QREF_CredibilityPhillipsCurve_2012
description
Transcript of QREF_CredibilityPhillipsCurve_2012
-
The Quarterly Review of Economics and Finance 52 (2012) 266 271
Contents lists available at SciVerse ScienceDirect
The Quarterly Review of Economics and Finance
j ourna l ho me pa ge: www.elsev ier .co
Moneta iew
ChristopSchool of Econo nnesbu
a r t i c l
Article history:Received 2 AuReceived in reAccepted 24 MAvailable onlin
JEL classicatioE43E52
Keywords:CredibilityInation expecMonetary poli
of mos curs. Tw
thesary po
fram cred
nive
1. Introdu
This paper uses a Phillips-curve based method to investi-gate the extent of central bank credibility in eight countries.Since Kydlaas an impoAccording t(2008) credibility also actions. Thi(1992), Berand Viegi (tral banks fact that, atary policy be allowedcontingenci
The moapplied is so-called crannouncemabruptly. Tunemploymination ra
CorresponE-mail add
eliverevision of ination expectations, thereby shifting the Phillipscurve downwards. Ultimately therefore, the test for credibilitymust involve a test of the extent to which ination expecta-tions are negatively related to changes in the short-term interest
1062-9769/$ http://dx.doi.ond and Prescott (1977), credibility has been viewedrtant ingredient in the conduct of monetary policy.o Blinder, Ehrmann, Fratzscher, De Haan, and Jansenibility helps with making disination less costly. Cred-helps the central bank gain public support for itss view is shared by, for example, Bertola and Caballerotola and Svensson (1993) and Demertzis, Marcellino,2008). However, empirical analyses suggest that cen-are not perfectly credible. This may be due to thes Lohman (1992) argues, in order to optimize mone-commitment and retain credibility, central banks must
to exercise exible policy responses to unforeseenes.st relevant place where the theory of credibility isthe Phillips curve. Blinder (2000) notes that theedibility hypothesis says that perfectly credible pre-ents of disination will reduce ination expectationsherefore if the central bank is credible relatively lowent is required to bring about a drastic fall in the
te. By implication, slight increases in the interest rate
ding author. Tel.: +27 11 717 8109; fax: +27 11 717 8081.ress: [email protected] (C. Malikane).
rate.There are at least two ways in which ination expectations
are extracted. One way relies on surveys as in Berk (1999), Aronand Muellbauer (2007), Ang, Bekaert, and Wei (2007), Arnold andLemmen (2008) and Henzel (2008), and the other extracts inationexpectations from the bond rate. In the latter case, by combin-ing the Fisher relation and the expectations theory of the termstructure, it can be shown that the bond rate contains informa-tion about future ination. Goodfriend (1993) and Mehra (1996)for example, argue that the term structure is useful in predict-ing movements of future ination rates for some periods forthe US.
In this paper, we investigate the extent to which monetary pol-icy is credible in eight countries: South Korea, South Africa, Mexico,New Zealand, Australia, Canada, the United States and the UnitedKingdom. The contribution of this paper is that it uses the Phillipscurve to extract ination expectations. The hypothesis that thispaper seeks to prove is that, if monetary policy is credible, therewill be a negative relationship between ination expectations andthe nominal interest rate, as pointed out by Blinder (2000).
The paper is structured as follows: Section 2 derives themodel that extracts ination expectations from the Phillipscurve, and uses the nominal interest rate to test for thepresence or absence of monetary policy credibility. Section
see front matter 2012 The Board of Trustees of the University of Illinois. Published by Elsevier B.V. All rights reserved.rg/10.1016/j.qref.2012.05.002ry policy credibility: A Phillips curve v
her Malikane , Tshepo Mokokamic and Business Sciences, University of the Witwatersrand, 1 Jan Smuts Avenue, Joha
e i n f o
gust 2011vised form 27 April 2012ay 2012e 1 June 2012
n:
tationscy
a b s t r a c t
The paper investigates the presence residuals from an augmented Philliprobust credibility effects across samplein the sample involving the 1990s, butof the countries do not exhibit monetadopted an explicit ination-targetingframework enhances monetary policyand the output gap are measured.
2012 The Board of Trustees of the U
ction must dm/locate /qre f
rg 2050, South Africa
netary policy credibility in eight countries by ltering theve. Two of the eight countries (US and New Zealand) exhibito countries (South Africa and the UK) exhibit credibility effectse effects disappear in the sample beginning in 2000. The restlicy credibility. Given that seven of the eight countries have
ework, we conclude that there is very weak evidence that thisibility. These results are however sensitive to how ination
rsity of Illinois. Published by Elsevier B.V. All rights reserved.
r drastic declines in ination through a downward
-
C. Malikane, T. Mokoka / The Quarterly Review of Economics and Finance 52 (2012) 266 271 267
3 provides empirical results of the tests and Section 4concludes.
