academy of economic studies bucharest doctoral school of finance and banking
DESCRIPTION
ACADEMY OF ECONOMIC STUDIES BUCHAREST DOCTORAL SCHOOL OF FINANCE AND BANKING. INFLATION DYNAMICS IN ROMANIA: A NEW KEYNESIAN PHILLIPS CURVE APPROACH. Student: Razvan Radu Supervisor: Professor Moisa Altar. Bucharest July 2006. Contents. Objectives Literature review The Model - PowerPoint PPT PresentationTRANSCRIPT
ACADEMY OF ECONOMIC STUDIES BUCHAREST
DOCTORAL SCHOOL OF FINANCE AND BANKING
INFLATION DYNAMICS IN ROMANIA: A NEW KEYNESIAN PHILLIPS CURVE
APPROACH
Student: Razvan Radu
Supervisor: Professor Moisa Altar
BucharestJuly 2006
Contents
• Objectives
• Literature review
• The Model
• Empirical Specification
• The Data
• Results
• Conclusions
Objectives
• To analyse the short-run inflation dynamics in Romania
• To identify the structural parameters that characterize the firms’ princing behavior
• To assess the extent to which forward and backward looking price setting characterize the inflationary process
• To compare the nature of Romanian inflation with that reported for other economies
The New Keynesian Phillips Curve
• Based on the seminal work by Taylor (1980), Calvo (1983)
• The NKPC is derived explicitly from an optimization process, assuming staggered price setting by forward looking, monopolistically competing firms
• Current inflation is related to future expected inflation and real marginal cost
• The parameters of the NKPC are directly linked to the behavior of agents and are thus exempt from the Lucas critique
The New Keynesian Phillips Curve• Gali and Gertler (1999) extend the NKPC,
incorporating inflation inertia• Gali and Gertler (1999) - US• Gali, Gertler and Lopez-Salido (2001) - Euro Area• Leith and Malley (2002) - G7 countries• Balakrishnan and Lopez-Salido (2002) - UK• Rumler (2005) - Euro Area countries• Ribon (2004) - Israel• Cespedes, Soto and Ochoa (2005) - Chile• Lendvai (2005) - Hungary
The Model• Gali and Gertler (1999)
• The supply side of the economy consists of a continuum of monopolistically competing firms the size of which is normalized to one
• Each period a random fraction of firms reset their price, while all others firms keep their price unchanged
• Only a fraction of firms adjust their prices optimally in a forward looking manner, the remaining firms are backward looking and set their prices based on the recent aggregate price behavior
1
1
1. Closed economy model
tttftbt mcE )( 11
b
where the coefficients of the lagged inflation, the expected inflation and the marginal cost are functions of the deep parameters:
,
f
)1)(1)(1(
)1(1
2. Open economy model
Open economy effects are incorporated by modeling imported goods as intermediate production goods
Assuming a Cobb-Douglas technology, the real marginal cost has the form:
tttt mawmc )1()( - real wage tw
tm
- labor-augmenting technologyta
- real cost of a unit of the imported good
- labor’s share in gross output
tm
ttl
tbttft mawE )(11
b
f
)1)(1)(1(
l
)1)(1)(1)(1(
m
)1(1
Empirical Specification
ttttftbt mcE )( 11
ttm
ttl
ttftbt qawE )()( 11
twhere is an inflation shock, which is an innovation with respect to information set dated or earlier. 1t
Assuming rational expectations,
111 )( tttt E
1twhere is the forecast error in predicting future inflation, and hence a innovation with respect to contemporaneous and past information
tttftbt emc 11
ttm
ttl
tftbt eqaw )(11
1 tftte
The following orthogonality conditions can be written, forming the basis for estimating the model by the Generalized Method of Moments (Hansen, 1982).
