The current issue and full text archive of this journal is available at www.emeraldinsight.com/0144-3585.htm
JES 36,4
Testing the impact of inflation targeting on inflation
326
School of Economics, Finance and Marketing, RMIT University, Melbourne, Australia
George B. Tawadros
Received 10 October 2007 Accepted 18 February 2008
Abstract Purpose – The purpose of this paper is to examine the impact of inflation targeting on inflation for 27 countries that have adopted an inflation-targeting regime. Design/methodology/approach – The paper uses intervention analysis in Harvey’s structural time series model to analyse the impact of inflation targeting on inflation, using quarterly observations. This approach provides the most useful framework for separating changes that occur to a series ordinarily over time from those happening due to exogenous events identified a priori, such as inflation targeting. Findings – The empirical evidence suggests that almost all of the central banks that have pursued this strategy have been unsuccessful at controlling inflation, with the results indicating that the adoption of an inflation-targeting regime has had the perverse effect on inflation for almost every country. Practical implications – The implication of the finding is that central banks which have adopted an inflation-targeting regime do not appear to have been particularly successful in reducing inflation in any significant way, as is regularly claimed in the extant literature. Originality/value – The paper provides further evidence against the adoption of an inflation-targeting regime using an unconventional approach for 27 countries that are regarded as “fully-fledged” inflation-targeting countries. Keywords Inflation, Monetary policy, Banking Paper type Research paper
1. Introduction Since the early 1990s, an increasing number of central banks have adopted an inflation targeting framework to support the implementation of monetary policy, igniting much debate on the benefits of such a framework. As the name suggests, the approach is characterised by the announcement of official target ranges for the inflation rate at one or more time horizons, and by explicit acknowledgement that low and stable inflation is the primary goal of monetary policy[1]. Scholars and practitioners broadly agree that other important features of inflation targeting include increased communication with the general public about the objectives of monetary authorities, and some mechanism for increased accountability of the monetary authorities in achieving these objectives. Advocates who espouse the adoption of inflation targeting highlight a number of advantages it confers to the implementation of monetary policy. First, in contrast to an Journal of Economic Studies Vol. 36 No. 4, 2009 pp. 326-342 q Emerald Group Publishing Limited 0144-3585 DOI 10.1108/01443580910973556
JEL classification – E31, E52, E58 The author would like to thank Ashton J. de Silva and Richard A. Heaney for their comments and suggestions that have helped improve the quality of an earlier draft. Any remaining errors are the responsibility of the author.
exchange rate peg, inflation targeting allows monetary policy to focus on domestic economic conditions, and be able to respond to external shocks (Bernanke et al., 1999). Second, in contrast to money supply targeting, inflation targeting does not depend on a stable money demand function to predict the effect of changes in money on inflation, but rather, uses all available information to determine the best settings for the instruments of monetary policy (Genberg, 2002). Third, inflation targeting is more easily understood by the general public, and is therefore highly transparent (Mishkin, 1999)[2]. Finally, because an explicit numerical target for inflation increases the accountability of the central bank, inflation targeting has the potential to reduce the possibility that the central bank will implement time-inconsistent monetary policy. Importantly, since the source of time-inconsistency is often found in political pressures on the central bank to implement overly expansionist monetary policy, inflation targeting has the advantage of focusing the political debate on what a central bank can do in the long-run, rather than what it cannot do, through monetary policy[3]. Despite the potential advantages of adopting an inflation targeting framework, critics of this framework have noted a number of disadvantages that it creates to the implementation of monetary policy. The first is that inflation targeting is too rigid. This is usually taken to mean that the adoption of inflation targeting would force the central bank to focus only on inflation, to the detriment of output stabilisation and other central bank objectives. In other words, inflation targeting is viewed as a move by the central bank towards the pursuit of an “inflation only” goal. As Friedman (2004, p. 148) states: . . . it is not too great a leap to conjecture that one consequence of constraining the discussion of monetary policy to be carried out entirely in terms of an optimal inflation trajectory will be that concerns for real outcomes will atrophy, or even disappear from policymakers consideration altogether.
Additionally, scholars such as, inter alia, Friedman and Kuttner (1996), Friedman (2003), and Kuttner (2004), have criticised inflation targeting because they believe it imposes a rigid policy rule on the conduct of monetary policy, which does not provide the monetary authorities with enough flexibility and discretion to respond to unforeseen shocks or changes in the structure of the economy. However, commentators, such as Mishkin(1999, 2000, 2001) and Bernanke and Mishkin (1997), suggest that implementing an inflation targeting regime employs the use of policy strategies that are “rule-like” in nature, which constrain the central bank from systematically engaging in policies that have undesirable long-run consequences, thereby alleviating the time-inconsistency problem, but also allows some discretion for dealing with unforeseen circumstances. They suggest that an inflation targeting regime is one such policy strategy that subjects the monetary authorities to “constrained discretion”[4]. Second, inflation is not as easily controlled as other instruments by the monetary authorities. This can be quite problematic for a developing economy that is trying to reduce inflation from a previously high level, and so is more likely to experience large inflation forecast errors. This suggests that hard targets for inflation should be phased in only after there has been some successful deflation achieved. Waiting to harden targets after some reduction in inflation had been achieved is consistent with what inflation targeting developed economies have done. Indeed, Masson et al. (1997), Mishkin and Posen (1997), and Bernanke et al. (1999) show that for every industrialised
The impact of inflation targeting 327
JES 36,4
328