2. Moneta
The objetion expectof monetaran all-encocurve, whicination, thtion. Our Phfound in Blition over anThe Phillips
t = et +
where t israte, xt reprqt is the logrency, zt is labour mark
and foodt aand t reprWe assumelagged resplence of consome expla
The NewSalido (200(2001), Lindmarginal cofor includinBardsen, JaFlaschel (20share as a cof excess dalso includeand Kennicits impact ogate in theand Svenssoreal fuel anvariables in
The speis a subjecow from tEt(t+1|t),at time t + 1formulate aof agents arforward-looGMM estimtion set is mination, inity prices. Hnot theoretto argue thlags appear
MavroeiNew Keynethat full inf
estimations, may be preferable because they are more robustto mis-specication problems. Using the Generalized EmpiricalLikelihood estimation, Martins and Gabriel (2009) also nd weak
catiow Ken et aly Rue usuillipion sion n.n empass
gap,nentcatiocatio
also is r008)
insigir (20catioe relecatioderaeve
ordoay bn raet = nente bacordoes innto anisanpectt muprovder
inaq. (1
xt1 +
t =ed iible,ges ancet raten ant rat
from expe care. Furre
and andsisteby Kordory policy and ination expectations
ctive of this paper is to construct a measure of ina-ations and to use this measure to test for the presencey policy credibility. Our method begins by specifyingmpassing empirical expectations-augmented Phillipsh includes a measure of demand pressure, import pricee labour share, money supply, fuel and food price ina-illips curve formulation is a modied version of the onender (2000) in that it adds other determinants of ina-d above ination expectations and demand pressure.
curve takes the following form:
xt1 + qt1 + zt1 + mt1 + fuelt1 + foodt1 + t,
(1)
the actual ination rate, et is the expected inationesents demand pressure measured by the output gap,
of the price of imports denominated in domestic cur-the labour share which represents cost push from theet, mt is the deviation of money supply from trend,
fuelt
re real fuel price and food price ination respectivelyesents a disturbance term that is serially uncorrelated., as noted by Rudebusch (2005), that there are inertialonses of ination to its determinants due to the preva-tracts and menu-costs. The above specication requiresnation.
Keynesian literature, e.g. Gali, Gertler, and Lpez-1), Gali, Gertler, and Lpez-Salido (2005), Woodford (2005), Sbordone (2005) among others, uses eitherst or the output gap but not both. Our justicationg both these variables can be found in Gordon (1998),nsen, and Nymoen (2004), Asada, Chen, Chiarella, and06) and Fair (2008). These authors interprete the labourost-push variable over and above standard measuresemand such as the unemployment or output gap. We
excess money supply as suggested by Ando, Brayton,kell (1992) and Mohanty and Klau (2004) to capturen the ination rate. The role of the monetary aggre-
Phillips curve is discussed by Nelson (2003), Gerlachn (2003), Ireland (2004) and Woodford (2006). Lastly,
d food price ination are specied as supply-side shock the triangle model of ination by Gordon (1998).cic way in which ination expectations are formedt of considerable debate. Two ways of specifying ethe work of New Keynesian economists. One way uses
which denotes an expectation at time t of ination, based on the information set t. Another way is to
hybrid specication which assumes that a fraction e backward-looking while the other fraction (1 ) isking. This leads to et = t1 + (1 )Et(t+1|t). Inations of New Keynesian Phillips Curves, the informa-ade up of the instrumental variables such as lags of
terest rate, output gap or unit labour cost and commod-owever Fair (2008) argues that the use of these lags isically appropriate. He says: To use these lags, one hasat the equation is part of a larger model in which the, but this is not very satisfying.dis (2005) further argues that the parameters of thesian Phillips Curve are weakly identied. He arguesormation methods, rather than single-equation GMM
identithe NeBardseSimilarthat arsian Phregressexpresinatio
At aencomoutputcompospecispeci(2010)(1999)Fair (2ers anand Faspecimay bspeciand mo
Howand Sbtion minatioet as compoture thand Sbbecomtaken i
Cogtion exelemenpolicy by Blinis thatwrite E
t =
wherecontainis credto chandisturbinteresinatiointeresarising
Wesistencliteratuaffect cmationFuhrerbe pernoted and Sbn, which casts serious doubts about the validity ofynesian Phillips Curve. The same result is obtained byl. (2004), on the basis of the rst-stage regression F-test.dd and Whelan (2005) show that (a) the instrumentsally used in the GMM estimations of the New Keyne-
s Curve imply that the parameters of the second-stagewill be downward-biased and (b) the reduced-formof the second-stage regression features only lagged
pirical level, Bardsen et al. (2004) nd that the all-ing model that features both labour share and the
and higher lags of ination, makes the forward-looking of the hybrid Phillips curve insignicant. Yet, thisn of the Phillips curve passes standard tests of mis-n in contrast to the NKPC. Boug, Cappelen, and Swensen
nd that the NKPC as formulated by Gali and Gertlermly rejected by the data for both the US and Euro-area.also shows that the FIML estimation of the NKPC deliv-nicant forward-looking component. Gordon (2011)08) nd that the NKPC fails to outperform traditionalns. Gordon (2011) in particular argues that the NKPCvant in high ination episodes whilst the traditionaln with only lagged ination may be relevant in stablete ination environments.r Kozicki and Tinsley (2003), Ireland (2007) and Cogleyne (2008) nd that the observed persistence of ina-e accounted for by the time-variation of underlyingther than lagged ination. In this case we may specifyt + (1 )Et(t+1|t), where t is the time-varying