0])[( 11 tttftbtt ZmcE
0]))([( 11 ttm
ttl
tftbtt ZqawE
tZ is a vector of instrumental variables which must be uncorrelated with te
GMM in the presence of “weak instruments”
• Stock, Wright and Yogo (2002), Kleibergen (2002), Dufour and Jasiak (2001), Dufour (2003), Dufour, Khalaf and Kichian (2005)
• When instruments are weakly correlated to endogenous variables, the sampling distributions of GMM statistics are in general non-normal and standard GMM estimates, hypothesis tests and confidence intervals are unreliable
• The empirical relevance of the results will also be examined using two type of tests (AR-statistic and K-statistic) the properties of which are robust in finite samples to the quality
of instruments
The AR statistic
Anderson-Rubin (1949), Dufour (2003), Dufour, Khalaf and Kichian (2005)
ukXYy 1
)1( Ty
)( mTY tmc1t
1t - vector of observations on dependent variable ( )t
- matrix of endogenous variables ( , )
)(11kT
X
- matrix of exogenous variables ( )
u - error term, which is an innovation with respect to exogenous variables
VXXY 2211
Assume that the reduced form for the endogenous variables is:
where is a matrix of error terms assumed to be
cross-correlated and correlated with and is the
matrix of instruments
)( mTV u
)(22kT
X
VupXpXy 2211
kp 11
22 p
Identification requires that the matrix have full rank 2
ukkXYkXYy )()( 01
01
00
00 ,: kkH o
)()]([)]()([ 00220
01101
0 VuXkkXkXYy
Given that the AR-test has low power when the number of instruments strongly exceeds the number of parameters to be estimated, Kleibergen (2002) propose to project the disturbances of the structural equation on the instrument set given by
22XZ
where is the coefficient of in the regression of on and .
2
2X
Y 1X2X
The AR-statistic assesses the exclusion of and in the regression of on and .
1X 2X
010 kXYy
1X 2X
The DataQuarterly data are used, covering the period 1998Q1:2005Q4. All series are in logarithms.
gdp_r Real GDP, in 2003 constant prices
cpi Consumer price index, 1990=100, quarterly average
coreCore price index, computed from CPI by exclusion of administered prices
l_infl_cpi Annualized quarter on quarter CPI inflation ratel_infl_core Annualized quarter on quarter core inflation rateppi Producer prices index, 2000=100, quarterly average
ind_prodIndustrial production index, 1993=100, quarterly average
wageGross average wage in industrial sector index, 2000=100, quarterly average
emplEmployment in industrial sector index, 1993=100, quarterly average
l_ulcReal unit labor cost in industrial sector, computed as a ratio between the real average gross wage index and productivity index
infl_cpi CPI inflation* infl_core Core inflation* ulc Unit labor cost* gap Output* rer Real exchange rate*
* Detrended using HP filterThe series are seasonally adjusted using X-12 procedure
.0
.1
.2
.3
.4
.5
.6
.7
.8
98 99 00 01 02 03 04 05
L_INFL_CORE
CORE INFLATION
.0
.1
.2
.3
.4
.5
.6
.7
.8
98 99 00 01 02 03 04 05
L_INFL_CPI
CPI INFLATION
-.3
-.2
-.1
.0
.1
.2
.3
.4
98 99 00 01 02 03 04 05
INFL_CORE
CORE INFLATION
Actual and detrended data
-.3
-.2
-.1
.0
.1
.2
.3
.4
98 99 00 01 02 03 04 05
INFL_CPI
CPI INFLATION
-.12
-.08
-.04
.00
.04
.08
.12
98 99 00 01 02 03 04 05
ULC
REAL ULC
-.03
-.02
-.01
.00
.01
.02
.03
.04
.05
.06
98 99 00 01 02 03 04 05
GAP
OUTPUT GAP
-4.9
-4.8
-4.7
-4.6
-4.5
-4.4
-4.3
98 99 00 01 02 03 04 05
L_ULC
REAL ULC
10.60
10.65
10.70
10.75
10.80
10.85
10.90
10.95
98 99 00 01 02 03 04 05
L_GDP_R
REAL GDP
2.7
2.8
2.9
3.0
3.1
3.2
98 99 00 01 02 03 04 05
L_RER
REAL EXCHANGE RATE
-.12
-.08
-.04
.00
.04
.08
.12
.16
98 99 00 01 02 03 04 05
RER
REAL EXCHANGE RATE
-.4
-.3
-.2
-.1
.0
.1
.2
.3
.4
.5
-.03
-.02
-.01
.00
.01
.02
.03
.04
.05
.06
1998 1999 2000 2001 2002 2003 2004 2005
INFL_CORE OUTPUT GAP
CORE INFLATION AND OUTPUT GAP
-.3
-.2
-.1
.0
.1
.2
.3
.4
-.12
-.08
-.04
.00
.04
.08
.12
.16
1998 1999 2000 2001 2002 2003 2004 2005
INFL_CORE ULC
CORE INFLATION AND ULC
In the theoretical model, the actual inflation rate is modeled as a deviation from a zero steady state. This assumption seems not appropiate in the case of the Romanian economy, as the inflation rate evolved along a downward convergence path.
Consequently, I choose to model the fluctuations of inflation around its trend, following Coenen and Wieland (2000, 2004), Coenen and Levin (2004), Ribon (2004), Rumler (2005).