country that has become an inflation targeter, such a regime was only implemented after substantial disinflation had been previously achieved. Third, because of the long lags associated with monetary policy, inflation outcomes are revealed only after a substantial lag. As such, inflation targeting does not provide immediate signals to both the public and the markets about the stance of monetary policy. Fourth, economic theory suggests that a commitment by the central bank to reduce and control inflation should improve its credibility, and thereby reduce both inflation expectations and the output losses associated with disinflation. However, the historical experience and empirical evidence do not support this contention. Inflation expectations do not immediately adjust downwards following the adoption of inflation targeting, and there appears to be very little reduction in the sacrifice ratio among alternative countries adopting inflation targeting. An additional limitation of inflation targeting is that it cannot prevent fiscal dominance or ensure fiscal discipline. Governments can still pursue irresponsible fiscal policies with an inflation targeting regime in place. As Masson et al. (1997) show, large fiscal deficits will cause an inflation targeting regime to break down in the long run. This is because large and increasing fiscal deficits will eventually have to be monetised, or the public debt eroded by a large depreciation, subsequently leading to high inflation. As such, the absence of outright fiscal dominance is, therefore, an important prerequisite for inflation targeting, and the construction of institutions that help regulate fiscal policy, are crucial to the success of an inflation targeting regime. Judging by its popularity, the adoption of an inflation targeting regime is widely perceived as a success[5]. Despite this, the question of whether inflation targeting has had any impact on reducing inflation is an empirical one. A number of studies have attempted to analyse the effect of inflation targeting on inflation, including those by, inter alia, von Hagen (1995), Svensson (1995), Bufman et al. (1995), Bernanke et al. (1999), Corbo et al. (2002), Johnson (2002, 2003), Neumann and von Hagen (2002), Levin et al. (2004), Ball and Sheridan (2003), Pe´tursson (2004), and Levin and Piger (2004). The results obtained by these studies examining whether inflation targeting has had any impact on inflation, and inflation expectations, is rather mixed. The objective of this study is to examine the impact of inflation targeting on inflation for 27 countries using quarterly observations for the period 1980:1 to 2006:3. Apart from using a more recent sample period, this paper differs significantly from that of the extant literature. In particular, this study follows the approach implemented by Angeriz and Arestis (2006), who utilise what is known as intervention analysis in Harvey’s (1985, 1989) structural time series model. The rationale for using this methodology is provided by, inter alia, Harvey and Durbin (1986), Harvey (1989, 1996), Koopman et al. (2000), and Durbin and Koopman (2001). First, this approach provides an ideal framework for the practitioner to identify the main observable characteristics of the phenomena under study, and incorporate an explicit allowance for each characteristic using intervention analysis. Second, it provides the most useful framework for separating changes that occur to a series ordinarily over time, from those happening due to exogenous events identified a priori, such as inflation targeting. The paper is organised as follows. Section 2 provides a review of the empirical evidence of the effect of inflation targeting on inflation, while section 3 describes the econometric methodology used to assess the impact of implementing an inflation
targeting regime. Section 4 presents the empirical results of applying intervention analysis to the structural time series models, while section 5 provides some concluding remarks. 2. Inflation targeting: the evidence The adoption of inflation targeting by New Zealand in 1990 has spawned a large and growing body of research that assesses whether inflation targeting matters for controlling inflation in various countries. For instance, Bufman et al. (1995) analyse Israel’s early experiences with inflation targets, while Svensson (1995) documents the Swedish experience. Similarly, von Hagen (1995) considers the performance of the Bundesbank as an inflation targeter, albeit with a short-run monetary strategy focusing on an annual monetary target, and an inflation objective formulated for the medium term. These studies, however, suffer from the deficiency that inflation targeting had only been in operation for, at most, half a decade. In more recent times, a number of studies have analysed the impact of inflation targeting using longer time periods and more data. For instance, Corbo et al. (2002) compare the policies and outcomes in countries that have adopted inflation targeting, to those who are potential inflation targeters and those who are not. They report a number of important findings. First, Corbo et al. (2002) show that the sacrifice ratio is significantly lower for Australia, Canada and the UK, but is higher for Sweden and New Zealand. Second, they find that inflation targeting has reduced inflation uncertainty and inflation forecast errors towards that of non-targeting countries. Finally, they find that inflation persistence is reduced substantially for those countries that have adopted inflation targeting. As such, Corbo et al. (2002) conclude that output volatility, inflation volatility, and the level of inflation forecast errors have fallen in both industrialised and developing countries after they had adopted inflation targeting, reaching levels similar to those countries that have not adopted inflation targeting. Similar results are provided by Johnson (2002), who analyses the effect of inflation targeting on inflation expectations. He shows that there is a substantial reduction in private expectations of inflation levels after the announcement that an inflation targeting regime has been adopted. However, Johnson (2002) does not find any significant reduction in the inflation forecast errors of targeting countries compared to non-targeting countries. This finding is in contrast to that by Corbo et al. (2002). In a latter study, Johnson (2003) compares the actual forecasts made by professional forecasters with that of predicted forecasts obtained from a model, for five consecutive 12-month periods after the announcement of inflation targets. He finds an immediate reduction in expected inflation for Sweden and New Zealand, but a slower and smaller response for Australia and Canada. For the UK, Johnson (2003) finds that the announcement of inflation targets do not have a significant effect. Levin et al. (2004) analyse the impact of inflation targeting on the persistence of inflation and inflation expectations for five developed countries that have adopted inflation targeting, and seven that have not. They examine the univariate properties of inflation for each country using a series of regressions, and report a number of findings. First, they find that actual inflation exhibits lower persistence in inflation targeting countries, leading them to suggest that low levels of inflation persistence prevented higher levels of inflation volatility in inflation targeting countries. Second, Levin et al. (2004) find a one-year-ahead expected inflation in
The impact of inflation targeting 329
JES 36,4
330
