Professional Documents
Culture Documents
QREF CredibilityPhillipsCurve 2012
QREF CredibilityPhillipsCurve 2012
a r t i c l e
i n f o
Article history:
Received 2 August 2011
Received in revised form 27 April 2012
Accepted 24 May 2012
Available online 1 June 2012
JEL classication:
E43
E52
a b s t r a c t
The paper investigates the presence of monetary policy credibility in eight countries by ltering the
residuals from an augmented Phillips curve. Two of the eight countries (US and New Zealand) exhibit
robust credibility effects across samples. Two countries (South Africa and the UK) exhibit credibility effects
in the sample involving the 1990s, but these effects disappear in the sample beginning in 2000. The rest
of the countries do not exhibit monetary policy credibility. Given that seven of the eight countries have
adopted an explicit ination-targeting framework, we conclude that there is very weak evidence that this
framework enhances monetary policy credibility. These results are however sensitive to how ination
and the output gap are measured.
2012 The Board of Trustees of the University of Illinois. Published by Elsevier B.V. All rights reserved.
Keywords:
Credibility
Ination expectations
Monetary policy
1. Introduction
This paper uses a Phillips-curve based method to investigate the extent of central bank credibility in eight countries.
Since Kydland and Prescott (1977), credibility has been viewed
as an important ingredient in the conduct of monetary policy.
According to Blinder, Ehrmann, Fratzscher, De Haan, and Jansen
(2008) credibility helps with making disination less costly. Credibility also helps the central bank gain public support for its
actions. This view is shared by, for example, Bertola and Caballero
(1992), Bertola and Svensson (1993) and Demertzis, Marcellino,
and Viegi (2008). However, empirical analyses suggest that central banks are not perfectly credible. This may be due to the
fact that, as Lohman (1992) argues, in order to optimize monetary policy commitment and retain credibility, central banks must
be allowed to exercise exible policy responses to unforeseen
contingencies.
The most relevant place where the theory of credibility is
applied is the Phillips curve. Blinder (2000) notes that the
so-called credibility hypothesis says that perfectly credible preannouncements of disination will reduce ination expectations
abruptly. Therefore if the central bank is credible relatively low
unemployment is required to bring about a drastic fall in the
ination rate. By implication, slight increases in the interest rate
Corresponding author. Tel.: +27 11 717 8109; fax: +27 11 717 8081.
E-mail address: christopher.malikane@wits.ac.za (C. Malikane).
1062-9769/$ see front matter 2012 The Board of Trustees of the University of Illinois. Published by Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.qref.2012.05.002
C. Malikane, T. Mokoka / The Quarterly Review of Economics and Finance 52 (2012) 266271
t1 +
t = te + xt1 + qt1 + zt1 + mt1 +
fuel
t1 + t ,
food
(1)
te
267
t1 +
t = xt1 + qt1 + zt1 + mt1 +
fuel
t1 +
t ,
food
(2)
where
t = te + t . Ination expectations in the Phillips curve are
contained in the term
t . We postulate that if monetary policy
is credible, then ination expectations contained in
t respond
to changes in the short term nominal interest rate. Note that the
disturbance term t does not respond to changes in the nominal
interest rate. The reason for this is that E(te t ) = 0. If both expected
ination and the disturbance term were affected by the nominal
interest rate, there would be some correlation between the two
arising from the movement in the nominal interest rate.
We expect that
t will exhibit some persistence. This persistence can be grounded in the two explanations found in the
literature. Firstly,
t may be persistent because past ination may
affect current ination through a combination of expectations formation and overlapping wage and price contracts as noted by
Fuhrer and Moore (1995) and Gordon (1998). Secondly,
t may
be persistent because of the time-variation of trend ination as
noted by Kozicki and Tinsley (2003), Ireland (2007) and Cogley
and Sbordone (2008). Our specication does not favour any of
268
C. Malikane, T. Mokoka / The Quarterly Review of Economics and Finance 52 (2012) 266271
t = (L)
t (L)rt + t ,
(3)
where rt is the change in the nominal interest rate, (L) measures
the extent to which
t is persistent, (L) > 0 captures the response
of
t to changes in the nominal interest rate and t is the component of the composite shock that does not respond to changes in
the nominal interest rate. Ireland (2007) and Cogley and Sbordone
(2008) specify the process that drives trend ination to follow a
driftless random walk. On the other hand, the Gordon-type specication would require that the coefcients of lagged ination sum
to unity. Accordingly, these two strands of literature support us in
specifying that (1) = 1. The larger is (L) the more credible is monetary policy. If (1) < 0, then monetary policy cannot deliver a low
ination environment. In this instance, ination expectations do
not respond in a desirable manner to movements in the nominal
interest rate. That is, increases in the nominal interest rate do not
lower ination expectations.