Results
f b J test
(p-value)
AR test
(p-value)
K test
(p-value)
0.491
(0.110)
0.442
(0.071)
0.161
(0.035)0.777 0.696 0.145
0.506
(0.043)
0.494
(0.043)
0.062
(0.025)0.812 0.603 0.078
1.1. Closed economy model
1. Reduced-form estimates
The instruments set includes three lags of: the core inflation, the deviation of unit labor cost from the trend and the output gap (from t-2 to t-4); and the lagged deviation of real exchange rate from the trend (t-2).
f b l mJ test
(p-value)
AR test
(p-value)
K test
(p-value)
0.571
(0.042)
0.261
(0.035)
0.428
(0.039)
0.726
(0.063)
0.847 0.211 0.516
0.729
(0.017)
0.271
(0.017)
1.359
(0.124)
0.336
(0.017)
0.759 0.129 0.194
1.2 Open economy model
The instruments set consists of three lags of: the core inflation, the output gap, the deviation of the real exchange rate from its trend and
the change in real exchange rate (from t-2 to t-4 ).
f b DJ test
(p-value)
AR test
(p-value)
K test
(p-value)
0.546
(0.050)
0.396
(0.062)
0.799
(0.158)
0.485
(0.114)
0.441
(0.074)
0.172
(0.045)
2.201
(0.244)0.779 0.701 0.150
0.552
(0.052)
0.404
(0.060)
0.812
(0.154)
0.491
(0.110)
0.442
(0.071)
0.161
(0.045)
2.232
(0.258)0.777 0.697 0.145
0.567
(0.071)
0.575
(0.042)0.99
0.493
(0.043)
0.504
(0.043)
0.070
(0.037)
2.312
(0.380)0.806 0.593 0.067
0.587
(0.075)
0.566
(0.042)0.99
0.506
(0.043)
0.492
(0.043)
0.065
(0.038)
2.421
(0.440)0.814 0.607 0.074
Instrumental variables include three lags of: the core inflation, the deviation of unit labor cost from the trend and the output gap (from t-2 to t-4 ); and the lagged deviation of real exchange rate from the trend (t-2).
2. Structural-form estimates
2.1 Closed economy model
US, Euro Area 0.80-0.90 Gali, Gertler, Lopez-Salido (2001)Spain 0.85-0.90 Gali, Lopez-Salido (2001)Hungary 0.50-0.65 Lendvai (2005)Israel 0.50-0.60 Ribon (2004)
Estimates for other economies
Euro Area 0.02-0.30 Gali, Gertler, Lopez-Salido (2001)US 0.40-0.45 Gali, Gertler, Lopez-Salido (2001)Spain 0.70 Gali, Lopez-Salido (2001)Hungary 0.30-0.55 Lendvai (2005)Israel 0.40-0.50 Ribon (2004)
Price stickiness ( )
Fraction of backward looking price setters ( )
2.1 Open economy model
f b l m DJ test
(p-value)
AR test
(p-value)
K test
(p-value)
0.389
(0.043)
0.116
(0.080)
0.733
(0.071)
0.859
(0.056)
0.578
(0.014)
0.235
(0.014)
0.675
(0.009)
0.109
(0.002)
1.636
(0.114)
0.954 0.670 0.396
0.403
(0.045)
0.171
(0.097)
0.675
(0.095)
0.879
(0.087)
0.493
(0.017)
0.310
(0.017)
0.573
(0.002)
0.078
(0.004)
1.676
(0.127)
0.947 0.560 0.293
The set of instrumental variables consists of three lags of: the core inflation, the deviation of unit labor cost from the trend, the deviation of real exchange rate from its trend and the change of the real exchange rate (from t-2 to t-4 ); the lagged output
gap (t-2).
f b l m DJ test
(p-value)
AR test
(p-value)
K test
(p-value)
0.249
(0.075)
0.157
(0.053)
0.990.761
(0.047)
0.607
(0.120)
0.387
(0.121)
0.895
(0.389)
0.274
(0.142)
1.331
(0.133)
0.8400.863 0.120
0.237
(0.016)
0.265
(0.021)
0.990.815
(0.012)
0.467
(0.037)
0.529
(0.037)
0.700
(0.019)
0.158
(0.012)
1.309
(0.028)
0.7840.743 0.054
Instruments include three lags of: the detrended core inflation, the deviation of the unit labor cost from its trend, the deviation of real exchange rate from the trend and
the output gap (from t-2 to t-4 ).
Price stickiness ( )Euro Area 0.68 Rumler (2005)US 0.55 Leith and Malley (2002)UK 0.60 Leith and Malley (2002)Canada 0.70 Leith and Malley (2002)France 0.70 Leith and Malley (2002)Japan 0.70 Leith and Malley (2002)Germany 0.80 Rumler (2005), Leith and Malley (2002)Italy 0.70 Rumler (2005)Hungary 0.50-0.70 Lendvai (2005)Israel 0.50-0.60 Ribon (2004)
Fraction of backward looking price setters ( )
Euro Area 0.50 Rumler (2005)Germany 0.21, 0.65 Leith and Malley (2002), Rumler (2005)Italy 0.43, 0.35 Leith and Malley (2002), Rumler (2005)US 0.30 Leith and Malley (2002)Canada 0.30 Leith and Malley (2002)Japan 0.30 Leith and Malley (2002)Hungary 0.30-0.50 Lendvai (2005)Israel 0.40-0.50 Ribon (2004)
Estimates for other economies
J test (p-value)
AR test (p-value)
K test (p-value)
0.300.471
(0.072)0.329
(0.122)0.584
(0.015)0.412
(0.015)0.071
(0.002)0.166
(0.004)1.890
(0.259)0.780 0.479 0.129
0.350.461
(0.072)0.321
(0.120)0.584
(0.015)0.411
(0.015)0.089
(0.002)0.166
(0.004)1.854
(0.247)0.642 0.479 0.130
0.400.450
(0.071)0.313
(0.118)0.585
(0.015)0.411
(0.015)0.110
(0.002)0.165
(0.004)1.817
(0.234)0.782 0.477 0.130
0.450.438
(0.070)0.303
(0.116)0.586
(0.014)0.410
(0.015)0.135
(0.003)0.165
(0.003)1.778
(0.222)0.782 0.476 0.130
0.500.424
(0.069)0.294
(0.114)0.586
(0.014)0.410
(0.015)0.164
(0.003)0.164
(0.003)1.737
(0.209)0.642 0.474 0.131
0.550.409
(0.069)0.284
(0.111)0.585
(0.016)0.410
(0.015)0.200
(0.004)0.164
(0.003)1.692
(0.196)0.781 0.470 0.130
0.600.391
(0.068)0.275
(0.109)0.583
(0.014)0.413
(0.015)0.244
(0.005)0.163
(0.003)1.642
(0.183)0.642 0.464 0.127
0.650.367
(0.068)0.269
(0.106)0.572
(0.014)0.424
(0.015)0.300
(0.005)0.162
(0.015)1.581
(0.170)0.770 0.448 0.117
0.700.211
(0.079)0.389
(0.084)0.348
(0.012)0.649
(0.012)0.446
(0.009)0.191
(0.004)1.267
(0.127)0.722 0.221 0.011
0.750.216
(0.074)0.366
(0.084)0.367
(0.011)0.630
(0.011)0.503
(0.009)0.167
(0.003)1.276
(0.121)0.723 0.237 0.155
0.800.243
(0.070)0.336
(0.090)0.416
(0.011)0.580
(0.011)0.526
(0.010)0.131
(0.002)1.322
(0.123)0.720 0.285 0.029
f b l m D
Influence of labor’s share on estimated parameters
Influence of labor’s share on estimated price stickiness
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80
0
0.5
1
1.5
2
2.5
3
0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80
D
Actual vs. estimated values
-.1
.0
.1
.2
.3
.4
1999 2000 2001 2002 2003 2004 2005
ACTUAL FITTED
REDUCED FORM ESTIMATES - CLOSED ECONOMY
-.1
.0
.1
.2
.3
.4
1999 2000 2001 2002 2003 2004 2005
ACTUAL FITTED
STRUCTURAL ESTIMATES - CLOSED ECONOMY
-.1
.0
.1
.2
.3
.4
1999 2000 2001 2002 2003 2004 2005
ACTUAL FITTED
REDUCED FORM ESTIMATES - OPEN ECONOMY
-.1
.0
.1
.2
.3
.4
1999 2000 2001 2002 2003 2004 2005
ACTUAL FITTED
STUCTURAL ESTIMATES - OPEN ECONOMY
Conclusions• The weights of lagged and future expected inflation are roughly
equal
• The inflationary process in Romania is more inertial than in the Euro Area and other economies characterized by stable and low rates of inflation
• The inertia of inflation can be explained by the fraction of firms that set their prices in a backward-looking manner, which ranges between 0.3 and 0.4
• Firms adjust their prices approximately every 1.3 to 2.4 quarters, which imply a lower price stickiness as compared to the Euro Area
• The estimated parameters are in line with those reported in the literature for other countries that experienced periods of high inflation