response to actual inflation is lower, in absolute terms, for inflation targeters. Finally, they show that private sector inflation forecasts, with horizons of five to ten years, are uncorrelated with past inflation rates for developed countries that target inflation, suggesting that inflation targeting can significantly anchor long-run inflation expectations. Using a similar approach to that of Levin et al. (2004), Levin and Piger (2004) analyse twelve developed countries and eleven developing countries, and find that inflation exhibits low persistence. However, they also find that inflation targeting has had very little impact on long-term expected inflation for the developing countries. These results support the findings of Benati (2004), who studies the dynamics of inflation for 20 developed countries. According to, inter alios, Kuttner and Posen (2001), Neumann and von Hagen (2002), and Kuttner (2004), many of the studies analysing the effects of inflation targeting suffer from a number of deficiencies. The first is that, despite extensive efforts, empirical studies consistently fail to produce convincing evidence that inflation targeting directly impacts on inflation performance, policy credibility and the sacrifice ratio. After all, the environment of the 1990s, when inflation targeting was first adopted, was generally a period friendly to increased price stability, implying that the practice of inflation targeting may have done very little to improve monetary policy outcomes over what any other reasonable strategy could have achieved. This has led many practitioners, including, inter alia, Debelle (1997), Cecchetti and Ehrmann (1999), Neumann and von Hagen (2002), and Ball and Sheridan (2003), to state that inflation targeting does not matter. For instance, Debelle (1997) compares average inflation rates in seven developed countries that have adopted inflation targeting, with G7 inflation targeting countries. He finds a larger fall in inflation rates and long-term interest rates in the former group, leading him to suggest that inflation targeting might be useful for countries facing a lack of policy credibility. However, Debelle (1997) shows that unemployment also increased for the non-G7 inflation targeting countries during the period of disinflation, suggesting that a reduction in inflation did not come without cost. Furthermore, he notes that other non-inflation targeting countries achieved similar reductions in inflation during the 1990s, making it difficult to conclude that inflation targeting matters. Cecchetti and Ehrmann (1999) make the observation that inflation rates generally fell in the 1990s compared with the 1980s, leading them to conclude that it was a decade conducive to price stability rather than being a product of adopting an inflation targeting regime. Similarly, Neumann and von Hagen (2002, p. 128) conclude that they “. . . cannot confirm the superiority of IT over other monetary policy strategies geared at price stability”. In a controversial study, Ball and Sheridan (2003) continually make the damning claim that inflation targeting does not matter. They find no evidence to suggest that inflation targeting improves a country’s economic performance. The second deficiency is that, despite the lack of any conclusive evidence to support the benefits of adopting an inflation targeting regime, its proponents strongly argue that the failure to adopt it puts at risk the ability of a central bank to deliver price stability. For instance, Bernanke and Mishkin (1997) and Bernanke et al. (1999) implore the Fed to adopt inflation targeting, arguing that it is critical to secure price stability for the US in the post-Greenspan era, despite presenting rather inconclusive evidence regarding the superiority of inflation targeting. Indeed, Mankiw (2001) suggests that the Fed may have been engaged in “covert inflation targeting” during the 1990s. Similarly, Galı´ et al. (2004) state that the European Central Bank could improve its
monetary policy outcomes by adopting a version of inflation targeting, although they do not provide any supporting evidence for this claim[6]. The third deficiency refers to the observation that in a number of countries, such as Australia, Canada, Sweden and the UK, inflation had been controlled well before the adoption of inflation targeting. For instance, Cecchetti and Ehrmann (1999) study the effects of inflation targeting on 23 countries, nine of which adopted inflation targeting in the 1990s. They make the observation that inflation rates generally fell during the 1990s compared with the 1980s, independent of the geographical region, their pursuit of an inflation target, or whether they were striving to enter the European Monetary Union. As such, Cecchetti and Ehrmann (1999) conclude that the 1990s provided an environment conducive to price stability rather than being a product of adopting an inflation targeting regime. In a provocative study, Ball and Sheridan (2003) analyse the impact of inflation targeting on the macroeconomic performance of 20 OECD countries, seven of which are inflation targeters. Initially, they find that inflation fell in countries that adopted inflation targeting, and that output growth stabilised during the post-inflation targeting period compared to the pre-inflation targeting period. Similar results are obtained for inflation volatility and inflation persistence, as well as interest rates and their variability. After increased scrutiny, however, the apparent benefit disappears altogether once the effect of regression to the mean is accounted for. Ball and Sheridan (2003, p. 5) state that “Just as short people have children who are, on average, taller than they are, countries with unusually high and unstable inflation tend to see these problems diminish, regardless of whether they adopt inflation targeting. Once we control for this effect, the apparent benefits of targeting disappear”. As such, Ball and Sheridan (2003) conclude that inflation targeting does not affect the behaviour of inflation, output or interest rates, and are highly suspicious of the results reported in the extant literature. Similarly, the evidence presented by Pe´tursson (2004) for the “longest history inflation targeting countries” (LHITCs), which includes Australia, Canada, Chile, New Zealand, Sweden and the UK, reaches similar conclusions, providing weak evidence of the effects of inflation targeting on inflation. Pe´tursson (2004) attributes this result to the fact that inflation targeting countries had completed a substantial part of the disinflation process before adopting inflation targeting. He concludes that implementing this strategy would be to lock in previous successes in controlling inflation rather than to facilitate the process of disinflation. 3. Econometric methodology The structural time series model of Harvey (1985, 1989) is used to assess the impact of inflation targeting, using a technique originally proposed by Box and Tiao (1975), known as intervention analysis. This model may be written as: yt ¼ mt þ wt þ gt þ avt þ 1t
ð1Þ
where yt is the actual observed inflation rate, mt is the trend component, wt is the cyclical component, gt is the seasonal component, vt is the intervention variable, a is its coefficient, and 1t is the irregular component. The trend, cyclical and seasonal components are assumed to be uncorrelated, while 1t is assumed to be white noise. The trend component, which represents the long-term movement of a series, is assumed to be stochastic and linear. This component can be represented by the following equations:
The impact of inflation targeting 331
JES 36,4
332
mt ¼ mt21 þ bt21 þ ht
ð2Þ
bt ¼ bt21 þ zt
ð3Þ
