI)

72.1 63.7 55.8

Crude oil price 22.0 20.1 19.2 Output gap 2.0 2.1 3.1 Fiscal deficit 1.7 4.8 12.6 Policy rate 2.2 9.3 9.3 Based on authors’ calculations using data on ‘Wholesale Price Index’, ‘GDP at constant factor cost’, ‘Fiscal deficit of central



Table 2: Forecast error variance decomposition - WPI inflation

http://ees.elsevier.com/asieco/download.aspx?id=18840&guid=b2926bf6-454f-4d95-873c-57c6d4f0bff6&scheme=1

of 29

Accep

ted

Man

uscr

ipt

Table 3: Forecast error variance decomposition – GDP deflator inflation

Variables Variation in inflation explained- 2 quarters ahead (%)



Inflation (GDP deflator)

73.6 54.8 51.3

Crude oil price 21.5 21.2 19.8 Output gap 3.4 3.2 4.8 Fiscal deficit 0.5 7.7 11.0 Policy rate 1.0 13.1 13.1 Based on authors’ calculations using data on ‘GDP at current factor cost’, ‘GDP at constant factor cost’, ‘Fiscal deficit of central



Table 3: Forecast error variance decomposition - GDP inflation

http://ees.elsevier.com/asieco/download.aspx?id=18841&guid=d27b344f-06f0-41e4-9793-adc25a37cb7b&scheme=1

Determinants of inflation in India

Documents

Transcript of Determinants of inflation in India