of trend-ination. This specication does not fea-kward-looking component because empirically, Cogleyne (2008) nd that the backward-looking componentsignicant when time-variation of trend-ination isccount.t of this on-going debate, our paper argues that if ina-ations have a forward-looking element, then such anst be inuenced by the prevailing stance of monetaryided there is some level of credibility, as pointed out(2000). Therefore the basic assumption that we maketion expectations are time-varying. Note that we can) as follows:
qt1 + zt1 + mt1 + fuelt1 + foodt1 + t, (2)
et + t . Ination expectations in the Phillips curve aren the term t. We postulate that if monetary policy
then ination expectations contained in t respondin the short term nominal interest rate. Note that the
term t does not respond to changes in the nominal. The reason for this is that E(et t) = 0. If both expectedd the disturbance term were affected by the nominale, there would be some correlation between the two
the movement in the nominal interest rate.ect that t will exhibit some persistence. This per-n be grounded in the two explanations found in theirstly, t may be persistent because past ination maynt ination through a combination of expectations for-
overlapping wage and price contracts as noted by Moore (1995) and Gordon (1998). Secondly, t maynt because of the time-variation of trend ination asozicki and Tinsley (2003), Ireland (2007) and Cogleyne (2008). Our specication does not favour any of
-
268 C. Malikane, T. Mokoka / The Quarterly Review of Economics and Finance 52 (2012) 266 271
these explanations. Instead, it encompasses both in the sensethat embedded in t is a persistent process arising from et sinceE(ttj) = 0.
If the central bank is credible, the et component of t is expectedto respond signicantly to changes in the nominal interest rate. Fol-lowing the literature, we assume that t is persistent. This allowsus to formumonetary p
t = (L)twhere rt ithe extent tof t to chanent of thethe nomina(2008) specdriftless rancation woto unity. Acspecifying tetary policyination ennot respondinterest ratlower inat
3. Empiric
The econity involvesrst stage wSbordone (2lated. In ordparametersproposed bwe estimat295) to obtre-estimateunbiased ancurve. The s
t = xt1
+ ftwhere t is
(L) denoteuncorrelateof t as t =which testscan be viewerrors whosstep regressand Rubinfeet al. (1985,
An impoof identicMavroeidis(2009). Marcondence
1 For a myt yt1 = (xgeneralized to
for weak identication. Mavroeidis (2005) uses the so-called con-centration parameter, which is the minimum eigenvalue of theconventional concentration matrix. A concentration parameter lessthan 10 is cwe approacby Bardsen
egre for uld ility age votenneitrt-templatid to cially ecify
+
ore inbecae thaWaldientsermsriod t shi
are e of
poi and couned w
starn, ance thfter twoulork
est1 to 2h cou
the es 1 or thond the i
centrget,e, oured ud foos, foolia, So
thelia wt rate, respred ule 1
usin-squg thelate the following relationship to capture the effects ofolicy credibility on the Phillips curve:
(L)rt + t, (3)s the change in the nominal interest rate, (L) measureso which t is persistent, (L) > 0 captures the responsenges in the nominal interest rate and t is the compo-
composite shock that does not respond to changes inl interest rate. Ireland (2007) and Cogley and Sbordoneify the process that drives trend ination to follow adom walk. On the other hand, the Gordon-type speci-
uld require that the coefcients of lagged ination sumcordingly, these two strands of literature support us inhat (1) = 1. The larger is (L) the more credible is mon-. If (1) < 0, then monetary policy cannot deliver a lowvironment. In this instance, ination expectations do
in a desirable manner to movements in the nominale. That is, increases in the nominal interest rate do notion expectations.
al strategy and results
ometric strategy we employ in our test for credibil- a two-stage procedure in the extraction of t. In thee estimate Eq. (2). From Ireland (2007), Cogley and008) and Gordon (1998) we know that t is autocorre-er to generate unbiased and efcient estimates of the
of the Phillips curve, we follow the two-step procedurey Pindyck and Rubinfeld (1998, p. 590). In the rst stepe Eq. (2) using OLS in line with Judge et al. (1985, p.ain errors which we denote t . In the second step, we
Eq. (2) augmented with lags of t in order to generated efcient estimates of the parameters of the Phillipsecond-step regression takes the form:
+ qt1 + zt1 + mt1 + fuelt1ood1 + (L)t + t, (4)
the error term obtained from the rst-step regression,s the coefcients of the lags of t and t is a seriallyd error-term. From Eq. (4) we then obtain the estimate
(L)t . It is this estimate of t that we take to Eq. (3), for central bank credibility. The method outlined aboveed as 2SLS, in which the rst-step regression generatese lags are then used as instruments for t in the second-ion. Alternatively the procedure mentioned by Pindyckld (1998, p. 590) can be viewed as NLS (see also Judge
chap. 8) and Davidson and McKinnon (2004, chap. 7)).1
rtant issue with Phillips curve estimations is thatation, as brought to the fore by Mavroeidis (2004),
(2005), Bardsen et al. (2004) and Martins and Gabrieltins and Gabriel (2009) compute identication-robustsets for the parameters of the NKPC in order to test
odel yt = xt + ut with ut = ut1 + t , we can show thatt xt1) + t is the same as yt = xt + ut1 + t . This can be
models with higher-order autocorrelation.
stage rto testthis woprobabrst-st
A pendogethe shofor exatains inresponpotentrst sp
rt = r0Therefof rtest ratof the coefc
In tthe penicanresultsthe casturning(2000)of the occurrfore weGodfajeviden1989 aless it framew
The1980QFor eactest forin Tablsions fthe secculate all theas a tabill ratmeasufuel animportAustraM2 andAustrainteressupplymeasu
Tabtion bythe Chiby usinonsidered to indicate weak identication. In this paperh this question by using the simple procedure proposed
et al. (2004). In this procedure, F-statistic of the rst-ssion of the instrumental variable estimation is usedweak identication. If this F-statistic is less than 10be indicative of weak identication. We also report theof the F-statistic to check the joint signicance of theariables.tial problem with the estimation of Eq. (3) is they of rt. It is now common for central banks to userm nominal interest rate as a policy instrument (see
e, Clarida, Gali, and Gertler (2000)). Because t con-on expectations, a forward-looking central bank wouldhanges in ination expectations thereby rendering rtendogenous. In order to account for this possibility, we
a regression of the following form:
t + t. (5) running the regression for Eq. (3), we use t insteaduse t is that component of the changes in the inter-t is purged of inuences from t. Lastly, we make use-statistic to test for the signicance of the sum of the
of the interest rate change in Eq. (3). of sample selection, we estimate the equations forwhere the literature suggests there has been a sig-ft in monetary policy and we also test whether thealtered for the sample in the 2000s. For example inthe US, the literature suggests that 1979 represents ant in the conduct of monetary policy, see Clarida et al.Goodfriend and King (2005) in this regard. In the resttries a recent and signicant shift in monetary policyhen these countries adopted ination targeting. There-t the sample on the basis of the dates provided by Fraga,d Minella (2003). In the case of South Africa, there isat the country adopted implicit ination targeting inhe de Kock Commission recommendations. Neverthe-d be useful to see if the explicit adoption of this policy
strengthened central bank credibility there.imations are conducted using quarterly data from009Q4, sourced from the International Monetary Fund.ntry, we run the regressions for two sample periods to
robustness of the results to sample period. For exampleand 2 in the case of Australia, we estimate the regres-e rst sample 1996Q12009Q4 (denoted 96)*** andsample in 2000Q1 2009Q4 (denoted ****00). We cal-nation rate by using the consumer price index sinceral banks in our study have this measure of ination
the nominal short-term interest rate is the treasurytput is measured by real GDP and the output gap issing the HodrickPrescott lter. The prices for imports,d are measured using the consumer price indices ford, and fuel. In relation to money supply, we use M3 foruth Africa, New Zealand and Mexico. For Korea we use
money market rate for the short-term interest rate. Fore also used the money market rate for the short-term. For the US and the UK we used M2 and M4 for moneyectively. The deviation of money supply from trend issing the HodrickPrescott lter.presents the results for Eq. (4). We test serial correla-g the LjungBox Q(4), the DurbinWatson statistic andared LM test. We test for conditional heteroskedasticity
ARCH Chi-squared test. While the regressions exhibit
-
C. Malikane, T. Mokoka / The Quarterly Review of Economics and Finance 52 (2012) 266 271 269
Australia Canada South Korea Mexico
-.01
.00
.01
.02
.03
.04
.05
.06
97 98 99 00
.05
.05
.06
01 02
.060
.065
Uni
-.01
.00
.01
.02
.03
.04
.05
95 96 97 98 99 90 92 9
expec
joint signilevel of theidenticatioexhibit an iproblem ofperiod.
On the
t = (L)tvariable. Thination exclosely trac
Table 2 instrumentAll the regrdiagnostic tity. Of interof (1). Onlare there rois present athe one begcredibility eeffects disa2000s the Sgeting. Witcentral ban(2003) amo
Our ndconsistent and Perrierpolicy has bIn the caseacquired crand Swansocompletelyof the Fed fothat beginssistent with
n 19 the ary p
ndonsi
datat forent watioctor , werder01 02 03 04 05 06 07 08 09
Vt
Inflation
-.01
.00
.01
.02
.03
.04
94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Vt
Inflation
.00
.01
.02
.03
.04
99 00
New Zealand United Kingdom
00 01 02 03 04 05 06 07 08 09
Vt
Inflation
-.02
.00
.02
.04
.06
.08
.10
.12
86 88 90 92 94 96 98 00 02 04 06 08
Vt
Inflation
-.02
-.01
.00
.01
.02
.03
.04
.05
.06
.07
84 86 88
Fig. 1. Ination and estimated ination
cance of the variables in the rst-stage regression, the rst-stage regression F-statistic shows that there is ann problem in Australia. Only the UK and the US do notdentication problem. For the rest of the countries the
identication appears to be dependent on the sample
basis of the estimations in Table 1, we then extract, which contains ination expectations. Fig. 1 plots thise gure illustrates the signicant role that is played bypectations in the dynamics of ination. The variable tks the actual ination rate.presents estimates of Eq. (3), where t is used as an
betweeble formonet
Ouris not csurveynicanconsistthe invate sein 1998
In o
for rt in order to deal with the issue of endogeneity.essions are adequate in the sense that they pass theests, except for the UK equations in relation to normal-est in the test for credibility is the sign and signicancey for three countries; New Zealand, the UK and the USbust credibility effects. In these countries, credibilitynd signicant in the sample containing the 1990s andinning in 2000. In the case of South Africa, signicantffects are present in the 1990s sample, however theseppear in the 2000s. This is interesting because in theouth African Reserve Bank adopted explicit ination tar-h an explicit ination target, we would have expectedk credibility to be enhanced, as noted by Fraga et al.ng others.ing that monetary policy is not credible in Canada is notwith Amano, Fenton, Tessier, and Van Norden (1997)
and Amano (2000), who nd that Canadas monetaryeen credible since the adoption of ination targeting.
of the US, Goodfriend (1993) nds that the Fed hadedibility since 1983, while Grkaynak, Levin, Marder,n (2007) nd that ination expectations have not been
anchored by the Fed. We nd stronger credibility effectsr the sample beginning in 1983 than the recent sample
in 2000. For the UK and Australia our results are con- Johnson (1998), who uses survey data and nds that
tive measurre-estimatetor and outpwe nd thawith the onbased on thare reporteNeverthelesour samplemeasure of
In orderbetween ouformed twoour measurveys and ththe US, UK, (3) using suif we arrivethe three mcient betweis 0.11, betwbetween outests basedresults thatsistency be03 04 05 06 07 08 09
Vt
Inflation
.030
.035
.040
.045
.050
.055
2002 2003 2004 2005 2006 2007 2008 2009
Vt
Inflation
ted States South Africa
4 96 98 00 02 04 06 08
Vt
Inflation
-.04
-.02
.00
.02
.04
.06
.08
.10
.12
.14
95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Vt
Inflation
tations (vt).
84 and 1995 monetary policy targets were not credi-two countries. However, like Johnson (1998), we ndolicy credibility for New Zealand.ing on the credibility of monetary policy for South Africastent with Aron and Muellbauer (2007) who nd, using, that monetary policy credibility effects have been sig-
the country since 2000. Our nding for Mexico is notith Schmidt-Hebbel and Werner (2002). They nd that
n target exerts a credible and strong inuence on pri-expectations. Despite its adoption of ination targeting
nd that the Bank of Korea is not credible. to check whether these ndings are robust to alterna-
es of ination and measurement of the output gap, wed Eqs. (2) and (3) using ination based on the GDP dea-ut detrended using a polynomial time trend. Generallyt the results based on the deator are not consistentes that are contained in this paper. However, the resultse de-trended output are consistent with the results thatd in this paper, except for New Zealand (20002009).s, in view of the fact that all the countries that are in
target CPI-based ination, it is clear that an appropriate ination is the one based on the CPI.
to examine the possible sources of the inconsistencyr results and those found in the literature, we per-
procedures. Firstly, we computed correlations betweene of ination expectations and the ones based on sur-e yield curve for a selected number of economies (viz.Australia and South Africa). Secondly, we estimated Eq.rvey-based and yield curve based measures to check
at the same results. We nd weak correlation amongeasures. For example for the UK the correlation coef-en our measure and the one based on the yield curveeen the yield curve measure and surveys is 0.42, andr measure and surveys is 0.29. In terms of credibility
on the yield and survey expectations, only the US has are consistent with our measure. Therefore, the incon-tween our results and those found using alternative
-
270C.
Malikane,
T. M
okoka /
The Q
uarterly Review
of Econom
ics and
Finance 52 (2012) 266 271
Table 1Estimations of the Phillips curve (Eq. (4)) (standard errors in parentheses).
Coeff Aus Can Kor Mex NZ UK US SA
96 00 93 00 99 01 93 00 82 00 83 00 93 00
0.43 (0.13) 0.31 (0.13) 0.39 (0.08) 0.41 (0.10) 0.11 (0.05) 0.05 (0.03) 0 . 173(0.09) 0.10 (0.12) 0.87 (0.05) 0.72 (0.06) 0.23 (0.05) 0.42 (0.09) 0.61 (0.17) 0.32 (0.32) 0.004 (0.01) 0.05 (0.02) 0.03 (0.006) 0.05 (0.01) 0.01 (0.01) 0.02 (0.02) 0.07 (0.02) 0.07 (0.02) 0.12 (0.04) 0.06 (0.04) 0 . 253(0.38) 1 . 123(0.42) 0.05 (0.09) 0.06 (0.06) 0 . 012(0.03) 0.07 (0.02) 0.07 (0.03) 0.04 (0.02) 0.007 (0.009) 0.03 (0.01) 0.08 (0.05) 0.18 (0.06) 0.07 (0.03) 0.43 (0.05) 0.63 (0.09)
R2 0.90 0.70 0.60 0.63 0.72 0.81 0.77 0.47 0.93 0.88 0.89 0.92 0.87 0.972(4) 0.26 0.64 0.17 0.53 0.69 0.25 0.19 0.16 0.29 0.40 0.18 0.16 0.12 0.30Q(4) 0.78 0.31 0.93 0.36 0.19 0.60 0.93 0.71 0.87 0.96 0.68 0.68 0.77 0.57DW 1.59 1.57 2.14 2.11 1.96 1.23 1.91 2.03 2.11 1.95 2.04 2.02 1.61 2.292A(4) 0.90 0.44 0.95 0.44 0.29 0.57 0.93 0.86 0.82 0.93 0.48 0.86 0.73 0.60
JB 0.72 0.67 0.32 0.97 0.91 0.70 0.53 0.60 0.91 0.82 0.96 0.49 0.57 0.33Fprob 0.01 0.02 0.00 0.00 0.00 0.06 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00Fstat 3.60 4.47 6.74 14.19 22.72 2.44 10.34 6.84 20.86 42.53 11.63 32.96 3.91 10.71
Superscript above the estimate of the parameter denotes the lag of the variable. Probability.
Table 2Post ination targeting estimations of credibility (standard errors in parentheses).
Coeff Aus Can Kor Mex NZ UK US SA
96 00 93 00 99 01 93 00 82 00 83 00 93 00
1 1.14 (0.09) 1.45 (0.12) 1.00 (0.00) 0.68 (0.11) 1.00 (0.00) 1.00 (0.00) 0.75 (0.07) 0.96 (0.16) 1.41 (0.09) 1.43 (0.17) 1.49 (0.07) 0.87 (0.11) 1.20 (0.10) 1.22 (0.08)2 0.14 (0.09) 0.82 (0.09) 0.22 (0.08) 0.14 (0.07) 0.46 (0.13) 0.27 (0.25) 0.51 (0.13) 0.35 (0.18) 0.22 (0.08)3 0.17 (0.08) 0.16 (0.20) 0.08 (0.07) 0.59 (0.26) 0.03 (0.00) 0.36 (0.13) 0.04 (0.16)4 0.37 (0.09) 0.32 (0.11) 0.31 (0.00) 0.11 (0.00)5 0.49 (0.11)0 0.33 (0.18) 0.98* (0.23) 0.52* (0.10) 0.79* (0.23) 0.40* (0.17) 0.17** (0.08) 0.10* (0.04) 0.26* (0.12) 0.45* (0.03) 0.46* (0.13) 0.57* (0.07) 0.03* (0.01) 0.56 (0.12) 0.37 (0.25)1 0.17 (0.24) 1.03*(0.23) 0.81* (0.25) 0.30* (0.05) 0.33*(0.15) 0.40* (0.05) 1.17*(0.28)2 0.39 (0.18)(1) 0.22 (0.22) 0.05 (0.30) 0.10 (0.28) 0.15* (0.06) 0.13 (0.29) 0.17* (0.05) 0.80* (0.30)R2 0.85 0.75 0.56 0.53 0.43 0.73 0.90 0.47 0.95 0.73 0.97 0.74 0.86 0.812(4) 0.15 0.28 0.48 0.39 0.68 0.19 0.81 0.11 0.39 0.36 0.47 0.91 0.16 0.48Q(4) 0.56 0.45 0.32 0.20 0.83 0.17 0.48 0.55 0.56 0.67 0.89 0.91 0.62 0.87DW 1.59 2.26 2.14 2.09 2.4 1.93 1.74 1.88 2.04 1.97 1.79 1.82 1.63 1.542A(4) 0.69 0.39 0.30 0.22 0.67 0.75 0.27 0.34 0.63 0.79 0.91 0.56 0.63 0.91
JB 0.63 0.25 0.53 0.63 0.96 0.95 0.17 0.65 0.00 0.00 0.35 0.38 0.40 0.21
* 5% signicance.** 10% signicance. Probability.
-
C. Malikane, T. Mokoka / The Quarterly Review of Economics and Finance 52 (2012) 266 271 271
measures of expectations may have to do with the way inationexpectations are measured.2
Therefore the conclusion that we draw from these results is that,from the Phillips curve perspective, the evidence that the inationtargeting framework enhances the credibility of monetary policyis mixed. Furthermore, among the countries where there are cred-ibility effects, New Zealand appears to be the only one that hasincreased itstandard eralso observicant in thecase of the the size of tby the stand
4. Conclus
This papcurve in ordcountries. Wlate into a nnominal intnicant andin our samp
This ndpolicy, espein credibilitpose that fucurve approon surveysalternative Phillips cur
References
Amano, R., Fentary policyEconWPA
Ando, A., Braydirect effe
Ang, A., Bekaeforecast in
Arnold, I. J. M.,tainty in th144, 3253
Aron, J., & Mu1994. Jour
Asada, T., Chenwage price28, 90130
Bardsen, G., JaKeynesianment), 030
Berk, J. M. (199Economics
Bertola, G., & Cnomic Rev
Bertola, G., & St of targe
Blinder, A. S. (American E
Blinder, A. S., Ebank comNBER wor
Boug, P., Capprevisited. J
Clarida, R., Galstability: E147180.
2 Results foroutput gap arresults for the
Cogley, T., & Sbordone, A. M. (2008). Trend ination, indexation and ination per-sistence in the New Keynesian Phillips Curve. American Economic Review, 98,21012126.
Davidson, R., & Mackinnon, J. G. (2004). Econometric theory and methods. New York:Oxford Un
Demertzis, M.us moneta
Fair, R. C. (2008Fraga, A., God
omies., & M
1271 Gertlal of Mertler
EconoGertlehybrid111
S., & Sonetnd, M199nd, Metary E
R. J. (1-vary
R. J. (omicaak, R. the anrve BaS. (20casts?P. (200encesP. (20it and, D. (19urveys. G., GpracticS., & Tice ad
of Ca, F. E., &timal2005)ihood , S. (1bility.
L. F., ticatroecondis, S.
econo904dis, S.
an ap), 421. P. (1
of Ri, M. omiesE. (20al of M
, D. L., &: McGP., & ew Spch, G.al of M& Wheonetar-HebbMexic9.e, A. Mal of M
rd, M.ic Revird, M. r prep.s credibility over the two sample periods, although theror of the credibility parameter has also increased. Wee that the credibility of UK monetary policy is not signif-
2000s while it was signicant in earlier periods. In theUS, credibility seems to have declined when we look athe credibility parameter although its quality, measuredard error, has improved.
ion
er extracts ination expectations from the Phillipser to test for the credibility of monetary policy in eighte argue that the presence of credibility should trans-
egative relationship between changes in the short-termerest rate and ination expectations. We do not nd sig-
robust credibility of monetary policy in the countriesle except for the UK, US and New Zealand.ing casts serious doubt on the claims that monetarycially with the advent of ination targeting, has gainedy by way of affecting ination expectations. We pro-ture research be undertaken to reconcile the Phillipsach that we use in this paper and the one that is based. Furthermore, future research may have to exploreways in which credibility effects nd expression in theve.
ton, P., Tessier, D., Van Norden, S. (1997). The credibility of mone-: A survey of the literature with some simple application to Canada.meeting series papers no. 9610001.ton, F, Kennickell, A. B. (1992). Reappraisal of the Phillips curve andcts of money supply on ination. NBER working paper no. R1726.rt, G., & Wei, M. (2007). Do macro variables, asset markets or surveysation better? Journal of Monetary Economics, 54, 11631212.
& Lemmen, J. J. G. (2008). Ination expectations and ination uncer-e Eurozone: Evidence from survey data. Review of World Economics,46.ellbauer, J. (2007). Review of monetary policy in South Africa sincenal of African Economies, 16, 705744., P., Chiarella, C., & Flaschel, P. (2006). Keynesian dynamics and the
spiral: A baseline disequilibrium model. Journal of Macroeconomics,.nsen, E. S., & Nymoen, R. (2004). Econometric evaluation of the New
Phillips Curve. Oxford Bulletin of Economics and Statistics, 66(Supple-59049.9). Measuring ination expectations: A survey data approach. Applied, 31, 14671480.aballero, R. J. (1992). Target zones and realignments. American Eco-
iew, 82, 520536.vensson, L. E. O. (1993). Stochastic devaluation risk and the empiricalt zone models. Review of Economic Studies, 60, 689712.2000). Central bank credibility: Why do we care how do we build it?conomic Review, 90, 14211431.hrmann, M., Fratzscher, M., De Haan, J., Jansen, D. J. (2008). Central
munication and monetary policy: A survey of theory and evidence.king paper series no. 13932.elen, A., & Swensen, A. R. (2010). The New Keynesian Phillips Curveournal of Economic Dynamics and Control, 34, 858878.i, J., & Gertler, M. (2000). Monetary policy rules and macroeconomicvidence and some theory. Quarterly Journal of Economics, February,
the measure of ination based on the deator and time de-trendede available upon request. Furthermore correlation and regression
US, UK, Australia and South Africa are also available upon request.
econFuhrer, J
110,Gali, J., &
JournGali, J., G
peanGali, J.,
the 1107
Gerlach,for m
Goodfrie1979
GoodfrieMon
Gordon,time
Gordon,Econ
Grkaynand Rese
Henzel, fore
Ireland, sequ
Ireland, Cred
Johnsonon s
Judge, Gand
Kozicki, In PrBank
Kydlandof op
Lind, J. (likel
Lohmanexi
Martins,idenMac
Mavroeitary0305
Mavroeiwith37(3
Mehra, Ybank
Mohantyecon
Nelson, Journ
PindyckYork
Perrier, Revi
RudebusJourn
Rudd, J., of M
Schmidtand 318
SbordonJourn
Woodfonom
WoodfoPape2006iversity Press., Marcellino, M., Viegi, N. (2008). A measure for credibility: Trackingry developments. DNB working paper no. 187/2008.). Testing price equations. European Economic Review, 52, 14241437.
fajn, I., & Minella, A. (2003). Ination targeting in emerging market. NBER Macroeconomics Annual, 18, 365418.oore, G. (1995). Ination persistence. Quarterly Journal of Economics,59.er, M. (1999). Ination dynamics: A structural econometric analysis.onetary Economics, 44, 195222.
, M., & Lpez-Salido, J. D. (2001). European ination dynamics. Euro-mics Review, 45, 12371270.r, M., & Lpez-Salido, J. D. (2005). Robustness of the estimates of
New Keynesian Phillips curve. Journal of Monetary Economics, 52,8.vensson, L. E. O. (2003). Money and ination in the Euro-Area. A caseary indicators? Journal of Monetary Economics, 50, 16491672.. (1993). Interest rate policy and the ination scare problem:
2. Federal Reserve Bank of Richmond Economic Quarterly, 79, 124.., & King, R. G. (2005). The incredible Volker disination. Journal ofconomics, 52, 9811015.998). Foundations of the goldilocks economy: Supply shocks and theing NAIRU. Brookings Papers on Economic Activity, 2, 97346.2011). The history of the Phillips curve: Consensus and bifurcation., 78, 1050.S., Levin, A. T., Marder, A. N., & Swanson, E. T. (2007). Ination targetingchoring of ination expectations in the Western Hemisphere. Federalnk of San Francisco Economic Review, 2547.08). Learning trend ination Can signal extraction explain survey
IFO institute for economic research working papers no. 55.7). Changes in the federal reserves ination target: Causes and con-. Journal of Money, Credit and Banking, 39, 18511882.04). Moneys role in the monetary business cycle. Journal of Money,
Banking, 36, 969983.98). The credibility of monetary policy: International evidence based
of expected ination. Bank of Canada conference paper.riths, W. E., Hill, R. C., Lutkepohl, H., & Lee, C. G. (1985). The theorye of econometrics (2nd ed.). New York: John Wiley & Sons.insley, P. A. (2003). Alternative sources of lag dynamics of ination.justment and monetary Policy: Proceedings of the conference held by thenada November 2002, (pp. 347).
Prescott, E. C. (1977). Rules rather than discretion: The inconsistence plans. Journal of Political Economy, 85, 473491.. Estimating New-Keynesian Phillips Curves: A full information maximumapproach, 52, 11351149.992). Optimal commitment in monetary policy: Credibility versus
The American Economic Review, 82, 273286.& Gabriel, V. J. (2009). New Keynesian Phillips Curves and potentialion failures: A generalized empirical likelihood analysis. Journal ofomics, 31, 561571.
(2004). Weak identication of forward-looking models in mone-mics. Oxford Bulletin of Economics and Statistics, 66(Supplement),
9. (2005). Identication in forward-looking models estimated by GMM,plication to the Phillips curve. Journal of Money, Credit and Banking,448.996). The bond rate and actual future ination. The Federal reservechmond working paper 97-03.S., Klau, M. (2004). Monetary policy rules in emerging market: Issues and evidence. BIS working paper 149.
03). The future of monetary aggregates in monetary policy analysis.onetary Economics, 50, 585608.
Rubinfeld, R. S. (1998). Economic models and economic forecasts. Newraw Hill.Amano, R. (2000). Credibility and monetary policy. Bank of Canadaring., pp. 1117.
D. (2005). Assessing the Lucas critique in monetary policy models.oney, Credit and Banking, 37, 245272.lan, K. (2005). New tests of the New Keynesian Phillips Curve. Journaly Economics, 52, 11671181.el, K., & Werner, A. (2002). Ination targeting in Brazil, Chile,o: Performance, credibility, and the exchange rate. Economica, 2,
. (2005). Do expected future marginal costs drive ination dynamics.onetary Economics, 52, 11831197.
(2001). The Taylor rule and optimal monetary policy. American Eco-ew, 91, 232237.(2006). How important is money in the conduct of monetary policy.ared for the fourth ECB central banking conference, November 910,
Monetary policy credibility: A Phillips curve view1 Introduction2 Monetary policy and inflation expectations3 Empirical strategy and results4 ConclusionReferences