where ht~NIDð0; s2h Þ, and zt~NIDð0; s2h Þ. The trend component, mt , is a random walk with a drift factor, bt , which follows a first-order autoregressive process represented by equation (3). This process collapses to a simple random walk with drift if s2z ¼ 0, and to a deterministic linear trend if s2h ¼ 0 as well. If, however, s2h ¼ 0 while s2z – 0, the process will have a trend that changes relatively smoothly. The terms mt and bt represent the level and the slope of the trend component respectively, and are equivalent to a constant term and the coefficient on a deterministic time trend in a conventional regression model. The cyclical component, which is assumed to be a stationary linear process, may be represented by:
wt ¼ a cos ut þ b sin ut
ð4Þ
where t is time, and the amplitude of the cycle is given by ða 2 þ b 2 Þ1=2 . To make the cycle stochastic, the parameters a and b are allowed to evolve over time, while continuity is preserved by writing down a recursion for constructing wt before introducing the stochastic components. By introducing disturbances and a damping factor, we obtain: *
wt ¼ rðwt21 cos u þ wt21 sin uÞ þ vt *
*
ð5Þ *
wt ¼ rð2wt21 sin u þ wt21 cos uÞ þ vt *
ð6Þ
v*t
where w appears by construction, such that vt and are uncorrelated white noise disturbances with variances s2v and s2 * , respectively. The parameters 0 # u # p and v 0 # r # 1 are the frequency of the cycle and the damping factor on the amplitude, respectively. In order to make numerical optimisation easier, the constraint s2v ¼ s2 * v is imposed. Although there are a number of different specifications to model the seasonal component, the trigonometric specification is the one most preferred. Formally, this can be written as:
gt ¼
s=2 X
ð7Þ
gj;t
j¼1
where s is the number of seasons per year (four for quarterly data), and gj;t is given by: *
gj;t ¼ gj;t21 cos lj þ g j;t21 sin lj þ kj;t
ð8Þ
*
ð9Þ
*
*
g j;t ¼ 2gj;t21 sin lj þ g j;t21 cos lj þ kj;t where j ¼ 1; :::; s=2 2 1, lj ¼ 2pj=s and:
gj;t ¼ 2gj;t21 þ kj;t
ð10Þ
*
where j ¼ s=2, kj;t , NIDð0; s2k Þ and kj;t , NIDð0; s2 * Þ. In order to make numerical k optimisation easier, the constraint s2k ¼ s2 * is imposed. k The intervention variable, vt , is taken to be a step variable, defined as: ( 0; t , t vt ¼ ð11Þ 1; t $ t at the time of implementing an inflation targeting regime, t ¼ t. This specification for the intervention variable assumes that the underlying level of the series presents a sustained constant change after the adoption of inflation targeting, and is particularly relevant, given that all the countries that have adopted inflation targeting have not abandoned it over the period of analysis. The extent to which the trend, cyclical and seasonal components evolve over time depends on the values of s2h , s2z , s2v , s2k , s21 , u, and r, which are known as the hyperparameters. The hyperparameters and the components can be estimated by maximum likelihood once the model has been written in state space form. The state space representation of equation (1) is given by: yt ¼ Z ‘t At þ aW t þ 1t
ð12Þ
At ¼ Bt At21 þ vt
ð13Þ
Equation (12) is the measurement equation, where Z t is an m £ 1 fixed vector, At is an m £ 1 unobservable state vector, and W t is an m £ 1 vector, the first element of which is the intervention variable, vt , defined as in equation (11), while the remaining elements are zeros. The maximum likelihood estimator of the scalar parameter, a, is the appropriate estimator of the effect of the intervention when the information in the control groups is taken into account. Equation (13) is the transition equation with a non-stochastic Bt matrix of dimension m £ m, and an m £ 1 vector of disturbances, vt , such that vt~NIDð0; Mt Þ. Once the model is written in state space form, parameter estimates can be obtained by maximum likelihood, where the Kalman filter is used to update the unobserved components. If at21 is an estimate of At21 , with Rt21 as its covariance matrix, then the linear projections (with minimum mean square error) of at and Rt at time t 2 1 are given by: atjt21 ¼ Bat21
ð14Þ
Rtjt21 ¼ BRt21 B ‘ þ M t
ð15Þ
and
The Kalman filter updates atjt21 with the new information set contained in yt , by a process described by the following equations: at ¼ atjt21 þ Rtjt21 Z t yt 2 Z ‘t atjt21 =kt ð16Þ Rt ¼ Rtjt21 2 Rtjt21 Z t Z ‘t Rtjt21 =kt
ð17Þ
The impact of inflation targeting 333
JES 36,4
334
where kt ¼ Z ‘t Rtjt21 Z t þ s21 . Equation (16) states that the predictor, atjt21 , is updated by incorporating the prediction error, yt 2 Z ‘t atjt21 , weighted by Rtjt21 Z t =kt , the Kalman gain. Similarly, equation (17) states that Rt , the covariance matrix, is updated so that its new value is equal to the old value minus Z ‘t Rtjt21 , weighted by the Kalman gain, Rtjt21 Z t =kt . More efficient estimates can be obtained by estimating the individual equations jointly. This can be done by using the time series equivalent of Zellner’s (1963) seemingly unrelated regressions technique, known as seemingly unrelated time series equations (SUTSE). In this case, equation (1) may be written as: yt ¼ mt þ wt þ gt þ Fwt þ 1t
ð18Þ
where yt is a vector of prices, mt is a vector of trends, wt is a vector of cyclical components, gt is a vector of seasonal components, wt is a vector of interventions, and F is a matrix of coefficients. As such, ht , zt , vt , kt and 1t are vectors of disturbance terms, which are assumed to be serially and mutually uncorrelated, with zero expected values and covariance matrices given by Vh , Vz , Vv , Vk and V1 , respectively. In this model, different series represented by the vector yt are linked through the correlations of the disturbances driving the components, thus increasing the efficiency of the estimation. For more details about the multivariate model that utilises intervention analysis with control groups, see Harvey and Durbin (1986), Harvey (1989, 1996), and Koopman et al. (2000). 4. Empirical results The estimation technique described in the previous section is applied to twenty-seven countries that are regarded as fully-fledged inflation targeting countries since the early 1990s, but with different adoption dates. As in Angeriz and Arestis (2006), the USA and the European Union are chosen as the control groups, thereby producing 27 multivariate time series models, one for each inflation targeting country. Quarterly data on the consumer price index (CPI), for the period 1980:1 to 2006:3, has been used for the following countries: Australia, Brazil, Canada, Chile, Colombia, the Czech Republic (C. Republic), Finland, Hungary, Iceland, Indonesia, Israel, South Korea, Mexico, New Zealand (N. Zealand), Norway, Peru, the Philippines, Poland, Romania, the Slovak Republic (S. Republic), South Africa (S. Africa), Spain, Sweden, Switzerland, Thailand, Turkey, the UK, the USA, and the European Union[7]. As such, EU ‘ ; p is a vector of 3 £ n, where i stands for country i. yt ¼ pit ; pUS t t Tables I-III report the results of estimating the SUTSE models represented by equation (17). They present the estimated hyperparameters s2z , s2v , s2k and s21 , together with their respective q-ratios in parenthesis[8,9]. The estimated components of the state vector, mt , bt and wt , are also presented, along with their associated t-statistics in square parenthesis. Tables I-III also present a number of diagnostic tests, as well as measures of the goodness of fit. The goodness of fit is measured by the standard error, s~, and the modified coefficient of determination, R2S . The statistic R2S is the coefficient of determination calculated on the basis of the seasonal mean, which is more appropriate for seasonal data than the conventional R 2 . The measure of s~ is the standard error of the estimated equation, calculated as the square root of the one-step ahead predictive error variance.