3. Empirical strategy and results
The econometric strategy we employ in our test for credibility involves a two-stage procedure in the extraction of
t . In the
rst stage we estimate Eq. (2). From Ireland (2007), Cogley and
Sbordone (2008) and Gordon (1998) we know that
t is autocorrelated. In order to generate unbiased and efcient estimates of the
parameters of the Phillips curve, we follow the two-step procedure
proposed by Pindyck and Rubinfeld (1998, p. 590). In the rst step
we estimate Eq. (2) using OLS in line with Judge et al. (1985, p.
t . In the second step, we
295) to obtain errors which we denote
t in order to generate
re-estimate Eq. (2) augmented with lags of
t1
zt1 + mt1 +
t = xt1 + qt1 +
fuel
t + t ,
t1 +
(L)
food
(4)
t and t is a serially
(L) denotes the coefcients of the lags of
which tests for central bank credibility. The method outlined above
can be viewed as 2SLS, in which the rst-step regression generates
errors whose lags are then used as instruments for
t in the secondstep regression. Alternatively the procedure mentioned by Pindyck
and Rubinfeld (1998, p. 590) can be viewed as NLS (see also Judge
et al. (1985, chap. 8) and Davidson and McKinnon (2004, chap. 7)).1
An important issue with Phillips curve estimations is that
of identication, as brought to the fore by Mavroeidis (2004),
Mavroeidis (2005), Bardsen et al. (2004) and Martins and Gabriel
(2009). Martins and Gabriel (2009) compute identication-robust
condence sets for the parameters of the NKPC in order to test
1
For a model yt = xt + ut with ut = ut1 + t , we can show that
yt yt1 = (xt xt1 ) + t is the same as yt = xt + ut1 + t . This can be
generalized to models with higher-order autocorrelation.
for weak identication. Mavroeidis (2005) uses the so-called concentration parameter, which is the minimum eigenvalue of the
conventional concentration matrix. A concentration parameter less
than 10 is considered to indicate weak identication. In this paper
we approach this question by using the simple procedure proposed
by Bardsen et al. (2004). In this procedure, F-statistic of the rststage regression of the instrumental variable estimation is used
to test for weak identication. If this F-statistic is less than 10
this would be indicative of weak identication. We also report the
probability of the F-statistic to check the joint signicance of the
rst-stage variables.
A potential problem with the estimation of Eq. (3) is the
endogeneity of rt . It is now common for central banks to use
the short-term nominal interest rate as a policy instrument (see
for example, Clarida, Gali, and Gertler (2000)). Because
t contains ination expectations, a forward-looking central bank would
respond to changes in ination expectations thereby rendering rt
potentially endogenous. In order to account for this possibility, we
rst specify a regression of the following form:
rt = r0 +
t + t .
(5)
C. Malikane, T. Mokoka / The Quarterly Review of Economics and Finance 52 (2012) 266271
Australia
Canada
269
South Korea
.05
.06
.04
.05
.03
.04
.02
.03
.01
.02
Mexico
.065
.06
.05
.04
.060
.055
Vt
Inflation
.050
.03
.045
.02
.01
Vt
Inflation
.040
Vt
Inflation
.00
.00
-.01
-.01
97
98
99
00
01
02
03
04
05
06
07
08
.035
.030
.00
94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
09
Vt
Inflation
.01
New Zealand
99
United Kingdom
00
01
02
03
04
05
06
07
08
2002
09
2003
2004
United States
2005
2006
2007
2008
2009
South Africa
.14
.12
.05
.10
.04
.12
.07
.08
Vt
Inflation
.03
.06
.06
.10
.05
.08
.04
.06
.03
.02
.04
.04
.02
.02
.01
Vt
Inflation
.00
-.01
.01
.02
.00
.00
97 98 99 00 01 02 03 04 05 06 07 08 09
88
90
92
94
96
98
00
02
04
06
08
Vt
Inflation
-.04
-.02
86
.00
-.02
-.01
-.02
95 96
Vt
Inflation
84 86 88 90 92 94 96 98 00 02 04 06 08
95
96
97
98
99 00
01
02
03
04
05 06
07
08
09
t = (L)
between 1984 and 1995 monetary policy targets were not credible for the two countries. However, like Johnson (1998), we nd
monetary policy credibility for New Zealand.
Our nding on the credibility of monetary policy for South Africa
is not consistent with Aron and Muellbauer (2007) who nd, using
survey data, that monetary policy credibility effects have been signicant for the country since 2000. Our nding for Mexico is not
consistent with Schmidt-Hebbel and Werner (2002). They nd that
the ination target exerts a credible and strong inuence on private sector expectations. Despite its adoption of ination targeting
in 1998, we nd that the Bank of Korea is not credible.
In order to check whether these ndings are robust to alternative measures of ination and measurement of the output gap, we
re-estimated Eqs. (2) and (3) using ination based on the GDP deator and output detrended using a polynomial time trend. Generally
we nd that the results based on the deator are not consistent
with the ones that are contained in this paper. However, the results
based on the de-trended output are consistent with the results that
are reported in this paper, except for New Zealand (20002009).
Nevertheless, in view of the fact that all the countries that are in
our sample target CPI-based ination, it is clear that an appropriate
measure of ination is the one based on the CPI.
In order to examine the possible sources of the inconsistency
between our results and those found in the literature, we performed two procedures. Firstly, we computed correlations between
our measure of ination expectations and the ones based on surveys and the yield curve for a selected number of economies (viz.
the US, UK, Australia and South Africa). Secondly, we estimated Eq.
(3) using survey-based and yield curve based measures to check
if we arrive at the same results. We nd weak correlation among
the three measures. For example for the UK the correlation coefcient between our measure and the one based on the yield curve
is 0.11, between the yield curve measure and surveys is 0.42, and
between our measure and surveys is 0.29. In terms of credibility
tests based on the yield and survey expectations, only the US has
results that are consistent with our measure. Therefore, the inconsistency between our results and those found using alternative
270
Table 1
Estimations of the Phillips curve (Eq. (4)) (standard errors in parentheses).
Coeff
Aus
Kor
Mex
NZ
96
00
Can
93
00
99
01
93
00
UK
82
00
US
83
00
SA
93
00
0.43 (0.13)
0.004 (0.01)
0.31 (0.13)
0.05 (0.02)
0.39 (0.08)
0.41 (0.10)
0.11 (0.05)
0.03 (0.006)
0.06 (0.04)
0.05 (0.03)
0.05 (0.01)
0 . 173 (0.09)
0.01 (0.01)
0 . 253 (0.38)
0.10 (0.12)
0.02 (0.02)
1 . 123 (0.42)
0.87 (0.05)
0.72 (0.06)
0.23 (0.05)
0.07 (0.02)
0.42 (0.09)
0.07 (0.02)
0.61 (0.17)
0.32 (0.32)
0.12 (0.04)
0 . 012 (0.03)
0.07 (0.02)
0.08 (0.05)
0.07 (0.03)
0.18 (0.06)
0.04 (0.02)
0.007 (0.009)
0.07 (0.03)
0.03 (0.01)
R2
2 (4)
Q(4)
DW
2A (4)
JB
Fprob
Fstat
0.90
0.26
0.78
1.59
0.90
0.72
0.01
3.60
0.81
0.25
0.60
1.23
0.57
0.70
0.06
2.44
0.77
0.19
0.93
1.91
0.93
0.53
0.00
10.34
0.47
0.16
0.71
2.03
0.86
0.60
0.00
6.84
0.88
0.40
0.96
1.95
0.93
0.82
0.00
42.53
0.89
0.18
0.68
2.04
0.48
0.96
0.00
11.63
0.92
0.16
0.68
2.02
0.86
0.49
0.00
32.96
0.05 (0.09)
0.60
0.17
0.93
2.14
0.95
0.32
0.00
6.74
0.63
0.53
0.36
2.11
0.44
0.97
0.00
14.19
0.72
0.69
0.19
1.96
0.29
0.91
0.00
22.72
0.93
0.29
0.87
2.11
0.82
0.91
0.00
20.86
0.43 (0.05)
0.63 (0.09)
0.87
0.12
0.77
1.61
0.73
0.57
0.01
3.91
0.97
0.30
0.57
2.29
0.60
0.33
0.00
10.71
Superscript above the estimate of the parameter denotes the lag of the variable.
Probability.
Table 2
Post ination targeting estimations of credibility (standard errors in parentheses).
Coeff
1
2
3
4
5
0
1
2
(1)
2
R
2 (4)
Q(4)
DW
2A (4)
JB
*
**
Aus
Kor
Mex
NZ
96
00
Can
93
00
99
01
93
00
82
00
83
00
93
00
1.14 (0.09)
0.14 (0.09)
1.45 (0.12)
0.82 (0.09)
1.00 (0.00)
0.68 (0.11)
1.00 (0.00)
1.00 (0.00)
0.75 (0.07)
0.22 (0.08)
0.17 (0.08)
0.96 (0.16)
0.14 (0.07)
0.16 (0.20)
1.41 (0.09)
0.46 (0.13)
0.08 (0.07)
0.31 (0.00)
1.43 (0.17)
0.27 (0.25)
0.59 (0.26)
1.49 (0.07)
0.51 (0.13)
0.03 (0.00)
0.87 (0.11)
1.20 (0.10)
0.35 (0.18)
0.04 (0.16)
0.11 (0.00)
1.22 (0.08)
0.22 (0.08)
0.10* (0.04)
0.26* (0.12)
0.45* (0.03)
0.30* (0.05)
0.46* (0.13)
0.33* (0.15)
0.57* (0.07)
0.40* (0.05)
0.56 (0.12)
0.37 (0.25)
1.17* (0.28)
0.15* (0.06)
0.13 (0.29)
0.17* (0.05)
0.95
0.39
0.56
2.04
0.63
0.00
0.73
0.36
0.67
1.97
0.79
0.00
0.97
0.47
0.89
1.79
0.91
0.35
0.37 (0.09)
0.33 (0.18)
0.17 (0.24)
0.39 (0.18)
0.22 (0.22)
0.98* (0.23)
1.03* (0.23)
0.85
0.15
0.56
1.59
0.69
0.63
0.75
0.28
0.45
2.26
0.39
0.25
5% signicance.
10% signicance.
Probability.
UK
0.32 (0.11)
0.52* (0.10)
0.05 (0.30)
0.79* (0.23)
0.81* (0.25)
0.40* (0.17)
0.17** (0.08)
0.10 (0.28)
0.56
0.48
0.32
2.14
0.30
0.53
0.53
0.39
0.20
2.09
0.22
0.63
0.43
0.68
0.83
2.4
0.67
0.96
0.73
0.19
0.17
1.93
0.75
0.95
0.90
0.81
0.48
1.74
0.27
0.17
0.47
0.11
0.55
1.88
0.34
0.65
US
SA
0.36 (0.13)
0.49 (0.11)
0.03* (0.01)
0.80* (0.30)
0.74
0.91
0.91
1.82
0.56
0.38
0.86
0.16
0.62
1.63
0.63
0.40
0.81
0.48
0.87
1.54
0.91
0.21
C. Malikane, T. Mokoka / The Quarterly Review of Economics and Finance 52 (2012) 266271
0.70
0.64
0.31
1.57
0.44
0.67
0.02
4.47
0.06 (0.06)
C. Malikane, T. Mokoka / The Quarterly Review of Economics and Finance 52 (2012) 266271
2
Results for the measure of ination based on the deator and time de-trended
output gap are available upon request. Furthermore correlation and regression
results for the US, UK, Australia and South Africa are also available upon request.
271
Cogley, T., & Sbordone, A. M. (2008). Trend ination, indexation and ination persistence in the New Keynesian Phillips Curve. American Economic Review, 98,
21012126.
Davidson, R., & Mackinnon, J. G. (2004). Econometric theory and methods. New York:
Oxford University Press.
Demertzis, M., Marcellino, M., Viegi, N. (2008). A measure for credibility: Tracking
us monetary developments. DNB working paper no. 187/2008.
Fair, R. C. (2008). Testing price equations. European Economic Review, 52, 14241437.
Fraga, A., Godfajn, I., & Minella, A. (2003). Ination targeting in emerging market
economies. NBER Macroeconomics Annual, 18, 365418.
Fuhrer, J., & Moore, G. (1995). Ination persistence. Quarterly Journal of Economics,
110, 127159.
Gali, J., & Gertler, M. (1999). Ination dynamics: A structural econometric analysis.
Journal of Monetary Economics, 44, 195222.
Gali, J., Gertler, M., & Lpez-Salido, J. D. (2001). European ination dynamics. European Economics Review, 45, 12371270.
Gali, J., Gertler, M., & Lpez-Salido, J. D. (2005). Robustness of the estimates of
the hybrid New Keynesian Phillips curve. Journal of Monetary Economics, 52,
11071118.
Gerlach, S., & Svensson, L. E. O. (2003). Money and ination in the Euro-Area. A case
for monetary indicators? Journal of Monetary Economics, 50, 16491672.
Goodfriend, M. (1993). Interest rate policy and the ination scare problem:
19791992. Federal Reserve Bank of Richmond Economic Quarterly, 79, 124.
Goodfriend, M., & King, R. G. (2005). The incredible Volker disination. Journal of
Monetary Economics, 52, 9811015.
Gordon, R. J. (1998). Foundations of the goldilocks economy: Supply shocks and the
time-varying NAIRU. Brookings Papers on Economic Activity, 2, 97346.
Gordon, R. J. (2011). The history of the Phillips curve: Consensus and bifurcation.
Economica, 78, 1050.
Grkaynak, R. S., Levin, A. T., Marder, A. N., & Swanson, E. T. (2007). Ination targeting
and the anchoring of ination expectations in the Western Hemisphere. Federal
Reserve Bank of San Francisco Economic Review, 2547.
Henzel, S. (2008). Learning trend ination Can signal extraction explain survey
forecasts? IFO institute for economic research working papers no. 55.
Ireland, P. (2007). Changes in the federal reserves ination target: Causes and consequences. Journal of Money, Credit and Banking, 39, 18511882.
Ireland, P. (2004). Moneys role in the monetary business cycle. Journal of Money,
Credit and Banking, 36, 969983.
Johnson, D. (1998). The credibility of monetary policy: International evidence based
on surveys of expected ination. Bank of Canada conference paper.
Judge, G. G., Griths, W. E., Hill, R. C., Lutkepohl, H., & Lee, C. G. (1985). The theory
and practice of econometrics (2nd ed.). New York: John Wiley & Sons.
Kozicki, S., & Tinsley, P. A. (2003). Alternative sources of lag dynamics of ination.
In Price adjustment and monetary Policy: Proceedings of the conference held by the
Bank of Canada November 2002, (pp. 347).
Kydland, F. E., & Prescott, E. C. (1977). Rules rather than discretion: The inconsistence
of optimal plans. Journal of Political Economy, 85, 473491.
Lind, J. (2005). Estimating New-Keynesian Phillips Curves: A full information maximum
likelihood approach, 52, 11351149.
Lohman, S. (1992). Optimal commitment in monetary policy: Credibility versus
exibility. The American Economic Review, 82, 273286.
Martins, L. F., & Gabriel, V. J. (2009). New Keynesian Phillips Curves and potential
identication failures: A generalized empirical likelihood analysis. Journal of
Macroeconomics, 31, 561571.
Mavroeidis, S. (2004). Weak identication of forward-looking models in monetary economics. Oxford Bulletin of Economics and Statistics, 66(Supplement),
03059049.
Mavroeidis, S. (2005). Identication in forward-looking models estimated by GMM,
with an application to the Phillips curve. Journal of Money, Credit and Banking,
37(3), 421448.
Mehra, Y. P. (1996). The bond rate and actual future ination. The Federal reserve
bank of Richmond working paper 97-03.
Mohanty, M. S., Klau, M. (2004). Monetary policy rules in emerging market
economies: Issues and evidence. BIS working paper 149.
Nelson, E. (2003). The future of monetary aggregates in monetary policy analysis.
Journal of Monetary Economics, 50, 585608.
Pindyck, D. L., & Rubinfeld, R. S. (1998). Economic models and economic forecasts. New
York: McGraw Hill.
Perrier, P., & Amano, R. (2000). Credibility and monetary policy. Bank of Canada
Review Spring., pp. 1117.
Rudebusch, G. D. (2005). Assessing the Lucas critique in monetary policy models.
Journal of Money, Credit and Banking, 37, 245272.
Rudd, J., & Whelan, K. (2005). New tests of the New Keynesian Phillips Curve. Journal
of Monetary Economics, 52, 11671181.
Schmidt-Hebbel, K., & Werner, A. (2002). Ination targeting in Brazil, Chile,
and Mexico: Performance, credibility, and the exchange rate. Economica, 2,
3189.
Sbordone, A. M. (2005). Do expected future marginal costs drive ination dynamics.
Journal of Monetary Economics, 52, 11831197.
Woodford, M. (2001). The Taylor rule and optimal monetary policy. American Economic Review, 91, 232237.
Woodford, M. (2006). How important is money in the conduct of monetary policy.
Paper prepared for the fourth ECB central banking conference, November 910,
2006.