t-value p-value
u r s~ R2S N(2) Q(11,6) H(34) DOV
wt
bt
q-ratio s2z q-ratio s2z (q-ratio) s2z q-ratio mt
s2z
1.141 £ 10 21 7.444 £ 102 7 20.065 1.023 £ 102 8 (9.000 £ 102 4) 7.418 £ 102 6 20.65 4.798 * [1796.000] 1.132 £ 102 2 * [2.711] 2.400 £ 102 4 [0.239] 1.432 0.732 6.395 £ 102 3 0.513 19.959 * 6.222 0.96 1993:1 7.882 £ 102 3 21.382 [0.170]
25
Australia 22
2.211 £ 10 21 2.317 £ 102 5 (1.000 £ 102 3) 0 0 0 0 5.077 * [406.120] 3.262 £ 102 2 [0.219] 9.853 £ 102 5 [0.016] 1.331 0.621 0.145 0.764 35.336 * 19.051 * 0.179 1999:2 2.164 £ 102 2 20.201 [0.841]
Brazil 22
1.653 £ 10 20.45 3.670 £ 102 2 21 1.251 £ 102 4 (3.400 £ 102 3) 1.507 £ 102 2 20.411 114.860 * [464.710] 0.545 * [2.936] 29.272 £ 102 2 [0.397] 0.83 0.671 0.419 0.237 8.849 * 7.592 4.019 * 1991:1 1.529 * 24.188 [0.000]
Canada 24
1.118 £ 10 21 3.401 £ 102 6 (3.040 £ 102 2) 1.083 £ 102 7 (1.000 £ 102 3) 3.607 £ 102 5 20.323 4.776 * [753.740] 1.154 £ 102 2 [0.925] 5.292 £ 102 5 [0.022] 1.482 0.741 0.017 0.477 31.770 * 28.867 * 0.07 1991:1 23.150 £ 102 2 * 22.153 [0.034]
Chile 2.561 £ 10 21 1.646 £ 102 5 20.643 2.253 £ 102 7 (8.800 £ 102 3) 2.395 £ 102 6 (9.350 £ 102 2) 4.967 * [957.260] 1.207 £ 102 2 [1.797] 1.549 £ 102 3 [0.320] 1.321 0.707 0.011 0.687 14.024 * 11.716 0.294 1999:3 29.389 £ 102 4 (9.667 £ 102 2) [0.923]
25
Colombia 1.439 £ 10 2 5 21 5.470 £ 102 6 2 0.38 6.629 £ 102 8 (4.600 £ 102 3) 3.363 £ 102 6 2 0.234 4.748 * [1360.300] 6.916 £ 102 3 [0.164] 2 3.045 £ 102 4 [0.102] 1.365 0.776 7.727 £ 102 3 0.481 15.747 * 5.624 0.288 1998:1 3.016 £ 102 2 * 4.407 [0.000]
C. Republic 26
3.835 £ 10 21 1.013 £ 102 6 20.264 1.076 £ 102 7 (2.800 £ 102 2) 3.949 £ 102 7 20.103 4.560 * [2586.000] 3.816 £ 102 3 [1.527] 27.296 £ 102 4 [0.530] 1.152 0.808 4.040 £ 102 3 0.792 0.295 9.654 0.599 1993:1 4.954 £ 102 3 1.544 [0.127]
Finland 9.231 £ 10 21 3.239 £ 102 2 20.351 4.403 £ 102 4 (4.800 £ 102 3) 1.008 £ 102 3 (1.090 £ 102 2) 138.570 * [554.050] 1.934 * [5.021] 0.680 * [2.765] 1.425 0.815 0.593 0.712 3.755 10.707 21.050 * 2001:3 20.460 20.95 [0.345]
22
Hungary
2.961 £ 102 4 20.1 1.364 £ 102 5 (4.610 £ 102 2) 0 0 2.972 £ 102 5 20.1 4.884 * [705.310] 2.283 £ 102 2 [0.229] 1.026 £ 102 3 [0.241] 1.415 0.701 0.022 0.731 54.192 * 22.470 * 0.061 2001:1 21.307 £ 102 2 20.707 [0.481]
Iceland
The impact of inflation targeting 335
Table I. Maximum likelihood estimates of the intervention models
t-value p-value
u r s~ R2S N(2) Q(11,6) H(34) DOV
wt
bt
s2z q-ratio s2z q-ratio s2z q-ratio s2z q-ratio mt
1.395 £ 102 3 21 8.765 £ 102 5 (6.280 £ 102 2) 1.201 £ 102 7 (1.000 £ 102 4) 7.619 £ 102 5 (5.460 £ 102 2) 4.715 * [309.470] 21.373 102 4 [0.997] 7.572 102 5 [0.006] 1.47 0.701 0.048 0.819 60.526 * 19.035 * 0.03 1992:1 3.048 102 3 (7.680 102 2) [0.939]
5.512 £ 102 4 21 9.311 £ 102 6 (1.690 £ 102 2) 2.095 £ 102 7 (4.000 £ 102 4) 4.986 £ 102 6 (4.610 £ 102 2) 5.181 * [926.460] 1.316 £ 102 2 [0.594] 2 1.981 £ 102 3 [0.474] 1.303 0.72 0.026 0.257 152.570 * 10.302 3.216 * 2005:3 2 3.976 £ 102 2 * 2 2.034 [0.045]
Table II. Maximum likelihood estimates of the intervention models Israel 7.814 £ 102 2 21 6.683 £ 102 2 2 0.855 7.259 £ 102 5 (9.000 £ 102 4) 3.471 £ 102 2 2 0.444 121.140 * [541.830] 0.777 * [2.625] 2.598 102 2 [0.207] 1.232 0.654 0.643 0.454 11.587 * 6.062 1.364 1998:1 4.006 * 2 9.135 [0.000]
Korea 8.454 £ 102 4 21 0 0 2.497 £ 102 7 (3.000 £ 102 4) 1.404 £ 102 6 (1.700 £ 102 3) 4.887 * [1402.900] 2.016 102 2 [0.496] 0 [0.000] 1.374 0.72 0.029 0.799 65.917 * 8.732 0.073 1999:1 7.153 102 3 20.332 [0.741]
Mexico 0 0 6.834 21 0 0 0.189 (2.770 £ 102 2) 133.300 * [28.166] 0.245 [0.542] 215.732 * [3.320] 0.14 0.959 2.748 20.019 473.700 * 2.284 23.707 * 1990:1 1.226 20.416 [0.678]
N. Zealand 1.236 £ 102 2 20.194 6.378 £ 102 2 21 1.961 £ 102 4 (3.100 £ 102 3) 3.466 £ 102 2 20.544 111.870 * [385.670] 0.653 * [3.513] 0.124 [0.472] 0.889 0.604 0.513 0.164 89.836 * 6.408 4.223 * 2001:1 0.796 21.693 [0.094]
Norway
1.677 £ 102 2 21 1.813 £ 102 3 20.108 0 0 7.764 £ 102 3 20.463 4.711 * [50.862] 22.149 £ 102 2 [0.137] 3.060 £ 102 3 [0.070] 1.395 0.661 0.226 0.487 74.647 * 23.756 * 0.051 2002:1 4.463 £ 102 2 20.221 [0.826]
Peru
2.192 £ 102 4 21 0 0 8.573 £ 102 8 (4.000 £ 102 4) 2.544 £ 102 5 20.116 4.935 * [950.670] 9.595 £ 102 3 [0.553] 0 [0.000] 1.378 0.722 0.018 0.439 38.151 * 18.195 * 0.259 2002:1 28.053 £ 102 4 20.053 [0.958]
Philippines
336
Indonesia
8.647 102 3 21 1.224 £ 102 6 (1.000 £ 102 4) 0 0 0 0 4.773 * [680.540] 3.373 £ 102 2 [0.361] 22.771 £ 102 5 [0.022] 1.353 0.633 0.091 0.543 209.440 * 25.429 * 0.155 1998:3 1.302 £ 102 2 20.194 [0.847]
Poland
JES 36,4
t-value p-value
u r s~ R2S N(2) Q(11,6) H(34) DOV
wt
bt
q-ratio s2z q-ratio s2z q-ratio s2z q-ratio mt
s2z
23
4.335 £ 10 21 0 0 0 0 1.318 £ 102 4 (3.040 £ 102 2) 5.525 * [506.260] 2.310 £ 102 2 [0.731] 27.181 £ 102 4 [0.507] 1.533 1 0.065 0.668 31.036 * 7.01 0.166 2005:3 1.335 £ 102 2 20.265 [0.792]
Romania 25
2.366 £ 10 21 6.136 £ 102 6 2 0.259 1.126 £ 102 6 (4.760 £ 102 2) 1.975 £ 102 5 2 0.835 4.940 * [730.390] 1.145 £ 102 2 [1.681] 2 9.039 £ 102 5 [0.020] 0.82 0.877 0.013 0.119 19.085 * 12.338 2.502 * 2005:1 2 2.158 £ 102 2 2 1.834 [0.073]
S. Republic 25
3.392 £ 10 21 7.777 £ 102 6 20.229 7.391 £ 102 9 0 2.947 £ 102 6 (8.690 £ 102 2) 4.908 * [1264.800] 1.861 £ 102 2 * [2.665] 1.619 £ 102 3 [0.462] 1.338 0.693 9.415 £ 102 3 0.471 0.682 24.530 * 1.179 2001:1 1.648 £ 102 3 20.205 [0.838]
S. Africa 3.963 £ 21 2.081 £ 20.525 3.088 £ (7.800 £ 3.648 £ 20.921 4.547 * [1844.800] 5.127 £ [1.855] 21.085 £ [0.548] 1.399 0.799 5.345 £ 0.711 3.315 8.642 0.411 1994:4 1.104 £ 20.023 [0.982]
Spain
28
26
102 4
102 3
102 3
102 3
10 102 3) 102 6
10
10
26
22
6.473 £ 10 21 3.498 £ 102 2 20.54 3.746 £ 102 4 (5.800 £ 102 3) 2.906 £ 102 2 20.449 109.370 * [224.730] 0.446 [1.705] 2.489 £ 102 2 [0.055] 0.304 0.857 0.567 0.454 32.567 * 14.673 * 1.675 1993:1 2.147 * 24.798 [0.000]
Sweden 1.971 £ 10 21.053 1.872 £ 102 2 21 1.025 £ 102 3 (5.480 £ 102 2) 1.903 £ 102 3 20.102 105.770 * [532.090] 0.312 * [1.582] 5.895 £ 102 3 [0.034] 0.863 0.674 0.374 0.424 0.133 7.578 0.513 2001:1 0.247 20.788 [0.433]
22
Switzerland 2.427 £ 21 2.921 £ 20.12 5.381 £ (2.200 £ 6.427 £ 20.265 4.767 * [1429.600] 5.713 £ [0.964] 9.077 £ [0.413] 1.394 0.758 8.465 £ 0.243 1.818 16.305 * 0.741 2000:2 28.023 £ 21.133 [0.260] 26
25
102 3
102 3
102 4
102 3
102 8 102 3) 102 6
10
10
Thailand 24
6.174 £ 10 21 4.367 £ 102 5 (7.070 £ 102 2) 5.294 £ 102 8 (1.000 £ 102 4) 3.323 £ 102 4 2 0.538 5.910 * [314.080] 3.673 £ 102 2 [1.184] 9.559 £ 102 4 [0.113] 1.425 0.786 0.044 0.377 20.453 * 16.038 * 0.463 2006:1 2 2.312 £ 102 2 2 0.565 [0.574]
Turkey
3.289 £ 102 2 21.45 2.268 £ 102 2 21 4.935 £ 102 5 (2.200 £ 102 3) 1.278 £ 102 3 (5.630 £ 102 2) 116.830 * [618.420] 1.076 * [4.584] 0.168 [0.924] 0.929 0.752 0.337 0.304 2.113 15.020 * 1.027 1992:4 0.35 21.232 [0.221]
UK
The impact of inflation targeting 337
Table III. Maximum likelihood estimates of the intervention models
JES 36,4
338
The diagnostic test statistics reported include the Ljung and Box (1978) QðP; d Þ statistic based on the first P residual autocorrelations, and is approximately distributed as x2d , and the Doornik and Hansen (1994) N ðd Þ statistic for normality based on the third and fourth moments of the distribution, and is approximately distributed as x22 under the null hypothesis. The last diagnostic test statistic reported is the simple non-parametric H ðhÞ test used to detect the presence of heteroscedasticity. It is calculated as the ratio of the squares of the last h residuals to the squares of the first h residuals, where h is the closest integer to one-third of the sample size, and is approximately distributed as Fðh; hÞ[10]. The results show that the estimated SUTSE models perform very well, as indicated by the measures of the goodness of fit. All models appear to be well determined, except that for New Zealand, which has the largest standard error, and an R2S of almost zero, implying that the model is no better than a seasonal random walk model. The estimated models also perform reasonably well based on the battery of diagnostic tests, with only 12 countries (Brazil, Chile, Iceland, Israel, Peru, the Philippines, Poland, South Africa, Sweden, Thailand, Turkey and the UK) failing the test for serial correlation, suggesting that the model is misspecified for these countries. Similarly, only six countries (Cananda, Hungary, Indonesia, New Zealand, Norway and the Slovak Republic) fail the test for heteroscedasticity. However, only seven countries (Finland, Hungary, South Africa, Spain, Switzerland, Thailand and the UK) pass the test for normality. Furthermore, only three countries (Finland, Spain and Switzerland) pass all three tests for normality, serial correlation and heteroscedasticity. The results also show that the level of the trend is significantly positive in all cases, whereas the slope is significant for eight countries (Australia, Canada, Hungary, Korea, Norway, South Africa, Switzerland and the UK), suggesting that all price series are at least I(1), and may even be I(2) for these countries. Also reported in Tables I-III are the dates that all 27 countries had adopted an inflation targeting regime (DOV), along with the coefficients on the intervention variable and their levels of significance. The 27 countries are separated into four categories. The first category includes Australia, Brazil, Finland, Israel, Mexico, New Zealand, Norway, Peru, Poland, Romania, South Africa, Spain, Switzerland and the UK. In these countries, the coefficient on the intervention variable is positive, but not significant at the 5 per cent level. This indicates that the introduction of inflation targeting had a perverse effect on inflation, pushing the already decreasing trend in inflation upwards rather than downwards. A second group, consisting of Canada, South Korea, Sweden and the Czech Republic, shows a significant perverse effect on inflation. For these four countries, the adoption of inflation targeting led to a significant rise, rather than a fall, in inflation, as shown by a significantly positive coefficient on the intervention variable. A third group of countries strongly supports the contention that inflation targeting was introduced well after inflation had begun on its downward trend. The impact of adopting inflation targeting on inflation is negative for Colombia, Hungary, Iceland, the Philippines, the Slovak Republic, Thailand and Turkey. However, the movement downwards in inflation is already in progress well before the introduction of inflation targeting. This is evidenced by the fact that, although the coefficient on the intervention variable is negative, it is, nonetheless, insignificant.
The fourth group consists of the only two countries that may be considered as “successes”. These are Chile and Indonesia. It is worth noting that these are the only two cases where the coefficient of the intervention variable is both negative and significant at the 5 per cent level. It is also worth pointing out that in the case of Indonesia, the coefficient on the intervention variable is only marginally significant, and could very well be included in the third category. Despite the exercise just attempted, we may conclude that inflation targeting appears to have been introduced after the countries under investigation had already controlled inflation and its variability. The adoption of inflation targeting does not appear to have had a significant effect on inflation. 5. Concluding remarks In this paper, intervention analysis with a control group was utilised in Harvey’s (1985, 1989) multivariate structural time series model to assess the empirical performance of inflation targeting for 27 countries. The empirical evidence obtained would suggest that central banks who have adopted an inflation targeting regime do not appear to have been particularly successful in reducing inflation in any significant way, as is regularly claimed in the extant literature. Based on the evidence obtained in this study, the often made claim that the reduction of inflation in countries that have adopted an inflation targeting regime is far greater than would otherwise have been, is not supported. The initial impact on inflation of implementing an inflation targeting strategy is negative for nine countries, and significant for only Chile and Indonesia, suggesting that inflation targeting may have been successful in controlling inflation for only these two countries. For these nine countries, the results show that the adoption of an inflation targeting strategy was introduced well after inflation had begun its downward trend. On the other hand, the impact of inflation targeting on inflation is positive for 18 countries, and significant in four of these cases, suggesting that such a strategy had a perverse effect on inflation. This result indicates that the adoption of an inflation targeting regime has had a perverse effect in most cases. Notes 1. See, for instance, Bernanke and Mishkin (1997), Mishkin (1999), and Bernanke et al. (1999). 2. de Mendonc¸a and Filho (2007) analyse the impact of greater central bank transparency on the effectiveness of monetary policy, as well as the effect of greater transparency on accountability, central bank credibility, the average level of inflation, and the output gap. They find that an improvement in the clarity of information results in a significant change in the rate of readjustment of market expectations, and helps anchor the public’s expectation of long-run inflation. 3. See, for instance, Bernanke and Mishkin (1997), Mishkin (2000), and Genberg (2002). 4. See, for instance, Akay and Nargelecekenler (2007). 5. Sterne (2002), for instance, documents that 54 countries pursued some form of inflation targeting by 1998, compared to only 6 in 1990. However, Rose (2007) provides an annotated listing of 27 countries that are considered as “fully-fledged” inflation targeters, having more than 150 years of collective experience with implementing an inflation targeting regime.
The impact of inflation targeting 339
JES 36,4
340
6. In a recent study, Domac and Kandil (2002) assess the performance and practicalities of nominal income targeting in Germany, and its potential adoption by the European Central Bank, given Germany’s successful experience with implementing a nominal income targeting regime. They conclude that the scepticism regarding the use of a monetary aggregate as an intermediate target, which arises mainly from the US experience, may be attributed to the timing and deregulation of financial markets. As such, Domac and Kandil (2002) suggest that the monetary policy framework implemented by the Bundesbank should serve as a template for the implementation of monetary policy by the European Central Bank. 7. For Finland and Spain, the period of estimation is 1980:1 to 1998:2. 8. The inclusion of a stochastic trend fails to achieve convergence in many instances. As such, a deterministic trend and a stochastic slope (s2h ¼ 0 and s2z – 0) are chosen for all countries to achieve convergence. 9. The q-ratio of a hyperparameter is obtained by dividing its value by the value of the largest estimated hyperparameter. It indicates the relative significance, or magnitude, of the hyperparameter. 10. For Finland and Spain, h ¼ 23 so that the H(h) test is distributed as F(23,23).
References Akay, H.K. and Nargelecekenler, M. (2007), “Is there the time-inconsistency problem in Turkey?”, Journal of Economic Studies, Vol. 35 No. 5, pp. 389-400. Angeriz, A. and Arestis, P. (2006), “Has inflation targeting had any impact on inflation?”, Journal of Post-Keynesian Economics, Vol. 28 No. 4, pp. 559-71. Ball, L. and Sheridan, N. (2003), “Does inflation targeting matter?”, NBER Working Paper No. 9577, National Bureau of Economic Research, Cambridge, MA. Benati, L. (2004), “International evidence on inflation persistence”, mimeo, Bank of England Working Paper Series, Bank of England, London. Bernanke, B.S. and Mishkin, F.S. (1997), “Inflation targeting: a new framework for monetary policy?”, Journal of Economic Perspectives, Vol. 11 No. 2, pp. 97-116. Bernanke, B.S., Laubach, T., Mishkin, F.S. and Posen, A.S. (1999), Inflation Targting: Lessons from the International Experience, Princeton University Press, Princeton, NJ. Box, G.E.P. and Tiao, G.C. (1975), “Intervention analysis with applications to economic and environmental problems”, Journal of the American Statistical Association, Vol. 70 No. 349, pp. 70-9. Bufman, G., Leiderman, L. and Sokoler, M. (1995), “Israel’s experience with explicit inflation targets: a first assessment”, in Leiderman, L. and Svensson, L.E.O. (Eds), Inflation Targets, CEPR, London, pp. 169-91. Cecchetti, S.G. and Ehrmann, M. (1999), “Does inflation targeting increase output volatility? An international comparison in policymakers’ preferences and outcomes”, NBER Working Paper No. 7426, National Bureau of Economic Research, Cambridge, MA. Corbo, V., Landerretche, O´. and Schmidt-Hebbel, K. (2002), “Does inflation targeting make a difference?”, in Loayza, N. and Soto, R. (Eds), Inflation Targeting: Design, Performance, Challenges, Central Bank of Chile, Santiago, pp. 221-69. de Mendonc¸a, H.F. and Filho, J.S. (2007), “Economic transparency and effectiveness of monetary policy”, Journal of Economic Studies, Vol. 34 No. 6, pp. 497-514.
Debelle, G. (1997), “Inflation targeting in practice”, IMF Working Paper No. 97/35, International Monetary Fund, Washington, DC. Domac, I. and Kandil, M. (2002), “On the practice and practicality of nominal GDP targeting in Germany”, Journal of Economic Studies, Vol. 29 No. 3, pp. 179-204. Doornik, J.A. and Hansen, H. (1994), “An omnibus test for univariate and multivariate normality”, Nuffield College Working Paper No. W4&91, University of Oxford, Oxford. Durbin, J. and Koopman, S.J. (2001), Time Series Analysis by State Space Methods, Oxford University Press, Oxford. Friedman, B.M. (2003), “The use and meaning of words in central banking: inflation targeting, credibility, and transparency”, in Mizen, P. (Ed.), Central Banking, Monetary Theory and Practice: Essays in Honour of Charles Goodhart, Vol. 1, Edward Elgar Publishing, Cheltenham, pp. 111-24. Friedman, B.M. (2004), “Is inflation targeting best-practice monetary policy?”, The Federal Reserve Bank of St Louis Review, Vol. 86 No. 4, pp. 145-50. Friedman, B.M. and Kuttner, K.N. (1996), “A price target for US monetary policy? Lessons from the experience with money growth targets”, Brookings Papers on Economic Activity, Vol. 1, pp. 77-146. Galı´, J., Gerlach, S.H.M., Rotemberg, J.J., Uhlig, H. and Woodford, M. (2004), The Monetary Policy Strategy of the ECB Reconsidered: Monitoring the European Central Bank, CEPR, London. Genberg, H. (2002), “Inflation targeting the Holy Grail of monetary policy?”, Journal of Economic Policy Reform, Vol. 5 No. 3, pp. 161-71. Harvey, A.C. (1985), “Trends and cycles in macroeconomic time series”, Journal of Business and Economic Statistics, Vol. 13, pp. 216-27. Harvey, A.C. (1989), Forecasting, Structural Time Series Models and the Kalman Filter, Cambridge University Press, Cambridge. Harvey, A.C. (1996), “Intervention analysis with control groups”, International Statistical Review, Vol. 64 No. 3, pp. 313-28. Harvey, A.C. and Durbin, J. (1986), “The effects of seat belt legislation on British road casualties: a case study in structural time series modelling”, Journal of the Royal Statistical Society, Series A, Vol. 149 No. 3, pp. 187-227. Johnson, D.R. (2002), “The effect of inflation targeting on the behaviour of expected inflation: evidence from an 11 country panel”, Journal of Monetary Economics, Vol. 49 No. 4, pp. 1521-38. Johnson, D.R. (2003), “The effect of inflation targets on the level of expected inflation in five countries”, The Review of Economics and Statistics, Vol. 85 No. 4, pp. 1076-81. Koopman, S.J., Harvey, A.C., Doornik, J.A. and Shephard, N. (2000), STAMP 6.30: Structural Time Series Analyser, Modeller and Predictor, Timberlake Consultants Press, London. Kuttner, K.N. (2004), “The role of policy rules in inflation targeting”, The Federal Reserve Bank of St Louis Review, Vol. 86 No. 4, pp. 89-112. Kuttner, K.N. and Posen, A.S. (2001), “Does talk matter after all? Inflation targeting and central bank behaviour”, IIE Working Paper Series No. 99/10, Institute for International Economics, Washington, DC. Levin, A.T. and Piger, J.M. (2004), “The macroeconomic effects of inflation targeting”, ECB Working Paper Series No. 334, European Central Bank, Frankfurt. Levin, A.T., Natalucci, F.M. and Piger, J.M. (2004), “The macroeconomic effects of inflation targeting”, The Federal Reserve Bank of St Louis Review, Vol. 86 No. 4, pp. 51-80.
The impact of inflation targeting 341
JES 36,4
342
Ljung, G.M. and Box, G.E.P. (1978), “On a measure of lack of fit in time series models”, Biometrika, Vol. 65 No. 2, pp. 297-303. Mankiw, G.N. (2001), “US monetary policy during the 1990s?”, NBER Working Paper No. 8471, National Bureau of Economic Research, Cambridge, MA. Masson, P.R., Savastano, M.A. and Sharma, S. (1997), “The scope for inflation targeting in developing countries”, IMF Working Paper No. 97/130, International Monetary Fund, Washington, DC. Mishkin, F.S. (1999), “International experiences with different monetary policy regimes”, Journal of Monetary Economics, Vol. 43 No. 3, pp. 579-605. Mishkin, F.S. (2000), “Inflation targeting in emerging-market countries”, American Economic Review Papers and Proceedings, Vol. 90 No. 2, pp. 105-9. Mishkin, F.S. (2001), “Issues in inflation targeting”, in Crawford, A. (Ed.), Price Stability and the Long-Run Target for Monetary Policy, Bank of Canada, Ottawa, pp. 203-22. Mishkin, F.S. and Posen, A.S. (1997), “Inflation targeting: lessons from four countries”, The Federal Reserve Bank of New York Economic Policy Review, Vol. 3 No. 4, pp. 9-110. Neumann, M.J.M. and von Hagen, J. (2002), “Does inflation targeting matter?”, The Federal Reserve Bank of St Louis Review, Vol. 84 No. 4, pp. 127-48. Pe´tursson, T.G. (2004), “The effects of inflation targeting on macroeconomic performance”, Working Paper No. 23, Central Bank of Iceland, Reykjavik. Rose, A.K. (2007), “A stable international monetary system emerges: inflation targeting is Bretton Woods, reversed”, Journal of International Money and Finance, Vol. 26 No. 5, pp. 663-81. Sterne, G. (2002), “Inflation targets in a global context”, in Loayza, N. and Soto, R. (Eds), Inflation Targeting: Design, Performance, Challenges, Central Bank of Chile, Santiago, pp. 23-77. Svensson, L.E.O. (1995), “The Swedish experience of an inflation target”, in Leiderman, L. and Svensson, L.E.O. (Eds), Inflation Targets, CEPR, London, pp. 69-89. von Hagen, J. (1995), “Inflation and monetary targeting in Germany”, in Leiderman, L. and Svensson, L.E.O. (Eds), Inflation Targets, CEPR, London, pp. 107-21. Zellner, A. (1963), “Estimators for seemingly unrelated regression equations: some exact finite sample results”, Journal of the American Statistical Association, Vol. 58 No. 304, pp. 977-92. Corresponding author George B. Tawadros can be contacted at:
[email protected]
To purchase reprints of this article please e-mail:
[email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints
