phd dissertation horvath

Empirical Essays on Monetary Economics

Roman Horváth

Doctoral Dissertation

2008

Institute of Economic Studies Faculty of Social Sciences, Charles University, Prague

Acknowledgments To my family; to my son David.

Contents - 3 -

Contents Introduction ................................................................................................................................ 4 1 The Time-Varying Policy Neutral Rate in Real Time: A Predictor for Future Inflation?............ 6

1.1 Introduction .................................................................................................................. 6 2 Related Literature ............................................................................................................. 9 2.1 Methodological Background .......................................................................................... 9 2.2 Methods for Natural Rate of Interest Estimation ......................................................... 11 3 Data and Empirical Methodology ................................................................................... 13 3.1 Monetary Policy Rules ................................................................................................. 13 3.2 Data............................................................................................................................. 19 4 Results ............................................................................................................................ 21 4.1 Time-varying Equilibrium Interest Rates...................................................................... 21 4.2 Monetary Policy Stance and Inflation Developments ................................................... 25 5 Conclusions .................................................................................................................... 31 References ......................................................................................................................... 33

2 Price Setting and Market Structure: An Empirical Analysis of Micro Data ............................. 43 2.1 Introduction ................................................................................................................ 43 2.2 Price Setting Behavior.................................................................................................. 44 2.3 Empirical analysis ........................................................................................................ 46 2.4 Conclusions ................................................................................................................. 62

3 Assessing Inflation Persistence: Micro Evidence on an Inflation Targeting Economy............. 80 3.1 Introduction ................................................................................................................ 80 3.2 Estimating inflation persistence ................................................................................... 84 3.3 Data............................................................................................................................. 88 3.4 Results ......................................................................................................................... 90 3.5 Conclusions ............................................................................................................... 103

4 The Effects of Monetary Policy in the Czech Republic: An Empirical Study ....................... 126 4.1 Introduction .............................................................................................................. 126 4.2 Related VAR Literature.............................................................................................. 128 4.3 Data........................................................................................................................... 133 4.4 Identification ............................................................................................................. 134 4.5 Results ....................................................................................................................... 137 4.6 Concluding Remarks.................................................................................................. 142 References ....................................................................................................................... 144

Introduction - 4 -

Introduction This dissertation contains four empirical essays on monetary economics. Each of them is self-

contained and independent of the others. Nevertheless, they all deal with various aspects of

monetary issues in the Central Europe and may be of interest outside of our region, too. More

specifically, all chapters look at the nexus of interest rates and prices and examine the impact of

interest rates on price developments or the behavior of prices itself.

Chapter 1 focuses on estimation of time-varying neutral interest rate. It applies a structural time-

varying parameter model with endogenous regressors to estimate a monetary policy rule in the

Czech Republic. Literature pointed out that the estimates of neutral interest rates are uncertain

and thus difficult to apply in daily policy work. I provide a counter-argument to this literature

showing that my estimates of neutral interest rates are able to explain future inflation

developments. The chapter is forthcoming in Economic Modelling.

Chapter 2 explores price setting behavior in Slovakia using a unique micro-level dataset about 5

millions observations. I argue that the effect of market structure on persistence of inflation

results from two conflicting forces. Increased competition may reduce persistence by increasing

the frequency of price changes. In contrast, higher competition may increase persistence through

inertial behaviour induced by the strategic complementarity among price setters. In this case

study, I find that the latter effects dominate. Chapter 2 is a joint work with Fabrizio Coricelli and

has been published as a discussion paper at the Centre for Economic Policy Research and it is

forthcoming in a special issue on price setting in the EU in Managerial and Decision Economics.

Chapter 3 studies inflation dynamics at the microeconomic level and adds further evidence in

support to the findings in Chapter 2. It discusses issues such as aggregation bias, the link between

the adoption of inflation target and inflation persistence and addresses services inflation

persistence puzzle. Chapter 3 is a joint work with Ian Babetskii and Fabrizio Coricelli and has

been published as a working paper at the Czech National Bank and CERGE-EI.

Chapter 4 analyzes the monetary policy effects within vector autoregression framework. It

documents a well-functioning monetary transmission mechanism in the Czech Republic similar

to the euro area countries, especially in terms of persistence of monetary policy shocks. Chapter 4

Introduction - 5 -

is a joint work with Magdalena Morgese Borys and working paper version is available at Czech

National Bank, CERGE-EI or at our institution. Currently, it is second revise-resubmit at

Empirica.

During writing this thesis, I benefited from numerous comments from my co-authors, colleagues

at the Czech National Bank and Charles University as well as the participants at various seminars

and conferences. I wish to thank to all of them.

Time-Varying Policy Neutral Rate - 6 -

1 The Time-Varying Policy Neutral Rate in Real Time: A Predictor for Future Inflation? *

1.1 Introduction Inflation targeting regimes are increasingly popular around the world. For example, regarding

Central and Eastern Europe, while the first two countries adopted explicit inflation targeting

regime back in 1998, there were already seven countries conducting inflation targeting in 2006

and others were contemplating doing so (International Monetary Fund, 2006).1 A characteristic

feature of inflation targeting is that central banks set the short-term nominal interest rate (path)

so as to get inflation and output to their targeted levels. When inflation and output is at their

targeted level, actual interest rate is equal to the equilibrium rate, which we label as the policy

neutral rate hereinafter.

In this regard, Woodford (2003) notes that central banks should on average track the policy

neutral rate to stabilize the economy. In a similar fashion, Taylor (1999) emphasizes that the

measurement of the policy neutral rate is one of the key issues for countries targeting inflation. In

this respect, it is of great importance for central banks to identify as precisely as possible the

policy neutral rate. This is quite an intricate exercise, as the policy neutral rate is unobservable;

however its mis-measurement is high-priced, as it may result in over- or undershooting the

inflation target.2

* I thank an anonymous referee, Juraj Antal, Ian Babetskii, Péter Benczúr, Martin Cincibuch, Woon Gyu Choi, Dana Hájková, Tomáš Holub, Anca Podpiera, Michal Ježek, Pavol Prievozník, Michal Skořepa, Guntram Wolff and the seminar participants at the Econometric Society - European Meeting, 11th International Conference on Macroeconomic Theory and Policy (Greece), DIW (Berlin), Czech National Bank, Czech Economic Society and Charles University (Prague) for helpful comments. All remaining errors are entirely my own. I am grateful to the Czech Economic Society for supporting this research. 1 The Czech Republic and Poland adopted inflation targeting in 1998, followed by Hungary in 2001, Romania and Slovakia in 2005 and Armenia and the Serbia in 2006 (note this is an updated list of Table 1 in IMF, 2006). Ukraine is likely to adopt inflation targeting in near future (IMF, 2006). 2 In case that trend productivity growth is positively correlated with policy neutral rate Trehan and Wu (2007) show that when the central bank mis-measures the trend productivity growth, it is in consequence likely to mis-measure policy neutral rate, too. But these two policy errors will tend to offset each other. Policy would be loose, when


In this light, it is quite striking that remarkably little evidence is available for Central and Eastern

European Countries (CEECs) on the estimation of the policy neutral rate. While there are dozens

of studies on equilibrium exchange rates in the new EU members, there is surprisingly very little

evidence on equilibrium interest rates (Brzoza-Brzezina, 2006, seems to be the only exception, with

evidence on Poland). This imbalance is rather paradoxical, as half the new EU members target

inflation (Czech Republic, Hungary, Poland, Romania and Slovakia), for which the concept of the

policy neutral rate is of primary importance for the conduct of monetary policy.3 Consequently,

this paper tries to bridge this gap.

This paper addresses the issue of policy neutral rate estimation in one of the new EU member

states, the Czech Republic, based on various specifications of simple Taylor-type monetary policy

rules. This former transition country provides an interesting case to evaluate the policy neutral

interest rate, as one can expect a certain pattern in the path of nominal and real equilibrium

interest rates over the longer term (note that the policy neutral rate is in fact the short-term

nominal equilibrium interest rate; more on definitions below).

Lipschitz et al. (2006) point out that at the outset of the transition the capital/labor ratios were

much lower than those in Western Europe and therefore the marginal product of capital and for

that reason the real equilibrium interest rate were rather high. Given the capital accumulation

over the course of the transition, there should be a tendency for the real equilibrium interest rate

to decrease. From an open economy perspective, the new EU members exhibited falling

exchange rate risk premium during their transition process to a market economy (Beneš and

central bank fails to recognize, that trend grows has slowed down and at the same time, policy would be restrictive, as the policy neutral rate has fallen. As a result, mismeasurement of the equilibrium real interest rate due to mismeasurement of productivity growth is likely to have little net impact on monetary policy. 3 See Coats et al. (2003) and Kotlan and Navratil (2003) on Czech monetary policy.


N’Diaye, 2004), which also puts downward pressure on real equilibrium interest rates (Archibald

and Hunter, 2001). Analogously, it is a well-documented empirical regularity that these countries

exhibit real equilibrium exchange rate appreciation (see Égert et al., 2006 for a comprehensive

survey of the sources of appreciation). A decrease in the foreign equilibrium interest rate, which

is reported by several authors for the euro area (e.g. Wintr et al., 2005), may, especially in a small

open economy, reduce the level of the domestic equilibrium interest rate as well. Additionally, the

path of nominal equilibrium interest rates should reflect not only the decrease of real equilibrium

rates, but also successful disinflation in transition countries (see Korhonen and Wachtel, 2006).

All in all, the aforementioned arguments provide a rationale for modeling the policy neutral rate

as time-varying.

In this paper we provide first an estimation of monetary policy rules with a time-varying intercept

to assess the fluctuations of the policy neutral interest rate over time. The novelty of our

approach lies in estimation of the policy neutral rate by the time-varying parameter model with

endogenous regressors (Kim, 2006).4 Unlike the ‘conventional’ time-varying parameter model5,

this approach is robust to endogeneity of the explanatory variables, which is indeed relevant

when estimating the monetary policy rules. While endogeneity is almost always accounted for in

the literature on time-invariant monetary policy rules, as it is estimated by the generalized method

of moments (GMM), it is almost never addressed in studies on time-varying policy rules. An

additional feature of this paper is that we utilize ex-post as well as real-time based data (see e.g.

Orphanides, 2001, on real-time data analysis within a monetary policy rules framework),

specifically the real-time output gap and real-time inflation forecast of the Czech National Bank

(CNB), to estimate the monetary policy rules.

4 Note that in the working paper version of Kim (2006), this procedure is also labeled as the augmented Kalman filter. 5 We label the time-varying parameter as conventional, when it does not account for endogeneity in the regressors hereinafter.


One of our primary policy applications, except for measuring the policy neutral rate by a novel

technique, is also to propose a measure of the monetary policy stance based on the difference

between the actual interest rate and the estimated policy neutral rate. Anticipating our results, we

find this measure of the monetary policy stance quite useful in predicting both the level and

change of the future inflation rate.

The paper is organized as follows. Section 2 discusses related literature. Section 3 describes our

data and empirical methodology. Section 4 gives the results of the estimation of time-varying

estimates of the policy neutral rate as well as an analysis of the ability of the monetary policy

stance to predict future inflation developments. Section 5 concludes. An appendix with additional

results follows.

2 Related Literature

2.1 Methodological Background It has been acknowledged in monetary economics for a long time that there exists some

unobservable real interest rate that equilibrates aggregate demand and aggregate supply

(Woodford, 2003). When the actual real interest rate is equal to the unobservable one, price

stability is achieved. This unobservable rate is often labeled as the natural rate of interest or

equilibrium interest rate. Equivalently, it has been noted that the equilibrium interest rate is the

real interest rate that prevails when prices are fully flexible in all markets (Neiss and Nelson,

2003; Woodford, 2003).

Consequently, the equilibrium interest rate or natural rate of interest is a fairly general concept

and in principle it may be well associated with short-term, medium-term or long-term interest

rates. In this context, it is worth pointing out that the determinants of the equilibrium interest

rate are likely to differ according to time horizon (different frequency movements). In the long-


term, the level of the equilibrium interest rate is influenced by supply-side structural

characteristics of economy such as long-run growth potential, which in turn depends on

technological progress, population growth and inter-temporal substitution of consumption

(Crespo-Cuaresma et al., 2005). In the medium-term, the equilibrium interest rate is associated

with business cycle fluctuations. In the short-term, the equilibrium interest rate is linked mainly

to demand factors related to monetary policy and its targeting horizon (Archibald and Hunter,

2001). Here monetary policy may systematically influence inflation expectations and in turn the

level of the short-term nominal equilibrium rate.

For the purposes of monetary policy conduct, it is vital to know which interest rate level the

monetary authority should set in order to achieve price stability (i.e. neutral policy stance). As the

primary monetary policy instrument is the level of the short-term interest rate, the equilibrium

interest rate in this context is a rather short-term concept and is often labeled as the policy

neutral rate (Coats et al., 2003; Lam and Tkacz, 2004; Beneš et al., 2005). The policy neutral rate

thus represents the nominal equilibrium interest rate and is defined as the real equilibrium interest

rate plus expected inflation (Coats et al., 2003). In other words, the policy neutral rate is linked to

the short-term nominal interest rate, over which the central bank has substantial control, and thus

the policy neutral rate may be understood as a rather narrower concept in comparison to

equilibrium interest rate and natural rate of interest.6

Should interest rate policy of the monetary authority strictly follow the neutral rate when

targeting inflation? Not necessarily. The first point is that obviously there is uncertainty in policy

neutral rate measurement. Second, and more importantly, there are shocks to which it is sub-

optimal for the authority to react. More specifically, central banks may wish deliberately not to

6 For convenience, we use policy neutral rate, natural rate of interest and equilibrium interest rate in the following text interchangeably to a certain extent. However, when we want to emphasize the short-run concept of it, we always use the term policy neutral rate.


react to the first-round effects of cost-push shocks such as those stemming from value added tax

changes, as this can be destabilizing to the economy in the short run. This may, however, alter

the inflation expectations of economic agents, if some fraction of them are myopic, and as a

result, induce a change in the policy neutral rate. In such case, the central bank’s interest rate

policy may temporarily deviate from the policy neutral rate.

2.2 Methods for Natural Rate of Interest Estimation Generally, there are several main methods for estimating the natural rate of interest (see e.g.

Giammarioli and Valla, 2004, for a survey).7 The simplest is to assume that the equilibrium is

captured reasonably well by some univariate trend such as an HP filter. Nevertheless, a number

of papers document that the estimates based on these filters are often misleading (Clark and

Kozicki, 2005). In general, the limitations of the univariate methods have been pointed out by

many authors (e.g. Canova, 1998).

Another method for deriving equilibrium interest rates is based on the estimation of a simple

monetary policy rule of the central bank (Taylor, 1993). The reaction function typically associates

short-term interest rate with its lagged value, the difference between inflation (forecast) and its

target, and the output gap. The intercept of the estimated reaction function can be interpreted as

the nominal equilibrium interest rate (that is, the interest rate that would prevail when inflation

and output are at their targeted values). This method has been applied to estimate the equilibrium

interest rates by e.g. Clarida et al. (1998, 2000) and Orphanides (2001) for the United States and

Germany, Adam et al. (2005) for the United Kingdom, and Gerdesmeier and Roffia (2004, 2005)

for the euro area. Nevertheless, the assumption of constant equilibrium interest rates is often

7 Note that we do not present the exhaustive list of methods for equilibrium interest rate estimation, for example Brzoza-Brzezina (2006) proposes a structural vector autoregression model in this regard. In general, the role of the equilibrium interest rate for monetary policy conduct is discussed extensively by Taylor (1993), Woodford (2003) and Amato (2005).


found too restrictive over the longer term. Consequently, a number of studies model the

equilibrium interest rate, or more generally the monetary policy rule as time-varying (see Plantier

and Scrimgeour, 2002, Elkhoury, 2006, and Kim and Nelson, 2006). Typically, these studies find

a rationale for modeling the rule as time-varying, given that the equilibrium interest rate

sometimes fluctuates considerably over longer time horizons (as do other parameters in the

policy rule). Generally, the monetary policy rules approach measures the behavior of the central

bank and assumes that the central bank estimates equilibrium interest rates correctly. In case of

systematic mis-measurement of equilibrium rates by the central bank, it is likely that the

equilibrium rates retrieved from the estimation of the reaction function are mis-measured as well.

Structural time series models represent another common method for measuring equilibrium

interest rates. The primary contribution in this area is Laubach and Williams (2003), who

formulate a simple empirical model containing an IS curve, a Phillips curve and an equation

linking the equilibrium interest rate to trend growth, and model equilibrium interest rates and

potential output as unobserved components. Their method has gained popularity recently and

has been applied by Manrique and Marques (2004) for the U.S. and Germany, Mesonnier and

Renne (2007) for the euro area, and Wintr, Guarda and Rouabah (2005) for the euro area8 and

Luxembourg as well. In principle, the joint estimation of equilibrium interest rates and the output

gap is an advantage of this approach; however it also reduces the degrees of freedom, which may

be an issue for transition countries with rather short time series.

Equilibrium interest rates can also be estimated within stochastic dynamic general equilibrium

models. The advantage of this type of literature is that it specifies the structure of the economy

and thus in principle allows identification of a variety of shocks hitting the economy. On the

other hand, Levin et al. (1999) find that more complex models seem to be less robust to model

8 See Crespo-Cuaresma et al. (2004) on related estimates for the euro area using a somewhat different methodology.


uncertainty (see also Giammarioli and Valla, 2004). Consequently, these model outcomes may be

quite sensitive to model assumptions. The recent examples of this approach to estimating

equilibrium interest rates include Giammarioli and Valla (2003), Neiss and Nelson (2003) and

Smets and Wouters (2003).

The last major stream of literature estimates equilibrium interest rates from the yield curve and

asset pricing models. Bomfim (2001) uses inflation linked bonds in order to eliminate the

distortions from inflation expectations and retrieves equilibrium interest rates from the realized

yields on U.S. Treasury inflation-indexed securities. In this regard, Giammarioli and Valla (2004)

discuss equilibrium interest rate estimates in relation to consumption capital asset pricing models.

Generally, this stream of literature hinges on the notion of liquid financial markets and thus this

approach is viable mainly for countries with developed financial markets.

3 Data and Empirical Methodology In this part, we discuss the methodology and dataset we employ to evaluate the policy neutral rate

fluctuations in the Czech Republic. Specifically, we estimate a variety of backward or forward

looking monetary policy rules with a time-varying policy neutral rate.

3.1 Monetary Policy Rules A starting point for a formal derivation of the monetary policy rule is the reasonable assumption

that the central bank targets to set the nominal interest rate in line with the state of the economy

(see Clarida et al., 1998, 2000), as postulated in Eq. (1):

( ) ttittitt xEErr Ω+−Ω+= ++ βππα *_

* (1)


where *tr denotes the targeted interest rate,

_r is the policy neutral rate, it+π stands for the

central bank forecast of the yearly inflation rate i periods ahead, and *it+π is the central bank’s

inflation target9. tx represents a measure of the output gap. (.)E is the expectation operator and

tΩ is the information set available at the time when interest rates are set. Hereinafter, we set i

either equal to 12 months to reflect the CNB’s actual targeting horizon10 or alternatively equal to

0, i.e. using the current inflation for the sensitivity analysis. Therefore, Eq. (1) links targeted

nominal interest rates to a constant (i.e. the interest rate – policy neutral rate – that would prevail

when expected inflation and output are at their targeted levels), the deviation of expected

inflation from the target and the output gap.

Nevertheless, Eq. (1) is often argued to be too restrictive, as it does not account for interest rate

smoothing of central banks. Clarida et al. (1998) assume that the central bank adjusts the interest

rate sluggishly to the targeted value. This is so for a number of reasons. For example, Goodfriend

(1991) puts forward concerns over the stability of financial markets. Sack (1997) highlights

uncertainty about the effects of interest rate changes on the economy.11 Instead of an explicit

listing of various factors behind the interest rate smoothing, Clarida et al. (1998) assume for

simplicity that the actual policy interest rate is a combination of its lagged value and the targeted

policy rate as in Eq. (2).

9 At the beginning of our sample period, the target has been defined as the band decreasing from 3%-5% to 2%-4% at the end of 2005. For this period, we take the mid-points of target. From 2006 onwards, the target is set in terms of yearly change of headline inflation of 3%. The evolution of target is depicted in Chart A.1 in the Appendix. 10 This is in line with the CNB main forecasting model – the Quarterly Prediction Model; see Coats et al., 2003. The actual targeting horizon is 12-18 months, but due to data limitations we prefer to work with 12 months. In general, see Batini and Nelson, 1999, for contributions on the optimal targeting horizon. Note also that the policy neutral rate is defined as the real rate plus the expected inflation in period t+k, where k is given by the maturity of the interbank rate (in our case k=3). k is thus different from the forecasting horizon i. As argued by Clarida et al. (2000), this is not very relevant in practice, as the short-term interbank interest rates at various maturities are strongly linked together. Indeed, the correlation of 3M PRIBOR and 12M PRIBOR – to reflect that i=12 – stands at 0.99 in our sample. 11 Nevertheless, Rudebusch (2006) recently questioned the extent of monetary policy inertia and argued that the inertia is rather low.


( ) tttt rrr νρρ +−+= −*

1 1 (2),12

where [ ]1,0∈ρ . In line with Clarida et al. (1998), substituting Eq. (2) into Eq. (1) and

eliminating unobserved forecast variables results in Eq. (3):

( ) ( ) tttititt rxrr ερβππαρ ++⎥⎦⎤

⎢⎣⎡ +−+−= −++ 1

*_

1 (3)

The disturbance term tε is a combination of forecast errors (i.e.,

[ ])()()1( ttttititt xExE Ω−+Ω−−−= ++ βππαρε ) and is thus orthogonal to all

information available at time t ( tΩ ).

Next, in order to estimate the time-varying neutral policy rate we apply a structural time-varying

coefficient model with endogenous regressors. Kim (2006) shows that the conventional time-

varying parameter model delivers inconsistent estimates when explanatory variables are correlated

with the disturbance term, which is indeed relevant when estimating policy rules. It is interesting

to note that the correlation of it+π and tx with tε in Eq. (3) is almost always taken into

account in empirical work on time-invariant rules (as typically estimated by the GMM), while it is

almost never considered in the literature on time-varying monetary policy rules (Kim and Nelson,

2006, seem to be the exception). Subsequently, Kim (2006) derives a consistent estimator of the

time-varying parameter model when regressors are endogenous. In line with Kim (2006), we

estimate the following empirical model:

( ) ( ) tttitittt rxrr ερβππαρ ++⎥⎦⎤

⎢⎣⎡ +−+−= −++ 1

*_

1 (4)

ttt rr ϑ+= −1

__, ( )2,0...~ ϑσϑ Ndiit (5)

tjtit Z ϕσξπ ϕ+= −+' , ( )1,0...~ Ndiitϕ (6)

12 We estimated the monetary policy rules including higher lags of interest rates, but failed to find them significant.


tjtt Zx νσψ ν+= −' , ( )1,0...~ Ndiitν (7)

The measurement equation (4) is a Taylor rule with the policy neutral rate, tr_

, as outlined above.

However, we relax here the assumption of a constant policy neutral rate and let it vary over time,

tr_

, as specified in the transition equation (5), assuming a random walk process without drift.13

Given the data limitations and the fact that our sample is characterized by a relatively stable

institutional structure as well as actual conduct of monetary policy, we do not allow α , β and

ρ to be time-varying. The ‘first-stage’ Eqs. (6) and (7) lay out the relationship between the

endogenous regressors ( it+π and tx ) and their instruments, tZ . The list of instruments, jtZ − , is

as follows: 1−tπ , 12−tπ , 1−tx , 2−tx , 1−tr and +tr (foreign interest rate – 1YEURIBOR). We assume

that the parameters in Eqs. (6) and (7) are time-invariant. Next, the correlation between the

standardized residuals tϕ and tν and tε is εϕκ , and ενκ , , respectively (note that ϕσ and νσ are

standard errors of tϕ and tν , respectively). The consistent estimates of the coefficients in Eq. (4)

are then obtained in two steps. In the first step, we estimate the equations (6) and (7) and save

the standardized residuals tϕ and tν . In the second step, we estimate Eq. (8) along with Eq. (5)

using maximum likelihood via the Kalman filter. Note that (8) now includes bias correction

terms, (standardized) residuals from Eqs. (6) and (7), to address the aforementioned endogeneity

of the regressors. Consequently, the estimated parameters in Eq. (8) are consistent, as tι is

uncorrelated with the regressors.

13 We also experimented with an AR(1) structure in Eq. (5), but it just marginally reduced the likelihood and the estimated AR parameter was very close to one, anyway. In general, it would be possible to link time variation of policy neutral rate to the inflation target, but this would be more demanding on degrees of freedom. I leave this generalization for further research.


( ) ( ) ttε,t,εtε,tν,εtt*ititt

_

t ισκνσκρrβxππαrρr ++++⎥⎦⎤

⎢⎣⎡ +−+−= −++ ϕϕ11 ,

( )2,

2,

2, )1(,0~ tvt Nι εεϕε σκκ −−

(8)

Several authors (see for example Gerdesmeier and Roffia, 2004) raised the issue of including

additional economic variables such as the (real) exchange rate or money growth in Eq. (3) to try

to capture the state of the economy in a fuller manner. Nevertheless, this is typically done in an

ad hoc manner. On the contrary, when the literature assumes that interest rates depend only on

inflation and output, obviously it does not mean that other variables are neglected. For example,

as Taylor (2001, p. 266) puts it for the exchange rate: “Although the policy rule ... may not appear to

involve interest rate reaction to exchange rate, it implies such a reaction. What might appear to be a closed economy

policy rule is actually just as much an open economy rule as if the exchange rate appeared directly.” In other

words, additional variables that do not enter Taylor rule directly may influence inflation and

output and therefore the interest rate setting indirectly. It is also important to emphasize that CPI

inflation targeting, as opposed to domestic inflation targeting, has an implicit concern for foreign

shocks given the composition of the consumer basket (Svensson, 2000). Therefore, in our paper

we do not introduce foreign disturbances explicitly in the monetary policy rule, but we will

employ them as the instruments in our empirical specification.

Additionally, in a book describing the current forecasting and policy analysis process in the CNB,

Coats et al. (2003) report that no other variables than inflation and the output gap enter into the

monetary policy rule in the CNB Quarterly Projection Model (QPM hereinafter). In particular, as

regards exchange rate fluctuations the CNB has stated several times that it does not directly react

to them. It has acknowledged that the exchange rate plays an important role for inflation

developments in small open economies and that it might react indirectly to exchange rate


fluctuations if they jeopardize inflation developments (Kotlán and Navrátil, 2003). Regarding

money aggregates, one can expect that an inflation targeting central bank in general views them as

only supplementary information about the degree of economic activity and/or inflationary

pressures, and also does not directly react to them.

Originally, the design of the Taylor rule lacked a forward-looking element characteristic of the

modern monetary policy conduct. An additional way of assessing the sensitivity of our results is

to estimate of both backward- and forward-looking Taylor-type rules. Therefore, we formulate

the monetary policy rule in Eq. (3) in the case of backward-looking policy rule such that we set

0=i , i.e. we use current inflation rate instead of its forecasted value (which is utilized for the

forward-looking policy rule). Another important point has been raised about the timeliness of

information in monetary policy conduct (Orphanides, 2001). Output data are typically revised at a

later stage, but monetary policy is conducted based on the information available at the time.

Therefore, we collect real-time based CNB output gap estimates (note that inflation is not revised

at a later stage by the Czech Statistical Office) and re-estimate the monetary policy rule with the

real-time output gap. Analogously, we use one year ahead CNB’s real-time inflation forecast in

estimating the monetary policy rule.14

There are further modeling issues stemming from the fact that the policy interest rate is not

changed in a continuous fashion. For instance, the CNB Bank Board meets on a monthly basis to

discuss the policy interest rate settings. Besides, the policy rate change itself is not continuous.

Typically, if the rates are changed, the respective magnitude is 0.25 percentage points (or some

multiple of 0.25), even if the change maximizing economic stability according to the model-based

forecast might be of a (slightly) different magnitude. In consequence, the policy rate is not only

14 In general, the specification with inflation forecast and real-time output gap, the rule is using information known to the central bank at the time. However, as we interpolate quarterly output gap figures to the monthly frequency, we introduce information not known at the time of policy decision. As a result, conventional time-varying parameter model is not appropriate even in this case.


discrete, but also censored. Given the inherent censoring of policy interest rates, the majority of

authors, such as Clarida et al. (1998, 2000) and Adam et al. (2005), rely on using the 3-month

interbank rate as an approximation of the censored policy rate.

On the other hand, Choi (1999) and Carstensen (2006) put forward modeling censoring in the

policy rate directly by employing for example a modified Tobit model and an ordered probit

model, respectively. The advantage of this approach is that it models the interest rate setting

more realistically and does not have to make a simplifying assumption by utilizing the short-term

interbank rate. On the other hand, this stream of literature so far models only censoring in the

policy rate, but it has been stressed that there is also censoring in the policy rate change

(Podpiera, 2006). In addition, censored models are known to be less efficient and the results

based on them do not seem to stand up in sharp contrast to those using short-term interbank

rates (for example, consider the extent of interest rate smoothing). An additional drawback of

this approach in our case is that our main estimation technique is a time-varying parameter model

with endogenous regressors (Kim, 2006) and to our knowledge this technique is simply not

available with a censored dependent variable. Besides, the CNB’s QPM also uses the 3-month

interbank rate instead of the 2-week policy rate as well. Having all the pros and cons of these two

approaches – short-term interbank rate vs. policy interest rate – in mind, we opt to use the short-

term interbank rate in the estimation of the monetary policy rules.

3.2 Data Our sample contains monthly data over the period 2001:1-2006:09 on yearly CPI inflation

( 12−−= ttt ppπ , where tp is the log of the price level at time t), yearly net inflation (price

indexes of regulated goods excluded from the price index; thus nett

nett

nett pp 12−−=π , where net

tp is

the log of the net price level at time t), the output gap ( tx ,the difference between actual and


potential GDP growth, defined as below), tr , the short-term interbank rate (3M PRIBOR), the

real effective exchange rate, ( )treer , and +tr , the foreign interest rate (1YEURIBOR). We also

use the real-time CNB internal forecasts of CPI ( )ft 12+π and net inflation ( )fnet

t,

12+π , and the output

gap. Three different estimates of the output gap are employed: a) the estimate using HP filter15, b)

the ex-post revised output gap from the CNB’s QPM as of their October 2006 forecast round

and c) the real-time based output gap collected from the CNB’s QPM. The source of our data is

the CNB public database system ARAD (except for inflation forecasts and the two

aforementioned output gap measures – b) and c), which are not available publicly).

All our variables are available on a monthly basis, except the output gap. Following Adam et al.

(2005), we linearly interpolate quarterly estimates of the output gap to monthly values.16 We use

the mid-points of the CNB inflation target. The choice of the 2001-2006 period is motivated to

have as long a sample period as possible, while not rejecting stationarity of all variables at the 5%

significance level (using the KPSS test). More importantly, the real-time output gap and inflation

forecast are not available before 2001. As a further robustness check, we also estimate the

monetary policy rules with net inflation (based on price indexes excluding regulated goods from

the consumer basket) instead of CPI inflation.17

15 The standard smoothing parameter of 1600 has been used. Different smoothing parameters, such as the one suggested by Ravn and Uhlig (2002), had very little impact on the resulting estimates of the policy neutral rate. The output gap derived from HP filter estimates of potential output differs considerably from those used by the CNB – see Chart B.2 in the Appendix. Note that the output gap from the CNB’s QPM is constructed using a multivariate Kalman filter. Generally, output gap estimates based on the HP filter can be viewed as less reliable, but we keep them in our empirical work, as this gap is replicable based on publicly available data, which is not the case with the two other output gap measures derived from the CNB’s QPM. Note also that the HP filter is known to suffer from end-point bias, making it cumbersome for real-time analysis. Another option would be to derive the output gap using some filtering technique in time series domain such as band-pass filter, see Baxter and King (1999). 16 We also used a quadratic match procedure for interpolation. This yields only little differences for the resulting output gap estimates. We are aware that interpolation introduces information not available at the time in the case of the output gap. 17 These results are available upon request. As net inflation is largely correlated with headline CPI inflation (a value of 0.93 in our sample), it is not surprising that the results are quite similar.


4 Results

4.1 Time-varying Equilibrium Interest Rates Generally, we find that the policy neutral rate decreases over time, as depicted in Chart 1. This is

in line with our conjecture laid out in the introduction. We report all specifications, from the

backward-looking policy rule with the output gap estimated by the HP filter to the forward-

looking rule with the real-time output gap.


Chart 1 –Policy Neutral Rate Estimates

Time-varying parameter model with endogenous regressors Backward-looking policy rule

1

2

3

4

5

6

2002 2003 2004 2005 2006

1

2

3

4

5

6

2002 2003 2004 2005 2006

1

2

3

4

5

6

2002 2003 2004 2005 2006

Current inflation, output gap – HP filter Current inflation, output gap – ex-post Current inflation, output gap – real-time Forward-looking policy rule

1

2

3

4

5

6

2002 2003 2004 2005 2006

1

2

3

4

5

6

2002 2003 2004 2005 2006

1

2

3

4

5

6

2002 2003 2004 2005 2006

Inflation forecast – real-time, output gap – HP filter

Inflation forecast – real-time, output gap – ex-post

Inflation forecast – real-time, output gap – real-time

‘Conventional’ time-varying parameter model Backward-looking policy rule

1

2

3

4

5

6

7

2002 2003 2004 2005 2006

1

2

3

4

5

6

7

2002 2003 2004 2005 2006

1

2

3

4

5

6

7

2002 2003 2004 2005 2006

Current inflation, output gap – HP filter Current inflation, output gap – ex-post Current inflation, output gap – real-time Forward-looking policy rule

1

2

3

4

5

6

7

2002 2003 2004 2005 2006

1

2

3

4

5

6

7

2002 2003 2004 2005 2006

1

2

3

4

5

6

7

2002 2003 2004 2005 2006




Note: The policy neutral rate ±2 standard errors are reported. The measure of the output gap and inflation used for estimation of the policy neutral rate is reported below each chart. The time-varying parameter model is labeled as conventional when it does not account for endogeneity of regressors.


The results in Chart 1 unambiguously indicate that the policy neutral rate gradually decreased

from around 5% to values around 2% at the end of 2005 and subsequently slightly increased to

around 2.5% over the course of 2006.18 This confirms substantial interest rate convergence to

levels comparable to the euro area countries. For example, Messonier and Renne (2007) estimate

the euro area real equilibrium interest rate around 1% at the end of their sample (i.e. 2002) and

Wintr et al. (2005) find it to have been just below 1% in 2004. If we add to these estimates 2% for

expected inflation – to reflect the European Central Bank definition of price stability, we receive

an estimate of the policy neutral rate of about 3% for the euro area. The results actually suggest

that the Czech policy neutral rate seems to be slightly below the euro area level. This should not

come as a surprise, as actual short-term interest rates were also often below those in the euro area

(except for several months in 2004, Czech rates appear to be below euro area rates from around

mid-2002 onwards).

Detailed parameter estimates of the monetary policy rules are presented in the Appendix – Table

A.2. The degree of interest rate smoothing falls considerably when allowing for time-varying

parameter specification. Compared to the case of the constant policy neutral assumption, which

is estimated by the GMM19, the value of the interest rate smoothing parameter falls from around

0.9 to 0.4. Time-invariant rules thus appear to overestimate substantially the degree of interest

rate smoothing. This complies with the results of Rudebusch (2006), who stresses that once one

accounts for expectations about future monetary policy, the actual policy rate inertia is in fact

quite low. The standard errors of the estimates are in some cases large, probably reflecting the

smaller sample size. The coefficient on inflation is around 0.3 in the case of the backward-looking

18 The increase at the end of sample period is likely to reflect higher inflation expectations of economic agents. The CNB conducts regularly a survey on inflation expectations of financial markets, households and non-financial firms (actual numbers are easily available from CNB website within their public database ARAD). Inflation expectations of financial markets for the 1 year horizon have risen from some 2.5% in mid 2005 to 3% over the course of 2006. Similar pattern is visible also for household’s and firm’s expectations. 19 The GMM results are available upon request.


specifications (interestingly, it is significant only when we introduce bias correction terms, tϕ and

tν ), and insignificant for the forward-looking specifications. This suggests that central bank

might have found inflation forecasts as uncertain and considered current inflation as more

informative for future inflation developments. This is probably due to structural changes in

transition economy and the associated higher uncertainty. The insignificance of output gap might

stem from similar reasons, too. The coefficient on the output gap is around 0.6 when it is

significant; otherwise it is smaller for other specifications.

Additionally, the results support the usefulness of applying the time-varying parameter model

with endogenous regressors. The bias correction terms, tϕ and tν , in Eq. (8), are typically

significant and the log likelihood improves after their inclusion. Comparing the estimated policy

neutral rate with those implied by the conventional time-varying parameter model, we find that

the resulting difference between these two varies according to the specification of the policy rule

as well as over time. While the median difference is only 0.05 p.p. in absolute terms, the

maximum difference that the inclusion of bias correction terms amounts to is 1.8 p.p.

Chart 3 in the Appendix presents a comparison of the policy neutral rate based on an identical

specification of the policy rule, but estimated either by the time-varying parameter model with

endogenous regressors or by the conventional time-varying parameter model. Denoting the

policy neutral rate estimated by the former method, ertr ,

_, and the latter method, ctr ,

_, Chart A.3

reports the difference between these two. Obviously, if 0,

_

,

_=− ctert rr , the bias correction terms

do not matter at all.20 However, we can see from the results that although the two methods yield

in general rather similar estimates of the policy neutral rate path, there are periods when the bias

20 Obviously, it might be the case that both bias correction terms are statistically significant, but they just “cancel out” their impact on the estimated policy neutral rate. The probability of “canceling out” for each month is virtually zero.


correction terms matter considerably, i.e. when the policy neutral rate estimates by the

conventional time-varying parameter model do not even lie inside the confidence interval of the

policy neutral rate estimated by the time-varying parameter model with endogenous regressors, as

shown in Chart A.3.

4.2 Monetary Policy Stance and Inflation Developments There is a discussion in the literature about to what extent monetary policy rules provide a useful

framework to evaluate the monetary policy stance and its impact on subsequent inflation. This is

typically done by comparing the actual interest rate setting with the one implied by the rule and

inflation outcomes. For example, Taylor (1999, p. 340) labels the difference as the policy mistakes

(i.e. the residual from the policy rule) and shows that they are well associated with high inflation

or low capacity utilization with the U.S. data. On the other hand, Reynard (2007), analyzing the

U.S. and Swiss data, questions the reliability of the so-called policy mistakes, as he observes rather

weak link between so-called policy mistakes and inflation relative to the inflation target.

Here we analyze a rather different framework to evaluate the usefulness of policy rules. Instead

of focusing on the residual from the policy rule, we analyze the difference between actual interest

rates and the policy neutral rate (“equilibrium rate”) and its impact on subsequent inflation. We

label the difference between the actual interest rate and the policy neutral rate as the monetary

policy stance hereinafter. Amato (2005) shows that there should be negative correlation between

policy stance and future inflation (at the monetary policy horizon), if private sector is tightly

constrained by past outcomes or mostly backward looking. Similarly, Neiss and Nelson (2003)

develop a DSGE model; their results show that real interest rate gap is valuable as an inflation

indicator.


We leave examination of the link between policy mistakes and inflation outcomes for further

research, as we do not concentrate in this paper on whether the estimated monetary policy rule

provides an accurate description of CNB policy, but rather on the estimation and evolution of

the policy neutral rate. Implicitly, by carrying out this estimation we also address the issue of the

uncertainty surrounding with equilibrium interest rate measurement (see Laubach and Williams,

2003, or Herrmann, Orphanides and Siklos, 2005, on this uncertainty). In this regard, Amato

(2005) emphasizes that the uncertainty in neutral rate estimates makes it difficult to use them as a

reliable indicator of excess demand pressures, but the theories of neutral interest rates still remain

a useful concept for the formulation of monetary policy.

So, our simple test here is to examine whether our estimated policy neutral rate is helpful in

predicting future inflation developments. If the policy neutral rate is too uncertain a measure,

then it is more likely that it does not provide information for subsequent inflation. Our

supposition is thus that when the actual interest rate is above the policy neutral rate, the future

inflation rate is then likely to fall, as monetary policy can be considered restrictive, and vice versa.

We label the difference between the actual interest rate and the policy neutral rate as the

monetary policy stance hereinafter.

The generic monetary policy stance together with current inflation is plotted in Chart 2.21 As we

also have confidence intervals for the policy neutral rate, it is possible to evaluate whether the

stance was statistically different from zero. The results indicate that monetary policy during the

sample period can be regarded as relatively easy, especially around the years 2002-2003. This

should not come as a surprise, since inflation was well below the target and even got into

21 The monetary policy stance presented in Chart 2 is based on a monetary policy rule specification with an inflation forecast and an ex-post GDP gap. Different specifications play a rather minor role in the overall assessment of the monetary policy stance. In addition, it is interesting to note that the significance of monetary policy stance is not affected by whether we employed the backward or the forward looking policy rule. In addition, Charts A.4 plots monetary policy stance and inflation and Chart A.5 provides a comparison of actual and neutral rate.


negative numbers for several months in 2003 and the output gap was negative, reaching its

minimum in mid-2003 according to the CNB output gap estimates (see Chart A.1).

Chart 2 – Monetary Policy Stance and Inflation

-1.2

-0.8

-0.4

0.0

0.4

0.8

-2

0

2

4

6

2001 2002 2003 2004 2005 2006

Policy stance Inflation

Note: Positive values refer to monetary policy tightening, while negative values point to policy easing. Policy stance values are on the left axis, while inflation is on the right axis.

As there are transmission lags between a monetary policy action and its impact on the bank’s

targeted variables, we assume that the current monetary policy stance affects inflation after 12 to

24 months. This coincides well with the CNB monetary policy horizon, as the bank

acknowledges that “…interest rate changes have their greatest impact on inflation some 12 to 18 months …”

(CNB). It is also supported by the empirical findings of Borys-Morgese and Horvath (2007).

Within their factor augmented VAR framework, they find that the peak response of inflation to

interest rate shocks is around a year or so (note the maximum reaction of non-tradable inflation

is found to be close to two years). Based on this evidence, it seems to be fruitful to analyze the

horizon between about one and two years. Here we broadly follow the empirical specification by

Moser et al. (2007), which is a variant of Stock and Watson (1999). While these two studies

examine the role of factor models for inflation forecasting, we analyze the impact of the


monetary policy stance instead. More specifically, we test the significance of monetary policy

stance indicator, including it in an autoregressive-type model for the inflation process.

Our estimation framework begins with the following regression:

itt

_

tit rrπ ++ +⎟⎠⎞

⎜⎝⎛ −+= υφφ 10 (9)

where itπ + is yearly inflation i months ahead, where i=12,….,24. Next, we control for the lagged

inflation terms:

it

n

hhtht

_

tit πrrπ +=

−++ ∑ ++⎟⎠⎞

⎜⎝⎛ −+= υφφφ

1110 (10)

where for simplicity we set n=4.22

Using Eqs. (9) and (10), we investigate the information content of the monetary policy stance on

the future level of inflation. We also re-specify the above equations to address the future change

in the inflation rate as follows:

itt

_

ttit rrππ ++ +⎟⎠⎞

⎜⎝⎛ −+=− υφφ 10 (11)

it

n

hhtht

_

ttit πrrππ +=

−++ ∑ ++⎟⎠⎞

⎜⎝⎛ −+=− υφφφ

1110 (12)

22 We also included higher lags, but with little impact on the results.


Table 1 – Monetary Stance and Future Level of Inflation

itt

_

tit rrπ ++ +⎟⎠⎞

⎜⎝⎛ −+= υφφ 10

Note: Robust standard errors. ***, ** and * indicate significance at 1, 5 and 10%, respectively.

Table 2 – Monetary Stance and Future Level of Inflation,

Controlling for Lagged Inflation

it

n

hhtht

_

tit πrrπ +=

−++ ∑ ++⎟⎠⎞

⎜⎝⎛ −+= υφφφ

1110

i 0φ 1φ 2φ 3φ 4φ 5φ Adj. R2

12 3.11*** 0.72 0.08 0.07 0.02 -0.67** 0.30 13 3.06*** 0.52 -0.06 0.23 0.13 -0.54* 0.35 14 2.88*** 0.18 -0.11 0.07 0.08 -0.53* 0.40 15 2.64*** -0.19 -0.27* 0.24 -0.06 -0.33 0.45 16 2.42*** -0.45** -0.13 -0.12 -0.20 0.02 0.47 17 2.37*** -0.28* -0.27 -0.19 0.18 -0.06 0.49 18 2.31*** -0.62** -0.59** 0.19 0.16 -0.08 0.49 19 2.18*** -0.76** -0.62** 0.16 0.33 -0.15 0.49 20 2.15*** -0.77** -0.66** 0.31 0.26 -0.16 0.45 21 2.00*** -0.91*** -0.55 0.27 0.21 -0.07 0.45 22 1.84*** -1.07*** -0.50* 0.22 0.01 0.15 0.47 23 1.77*** -1.07*** -0.35 -0.06 0.04 0.31 0.45 24 1.77*** -1.03*** -0.44 0.02 0.01 0.36 0.42

Note: Robust standard errors. ***, ** and *indicate significance at 1, 5 and 10%, respectively.

i 0φ 1φ Adj. R2

12 1.29*** -1.23** 0.16 13 1.19*** -1.47** 0.24 14 1.16*** -1.60*** 0.28 15 1.16*** -1.68*** 0.31 16 1.16*** -1.76*** 0.35 17 1.16*** -1.82*** 0.36 18 1.21*** -1.77*** 0.36 19 1.26*** -1.69*** 0.34 20 1.33*** -1.57** 0.30 21 1.44*** -1.39*** 0.25 22 1.53*** -1.25** 0.22 23 1.62*** -1.10* 0.19 24 1.71*** -0.96 0.16


Table 3 – Monetary Stance and Future Change of Inflation

itt

_

ttit rrππ ++ +⎟⎠⎞

⎜⎝⎛ −+=− υφφ 10

i 0φ 1φ Adj. R2

12 -2.18*** -5.23*** 0.69 13 -2.26*** -5.48*** 0.71 14 -2.28*** -5.62*** 0.72 15 -2.28*** -5.71*** 0.74 16 -2.28*** -5.79*** 0.75 17 -2.27*** -5.84*** 0.76 18 -2.22*** -5.81*** 0.74 19 -2.19*** -5.73*** 0.73 20 -2.12*** -5.62*** 0.72 21 -2.03*** -5.43*** 0.69 22 -1.96*** -5.30*** 0.72 23 -1.91*** -5.19*** 0.72 24 -1.84*** -5.07*** 0.72


Table 4 – Monetary Stance and Future Change of Inflation,

Controlling for Lagged Inflation

it

n

hhtht

_

ttit πrrππ +=

−++ ∑ ++⎟⎠⎞

⎜⎝⎛ −+=− υφφφ

1110

i 0φ 1φ 2φ 3φ 4φ 5φ Adj. R2

12 2.93*** 0.60 -1.08*** -0.03 0.46 -0.77** 0.81 13 2.85*** 0.41 -1.24*** 0.12 0.34 -0.64* 0.83 14 2.68*** 0.07 -1.28*** -0.05 0.55 -0.62* 0.84 15 2.48*** -0.28 -1.45*** 0.11 0.44 -0.47 0.86 16 2.26*** -0.54** -1.30*** -0.20 0.33 -0.12 0.86 17 2.20*** -0.63** -1.46*** -0.32 0.71 -0.21 0.88 18 2.16*** -0.71** -1.78*** 0.08 0.67 -0.22 0.87 19 2.01*** -0.86*** -1.82*** 0.06 0.83 -0.28 0.86 20 1.97*** -0.87*** -1.87*** 0.21 0.76 -0.29 0.85 21 1.84*** -1.00*** -1.75*** 0.08 0.74 -0.21 0.85 22 1.67*** -1.17*** -1.72*** 0.12 0.52 0.02 0.85 23 1.62*** -1.16*** -1.56*** -0.10 0.43 0.21 0.86 24 1.61*** -1.12*** -1.65*** -0.03 0.42 0.28 0.86


The results from Eq. (9) are given in Table 1. Our definition of the monetary policy stance seems

to be informative for future inflation, explaining typically about 1/4-1/3 of its variance. These

results are largely confirmed when controlling for the lagged inflation terms, as suggested by the


estimation of Eq. (10) presented in Table 2. Similarly, the policy neutral rate seems to be a

relatively good predictor of the future change of the inflation rate, as presented in Table 3. This

result is largely robust to the inclusion of lagged inflation as well (see Table 4).23 All in all, the

results support the usefulness of the policy neutral rate in understanding the future behavior of

inflation.24

5 Conclusions This paper analyzes the policy neutral rate in the Czech Republic. In order to do so, we estimate

various specifications of simple Taylor-type monetary policy rules at monthly frequency from

2001:1 to 2006:9. To address the sensitivity of results, the specifications differ based on whether

we include real-time or ex-post revised data, employ backward or forward-looking monetary

policy rules or vary the measure of the output gap.

As we focus on the fluctuations of the policy neutral rate over time, we use a time-varying

parameter model with endogenous regressors (Kim, 2006). This approach is especially appealing

for estimating monetary policy rules, as it addresses the endogeneity of inflation (forecast) and

output gap. Indeed, the results support the usefulness of applying the time-varying parameter

model with endogenous regressors. The bias correction terms, accounting for the endogeneity of

the regressors, are typically significant. The log likelihood improves after their inclusion and the

23 Several papers on the predictability of asset returns emphasize issues related to small sample inference (Kirby (1997), Kilian (1999) or Ang and Bekaert (2007)). In these applications, results based on conventional estimators of standard errors that appear significant often prove to be spurious. Typically, the bias grows with the forecasting horizon. On the other hand, our results do not suggest that the significance of policy stance and inflation increases with the forecasting horizon, but the significance rather reflects the monetary policy horizon of the CNB. We thank anonymous referee for this point. 24 We also tested the robustness of our results by including other macroeconomic variables in Eqs. (10) and (12) such as the real effective exchange rate, credit and monetary aggregates. The results remain largely unchanged and are available upon request. We also assess, if the predictive content of neutral rate comes from the “trend component” of monetary policy rule. In order to do so, we estimate a policy rule without inflation and output included. The resulting monetary policy stance fluctuates around zero with no clear pattern and also has no predictive content for inflation.


estimated path of the policy neutral rate is for certain periods considerably different from the

estimates ignoring the endogeneity.

The results indicate that the policy neutral rate decreases gradually over the course of the sample

period from around 5% in 2001 to about 2.5% in 2006, showing substantial interest rate

convergence to levels comparable to the euro area. Over the longer term, the decrease may be

supported by a number of factors such as capital accumulation, decrease in the risk premium, real

equilibrium exchange rate appreciation as well as successful disinflation of the Czech economy

and well-anchored inflation expectations.

One of our primary policy applications, besides measuring the policy neutral rate by a novel

technique, is also to propose a measure of the monetary policy stance based on the difference

between the actual interest rate and the policy neutral rate. Our results indicate that this measure

is quite useful in predicting future inflation developments, i.e. the monetary policy stance

indicator affects the level as well as the change of the future inflation rate. In terms of future

research, it would be interesting to see more evidence on other inflation targeting countries to

uncover whether our proposed monetary policy stance measure remains a useful predictor of

future inflation developments, as we find in the case of the Czech Republic.


References Adam, Christopher, Cobham, David, Girardin, Eric, 2005. Monetary Frameworks and

Institutional Constraints: UK Monetary Policy Reaction Functions, 1985-2003. Oxford

Bulletin of Economics and Statistics 67, 497-516.

Amato, Jeffery, 2005. The Role of Natural Rate of Interest in Monetary Policy. CESifo Economic

Studies 51, 729-755.

Ang, Andrew, Bekaert, Geert, 2007. Stock Return Predictability: Is It There?. Review of Financial

Studies 20, 651-707.

Archibald, Joanne, Hunter, Leni, 2001. What is the Neutral Interest Rate, and How Can We Use

It? Reserve Bank of New Zealand Bulletin 64, 15-28.

Batini, Nicoletta, Nelson, Edward, 1999. Optimal Horizons for Inflation Targeting. Journal of

Economic Dynamics and Control 25, 891-910.

Baxter, Marriane, King, Robert, 1999. Measuring Business-cycles: Approximate Band-Pass Filters

for Economic Time Series. The Review of Economics and Statistics, 81, 575-593.

Beneš, Jaromír, N’Diaye, Papa, 2004. A Multivariate Filter for Measuring Potential Output and

the NAIRU: Application to the Czech Republic. IMF Working Paper, No. 04/45.

Beneš, Jaromír, Hlédik, Tibor, Kumhof, Michael, Vávra, David, 2005. An Economy in Transition

and DSGE: What the Czech National Bank’s New Projection Model Needs. Czech

National Bank Working Paper Series, No. 12/2005.

Bomfim, Antulio N., 2001. Measuring Equilibrium Real Interest Rates: What Can We Learn from

Yields on Indexed Bonds? Board of Governors of the Federal Reserve System, mimeo.

Borys Morgese, Magdalena, Horvath, Roman, 2007. The Effects of Monetary Policy in the Czech

Republic: An Empirical Study. IES Working Paper, Charles University, No. 26/2007.

Brzoza-Brzezina, Michal, 2006. The Information Content of Neutral Rate of Interest. Economics

of Transition 14, 391-412.

Canova, Fabio, 1998. Detrending and Business Cycle Facts. Journal of Monetary Economics 41,

475-512.

Carstensen, Kai, 2006. Estimating the ECB Policy Reaction Function. German Economic Review

7, 1-34.

Choi, Woon Gyu, 1999. Estimating the Discount Rate Policy Reaction Function of the Monetary

Authority. Journal of Applied Econometrics 14, 379-401.

Clarida, Richard, Gali, Jordi, Gertler, Mark, 1998. Monetary Policy Rules in Practice: Some

International Evidence. European Economic Review 42, 1033-1067.


Clarida, Richard, Gali, Jordi, Gertler, Mark, 2000. Monetary Policy Rules and Macroeconomic

Stability: Evidence and Some Theory. The Quarterly Journal of Economics 115, 147-180.

Clark, Todd E., Kozicki, Sharon, 2005. Estimating Equilibrium Real Interest Rates in Real Time.

North American Journal of Economics and Finance 16, 395-413.

Coats, Warren, Laxton, Douglas, Rose, David (Eds.), 2003. The Czech National Bank’s

Forecasting and Policy Analysis System. Czech National Bank (available at www.cnb.cz).

Crespo-Cuaresma, Jesús, Gnan, Ernest, Ritzberger-Gruenwald, Doris, 2004. Searching for the

Natural Rate of Interest: A Euro Area Perspective. Empirica 31, 185-204.

Crespo-Cuaresma, Jesús, Gnan, Ernest, Ritzberger-Gruenwald, Doris, 2005. The Natural Rate of

Interest - Concepts and Appraisal for the Euro Area. Monetary Policy and the Economy

4/2005, Oesterreichische Nationalbank.

Égert, Balázs, Halpern, Lászlo, MacDonald, Ronald, 2006. Equilibrium Exchange Rates in

Transition Economies: Taking Stock of the Issues. Journal of Economic Surveys 20, 257-

324.

Elkhoury, Marwan, 2006. Time-Varying Parameter Model of a Monetary Policy Rule in the

Switzerland. HEI Working Paper, No.01, Geneva.

Gerdesmeier, Dieter, Roffia, Barbara, 2004. Empirical Estimates of Reaction Functions for the

Euro Area. Swiss Journal of Economics and Statistics 140, 37-66.

Gerdesmeier, Dieter, Roffia, Barbara, 2005. The Relevance of Real-Time Data in Estimating

Reaction Functions for the Euro Area. North American Journal of Economics and Finance

16, 293-307.

Giammarioli, Nicola, Valla, Natacha, 2003. The Natural Real Rate of Interest in the Euro Area.

European Central Bank Working Paper, No. 233.

Giammarioli, Nicola, Valla, Natacha, 2004. The Natural Real Interest Rate and Monetary Policy:

A Review. Journal of Policy Modeling 26, 641-660.

Goodfriend, Marvin, 1991. Interest Rates and the Conduct of Monetary Policy. Carnegie-

Rochester Conference Series on Public Policy 34, 7-30.

Herrmann, Heinz, Orphanides, Athanasios, Siklos, Pierre, 2005. Real-Time Data and Monetary

Policy. North American Journal of Economics and Finance 16, 271-276.

International Monetary Fund, 2006. Inflation Targeting and the IMF. IMF: Washington D.C.

Inflation Reports, Czech National Bank, various issues.

Kilian, Lutz, 1999. Exchange Rates and Monetary Fundamentals: What Do We Learn from

Long- Horizon Regressions?. Journal of Applied Econometrics 14, 491-510.


Kim, Chang-Jin, 2006. Time-Varying Parameter Models with Endogenous Regressors.

Economics Letters 91, 21-26.

Kim, Chang-Jin, Nelson, Charles R., 2006. Estimation of a Forward-Looking Monetary Policy

Rule: A Time-Varying Parameter Model Using Ex-post Data. Journal of Monetary

Economics 53, 1949-1966.

Kirby, Chris 1997. Measuring the Predictable Variation in Stock and Bond Returns, Review of

Financial Studies 10, 579-630.

Korhonen, Iika, Wachtel, Paul, 2006. Observations on Disinflation in Transition Economies. In:

Altig, David, Nosal, Ed, (Eds.), Monetary Policy in Low Inflation Economies. Cambridge

University Press: Cambridge.

Kotlán, Viktor, Navrátil, David, 2003. Inflation Targeting as a Stabilization Tool: Its Design and

Performance in the Czech Republic. Finance a uver – Czech Journal of Economics and

Finance 53, 220-242.

Kwiatkowski, Denis, Phillips, Peter C.B., Schmidt, Peter, Shin, Yongcheol, 1992. Testing the

Null Hypothesis of Stationarity against the Alternative of a Unit Root: How Sure Are We

that Economic Time Series Have a Unit Root? Journal of Econometrics 54, 159-178.

Lam, Jean-Paul, Tkacz, Greg, 2004. Estimating Policy Neutral Rate for Canada Using a Dynamic

Stochastic General Equilibrium Framework. Swiss Journal of Economics and Statistics 140,

89-126.

Laubach, Thomas, Williams, John C., 2003. Measuring the Natural Rate of Interest. The Review

of Economics and Statistics 85, 1063–1070.

Levin, Andrew, Wieland, Volker, Williams, John, C., 1999. Robustness of Simple Monetary

Policy Rules under Model Uncertainty. In: J. B. Taylor (Ed), Monetary Policy Rules.

Chicago: University of Chicago.

Lipshitz, Leslie, Lane, Timothy, Moumouras, Alex, 2006. Capital Flows to Transition Countries:

Master or Servant. Finance a úvěr – Czech Journal of Economics and Finance 56, 202-222.

Manrique, Marta, Marques, José M., 2007. An Empirical Approximation of the Natural Rate of

Interest and Potential Growth. Banco de Espana Working Paper, No. 0416.

Mesonnier, Jean-Stephane, Renne, Jean-Paule, 2007. Time Varying “Natural” Rate of Interest in

the Euro Area. European Economic Review 51, 1768-1784.

Moser, Gabriel, Rumler, Fabio, Scharler, Johann, 2007. Forecasting Austrian Inflation. Economic

Modelling 24, 470-480.

Neiss, Katherine S., Nelson, Edward, 2003. The Real Interest Rate Gap as an Inflation Indicator.

Macroeconomic Dynamics 7, 239-262.


Orphanides, Athanasios 2001. Monetary Policy Rules Based on Real-Time Data. American

Economic Review 91, 964-985.

Plantier, L. Christopher, Scrimgeour, Dean, 2002. Estimating a Taylor Rule for New Zealand

with a Time-Varying Neutral Real Rate. Reserve Bank of New Zealand Discussion Paper

Series, 06/2002.

Podpiera, Jiří, 2006. The Role of Policy Rule Misspecification in Monetary Policy Inertia Debate.

Czech National Bank, CERGE-EI Working Paper, No. 315, Charles University.

Ravn, Morten O., Uhlig, Harald, 2002. On Adjusting the Hodrick-Prescott Filter for the

Frequency of Observations. Review of Economics and Statistics 84, 371-375.

Reynard, Samuel, 2007. Maintaining Low Inflation: Money, Interest Rates and Inflation. Journal

of Monetary Economics 54, 1441-1471.

Rudebusch, Glenn, 2006. Monetary Policy Inertia: Fact or Fiction? International Journal of

Central Banking 2(4), 85-136.

Sack, Brian, 1997. Uncertainty, Learning and Gradual Monetary Policy, Board of Governors of

the Federal Reserve System. Finance and Economics Discussion Series, 1997-34.

Smets, Frank, Wouters, Raf, 2003. An Estimated Dynamic Stochastic General Equilibrium Model

of the Euro Area. Journal of the European Economics Association 1, 1123-1175.

Stock, James H. Watson, Mark W., 1999. Forecasting Inflation, Journal of Monetary Economics

44, 293-335.

Svensson, Lars E.O., 2000. Open Economy Inflation Targeting. Journal of International

Economics 50, 155-183.

Taylor, John B., 1993. Discretion versus Policy Rules in Practice. Carnegie-Rochester Conference

Series on Public Policy 39, 195-214.

Taylor, John B. (Ed.), 1999. Monetary Policy Rules. The University of Chicago Press: Chicago.

Taylor, John, B., 2001. The Role of Exchange Rate in Monetary-Policy Rules. American

Economic Review 91, 263-267.

Trehan, Bharat, Wu, Tao, 2007. Time-Varying Equilibrium Real Rates and Monetary Policy

Analysis. Journal of Economic Dynamics and Control 31, 1584-1609.

Wintr, Ladislav, Guarda, Paolo, Rouabah, Abdelaziz, 2005. Estimating the Natural Interest Rate

for the Euro Area and Luxembourg. Central Bank of Luxembourg Working Paper, No. 15.

Woodford, Michael, 2003. Interest and Prices: Foundations of a Theory of Monetary Policy.

Princeton University Press, Princeton: New Jersey.

Internet resources:

CNB – The Monetary Policy of the Czech National Bank, downloaded: June 10, 2007


http://www.cnb.cz/www.cnb.cz/en/about_cnb/missions_and_functions/mp_faktsheet_en.htm

l


APPENDIX

Chart A.1 – Interest Rate, Output Gap, Inflation and Inflation Target

-2

-1

0

1

2

3

4

5

6

2001 2002 2003 2004 2005 2006

Interest rateInflation

Inflation targetOutput gap

Note: This chart presents current inflation, the short-term interbank interest rate (3M PRIBOR) and the CNB output gap as of the October 2006 forecast round.

Chart A.2 – Comparison of Output Gap Estimates

-3

-2

-1

0

1

2

2001 2002 2003 2004 2005 2006

Gap - ex post Gap - real-time Gap - HP filter

Note: This chart presents the three measures of the output gap used in the paper: the output gap estimated by the CNB as of their October 2006 forecast round (Gap - ex post), the real-time based output gap estimated by the CNB (Gap - real-time) and the output gap calculated using the HP filter (Gap – HP filter) as the estimate of potential output.


Chart A.3 – Importance of Bias Correction Terms in Estimating Policy Rules

Backward-looking policy rule

-0.50

-0.25

0.00

0.25

0.50

0.75

1.00

1.25

2001 2002 2003 2004 2005 2006-.6

-.4

-.2

.0

.2

.4

.6

.8

2001 2002 2003 2004 2005 2006-2

-1

0

1

2

3

2001 2002 2003 2004 2005 2006

Current inflation, output gap – HP filter

Current inflation, output gap – ex-post

Current inflation, output gap – real-time

Forward-looking policy rule

-.8

-.6

-.4

-.2

.0

.2

.4

2001 2002 2003 2004 2005 2006-.8

-.6

-.4

-.2

.0

.2

.4

2001 2002 2003 2004 2005 2006-.8

-.6

-.4

-.2

.0

.2

.4

.6

.8

2001 2002 2003 2004 2005 2006




Note: The difference between the policy neutral rates estimated from the time-varying parameter

model with endogenous regressors, ertr ,

_, with its confidence intervals, and from the conventional

time-varying parameter model, ctr ,

_

. The measure of the output gap and inflation used for the

estimation of the policy neutral rate is reported below each chart. Consequently, if the confidence

intervals are different from zero, it means that ctr ,

_

does not lie within the confidence intervals of

ertr ,

_.


Chart A.4 – Actual Interest Rate and Policy Neutral Rate

1

2

3

4

5

6

2001 2002 2003 2004 2005 2006

Neutral rate Actual rate

Table A.1 – KPSS Test

Series Test statistic

PRIBOR 3M 0.355*

CPI Inflation 0.165

CPI Inflation forecast (t+12) 0.163

Net Inflation 0.147

Output gap – HP filtered 0.106

Output gap – Real-time 0.293

Output gap – Ex-post 0.214

M2 growth 0.168

Real effective exchange rate 0.447*

The null hypothesis is that the series is level stationary. Critical values for the null hypothesis: 10% - 0.347, 5% - 0.463, 1% - 0.739. Sample period: 2001:1-2006:09. *, **, *** denote significance at the 10, 5 and 1 percent level, respectively.


Table A.2 – Monetary Policy Rules Estimation

( ) ( ) tttitittt rxrr ερβππαρ ++⎥⎦⎤

⎢⎣⎡ +−+−= −++ 1

*_

1

ttt rr ϑ+= −1

__, ( )2,0...~ ϑσϑ Ndiit

ttε,t,εtε,tν,εt ισκνσκ ++= ϕε ϕ , ( )2,

2,

2, )1(,0~ tvt Nι εεϕε σκκ −−

Model Parameters 1 2 3 4 5 6 7 8 9 10 11 12

ρ 0.40*** 0.40*** 0.40*** 0.41*** 0.40*** 0.40*** 0.42*** 0.40*** 0.42*** 0.42*** 0.45*** 0.40***

(0.02) (0.09) (0.06) (0.06) (0.07) (0.09) (0.11) (0.10) (0.10) (0.09) (0.10) (0.15)

α 0.27*** 0.07 0.28*** 0.06 0.28*** 0.07 -0.15 -0.15 -0.18 -0.16 -0.17 -0.14

(0.07) (0.07) (0.08) (0.06) (0.07) (0.06) (0.12) (0.11) (0.11) (0.11) (0.10) (0.09)

β 0.21 0.11 -0.06 0.57** -0.06 -0.06 0.12 0.14 0.66** 0.66** 0.18 -0.02

(0.24) (0.24) (0.28) (0.29) (0.22) (0.23) (0.27) (0.25) (0.28) (0.31) (0.23) (0.19)

ν,εκ -0.06*** -0.07*** -0.06*** 0.01 0.02 0.02

(0.01) (0.02) (0.02) (0.01) (0.01) (0.01)

,εκϕ -0.02* -0.02* -0.02* 0.01 -0.01 -0.02***

(0.01) (0.01) (0.01) (0.01) (0.02) (0.01)

AIC -1.10 -1.00 -1.07 -1.00 -0.95 -0.94 -0.91 -0.94 -0.94 -0.96 -0.91 -0.88 Note: Robust standard errors in brackets. ***, ** and *indicate significance at 1, 5 and 10%, respectively. The models differ according to whether bias correction terms are included and the specification of it+π and tx . it+π is either the CNB inflation forecast one year ahead (abbreviated as IF below) or the current inflation rate (IC). tx is a measure of the output gap: 1. as estimated by HP filtering (HP), 2. CNB ex-post output gap measure based on multivariate Kalman filter procedure (EX), 3. CNB real-time output gap measure based on multivariate Kalman filter procedure (REAL). Models 1 and 2 = IC, HP; Models 3 and 4 = IC, EX; Models 5 and 6 = IC, REAL; Models 7 and 8 = IF, HP; Models 9 and 10 = IF, EX and Models 11 and 12 = IF, REAL;

Price Setting and Market Structure - 43 -

2 Price Setting and Market Structure: An Empirical Analysis of Micro Data *

2.1 Introduction

There is a large and growing body of empirical literature based on micro data on price setting in

Eurozone countries, with a view to uncover microeconomic sources of price inertia and possible

asymmetries across countries (e.g. the “Inflation persistence network”, see Dhyne et al., 2006). By

contrast, there is very little evidence based on comprehensive micro data on price setting in New

European Union (EU) members (or other emerging economies). In this paper, we analyze actual

prices for a wide range of products, which form a large proportion of the consumer basket in

Slovakia. In addition to the richness of our dataset, the analysis of a country like Slovakia has two

interesting implications.

First, in the sample period Slovakia experience an average rate of inflation near two-digit levels,

in contrast to previous studies focusing on low inflation countries. In principle, pricing policies

can be quite different from those in more stable macroeconomic environment (Calvo et al., 2002).

Second, there is a debate in Europe on the potentially large asymmetries between price rigidity in

the new as opposed to old EU members. These asymmetries would imply sharply asymmetric

effects of monetary policies for New member states when they will eventually join the Eurozone

(Elbourne and de Haan, 2006). The presence of strong asymmetries would call for delaying entry

in the Eurozone. One of the main objectives of our analysis is indeed to verify the view

according to which a country like Slovakia, undergoing a process of massive structural change

and market liberalization, is characterized by a more rigid price system, inducing a higher degree * We thank Ian Babetskii, Mihai Copaciu, Mario Holzner, Július Horváth, Attila Rátfai, Stanislav Vidovič†, Sangeeta Pratab, Zoltán Wolf, Petr Zemčík and the seminar participants at the European Economic Association Annual Congress, Hungarian National Bank 5th Macro Workshop, Czech Economic Association Annual Conference, the Deutsche Bundesbank, Inflation Persistence Network Follow-up Meeting, and Charles University (Prague) for valuable comments. We are especially grateful to Július Horváth for providing us with the data. This research was supported by a grant from the CERGE-EI Foundation under a programme of the Global Development Network. All opinions are those of authors and have not been endorsed by CERGE-EI or the GDN.


of inertia (persistence) in price dynamics (Dhyne et al., 2006). In fact, this view does not find

empirical support in the case of Slovakia. By comparing price inertia for different sectors,

characterized by different market structures, namely manufacturing vs. services, we find that a

lower degree of market competition is associated with higher price dispersion and lower

persistence. This should not come as a surprise, as in the well-known staggered price model of

Calvo (1983) as market competition increases inertia tends to increase rather than decrease (see

Calvo, 2000). This result has relevant implications as the process of integration of Slovakia in the

EU, and the attendant higher degree of competition in goods markets, is likely to increase rather

than decrease inflation inertia.

The paper is organized as follows. In section 2, we discuss a simple analytical framework, which

will serve as basis for the empirical analysis. In section 3, we study price setting behaviour of 423

narrowly defined products. More specifically, we estimate the frequency and magnitude of price

changes, price dispersion and inflation persistence at the level of price setter using a unique

dataset, covering a large component of the Slovak consumer price index (CPI) during the period

1997-2001. We identify the factors that affect price setting behaviour, namely, the determinants

of frequency and size of price changes, the level of inflation persistence and price dispersion.

Section 4 concludes and provides some policy implications. Appendix contains the formal

definitions of price setting descriptive statistics.

2.2 Price Setting Behavior Although there is disagreement on the relevant theoretical model for price setting, Calvo’s

staggered price model has become a benchmark in the literature. In its reduced form, Calvo’s

model gives a useful framework for empirical analysis.

Calvo price setting has several virtues: first, it can accommodate the case of perfect competition

and price flexibility as a limiting case; second, it highlights factors affecting inertia that are likely

to be relevant in most theories of price adjustment; third, it allows to distinguish factors related


to market structure from those linked to aggregate macroeconomic variables, that in turn are

affected by policy.

As an illustration of the framework, consider the well-known New Keynesian Phillips curve

(NKPC) obtained from a discrete time version of Calvo’s model. As in Calvo (1983), firms

change their prices at random intervals, so that at each point in time there is a fraction α of firms

that set new prices and a fraction 1-α that keep their prices unchanged. New prices are set

optimally by firms facing a downward-sloping demand function with price elasticity -θ. Denoting

with π the rate of inflation, such a NKPC relates current inflation to expected inflation and to

current output gap (Yt-Ytⁿ):

( )( )[ ] ( )ntttt YE −−−+= + t1 Y11 ςαβααπβπ

where β is the discount factor and ς a coefficient that characterizes the link between marginal

costs and output. It is easy to show that the parameter ς is a decreasing function of the parameter

θ, that measures the degree of competition in goods markets (see Woodford (2003) and Calvo

(2000)). As markets become more competitive, ς declines. Under perfect competition (θ→∞) ς

becomes zero.

The model can be closed by deriving the dynamics of the output gap and assuming a policy rule

determining the interest rate.

Focusing on two parameters of the model: α, the frequency of price adjustment and, θ, the

elasticity of substitution among goods, Calvo’s model has two main implications: first the higher

is α , the faster is the adjustment (lower inertia); second, and much less noted in the literature

(Calvo 2000 is the exception), the higher is θ, thus the more competitive are goods markets, the

slower is the adjustment, and thus the higher is inertia. Note that in the simplest version of the

model it is assumed that α is a constant. In Yun (1996) α is instead a function of the average

inflation rate. In addition, α could be affected by market structure, and increase with market

competition. Although more controversial, we assume in the empirical analysis that this channel

could be at work, and thus, through higher α market competition reduces inertia. At the same


time, market competition increases inertia through the parameter θ. The mechanism is related to

the so-called strategic complementarity in price behavior and thus has to do with interaction

among price setters (Woodford 2003). The intuition is that as competition increases firms will

tend to “follow the pack”, as deviations from the average price may push the firm out of the

market (see Calvo, 2000).25

Summing up, increasing market competition may increase or decrease inertia (persistence)

depending on the relative strength of two conflicting effects. One plausible assumption would be

that the effect of varying α with average inflation is rather weak: it is necessary to have a

substantial increase in average inflation to significantly modify price behavior in terms of

frequency of adjustment (Golosov and Lucas, 2007).

2.3 Empirical analysis 2.3.1 Dataset The dataset contains the price records of 604 products collected at monthly frequency by the

Slovak Statistical Office (SSO) in 38 districts in 1997-2001. For each record in the dataset, there

is information on the date (month and year), district, product category code and the price of item.

The data allow tracking individual price dynamics. The price for each product is collected

monthly at several stores in the district, but typically only from the capital town of the district.26

On average, about five stores are monitored in a particular district. As a result, each product

contains around 10,000 records. Considering all products, the dataset contains more than 5

millions observations. The dataset contains actual prices as opposed to quoted prices or price

indices. The final dataset has been reduced, excluding those products that featured a relatively

25 Note also that the deviation from the price of competitors has been found as one of the most important obstacles for price adjustment in surveys of euro area firms (see Fabiani et al., 2006). 26 See Horvath and Vidovic (2004) for more details on the methodology of data collection at the SSO.


high share of missing records and regulated prices that account for up to 20 percent of the total

CPI basket.27

Table 1 indicates 12 categories in which products are classified, and reports the original weight of

a given product category in the CPI basket, sample weight and the number of products in the

sample for each category in the dataset.

Table 1 – Coverage of the Dataset

Original Weights

Sample Weights

Number of products in the sample

Food and Non-Alcoholic Beverages 23.60 36.17 121 Alcoholic Beverages and Tobacco 6.98 12.05 10 Clothing and Footwear 7.51 8.43 68 Housing, Water, Gas and Electricity 21.50 4.85 14 Furnishing & Maintenance of Housing 5.18 3.6 70 Health Care Expenses 1.45 0 0 Transport 9.25 9.75 20 Communications 2.73 0 0 Leisure and Culture 7.21 9.19 56 Education 0.58 0 0 Hotels, Cafés and Restaurants 7.22 9.42 30 Miscellaneous Goods and Services 6.79 6.54 34 Total 100 100 423

2.3.2 Evidence This section contains evidence on price setting behavior in Slovakia. First, we characterize the

dynamics of Slovak inflation and analyze its persistence at various levels of aggregation. Second,

we estimate additional pricing statistics28 such as the frequency and size of price changes and

examine correlations among them. Third, we study the factors determining price setting behavior.

2.3.2.1 Inflation Dynamics and Persistence

27 In general, see Aucremanne and Dhyne (2004) for a comprehensive discussion of methodological issues related to the censored nature of dataset. 28 The detailed product-specific results are available upon request from the authors.


The average annual CPI inflation rate in Slovakia has been about 9% in the period 1998-2001.

There is a notable hike in the inflation rate starting in mid 1999, which has been caused mainly by

price deregulations and regulated prices increases. This one-off shock has merely changed the

price level and thus largely vanished after one year. From the end of 2000, there is a gradual

slowdown in the inflation rate to some 6% at the end of the sample period.

Figure 1 – Official CPI inflation and sample inflation, 1998-2001

2

4

6

8

10

12

14

16

18

98:3 99:1 99:3 00:1 00:3 01:1 01:3

FULL SAMPLE

Source: Slovak Statistical Office, own calculations

Figure 1 reports the Slovak official CPI inflation and our sample inflation. The sample contains

57% of the full CPI basket. The sample inflation is lower than aggregate inflation by 3.3

percentage points on average. The difference between official and sample inflation is attributable

chiefly to the rate of increase in regulated prices.29

Next, we examine the degree of inflation inertia by applying a non-parametric measure by

Marquez (2004) and Dias and Marquez (2005). This approach builds on the idea that a less

persistent inflation is more likely to cross its long-run (possibly time-varying) mean of inflation

29 For example, from January 1999 to January 2000 regulated prices increased by 33%. Given the weight in the CPI index of 17.8% (see Monetary Survey of National Bank of Slovakia, January 2000), this contributed 5.9 percentage points to the official inflation and as such, it almost fully explains the difference between official and our sample inflation.


rate. Specifically, a measure of persistence γ is calculated as follows: Tn /1−=γ , where n is

the number of times the series crosses its long-run mean and T is the number of observations.

Given the length of our sample, we opt for a simple univariate filter (Hodrick-Prescott filter) to

approximate the long-run time-varying mean.30

This non-parametric approach has several attractive features over more common parametric

measures, typically based on the sum of autoregressive parameters in the regression of inflation

on its lagged values. Dias and Marquez (2005) derive finite sample and asymptotic properties of

this non-parametric measure. When conducting Monte Carlo simulations, they find that the bias

of the estimate of persistence based on non-parametric approach is smaller for any sample size,

as compared to the parametric measure. Besides, they argue that non-parametric measure is more

robust to structural breaks and additive outliers, which is appealing in case of analysis of

emerging market economies. Marquez (2004) shows that the values of γ close to 0.5 indicate

absence of persistence in the series. Values significantly above 0.5 signal a positive

autocorrelation in the series, while values substantially below 0.5 signal negative autocorrelation.

Figure 2 - Inflation Persistence, 423 products

0

10

20

30

40

50

60

70

80

0.5 0.6 0.7 0.8 0.9 1.0

30 The standard smoothing parameter of 14400 has been used for all products. Different smoothing parameters, such as the one suggested by Ravn and Uhlig (2002), had very little impact on the resulting estimates.


The results presented in Figure 2 indicate that out of 423 products, only 4 of them display no

persistence (γ <0.66 in our case, as ( )5.02 −γT approximately distributed as N(0,1)), while the

remaining 419 products display positively persistent inflation, at the 5% significance level.

The results on the degree of inflation persistence are reported in Table 2. Given asymmetries

within the sectors, we average the product-level inflation persistence, instead of estimating

persistence on sectoral data directly, to reduce the potential aggregation bias (see Granger, 1980

and Zaffaroni, 2004). Indeed, the estimate of persistence of aggregate inflation is somewhat

greater than the average of estimates at the product-specific level (0.9 vs. 0.87). All sectors display

very high persistence, with the persistence in services inflation being somewhat lower.31 This is

surprising, as the services production is relatively more labor intensive and as such, it is expected

that the persistence should be greater than for the other sectors. It is noteworthy that also Clark

(2006) does not find persistence in the services sectors in the U.S. to be greater than other

components of consumer prices and, for some specifications, services show the smallest

persistence. Similarly, the results of Lunnemann and Matha (2004) for 15 EU countries indicate

that services dummy is significantly negative in nearly all their fixed effects regressions of the

determinants of persistence32 (see also Altissimo et al., 2007 for evidence that services often

exhibit smaller persistence than other products in the consumer basket). However, these papers

do not provide an explanation for this ”services inflation persistence puzzle”.

Even though we do not want to overestimate the robustness of this result, it is remarkable that

services, a sector typically characterized by a lower degree of competition, does not display higher

persistence in inflation. It is worth noting that Calvo’s model of staggered prices predicts such

result, as higher degree of competition in the market produces a process of “follow the pack”,

31 The difference in the results between the expenditure-weighted and non-weighted persistence in services inflation is largely driven by a single item ‘complete lunch in a factory canteen’. This item’s inflation persistence stands at 0.71 and its sample weight is 6.9%. 32 The same authors in a different study, Lunnemann and Matha (2005a) however report that if they exclude services from consumer prices then the estimated inflation persistence of the remaining basket declines in comparison to the full basket.


which is consistent with a high degree of price homogeneity across firms (Calvo, 2000). When

markets are highly competitive individual prices cannot diverge much from the average, as firms

would loose large shares of their market. In the limiting case of perfect competition, prices would

be exactly the same across firms. This effect arises because the degree of strategic

complementarity increases with higher competition, implying that the strategy of an individual

price setting firm is an increasing function of the average strategy (price) in the market (see also

Woodford, 2003). Therefore, although competition reduces costly dispersion in prices, from a

dynamic perspective it may increase persistence.33

Table 2– Inflation Persistence

No. of Products

Sample Weights

Inflation Persistence

Inflation Persistence – Weighed

Processed Goods 375 79.28 0.874 0.867 Raw Goods 48 20.72 0.846 0.875 Perishables 64 23.65 0.862 0.869 Durables 231 36.39 0.874 0.886 Non-durables 136 49.62 0.874 0.876 Services 56 13.99 0.851 0.796 Total 423 100 0.871 0.868

Notes: Raw goods category contains meats, fruits, vegetables, milk, cream, honey, eggs, salt, mineral water, gasoline, fuel oil, motor oil and coolants. The results in the last column are expenditure-weighted. Non-durables contain mainly food and beverages. Services include mainly the category ‘Hotels, cafés and restaurants’ and fees and repairs for various categories of products. Durables contain the remaining products. Perishables are a sub-category of food. 2.3.2.2 Frequency of Price Changes Table 3 shows that the estimated expenditure-weighted frequency of price changes is 0.34 in our

sample. This means that approximately one in every three consumer prices is changed in a given

month. It implies that the expenditure-weighted average duration of a price spell is 3.75 months

(and 4.2 months without CPI weights). As the distribution of the duration is asymmetric, the

median duration reaches 3.9 months.34 Thus, consumer prices in Slovakia change more often

than the one year frequency often found for the advanced market economies (see Dhyne et al.,

33 Coricelli (2005) discusses the implications of this issue for the conduct of monetary policy. An overview on inflation persistence and the conduct of monetary policy is presented in Levin and Moessner (2005). 34 Additional results using median, weighted median and simple average to estimate the frequency are available on a request.


2006). The greater frequency of price changes in Slovakia, as compared to advanced market

economies, is likely due to a higher inflation rate, as well as a smaller share of services in the

consumer basket. Next, frequent price adjustments may also reflect structural changes in the

economy. In such a context, demand tends to be more uncertain and, consequently, firms have to

experiment to find their optimal price to be set.35 The probability that the single price spell would

last longer then 12 months is essentially zero. More specifically, there are only 3 out of 423

products having the average duration of price spells longer than one year.

There is a considerable degree of heterogeneity in terms of the frequency of price changes.

Products such as fruits and vegetables or gasoline typically change their price less than bimonthly.

On the other hand, several services keep the price fixed for almost 2 years. The duration of a

price spell is more than 7 months for services. This is likely due to the fact that labor-intensive

services are typically less exposed to international competition. Furthermore, as noted by Bils and

Klenow (2004), the lower variability of demand for services can be behind their prolonged

inaction in price adjustment. At the other extreme, prices change most frequently for the raw

goods. Diversification of inputs for raw goods is typically limited, as compared to processed

goods and thus price changes are triggered more often.

Table 3 - Frequency of Price Changes

No. of products

Sample Weights

Average Frequency

Average Duration


Notes: Frequency refers to the frequency of price changes, i.e. empirical probability that price of the product will change. Duration indicates the number of months between price changes. CPI weights are used for weighting. See Table 2 for the classification of products into categories.

35 Rothschild (1974) presents a model in which monopolistic price setter learns gradually about optimal price in a noisily observed demand.


The frequency of price changes between 0.1-0.4 is far more common than other frequencies (see

Figure 3). There are only four products, truly flexible prices, changing the price more often than

in 80% of the cases. Overall, this evidence shows that one cannot simply refer to a given measure

price stickiness or price flexibility, as the degree of price stickiness varies dramatically across

products. The distribution of the frequency price changes is skewed to the right, similarly to what

has been found for other countries (see e.g. Diaz et al., 2004, for comparable evidence on

Portugal and Baharad and Eden, 2004, for Israel).

Figure 3 - Frequency of Price Changes, Histogram

0

10

20

30

40

50

60

0.25 0.50 0.75

Table 4 puts our results in an international perspective. While the distribution of price stickiness

seems to be relatively similar to more advanced countries, the frequency is somewhat higher

reflecting the higher rate of inflation in Slovakia. Note that, despite lower rates of inflation, the

frequency for the U.S. data is similar to that for Slovakia. However, as emphasized recently by

Nakamura and Steinsson (2007), the frequency of price changes for the U.S. is about half lower,

when sales are excluded in the calculation of the frequency of price changes. In this case, the

attendant frequency falls to about 0.1, which is a value similar to the euro area. Sales thus seem to


play more important role in the U.S. than in Europe, whereby the frequency is around 0.15

regardless whether sales are included or not.

Table 4 – Frequency of Price Changes, International Comparison

Slov

akia

Aus

tria

Belg

ium

Fran

ce

Italy

Luxe

mbo

urg

Net

herla

nds

Portu

gal

USA

Food and Non-Alcoholic Beverages 0.43 0.17 0.28 0.19 0.15 0.19 0.23 0.37 0.25Alcoholic Beverages and Tobacco 0.31 0.15 0.11 0.22 0.08 0.14 0.19 0.14 NAClothing and Footwear 0.25 0.12 0.03 0.18 0.05 0.20 0.21 0.27 0.29Housing, Water, Gas and Electricity 0.19 0.11 0.22 0.24 0.23 0.29 0.19 0.08 NAFurnishing & Maintenance of Housing 0.24 0.07 0.04 0.16 0.04 0.18 0.08 0.11 0.26Health Care Expenses NA 0.06 0.11 0.08 NA 0.03 NA 0.05 0.09Transport 0.59 0.36 0.21 0.36 0.28 0.21 0.88 0.26 0.39Communications NA 0.09 0.06 0.23 NA 0.04 NA 0.11 NALeisure and Culture 0.24 0.24 0.12 0.13 0.05 0.13 0.08 0.12 0.11Education NA 0.05 NA 0.06 NA 0.05 NA 0.08 NAHotels, Cafés and Restaurants 0.14 0.08 0.03 0.08 0.06 0.05 0.08 0.19 NAMiscellaneous Goods and Services 0.25 0.07 0.06 0.12 0.04 0.11 0.10 0.11 0.11 % of CPI 57 100 68 65 20 100 8 100 ---Total 0.34 0.15 0.17 0.19 0.09 0.17 0.17 0.22 0.26Average inflation rate, in %, yearly 9 2 2.2 1.5 2.5 2.5 2.9 2.6 2.4

Notes: Own calculations for Slovakia, the authors of results in other countries are as follows: Austria –Baumgartner et al. (2005), Belgium - Aucremanne and Dhyne (2004), France - Baudry et al. (2004), Italy – Veronese et al. (2005), Luxembourg – Lunnemann and Matha (2005b), Netherlands – Jonker et al. (2005), Portugal – Dias et al. (2004), USA – Bils and Klenow (2005). All averaged frequencies are expenditure-weighted. International comparison of the frequency of price changes in the Euro area countries based on 50 representative products is available in Dhyne et al. (2006).

2.3.2.3 Magnitude of Price Changes In this section, we estimate the average size of price increases and decreases. We find that the

magnitude of price changes is sizeable in both directions. The average size of expenditure-

weighted price increases and decreases is 12% and -11%, respectively. The corresponding size of

changes rises to 16% and -14% without CPI weights.

The results are comparable to the findings for other Euro area countries (see Dhyne et al., 2006),

which indicate that the magnitude of price changes is typically nearly 10%, both for price


increases and decreases. In general, the larger size of price changes in Slovakia may indicate lower

degree of market competition, as compared to more developed markets in Western Europe.

Figure 4 - Size of Price Changes

0

40

80

120

160

200

-0.4 -0.2 0.0 0.2 0.4 0.6

Although most price increases and decreases range between 5-15% in absolute terms (274 and

303 products, respectively), there are very long tails toward one (see Figure 4).

Looking at the product groups, Table 5 shows that services exhibit the greatest magnitude of

price changes, at near 15% in absolute terms. On the other hand, non-durables, raw goods and

perishable products tend to display much smaller changes in prices. Durables and processed

goods stand somewhere in between.


Table 5 - Size of Price Changes, Raw vs. Processed Goods, Perishables and Durables,

Non-Durables and Services

No. of products

Sample Weights

Increase – Weighted

Decrease - Weighted

Processed Goods 375 0.79 0.12 -0.12 Raw Goods 48 0.21 0.10 -0.08 Perishables 64 0.24 0.11 -0.10 Durables 231 0.36 0.13 -0.11 Non-durables 136 0.50 0.10 -0.10 Services 56 0.14 0.17 -0.15 Total 423 1 0.12 -0.11

Notes: See Table 2.

2.3.2.4 Price Dispersion

Price dispersion is one of the main characteristics of the market structure. While there is a variety

of theoretical models on price dispersion and market structure, recent empirical evidence suggest

that price dispersion decreases with market competition. The results of Caglayan et al. (2007)

show that price dispersion is lower in more competitive environment using micro level price data

from Turkey. Baye et al. (2004) analyze internet prices and find that price dispersion is greater

when smaller number of firms list their prices on internet price comparison site (this is also

found by Leiter and Warin (2007), who use different online shopping/price comparison site).

Similarly, Gerardi and Shapiro (2007) find a negative effect of competition on price dispersion,

with the effect being the most significant on the routes with more heterogeneous customer base.


Figure 5 –Product-Specific Price Dispersion

0

10

20

30

40

50

60

70

0.0 0.1 0.2 0.3 0.4 0.5 0.6

We examine to what extent prices of identical products sold in the same month differ across

different stores. Figure 5 shows large cross-sectional variation in price dispersion. Table 6

suggests that prices in services sector are by far the most dispersed, in comparison to non-

durables, raw goods and perishables is twice as much. Interestingly, our results appear to be in

line with Crucini et al. (2005), who find, using micro data from the EU-15 that the price

dispersion decreases with tradability of the product.

Table 6 – Price Dispersion

No. of products

Sample Weights

Dispersion –Weighted

Dispersion –No weights


Notes: See Table 2.

2.3.2.5 Trade-off between Frequency and Size of Price Changes? Menu costs models predict that when there are significant costs of price adjustment, price

adjustment should occur less frequently and the change should be sizeable (Mankiw, 1985). In


this regard, Carlton (1986) finds a negative correlation between the frequency of price changes

and the average absolute price change. In principle, the correlation may differ according to

whether the price increases or decreases. We find that the simple correlation between product-

specific frequency and size of price increases is -0.03, but fails to be significant. On the other

hand, frequency the size of price decreases are negatively correlated, with a coefficient oft -0.17,

significant at 5% level. This would suggest some mild support for the notion of trade-off

between the frequency and the magnitude of price changes.

To examine this issue further, we fit a spline among the frequency, size of price increases and

decreases. Interestingly, the results in Figure 6 point to a negative relationship between the

frequency and size only for more rigid prices. For some components of the basket, the

relationship is even reversed. These are products changing prices often and by large amounts

(such as fruits and vegetables). There also seem to be products with convex costs of price

adjustment, in line with Rotemberg (1982) model. For instance, products such as gasoline change

prices often, but only by a tiny magnitude.

We next try to identify the factors that affect the firm’s price behavior, analyzing the

determinants of inflation persistence, frequency of price changes and price dispersion.


Figure 6 – Frequency and Size of Price Changes

2.3.2.6 Determinants of Price Setting Behavior Table 7 presents the results on the determinants of inflation persistence. Price dispersion, as a

proxy for the level of market competition, tends to decrease persistence.36 This is in line with our

conjecture laid out in section 2; a more competitive environment is associated with greater

persistence of shocks. Furthermore, a greater frequency of price changes is associated with

smaller persistence. Galí (2004) emphasizes that this finding is likely to be a consequence of the

backward looking behavior of some fraction of price setters. Assuming a hybrid NKPC as in Gali

and Gertler (1999), he shows that inflation persistence is an increasing function of the fraction of

backward looking price setters. In this regard, Cecchetti and Debelle (2006) claim that greater

price rigidity generates more backward looking behavior and thus more persistence. It is worth

mentioning that here our results differ from Bils and Klenow (2004), who report a positive

correlation between the frequency of price changes and inflation persistence for their full sample

36 The list of our instrumental variables used in this section includes: raw goods, services, durables, perishables and expenditure weight. Note that the set of instruments differs across the tables and thus the degrees of freedom for Sargan-Hansen test varies correspondingly.


(and statistically insignificant correlation for the short sample). In contrast, we find a statistically

significant correlation of -0.14, which gives some support to Calvo pricing model.

Table 7 – Determinants of Inflation Persistence

(1) (2) (3) (4) (5) Price dispersion -0.18*** -0.41*** (0.07) (0.12) Frequency 0.04 -0.17*** (0.04) (0.06) Raw goods -0.03** -0.04** (0.01) (0.02) Services -0.03** -0.04*** (0.01) (0.01) Durables -0.01 (0.01) Perishables -0.002 (0.01) No. of observations 423 423 423 423 423 Adj. R-squared --- --- --- 0.03 0.04 Sargan-Hansen test 7.97(0.09) 13.9(0.01) 1.17(0.56) --- --- Estimation Method GMM GMM GMM OLS OLS

Note: Heteroscedasticity corrected standard errors & covariance. ***, **, and * - denotes significance at 1 percent, 5 percent, and 10 percent, respectively.

Staggered price setting models predict that in the steady state the products with less frequent

price changes exhibit greater price dispersion and smaller product-specific inflation rate. The

price dispersion among the homogenous products may occur because of the inability of firms to

adjust instantaneously their price to a given shock. The greater the inability, the greater is price

dispersion. Higher inflation rates erode the real price more quickly and therefore price adjustment

is triggered more often. Baharad and Eden (2004) try to discriminate between staggered price

models and uncertain and sequential trade (UST) models by running a regression between the

frequency of price changes and price dispersion and inflation rate.37 In contrast to the UST

model, the staggered price model implies a significant relationship between frequency of price

37 UST models assume that there is price dispersion in the equilibrium. There is a trade-off between high price and the probability of making a sale in the model. If seller quotes relatively low price, he increases the changes of selling his product. Seller may also charge high price, but at the expense that the probability of making a sale is then rather low. This implies that the seller does not have to change the price of a product, even if the inflation erodes his price, because it increases the probability of making a sale. Therefore, UST models claim that prices may be seemingly rigid, to a certain extent.


changes, price dispersion and inflation rate. In addition, we also examine the importance of

product characteristics in determining the length of inaction in price adjustment.

Table 8 contains the results of our analysis on the determinants of frequency of price changes.

The results suggest that product characteristics are the primary force triggering the price changes.

Individual inflation rate matters as well. Price dispersion seems to be associated with less frequent

price changes, but this result should be interpreted with caution, as the test for overidentifying

restrictions is rejected.38 We have also tested for the presence of any non-linear relationship

between frequency and inflation, but we failed to find any significant non-linearity. All in all, our

results give support to the staggered price setting model.

Table 8 – Determinants of Frequency of Price Changes

(1) (2) (3) (4) (5) Price dispersion -3.62*** -1.63*** (0.78) (0.21) Individual inflation 0.06** 0.02** (0.03) (0.01) Raw goods 0.26*** 0.17*** (0.03) (0.02) Services -0.10*** -0.17 *** (0.01) (0.02) Durables -0.10*** (0.01) Perishables 0.05*** (0.02) No. of observations 423 423 423 423 423 Adj. R-squared --- --- --- 0.47 0.59 Sargan-Hansen test 6.00(0.01) 2.25(0.13) 11.87(0.01) --- --- Estimation Method GMM GMM GMM OLS OLS


Table 9 contains our results on the determinants of price dispersion. We find that greater

frequency of price changes decreases the dispersion, while individual inflation increases it. The

38 The attendant OLS estimates, available upon request, show that a higher degree of competition is associated with less rigid prices, as found e.g. by Bils and Klenow (2004).


latter finding is in line with large literature represented by e.g. Lach and Tsidon (1992) and

Konieczny and Szkrypacz (2005), who find that inflation adds to price variability, while the

former comply with Caglayan et al. (2007). In addition, prices in services exhibit greater

dispersion. On the other hand, prices of raw goods are typically less dispersed.

Table 9 – Determinants of Price Dispersion

(1) (2) (3) (4) (5) Frequency -0.50*** -0.33*** (0.05) (0.09) Individual inflation 0.02*** 0.02*** (0.01) (0.01) Raw goods -0.04*** 0.01 (0.01) (0.01) Services 0.09*** 0.15 *** (0.02) (0.02) Durables 0.08*** (0.01) Perishables 0.01 (0.01) No. of observations 423 423 423 423 423 Adj. R-squared --- --- --- 0.12 0.25 Sargan-Hansen test 25.6(0.00) 3.43(0.18) 1.20 (0.27) --- --- Estimation Method GMM GMM GMM OLS OLS


2.4 Conclusions The analysis on price setting for a large micro dataset based on actual prices indicates that

Slovakia shares many of the features found in micro studies for advanced market economies.

All in all, not only average inflation, but also the frequency of price changes and the persistence

in inflation appear to be higher in Slovakia as compared to advanced economies. The significantly

higher persistence in emerging economies that are in the process of joining the European

Monetary Union is a cause for concern for policy-makers in the Eurozone, as it would imply

asymmetric effects of the common monetary policy, and persistent disequilibria in real exchange

rates across members of the Eurozone. Looking forward, higher market competition brought


about by the integration in the European Union and in the European Monetary Union are

unlikely to reduce such persistence, as our results indicate a positive correlation between market

competition and inflation persistence. By contrast, an increase in the importance of the service

sectors, that are at present compressed in emerging countries like Slovakia, may in fact reduce

persistence. Therefore, it is not easy to predict the tendency in persistence as Slovakia becomes

more and more integrated in the EU and its economic structure converges to that of EU

countries.


References Altissimo, F.; Mojon, B. and P. Zaffaroni, 2007. Fast Micro and Slow Macro: Can Aggregation

Explain the Persistence of Inflation? Federal Reserve Bank of Chicago, Working Paper,

No. 02/2007.

Aucremanne, L. and E. Dhyne (2004): How Frequently Do Prices Change? Evidence Based on

the Micro Data Underlying the Belgian CPI, European Central Bank WP, No. 331.

Baharad, E. and B. Eden (2004): Price Rigidity and Price Dispersion: Evidence for Micro Data,

Review of Economic Dynamics, pp. 613-641.

Barro, R. (1972): A Theory of Monopolistic Price Adjustment, Review of Economic Studies, pp. 17-

26.

Baudry, L.; Le Bihan, H.; Sevestre, P. and S. Tarrieu (2004): Price Rigidity in France – Evidence

from Consumer Price Micro-Data, European Central Bank WP, No. 384.

Baumgartner, J, Glatzer, E., Rumler, F. and A. Stiglbauer (2004): How Frequently Do Comsumer

Prices Change in Austria?, European Central Bank WP, No. 523.

Baye, M. R., Morgan, J. and P. Scholten (2004): Price Dispersion in the Small and in the Large:

Evidence from an Internet Price Comparison Site, Journal of Industrial Economics, 52, pp. 463-

496.

Bils, M. and P. Klenow (2004): New Evidence on the Importance of Sticky Prices, Journal of

Political Economy, pp. 947-985.

Caglayan, M., Filiztekin, A. and M.T. Rauh (2007): Inflation, Price Dispersion and Market

Structure, European Economic Review, forthcoming.

Calvo, G. (1983): Staggered Pricing in a Utility Maximizing Framework, Journal of Monetary

Economics, pp. 383-398.

Calvo, G. (2000): Notes on Price Stickiness: with Special Reference to Liability Dollarization and

Credibility, University of Maryland, manuscript, available at

http://www.bsos.umd.edu/econ/ciecrp.htm


Calvo, G.; Celasun, O. and Kumhof, M., 2002. A Theory of Rational Inflationary Inertia, in: P.

Aghion, R. Frydman, J. Stiglitz and M. Woodford, eds., Knowledge, Information and

Expectations in Modern Macroeconomics: In Honor of Edmund S. Phelps. Princeton:

Princeton University Press.

Clark, T. (2006): Disaggregate Evidence on the Persistence in Consumer Prices Inflation, Journal

of Applied Econometrics, pp. 563-587.

Coricelli, F. (2005): Central Bank Conservatism and Market Competition in a New Keynesian

Model, University of Siena, manuscript.

Crucini, M., Telmer, C. I. and M. Zachariadis (2005): Understanding European Real Exchange

Rates, American Economic Review, pp. 724-738.

Dias, M.; Dias, D. and P. Neves (2004): Stylised Features of Price Setting Behaviour in Portugal:

1992-2001, European Central Bank Working Paper Series No. 332.

Dias, D. and C. Marquez (2005): Using Mean Reversion as a Measure of Persistence, European

Central Bank Working Paper, No.450.

Dhyne, E.; Alvarez, L.; Le Bihan, H.; Veronese, G.; Dias, D.; Hoffman J.; Jonker, N.; Lunneman,

P.; Rumler, F. and J. Vilmunen (2006): Price Setting in the Euro Area: Some Stylized Facts

from Individual Consumer Price Data, Journal of Economic Perspectives, pp.171-192.

Elbourne, A. and J. de Haan (2006): Financial Structure and Monetary Policy Transmission in

Transition Countries, Journal of Comparative Economics, pp. 1-23.

Fabiani, S.;Druant, M.; Hernando, I.; Kwapil, C.; Landau, B.; Loupias, C.; Martins, F.; Matha, T.;

Sabbatini, R. and A. Stockman (2006): The Pricing Behavior of Firms in the Euro Area:

New Survey Evidence, International Journal of Central Banking, pp. 3-47.

Galí, J. and M. Gertler (1999): Inflation Dynamics: A Structural Econometric Analysis, Journal of

Monetary Economics, 44 (2), pp. 195-222.

Galí, J. (2004): Has the Inflation Process Changed? A Comment, presentation at the Third BIS

Annual Conference, Brunnen, Switzerland, 18-19 June.


Gerardi, K. and A.H. Shapiro (2007): The Effects of Competition on Price Dispersion in the

Airline Industry: A Panel Analysis, Federal Reserve Bank of Boston Working Paper, No.

07/2007

Golosov, M. and R. Lucas (2007): Menu Costs and Phillips Curves, Journal of Political Economy,

115, pp. 171-199.

Granger, C. W. J. (1980): Long Memory Relationships and the Aggregation of Dynamic Models,

Journal of Econometrics, 14, pp. 227-238.

Horvath, J. and S. Vidovic (2004): Price Variability and the Speed of Price Adjustment to Law of

One Price: Evidence from Slovakia, Bank of Finland Institute of Transition, Working Paper No.

3.

Jonker, N., Folkertsma, C. and H. Blijenberg (2005): An Empirical Analysis of Price Setting

Behavior in the Netherlands in the Period 1989-2003 Using Micro Data, European Central

Bank Working Paper, No. 413.

Konieczny, J. and A. Skrzypacz (2005): The Behaviour of Price Dispersion in a Natural

Experiment, Journal of Monetary Economics, pp.621-632.

Lach, S. and D. Tsiddon (1992): The Behavior of Prices and Inflation: An Empirical Analysis of

Disaggregated Data, Journal of Political Economy, pp. 349-389.

Leiter, D.B. and T. Warin (2007): An Empirical Study of Price Dispersion in Homogenous

Goods Markets, Middlebury College Economics Discussion Paper, No. 10/2007

Levin, A.T. and R. Moessner (2005): Inflation Persistence and Monetary Policy Design, European

Central Bank, Working Paper, No. 539.

Lunnemann, P. and T. Matha (2004): How Persistent in Disaggregate Inflation? An Analysis

Across EU15 Countries and HICP Sub-indices, European Central Bank Working Paper, No.

466.

Lunnemann, P. and T. Matha (2005a): Regulated and Services’ Prices and Inflation Persistence,

European Central Bank Working Paper, No. 466.


Lunnemann, P. and T. Matha (2005b): Consumer Price Behavior in Luxembourg: Evidence from

Micro CPI Data, European Central Bank Working Paper, No. 541.

Mankiw, G. (1985): Small Menu Costs and Large Business Cycles, Quarterly Journal of Economics,

100, pp. 529-538.

Marquez, C. (2004): Inflation Persistence: Facts or Artefacts?, European Central Bank Working

Paper, No. 371.

Monetary Survey of National Bank of Slovakia, various issues.

Nakamura, E. and J. Steinsson (2007): Five Facts about Prices: A Reevaluation of Menu Cost

Models, Harvard University, mimeo.

Ratfai, A. (2006): Linking Individual and Aggregate Price Changes, Journal of Money, Credit and

Banking, 38, 2199-2224.

Ravn, M. O. and H. Uhlig (2002): On Adjusting the Hodrick-Prescott Filter for the Frequency of

Observations, Review of Economics and Statistics, 84, pp. 371-375.

Rotemberg, J. (1982): Sticky Prices in the United States, Journal of Political Economy, 90, pp. 1187-

1211.

Rothschild, M. (1974): A Two-Armed Bandit Theory of Market Pricing, Journal of

Economic Theory, 9, pp. 185–202.

Sheshinski, E. and Y. Weiss (1977): Inflation and Cost of Price Adjustment, Review of Economic

Studies, pp. 287-303.

Taylor, J. (1999): Staggered Price and Wage Setting in Macroeconomics, in: Taylor, J. and M.

Woodford, eds.: Handbook of Macroeconomics, North-Holland, Amsterdam.

Veronese, G., Fabiani, S., Gattulli, A. and R. Sabbatini (2005): Consumer Price Behavior in Italy:

Evidence from Micro CPI Data, European Central Bank Working Paper, No. 449.

Woodford, M. (2003): Interest and Prices: Foundations of a Theory of Monetary Policy,

Princeton University Press.

Yun, T. (1996): Nominal Price Rigidity, Money Supply Endogeneity, and Business Cycles,


Journal of Monetary Economics, pp. 345-370.

Zaffaroni, P. (2004): Contemporaneous Aggregation of Linear Dynamic Models in Large

Economies, Journal of Econometrics, 120, pp. 75-102.


APPENDIX 1 – PRICING STATISTICS

In this Appendix we derive formally the pricing statistics used in the paper. This includes the

frequency of price changes, duration of single price spell, the size of price increases/decreases,

and price dispersion (inflation persistence is already defined formally in the main text).

To define the frequency of price changes and duration of single price spell formally, let istp

denote the price of product i in store s at time t , where Tt ,...,1= , Ss ,...,1= and Ii ,...,1= .

Let

1

1

1

0

−

−

≠

=⎩⎨⎧=

istist

istist

pp

pp

if

ifistx

As a result, the product-specific frequency of price changes, iµ , is computed as:

∑ ∑= =

=T

t

S

sisti x

ST 1 1

11µ

We compute average duration of single price spell, φ , as simple average over product-specific

frequencies ∑=

−=I

iiI 1

11

1 µφ or weighted average of product-specific frequencies

1

1 12

1−

= =

⎟⎠

⎞⎜⎝

⎛= ∑∑

I

i

W

wiiwI

µφ , where iw is the consumption weight of product i in the basket (note

that ∑=

=W

wiw

11). Alternatively, the duration can be defined in terms of medians instead of

averages.


The product-specific size of price increases, iλ , is computed as ∑ ∑= = −

−−=

T

t

S

s ist

ististi p

ppST 1 1 1

111λ ,

conditional on that 1−> istist pp . Similarly, the product-specific size of price decreases, iτ , is

computed as ∑ ∑= = −

−−=

T

t

S

s ist

ististi p

ppST 1 1 1

111τ , conditional on that 1−< istist pp . The average size of

price change can be computed as the simple or weighted average (or median) of product-specific

size of price changes.

Product-specific price dispersion is defined as follows. ( )∑=

=T

tijti pSD

T 1

log1σ , where ijtplog is a

logarithm of average price of product i at region j .


THE RESULTS ON PRODUCT-SPECIFIC DESCRIPTIVE STATISTICS (all 423 products)

CPI W

eight

s

Freq

uenc

y

Dur

atio

n (m

ean)

Price

disp

ersio

n

Infla

tion

rate

– y

-o-y

ch

ange

Size

- in

crea

se

Size

- de

crea

se

Infla

tion

pers

isten

ce

Rice 0.29 0.40 2.51 0.07 1.77 0.07 -0.07 0.88Rice in boiling packets 0.03 0.38 2.66 0.08 -9.98 0.15 -0.16 0.93Wheat Flour Half Fine 0.42 0.40 2.48 0.05 2.40 0.07 -0.08 0.95Farina 0.03 0.31 3.19 0.04 6.02 0.06 -0.07 0.85Rye bread 1.33 0.29 3.49 0.12 9.11 0.10 -0.10 0.88Bread white 0.80 0.25 3.97 0.13 8.70 0.10 -0.13 0.85White roll 1.02 0.22 4.63 0.09 5.03 0.13 -0.14 0.90Christmas cake 0.61 0.29 3.43 0.13 8.82 0.10 -0.12 0.85Biscuits without filling 0.17 0.35 2.88 0.10 9.38 0.09 -0.16 0.95Biscuits with filling 0.33 0.36 2.79 0.07 7.30 0.05 -0.07 0.90Waffles with filling 0.59 0.36 2.79 0.06 6.09 0.06 -0.07 0.90Wafers without flavor 0.21 0.39 2.54 0.06 7.23 0.08 -0.10 0.90Salted crackers 0.16 0.32 3.15 0.19 9.43 0.14 -0.14 0.90Pasta (with eggs) 0.42 0.38 2.63 0.13 6.64 0.09 -0.09 0.90Dumpling 0.31 0.30 3.33 0.11 7.24 0.08 -0.10 0.90Puff pastry (listkove) 0.10 0.36 2.76 0.07 4.85 0.11 -0.12 0.88Porridge (without flavor) 0.03 0.39 2.58 0.08 2.10 0.10 -0.10 0.80Chuck roast with bone 0.23 0.37 2.69 0.08 2.66 0.09 -0.09 0.90Boneless chuck roast 0.24 0.39 2.58 0.06 5.78 0.07 -0.08 0.85Beef rear without bone 0.36 0.40 2.50 0.06 4.33 0.06 -0.07 0.80Beef joint without bone, lower 0.09 0.32 3.13 0.06 4.80 0.07 -0.08 0.80Pork meat with bone 0.93 0.57 1.75 0.04 4.83 0.06 -0.06 0.73Pork neck with bone 0.43 0.54 1.84 0.04 5.04 0.07 -0.06 0.83Flank of bacon 0.29 0.54 1.84 0.05 5.96 0.11 -0.10 0.93Pork leg without bone 0.36 0.57 1.76 0.06 3.76 0.09 -0.08 0.83Pork shoulder without bone 0.47 0.56 1.78 0.05 4.34 0.07 -0.06 0.88Chicken without insides 1.16 0.53 1.90 0.03 6.15 0.05 -0.06 0.88Chicken portioned fresh and frozen 1.19 0.49 2.02 0.04 2.99 0.08 -0.08 0.90Pork diet salami 0.28 0.46 2.17 0.06 0.81 0.08 -0.08 0.90Fine frankfurters 0.61 0.47 2.13 0.05 1.62 0.07 -0.07 0.93Ham salami 0.52 0.47 2.12 0.04 0.90 0.06 -0.06 0.93Durable salami 0.59 0.38 2.62 0.06 4.50 0.06 -0.08 0.80Boneless smoked pork neck 0.16 0.58 1.73 0.06 7.39 0.13 -0.12 0.90Pork stewed ham 0.59 0.44 2.30 0.04 2.44 0.06 -0.06 0.93Smoked bacon with skin 0.26 0.37 2.69 0.13 10.20 0.16 -0.17 0.90Luncheon meat pork 0.03 0.35 2.86 0.10 1.78 0.11 -0.11 0.88Pork meat paste 0.55 0.40 2.48 0.12 3.40 0.11 -0.10 0.93Pork liver 0.10 0.33 3.05 0.08 0.78 0.09 -0.09 0.93File-fish not breaded 0.43 0.45 2.24 0.07 12.90 0.16 -0.17 0.93Carp (live and frozen) 0.07 0.22 4.51 0.28 4.18 0.16 -0.13 0.93


Smoked fish (mackerel with head) 0.05 0.33 3.05 0.08 10.94 0.08 -0.09 0.93Sardines in oil 0.24 0.38 2.60 0.07 6.82 0.07 -0.07 0.95Fish salad with mayonnaise 0.26 0.35 2.89 0.12 8.64 0.10 -0.14 0.88Fish in sour 0.09 0.36 2.78 0.08 10.11 0.13 -0.17 0.93Pasteurized half-fat milk 0.90 0.30 3.29 0.07 6.99 0.05 -0.05 0.88Thick milk Tatra without sugar 0.02 0.33 3.00 0.04 11.14 0.08 -0.13 0.83Dried Milk for Babies 0.09 0.39 2.60 0.06 6.15 0.15 -0.13 0.93Dried Milk Half-fat 0.02 0.41 2.45 0.06 3.87 0.10 -0.10 0.93Fruit yoghurt 1.02 0.39 2.54 0.07 -0.10 0.06 -0.08 0.95Sour milk (acidophilus) 0.16 0.27 3.74 0.24 79.07 0.40 -0.14 0.95Sweet cream 33% 0.24 0.31 3.25 0.05 5.86 0.05 -0.05 0.90Sour cream 0.29 0.32 3.16 0.07 6.48 0.08 -0.11 0.78Cheese Eidam, block 0.74 0.39 2.53 0.05 5.87 0.05 -0.06 0.88Spreadable cheese 0.68 0.39 2.60 0.07 -9.23 0.07 -0.11 0.95Ostiepok rolled smoked cheese 0.03 0.39 2.59 0.10 7.94 0.07 -0.08 0.83Mold cheese Niva 0.09 0.34 2.93 0.06 7.81 0.08 -0.11 0.98Bryndza sheep cheese 0.14 0.50 2.01 0.05 7.22 0.06 -0.08 0.87 Eggs (chicken, fresh) 0.80 0.65 1.54 0.07 4.81 0.11 -0.10 0.90Butter 0.68 0.45 2.20 0.05 -0.47 0.06 -0.06 0.93Margarine 0.66 0.38 2.60 0.08 4.15 0.07 -0.08 0.95Edible oil 0.85 0.39 2.56 0.05 4.60 0.06 -0.06 0.95Pork lard 0.03 0.50 1.98 0.14 5.05 0.16 -0.14 0.93Apples 0.59 0.77 1.30 0.11 6.45 0.17 -0.16 0.85Oranges 0.28 0.81 1.23 0.08 6.57 0.16 -0.12 0.90Mandarins 0.24 0.79 1.27 0.10 3.91 0.18 -0.14 0.85Lemons 0.12 0.77 1.29 0.07 2.27 0.12 -0.10 0.78Kiwi 0.07 0.77 1.30 0.17 6.11 0.33 -0.21 0.85Bananas 0.55 0.83 1.21 0.06 7.87 0.16 -0.13 0.54Dried grapes 0.05 0.42 2.39 0.07 6.41 0.11 -0.11 0.90Peanuts peeled salted 0.29 0.41 2.45 0.09 -0.97 0.10 -0.10 0.93Poppy seeds 0.05 0.39 2.54 0.11 6.04 0.17 -0.14 0.93Celery 0.02 0.61 1.64 0.17 5.91 0.31 -0.22 0.68Carrot 0.07 0.72 1.40 0.16 9.38 0.31 -0.22 0.78Parsley 0.03 0.75 1.33 0.17 19.20 0.37 -0.23 0.88Cauliflower 0.10 0.78 1.28 0.18 4.28 0.38 -0.23 0.51Cabbage (white) 0.10 0.66 1.51 0.15 4.64 0.33 -0.22 0.76Salad cucumbers 0.14 0.92 1.09 0.15 4.14 0.72 -0.33 0.71Paprika 0.21 0.90 1.11 0.18 5.64 0.52 -0.29 0.78Onion 0.10 0.68 1.47 0.12 4.59 0.24 -0.19 0.88Tomatoes 0.28 0.67 1.50 0.41 5.37 0.44 -0.28 0.63Beans white dried 0.03 0.36 2.75 0.10 3.70 0.09 -0.09 0.95Lentils (big) 0.03 0.35 2.85 0.08 3.54 0.08 -0.09 0.90Frozen vegetable mix 0.19 0.38 2.62 0.13 9.97 0.12 -0.13 0.85Frozen spinach 0.03 0.38 2.62 0.08 2.83 0.10 -0.10 0.93Sour cabbage (sterilized) 0.07 0.39 2.57 0.07 4.73 0.11 -0.10 0.90Sterilized peas in salty water 0.07 0.33 3.00 0.12 4.22 0.15 -0.14 0.93Paprika and tomatoes sterilized (without sausage) 0.02 0.48 2.10 0.07 -4.43 0.17 -0.19 0.88Potatoes 0.40 0.67 1.49 0.14 7.66 0.40 -0.20 0.78Potato chips 0.16 0.39 2.53 0.08 8.75 0.09 -0.09 0.93Frozen French fries 0.10 0.45 2.24 0.11 0.04 0.11 -0.11 0.95Crystal sugar 0.61 0.42 2.39 0.04 8.23 0.10 -0.08 0.88Ground sugar 0.14 0.42 2.37 0.05 9.17 0.11 -0.10 0.93Honey 0.12 0.46 2.15 0.10 1.92 0.11 -0.11 0.95Strawberry jam 0.03 0.32 3.11 0.06 3.02 0.07 -0.08 0.93


Milk chocolate 0.45 0.39 2.54 0.08 4.32 0.06 -0.07 0.83Cooking chocolate 0.12 0.37 2.68 0.05 3.96 0.08 -0.10 0.83Chocolate bar with filling 0.26 0.34 2.97 0.09 6.77 0.09 -0.12 0.90Dessert chocolates 0.29 0.40 2.48 0.10 8.31 0.07 -0.07 0.83Fruit jelly 0.07 0.34 2.90 0.11 7.77 0.09 -0.10 0.88Hard candies without filling 0.31 0.39 2.57 0.08 10.04 0.07 -0.07 0.90Chewing gum - slices 0.42 0.23 4.37 0.18 -2.47 0.11 -0.12 0.93Salt 0.07 0.25 3.94 0.04 5.28 0.05 -0.06 0.85Ground sweet paprika 0.09 0.38 2.62 0.10 6.70 0.13 -0.11 0.88Ground pepper 0.07 0.40 2.50 0.11 18.51 0.18 -0.13 0.85Caraway not ground 0.02 0.34 2.90 0.15 4.34 0.20 -0.15 0.78Vinegar 8% 0.10 0.34 2.94 0.06 1.82 0.10 -0.10 0.93Mustard full-fat 0.17 0.32 3.08 0.06 3.77 0.07 -0.08 0.76Ketchup 0.21 0.44 2.26 0.10 1.23 0.10 -0.10 0.83Mayonnaise 0.10 0.39 2.57 0.10 39.82 0.19 -0.09 0.95Baking powder 0.03 0.26 3.81 0.06 -2.52 0.16 -0.15 0.90Fresh yeast 0.07 0.27 3.72 0.31 11.92 0.21 -0.14 0.95Dehydrated soup (not instant) 0.19 0.39 2.56 0.17 2.49 0.10 -0.10 0.85Vanilla pudding 0.07 0.31 3.18 0.07 9.00 0.10 -0.09 0.88Dried vegetable flavoring 0.21 0.35 2.86 0.05 4.14 0.08 -0.07 0.85Coffee beans 0.90 0.43 2.33 0.08 -0.98 0.09 -0.09 0.95Instant coffee with caffeine 0.31 0.36 2.79 0.09 5.69 0.08 -0.09 0.85Black tea without flavor 0.28 0.32 3.17 0.17 3.36 0.12 -0.13 0.80Cocoa powder 0.05 0.34 2.90 0.05 0.38 0.09 -0.09 0.80Cocoa granko 0.12 0.36 2.81 0.04 5.04 0.06 -0.06 0.83Table mineral water 0.66 0.29 3.41 0.11 7.42 0.11 -0.12 0.93Fruit syrup 0.33 0.31 3.20 0.07 -0.04 0.06 -0.06 0.95Rum 38-40% 0.36 0.36 2.77 0.03 2.27 0.05 -0.06 0.95Vodka 38-40% 1.56 0.33 3.01 0.09 -4.92 0.07 -0.09 0.95Brandy 38-40% 1.26 0.37 2.71 0.08 -7.99 0.06 -0.11 0.95Bottled red wine 0.38 0.36 2.79 0.04 4.30 0.05 -0.05 0.90Bottled white wine 0.88 0.36 2.80 0.05 3.20 0.05 -0.05 0.85Bottled sparkling wine 0.68 0.31 3.18 0.03 3.49 0.05 -0.05 0.88Beer 10% bottled 1.11 0.28 3.57 0.09 7.01 0.08 -0.08 0.85Beer 12% bottled 0.78 0.29 3.45 0.09 6.20 0.07 -0.07 0.85Cigarettes "Mars" 20 pieces 4.41 0.30 3.31 0.03 10.94 0.07 -0.06 0.85Cigarettes "Dalila" 20 pieces 0.62 0.22 4.61 0.04 12.88 0.12 -0.24 0.88Cotton dress for women 0.16 0.26 3.78 0.17 16.00 0.26 -0.20 0.90Synthetic dress material 0.05 0.29 3.42 0.17 14.42 0.26 -0.19 0.95Wool dress for women 0.02 0.18 5.52 0.12 -0.20 0.12 -0.10 0.76Short underwear for men 0.07 0.35 2.85 0.20 8.55 0.13 -0.10 0.83Long knitted underwear for men 0.02 0.39 2.55 0.13 1.47 0.12 -0.11 0.76Undershirt for men 0.03 0.32 3.15 0.13 2.62 0.13 -0.11 0.80Pajamas for men (from fabric) 0.09 0.32 3.13 0.15 5.14 0.14 -0.12 0.76Shorts for men 0.03 0.25 3.96 0.14 3.74 0.14 -0.12 0.90Bathrobe for men 0.02 0.18 5.53 0.21 5.75 0.23 -0.19 0.66Panties for women 0.16 0.33 3.00 0.18 10.29 0.13 -0.12 0.88Women night-gown 0.03 0.37 2.72 0.16 4.18 0.13 -0.11 0.88Women slip 0.00 0.29 3.40 0.14 8.06 0.13 -0.11 0.95Pajamas for women (from fabric) 0.09 0.30 3.30 0.14 5.59 0.16 -0.14 0.88Ladies bra 0.23 0.34 2.95 0.18 9.72 0.12 -0.11 0.85Home dress for women 0.02 0.27 3.74 0.26 4.69 0.18 -0.14 0.73Shirt for babies (from fabric) 0.00 0.24 4.11 0.10 6.51 0.12 -0.11 0.66Cotton napkins for babies tetra 0.00 0.29 3.47 0.05 5.04 0.09 -0.10 0.93


Short sleeved shirt for babies 0.14 0.27 3.70 0.32 13.12 0.19 -0.16 0.83Panties for girls 0.02 0.30 3.31 0.12 4.31 0.12 -0.12 0.76Underwear for boys 0.02 0.31 3.23 0.16 2.57 0.13 -0.12 0.78Children pajamas 0.07 0.35 2.87 0.14 5.47 0.13 -0.12 0.85Shirt for babies without sleeves 0.02 0.27 3.69 0.13 4.77 0.12 -0.11 0.90Shorts for boys 0.00 0.18 5.60 0.18 8.72 0.29 -0.20 0.93Long winter coat for men 0.07 0.19 5.34 0.10 4.97 0.14 -0.13 0.93Winter jacket for men 0.31 0.24 4.16 0.18 4.72 0.17 -0.14 0.88Longer leather jacket for men 0.12 0.24 4.16 0.15 1.23 0.16 -0.13 0.83Spring jacket for men 0.09 0.18 5.56 0.15 4.72 0.21 -0.17 0.83Short sleeved shirt for men 0.42 0.24 4.12 0.22 1.30 0.14 -0.12 0.83Long winter coat for women 0.42 0.22 4.65 0.12 4.75 0.13 -0.12 0.93Winter jacket for women 0.24 0.20 5.03 0.21 4.88 0.17 -0.14 0.88Rabbit fur coat for women 0.09 0.18 5.46 0.13 3.00 0.13 -0.14 0.78Long spring coat for women 0.12 0.22 4.50 0.11 4.10 0.15 -0.12 0.95Thin costume for women 0.57 0.31 3.22 0.13 5.19 0.16 -0.14 0.88Summer dress for women 0.24 0.16 6.30 0.15 1.19 0.22 -0.18 0.93Tailoring of a dress for women 0.05 0.11 8.94 0.34 6.67 0.20 -0.24 0.90Spring children jacket 0.07 0.21 4.73 0.31 0.55 0.28 -0.21 0.88Winter jacket for boys 0.35 0.21 4.78 0.21 -4.23 0.21 -0.15 0.88Jeans for boys 0.40 0.19 5.40 0.36 12.59 0.18 -0.14 0.86Small baby coat 0.02 0.31 3.28 0.19 4.94 0.15 -0.13 0.88Baby stockings 0.02 0.31 3.23 0.20 3.69 0.14 -0.12 0.83Stockings for women 0.24 0.23 4.43 0.26 2.89 0.10 -0.09 0.93Stocking for children 0.07 0.28 3.61 0.12 0.11 0.14 -0.14 0.90Handkerchief for women 0.00 0.25 4.02 0.12 4.02 0.13 -0.12 0.88Shawl for adults 0.03 0.16 6.43 0.13 6.55 0.20 -0.17 0.93Felt hat for men 0.00 0.21 4.87 0.21 1.13 0.18 -0.17 0.93Fur cap for women 0.00 0.20 5.09 0.17 7.98 0.22 -0.17 0.90Knit cap for children 0.05 0.18 5.41 0.25 2.68 0.16 -0.14 0.83Knit gloves for children 0.02 0.21 4.78 0.17 6.33 0.21 -0.20 0.88Tie for men 0.05 0.23 4.30 0.22 4.93 0.13 -0.12 0.88Thread for sewing, Tebex 0.02 0.23 4.33 0.19 5.48 0.09 -0.09 0.80Imitation of sewing silk (Nora) 0.00 0.13 7.89 0.12 -3.43 0.12 -0.11 0.90Knit tread 0.05 0.21 4.76 0.14 -19.86 0.12 -0.13 0.95Elastic waistband 0.00 0.19 5.26 0.10 5.41 0.21 -0.18 0.95Metal zipper 0.02 0.18 5.41 0.15 6.43 0.12 -0.11 0.90Cleaning of trousers in 3 days 0.02 0.10 9.92 0.08 4.98 0.21 -0.16 0.90Cleaning of coats 0.03 0.12 8.42 0.37 11.94 0.15 -0.15 0.93Leather walking shoes for men 0.38 0.35 2.85 0.25 4.43 0.10 -0.10 0.88Leather walking shoes for men, sandals 0.07 0.22 4.63 0.11 8.87 0.24 -0.18 0.90Leather winter shoes 0.23 0.19 5.19 0.18 11.47 0.15 -0.13 0.93Leather sport shoes 0.38 0.29 3.51 0.11 7.92 0.21 -0.16 0.83Leather walking shoes for women 0.55 0.28 3.60 0.20 5.14 0.14 -0.12 0.71Leather shoes for women, sandals 0.36 0.23 4.27 0.16 11.12 0.19 -0.16 0.93Leather winter shoes for women 0.52 0.21 4.66 0.17 8.80 0.16 -0.13 0.85Textile indoor shoes for women, slippers 0.09 0.31 3.24 0.14 7.02 0.14 -0.12 0.88Baby leather shoes 0.00 0.24 4.11 0.11 4.85 0.19 -0.16 0.95Plastic winter shoes for children - boots 0.24 0.23 4.27 0.13 5.67 0.08 -0.09 0.73Leather summer shoes for children, sandals 0.10 0.23 4.36 0.08 5.68 0.21 -0.16 0.93Women's shoes heels repair 0.05 0.28 3.60 0.15 3.47 0.19 -0.19 0.90Paint (Primalex, Farmal, Permal etc.) 0.19 0.25 4.06 0.21 9.00 0.17 -0.14 0.90Basic synthetic paint 0.10 0.23 4.36 0.08 9.05 0.11 -0.11 0.93Synthetic and oil paint thinner 0.03 0.22 4.47 0.15 4.51 0.12 -0.15 0.85


Cement 0.57 0.22 4.52 0.07 10.10 0.09 -0.18 0.93Lime 0.05 0.18 5.66 0.06 9.91 0.09 -0.12 0.88Ceramic tiles, smooth, natural 0.64 0.22 4.59 0.09 7.94 0.14 -0.14 0.95Porous white and colored wall tiles 0.36 0.20 5.12 0.16 12.31 0.14 -0.13 0.78Wood (imitation of wood) board 0.23 0.10 9.88 0.16 6.94 0.12 -0.13 0.76Lever faucet 0.31 0.22 4.51 0.37 4.96 0.11 -0.10 0.90WC bowl with flusher 0.14 0.24 4.17 0.16 1.42 0.09 -0.08 0.85Installation services 0.21 0.13 7.97 0.09 5.44 0.15 -0.17 0.90Painting services 0.47 0.12 8.13 0.29 2.26 0.18 -0.18 0.95Varnishing of doors and windows 0.14 0.14 7.36 0.41 15.09 0.15 -0.16 0.90Glass services 0.12 0.16 6.31 0.26 11.36 0.15 -0.23 0.78Upholstered chair 0.14 0.23 4.36 0.19 5.78 0.12 -0.11 0.93Kitchen table 0.05 0.21 4.71 0.25 6.07 0.22 -0.21 0.71Two door closet for clothes 0.16 0.16 6.30 0.13 0.77 0.09 -0.09 0.88Kitchen cupboard 0.50 0.21 4.72 0.18 2.46 0.11 -0.11 0.73Bed with storage 0.12 0.15 6.62 0.11 -1.47 0.09 -0.09 0.93Bed for children with mattress 0.00 0.22 4.46 0.11 3.90 0.08 -0.08 0.90Furniture set for living room 0.33 0.21 4.87 0.18 -0.23 0.18 -0.15 0.93Upholstered set 0.54 0.24 4.17 0.10 1.44 0.11 -0.11 0.85Set of plastic garden furniture 0.03 0.20 5.02 0.22 -2.37 0.10 -0.10 0.95Synthetic carpet sewn-in 0.14 0.25 3.95 0.19 2.38 0.12 -0.10 0.83Plastic floor covering (pvc) 0.10 0.27 3.76 0.15 -1.96 0.16 -0.20 0.93Repair of upholstered sitting set 0.05 0.10 9.75 0.34 -2.76 0.34 -0.24 0.95Curtains 0.16 0.21 4.78 0.28 -9.76 0.12 -0.11 0.93Bed sheet 0.03 0.16 6.07 0.17 4.68 0.10 -0.09 0.76Bed linen for children - 1 bed 0.00 0.23 4.37 0.08 1.60 0.13 -0.12 0.90Bed linen for adults - 1 bed damask 0.02 0.27 3.66 0.12 3.08 0.15 -0.13 0.73Bed linen for adults - 1 bed 0.09 0.32 3.16 0.16 3.10 0.09 -0.09 0.80Turkish towel 0.05 0.31 3.27 0.10 2.82 0.15 -0.16 0.90Table cloth 0.07 0.28 3.54 0.15 5.87 0.13 -0.10 0.78Dish cloth 0.02 0.21 4.65 0.09 2.41 0.11 -0.10 0.66Big synthetic blanket (Larisa) 0.03 0.23 4.42 0.09 4.21 0.10 -0.10 0.93Comforter filled with synthetic material 0.07 0.19 5.17 0.12 2.73 0.11 -0.12 0.93Down comforter; quilt feather filling 0.00 0.18 5.62 0.09 -5.22 0.09 -0.13 0.93Refrigerator with freezer 260 liter 0.23 0.32 3.11 0.08 0.68 0.09 -0.09 0.88Freezer 130 liter 0.02 0.19 5.23 0.04 1.13 0.07 -0.07 0.88Air damper 0.02 0.23 4.27 0.33 -15.92 0.30 -0.29 0.93Electric suitcase sewing machine 0.05 0.35 2.84 0.14 1.19 0.19 -0.17 0.93Electric kitchen robot 0.03 0.22 4.50 0.20 1.40 0.13 -0.12 0.93Electric hand whipping tool 0.03 0.23 4.44 0.09 3.62 0.10 -0.09 0.93Electric juicer 0.02 0.19 5.26 0.18 -0.32 0.12 -0.11 0.78Electric fryer 0.02 0.21 4.70 0.13 0.01 0.15 -0.13 0.88Electric coffee maker with filter 0.02 0.25 4.04 0.15 2.10 0.09 -0.09 0.90Repair of electric refrigerator 0.03 0.23 4.33 0.21 5.09 0.12 -0.11 0.83Repair of automatic washing machine 0.05 0.17 5.99 0.20 5.20 0.12 -0.15 0.66Repair of electric vacuum cleaner 0.02 0.15 6.58 0.30 18.88 0.25 -0.26 0.90Repair of combined stove 0.00 0.09 11.69 0.35 6.08 0.13 -0.14 0.80Glass without holder 100ml 0.09 0.30 3.34 0.22 2.60 0.13 -0.14 0.90Lead crystal cup with holder 0.05 0.28 3.59 0.16 2.91 0.13 -0.12 0.63Plate set for 6 persons 0.12 0.25 3.99 0.15 7.24 0.12 -0.12 0.95Porcelain cup with decorations 0.05 0.25 4.02 0.11 5.59 0.10 -0.09 0.93Storage cans Omnia 720ml 0.02 0.21 4.71 0.08 5.42 0.15 -0.13 0.80Glass bowl from silex with cover 0.02 0.29 3.48 0.10 8.38 0.12 -0.13 0.93Kitchen pot 4 liters 0.07 0.30 3.35 0.26 12.21 0.14 -0.11 0.95


Enameled tea kettle 0.02 0.27 3.68 0.10 5.25 0.14 -0.13 0.85Cutlery for 6 persons - rustles 0.02 0.30 3.34 0.22 -0.52 0.17 -0.14 0.93Kitchen knife with plastic handle 0.03 0.27 3.77 0.20 2.95 0.11 -0.11 0.83Soup ladle - rustles 0.00 0.21 4.66 0.11 4.42 0.12 -0.11 0.95Pan without cover - tefal 0.09 0.30 3.39 0.16 4.26 0.14 -0.13 0.95Plastic bottle for babies 0.02 0.29 3.43 0.17 3.56 0.19 -0.15 0.90Kitchen scales - 1 bowl 0.03 0.22 4.54 0.07 7.69 0.13 -0.16 0.95Wooden ladle 0.03 0.23 4.42 0.13 4.39 0.14 -0.12 0.83Vacuum bottle without pump, 1 liter 0.09 0.28 3.61 0.08 5.08 0.10 -0.09 0.95Electric drilling machine - two speeds 0.10 0.24 4.13 0.14 2.98 0.13 -0.12 0.76Flat light switch 0.03 0.25 4.05 0.11 6.42 0.13 -0.16 0.93Electric adapter 0.02 0.20 5.03 0.07 2.94 0.12 -0.12 0.83Thin battery 1,5 v (alkaline) 0.17 0.25 3.99 0.13 5.02 0.12 -0.12 0.90Regular light bulb 60w 0.19 0.22 4.65 0.07 0.04 0.10 -0.09 0.93Tape measure, 2 meters 0.00 0.26 3.89 0.28 -5.32 0.24 -0.18 0.90Combination pliers (PVC handle) 0.00 0.23 4.30 0.26 5.20 0.28 -0.22 0.90Screw driver (PVC handle) 0.02 0.24 4.13 0.17 16.47 0.14 -0.12 0.95Metal rake without handle 0.00 0.19 5.27 0.09 2.75 0.14 -0.11 0.83Aluminum double ladder to 180cm 0.03 0.28 3.60 0.16 4.02 0.11 -0.10 0.85Household scissors 0.00 0.23 4.34 0.28 -0.85 0.34 -0.23 0.83Clothes drying rack 0.03 0.21 4.79 0.14 -1.92 0.09 -0.09 0.76Ironing board, holder and Teflon cover 0.02 0.24 4.23 0.19 1.18 0.22 -0.18 0.85Liquid detergent for dish-washing 0.19 0.28 3.54 0.08 5.14 0.10 -0.10 0.88Spray insecticide 0.02 0.24 4.18 0.07 7.18 0.08 -0.08 0.95Construction nails 70 mm 0.02 0.17 5.82 0.30 -5.29 0.10 -0.12 0.95Long screws 3x20mm 0.02 0.14 7.38 0.07 3.71 0.43 -0.23 0.95Mechanical carpet cleaning 0.03 0.15 6.66 0.34 6.48 0.17 -0.20 0.93Škoda Fabia 1,4 Classic (44 kW) 0.47 0.43 2.30 0.08 1.87 0.12 -0.10 0.90Škoda Felicia, 1999 0.17 0.45 2.23 0.16 3.28 0.20 -0.16 0.93Children bicycle 0.10 0.21 4.72 0.13 9.05 0.08 -0.08 0.90Radial car tire 0.57 0.28 3.59 0.10 2.29 0.31 -0.22 0.90Accumulator 0.17 0.44 2.27 0.26 7.53 0.28 -0.19 0.88Left front fender 0.05 0.30 3.32 0.22 5.32 0.17 -0.15 0.90Oil filter 0.03 0.31 3.20 0.38 4.07 0.45 -0.31 0.83Gasoline 91 octane 1.68 0.69 1.44 0.37 6.56 0.04 -0.03 0.90Gasoline 95 octane 5.02 0.69 1.45 0.01 10.35 0.04 -0.03 0.95Oil fuel 0.54 0.77 1.30 0.01 11.24 0.03 -0.03 0.95Motor oil 0.05 0.23 4.43 0.25 12.28 0.12 -0.19 0.90Gear box oil 0.00 0.22 4.52 0.22 12.77 0.10 -0.20 0.93Non freezing liquid for cooler 0.02 0.22 4.51 0.18 8.55 0.15 -0.13 0.88Complete repair of motor 0.38 0.14 6.95 0.47 6.54 0.57 -0.30 0.95Complete repair of brakes 0.09 0.24 4.17 0.61 12.85 0.40 -0.29 0.80Basic balancing of car wheels 0.02 0.20 4.98 0.46 5.68 0.34 -0.26 0.88Complete lacquer 0.24 0.11 8.96 0.32 11.13 0.16 -0.20 0.90Replacement of door frame 0.05 0.22 4.63 0.56 16.42 0.53 -0.31 0.80Car washing 0.05 0.12 8.65 0.24 7.17 0.21 -0.17 0.93Taxi - personal fare + fare for 5 km 0.03 0.15 6.89 0.19 9.37 0.16 -0.15 0.93Portable radio with tape, stereo 0.07 0.27 3.65 0.16 -4.89 0.22 -0.20 0.93Walkman 0.03 0.25 4.04 0.23 -4.07 0.26 -0.24 0.95Stereo set 0.36 0.25 4.03 0.14 10.84 0.16 -0.15 0.88TV set 0.54 0.23 4.30 0.10 -3.77 0.17 -0.12 0.93VCR - 6 heads 0.23 0.26 3.89 0.14 1.45 0.16 -0.11 0.90Camera with auto focus 0.05 0.22 4.49 0.20 1.43 0.19 -0.14 0.90Video camera 0.09 0.23 4.32 0.16 -3.05 0.17 -0.12 0.85


PC, Pentium - without accessories 0.52 0.44 2.29 0.19 5.64 0.30 -0.18 0.88Electronic pocket calculator 0.03 0.21 4.66 0.31 -4.22 0.26 -0.18 0.90Compact disc 0.21 0.22 4.46 0.18 7.36 0.14 -0.13 0.93Videotape - clean 0.10 0.20 5.02 0.16 -6.61 0.12 -0.10 0.83Tape for sound recording - clean 0.03 0.21 4.70 0.28 -3.22 0.14 -0.15 0.83Color film into the camera 0.16 0.19 5.19 0.15 3.80 0.16 -0.14 0.90Teddy bear 50 cm 0.07 0.33 3.06 0.23 4.62 0.14 -0.12 0.90Dressed doll with hair, PVC, from 40-50cm 0.10 0.29 3.43 0.33 10.04 0.21 -0.16 0.85Small bicycle for children 0.10 0.20 5.06 0.24 12.46 0.38 -0.25 0.93Children game "Clovece, nehnevaj sa" 0.05 0.26 3.89 0.19 7.68 0.17 -0.16 0.85Paper puzzle 0.03 0.22 4.53 0.17 7.39 0.24 -0.16 0.76Construction set Duplo 0.14 0.30 3.37 0.46 6.05 0.30 -0.18 0.85Downhill skis, 140 - 160 cm 0.14 0.20 4.91 0.30 11.37 0.30 -0.21 0.90Binding for downhill skiing 0.03 0.18 5.64 0.15 -1.84 0.15 -0.13 0.93Plastic bob sled with brakes 0.02 0.15 6.74 0.12 3.45 0.10 -0.10 0.90Ice-skating shoes 0.12 0.20 4.92 0.09 2.69 0.11 -0.11 0.95Ball for volleyball 0.02 0.22 4.56 0.19 1.87 0.14 -0.13 0.83Sleeping pack with a pack 0.09 0.18 5.56 0.25 5.68 0.20 -0.16 0.88Rose bush 0.19 0.15 6.54 0.20 6.73 0.14 -0.12 0.90Apple tree 1st class 0.28 0.12 8.07 0.15 9.89 0.14 -0.12 0.93Fertilizer 0.03 0.26 3.89 0.17 7.65 0.15 -0.15 0.93Karafiat (a flower) 0.17 0.39 2.60 0.12 4.98 0.15 -0.13 0.71Rose 0.42 0.54 1.84 0.12 8.89 0.14 -0.11 0.76Dog food 0.35 0.33 3.04 0.15 -0.41 0.13 -0.17 0.90Covered swimming pool ticket 0.05 0.26 3.85 0.27 16.53 0.34 -0.27 0.85Fee for exercises (in fitness center) 0.07 0.12 8.66 0.31 12.06 0.26 -0.30 0.90Dancing course fee 0.12 0.05 22.00 0.33 11.11 0.25 -0.19 0.90Cinema ticket 0.09 0.26 3.87 0.12 14.41 0.11 -0.10 0.76Videotape - 1 day borrowing 0.03 0.08 12.05 0.15 7.55 0.23 -0.25 0.95ID picture 0.03 0.09 11.16 0.26 3.29 0.21 -0.18 0.93Color film developing 0.12 0.10 10.44 0.26 8.31 0.31 -0.23 0.93Colored photo enlargement 9x13 cm 0.14 0.09 11.28 0.11 -3.73 0.14 -0.12 0.85Books for children, age: from 6 to 9 0.09 0.20 5.10 0.15 8.07 0.12 -0.11 0.71Pocket Dictionary Slovak-English 0.02 0.17 5.93 0.27 4.86 0.17 -0.21 0.83Book - prose - foreign author 0.29 0.21 4.68 0.10 9.20 0.10 -0.09 0.71Book - prose - Slovak author 0.09 0.27 3.65 0.15 11.72 0.14 -0.15 0.85Colored postcard, in envelope 0.07 0.24 4.24 0.24 7.42 0.13 -0.12 0.90Spiral calendar, size 30x20cm 0.03 0.18 5.71 0.14 3.51 0.14 -0.13 0.93Notebook - half thick 40 sheets 0.23 0.19 5.24 0.07 4.54 0.08 -0.08 0.85Note book A4 format 0.02 0.28 3.58 0.15 2.61 0.14 -0.12 0.88Black pencil 0.02 0.22 4.47 0.15 5.18 0.13 -0.15 0.90Ball pen - medium content 0.07 0.19 5.16 0.16 -1.18 0.12 -0.11 0.85Celluloid ruler, 30 cm 0.02 0.14 6.99 0.13 -1.16 0.17 -0.18 0.95DESIGN - A4 0.02 0.16 6.16 0.12 2.76 0.21 -0.19 0.80Color pencils 0.05 0.23 4.32 0.12 3.71 0.11 -0.10 0.95Recreation in Slovakia for 7 days - hotel B* 0.71 0.27 3.76 0.19 7.44 0.26 -0.18 0.78Spain 14 days, airplane 1.63 0.15 6.89 0.17 11.92 0.15 -0.11 0.93Italy 7 nights, by bus 0.36 0.18 5.45 0.11 9.73 0.17 -0.17 0.93Trip to the neighboring country, within 500km, by coach

0.05 0.16 6.43 0.38 5.05 0.17 -0.20 0.90

Beef bouillon with meat and noodles 0.05 0.23 4.40 0.20 7.95 0.23 -0.18 0.85Beef goulash 0.02 0.29 3.39 0.15 8.29 0.15 -0.17 0.83Joint with ham and egg 0.03 0.26 3.83 0.18 4.06 0.24 -0.19 0.73Roasted pork meat 0.10 0.21 4.66 0.15 6.97 0.09 -0.14 0.95


Fried pork meat (breadcrumbs) 0.19 0.26 3.82 0.14 5.45 0.16 -0.13 0.90Grilled or baked chicken 0.23 0.25 4.06 0.14 5.66 0.10 -0.15 0.88Pancakes with jam 0.03 0.34 2.97 0.20 9.90 0.19 -0.19 0.85Sheep cheese dumplings 0.03 0.21 4.83 0.14 7.42 0.11 -0.09 0.73Fried cheese 0.10 0.30 3.31 0.16 7.17 0.09 -0.12 0.80French fries 0.00 0.09 11.35 0.16 3.50 0.11 -0.10 0.88Dumplings, big 0.02 0.14 7.20 0.20 9.69 0.43 -0.26 0.73Stewed rice 0.02 0.16 6.10 0.25 15.80 0.21 -0.16 0.93Stewed cabbage 0.00 0.23 4.40 0.19 8.26 0.43 -0.28 0.80Cucumber salad 0.03 0.13 7.70 0.31 5.23 0.42 -0.30 0.88Caramel dessert "Veternik" 0.03 0.17 6.01 0.16 9.46 0.09 -0.11 0.85Ice cream 0.03 0.14 6.98 0.28 12.82 0.45 -0.34 0.83"Vlassky" salad 0.57 0.20 5.00 0.12 2.81 0.04 -0.04 0.95Sandwich with ham and vegetables 0.23 0.20 5.07 0.16 3.12 0.17 -0.13 0.88Coffee - 7 grams, 5 grams of sugar 0.10 0.19 5.27 0.16 8.57 0.31 -0.20 0.83Mineral water 0.02 0.18 5.43 0.34 12.94 0.46 -0.35 0.80Fruit soft drink 0.10 0.14 7.02 0.33 33.05 0.45 -0.30 0.95Cola soft drink 0.10 0.11 8.94 0.16 5.63 0.25 -0.19 0.68Beer 12%, light from barrel 0.09 0.16 6.28 0.16 7.33 0.16 -0.17 0.71Beer 12%, light bottled 0.24 0.12 8.68 0.20 6.80 0.25 -0.18 0.88White wine, bottled, domestic 0.03 0.13 7.76 0.14 4.67 0.26 -0.21 0.85Red wine, bottled, domestic 0.03 0.06 15.53 0.12 4.41 0.19 -0.18 0.85Dessert white wine, bottled, domestic 0.00 0.11 9.22 0.28 0.89 0.28 -0.18 0.76Slovak juniper brandy 40% 0.05 0.17 5.81 0.19 4.67 0.32 -0.27 0.85Brandy 40%, domestic production 0.02 0.15 6.52 0.25 4.86 0.34 -0.26 0.88Complete lunch in factory canteen 6.89 0.12 8.07 0.15 6.84 0.14 -0.14 0.71Electric hair dryer 0.02 0.21 4.81 0.20 4.03 0.28 -0.20 0.93Electric hair iron with accessories 0.02 0.20 4.93 0.32 2.98 0.27 -0.20 0.83Electric razor 0.05 0.24 4.23 0.21 0.80 0.22 -0.18 0.76Manual 2-blade metal shaving razor 0.03 0.22 4.65 0.23 13.33 0.19 -0.16 0.93Razor blade - 5 pieces n a pack 0.24 0.30 3.32 0.21 19.28 0.13 -0.10 0.95Shaving foam 0.09 0.27 3.67 0.12 9.35 0.11 -0.10 0.88Tooth brush 0.09 0.25 3.93 0.34 13.46 0.20 -0.18 0.80Tooth paste 0.54 0.27 3.67 0.17 5.19 0.12 -0.11 0.90Suntan milk with protective factor 0.07 0.19 5.32 0.23 81.34 0.74 -0.22 0.95Cosmetic alcohol Alpa 0.02 0.16 6.10 0.16 4.52 0.11 -0.12 0.85Body deodorant 0.35 0.29 3.51 0.40 7.53 0.15 -0.13 0.90Powder for children 0.00 0.24 4.24 0.05 7.36 0.08 -0.09 0.95Bath soap, higher quality 0.38 0.22 4.48 0.21 9.23 0.12 -0.10 0.95Folded bandage absorbent quality 0.05 0.31 3.18 0.09 4.52 0.09 -0.09 0.93Paper handkerchiefs (10 pieces) 0.16 0.22 4.60 0.11 -0.21 0.14 -0.12 0.85Disposable napkins for children 0.50 0.30 3.38 0.11 7.12 0.13 -0.11 0.95Sanitary napkins, 10 pcs in package 0.71 0.28 3.59 0.22 10.80 0.15 -0.13 0.95Toilet paper - 400 slips 0.62 0.20 5.07 0.05 -1.41 0.08 -0.08 0.66Hair shampoo 0.73 0.25 3.98 0.10 5.63 0.13 -0.11 0.95Golden wedding ring 0.54 0.19 5.16 0.11 0.95 0.11 -0.09 0.71Wrist watch for men 0.23 0.23 4.40 0.36 7.44 0.27 -0.18 0.83Alarm clock on battery 0.02 0.22 4.64 0.12 1.25 0.14 -0.13 0.85Wall clock of Quartz type 0.03 0.19 5.27 0.10 2.69 0.14 -0.15 0.93Golden chain - mechanically wrought 0.16 0.24 4.17 0.11 0.64 0.08 -0.08 0.83Complete repair of wrist watch for men 0.09 0.18 5.58 0.25 6.46 0.11 -0.10 0.80Leather bag for women 0.10 0.32 3.17 0.16 10.29 0.16 -0.12 0.80Plastic bag for women 0.19 0.34 2.97 0.15 5.17 0.11 -0.09 0.85Leather purse for women 0.07 0.28 3.58 0.20 11.98 0.14 -0.13 0.90


Plastic suitcase 40x60cm 0.03 0.27 3.75 0.24 6.05 0.19 -0.17 0.93School bag 0.10 0.27 3.73 0.10 5.72 0.10 -0.10 0.90Children stroller combined 0.09 0.25 4.07 0.13 5.75 0.10 -0.11 0.90Matches 0.02 0.10 10.23 0.11 4.03 0.19 -0.15 0.88Sun glasses with ultraviolet filter 0.12 0.20 5.09 0.63 11.10 0.19 -0.14 0.85Umbrella for women 0.10 0.27 3.75 0.16 4.60 0.15 -0.18 0.80

Assessing Inflation Persistence - 80 -

3 Assessing Inflation Persistence: Micro Evidence on an Inflation

Targeting Economy*

3.1 Introduction

The sensitivity of aggregate inflation to various macroeconomic disturbances has been

traditionally at the focus of attention of monetary authorities. Indeed, the transmission of

monetary policy actions to prices depends on a number of factors, including inter alia the degree

of nominal rigidities. Consequently, in the last 20 years or so, there has been substantial research

investigating the macroeconomic consequences of nominal rigidities for the working of an

economy in response to various shocks and for the design of monetary policy rules. The result of

this effort has been a number of micro-founded models with price or wage stickiness which

predict various types of inflation dynamics. Nevertheless, two standard models in their original

versions, Calvo (1983) and Taylor (1980), imply no role for the backward-looking dimension of

inflation. These models, while assuming price stickiness, do not imply intrinsic inflation

stickiness.39

Several models address this issue by introducing the lagged value of inflation into a new

Keynesian Phillips curve. The rationale behind the inclusion of the lagged value differs across the * We thank Oxana Babetskaia, Martin Čihák, Oldřich Dědek, Tomáš Holub, Mario Holzner, Vladislav Flek, Michal Franta, Ondřej Kameník, Evžen Kočenda, Sangeeta Pratab, Kateřina Šmídková and the seminar participants at the Czech Economic Association annual conference, the European Economic Association annual congress, Deutsche Bundesbank, Charles University (Prague) and Czech National Bank for valuable comments. We are grateful to Robert Murárik for providing us with some of the data we used. All remaining errors are entirely our own. The views expressed in this paper are not necessarily those of the Czech National Bank or the European Bank for Reconstruction and Development. This research was supported by a grant from the CERGE-EI Foundation under a programme of the Global Development Network. All opinions are those of authors and have not been endorsed by CERGE-EI or the GDN. This paper has been supported by the Czech National Bank Research Project No. E5/05. 39 Assuming the Galí and Gertler (1999) hybrid New Keynesian Phillips curve specification for inflation dynamics, Angeloni et al. (2006) distinguish between various sources of inflation persistence and label them accordingly. They define intrinsic inflation persistence as the persistence originating in past inflation, extrinsic inflation persistence as the persistence related to inertia in the output gap, and expectation-based inflation persistence as the persistence rooted in deviations from rational expectations due, for example, to learning.


models. Apart from simply assuming rule of thumb behavior (Galí and Gertler, 1999), Fuhrer

and More (1995) suggest that the relative wage structure might be a reason for the backward-

looking nature of inflation. Mankiw and Reis (2002) stress the significance of information

processing lags in price setting mechanisms. In addition, Erceg and Levin (2003) and Orphanides

and Williams (2003) explain persistence with adaptive learning of agents in response to changes

in monetary policy regime. In consequence, the ability of monetary policy to anchor long-term

inflation expectations induces agents to rely on past inflation to a lesser extent. In this regard,

Sargent (1999) studies extensively the interactions between the conduct of monetary policy and

inflation persistence. Nimark (2005) suggests that optimal price setting with firm-specific

marginal cost rationalizes the link between past and current inflation. Calvo, Celasun and

Kumhof (2002) show that in an environment of high steady state inflation, firms not only choose

their price today, but also set the rate at which they will update prices in the future (the firm-

specific inflation rate). Under a monetary policy shock, some firms will not reset their inflation

rate (and prices) and this gives rise to inflation inertia.

Recent empirical research has shown that inflation persistence is generally much lower than

previously thought (e.g. Cecchetti and Debelle, 2006). This is mainly associated with two factors.

First, inflation persistence did indeed decline in the 1990s as compared to the 1970s and 1980s

(O’Reilly and Whelan, 2005). Second, greater care has been undertaken in econometric work.

Levin and Piger (2004) find that inflation persistence falls considerably when structural breaks are

accounted for. Next, stability of the monetary policy regime and central bank credibility help to

anchor long-run inflation expectations and reduce the extent of backward-looking behavior.

Levin et al. (2004) find that the adoption of an explicit inflation target40 significantly reduces the

extent to which economic agents use backward-looking information in terms of their inflation

forecasting and thus puts downward pressure on the persistence of inflation. 40 See Kotlán and Navrátil (2003) on the design of the inflation targeting regime in the Czech Republic, and Jonas and Mishkin (2003) on the inflation targeting experience of transition countries in general.


There are various reasons why it is vital to study inflation persistence at a disaggregated level.

Disaggregated analysis generally uncovers smaller inflation persistence across the

individual/sectoral price indexes compared to aggregate inflation. This suggests that inflation

persistence observed at the aggregate level may arise due to aggregation bias (see Granger, 1980,

and Zaffaroni, 2004) and due to the fact that idiosyncratic shocks will tend to disappear when a

substantial number of series are aggregated (Altissimo, Mojon and Zaffaroni, 2007). Disaggregate

analysis is also fruitful for understanding which components of various price indexes exhibit

greater inflation persistence. In addition, the role of structural breaks in estimating inflation

persistence can be tackled in a fuller manner.

Additionally, several studies have raised the issue of which factors lie behind the fact that the

inflation process is relatively persistent. Cournede et al. (2005) argue that the lower responsiveness

of aggregate inflation to output developments in the euro area in comparison to the U.S. is

caused by more rigid structural policy settings and relate it to trade barriers in the European

services sector. Analogously, European Commission (2003) points out that low competition in

services enhances the sector’s inflation inertia as measured at the aggregated level. On the other

hand, studies employing disaggregated data, such as Lunnemann and Matha (2005) for several

EU countries and Clark (2006) for the U.S., find little evidence that services display greater

inflation persistence than goods. Similarly, Coricelli and Horvath (2006) report results for

Slovakia indicating that inflation inertia in the services sector is even lower than for goods and

put forward an explanation of why (labor intensive) services, where the degree of competition is

typically lower as services are often not exposed to international competition, may in fact exhibit

smaller persistence. The argument is based on Calvo (2000), who shows that greater competition

in the market may actually slow down the adjustment to shocks, as the degree of strategic


complementarity increases with higher competition. All these aforementioned issues give further

impetus for individual or sectoral level analysis of inflation persistence.

One of the interesting applications of inflation persistence analysis at the disaggregate level is

provided by Cutler (2001). Cutler constructs an alternative measure of core inflation –

persistence-weighted core inflation. The measure is constructed in a way giving larger weights to

items exhibiting higher inflation persistence. Using UK data, Cutler finds that in terms of ability

to predict headline inflation this measure outperforms some other standard measures of core

inflation, such as those using a trimmed mean or weighted median or those excluding food and

energy prices.41

In addition, it is noteworthy that there is still very little evidence on price setting behavior in the

New EU Member States (NMSs). Typically, the few available studies focus on aggregate inflation

dynamics. More detailed evidence on price setting is provided by Ratfai (2006), who studies the

linkages between individual price dynamics and aggregate inflation with Hungarian data.

Additionally, Konieczny and Skrzypacz (2005) analyze the price dynamics of about 50 products

in Poland. Among other things, they show that more intense search is associated with smaller

price dispersion. Coricelli and Horvath (2006) give evidence on the empirical stylized features of

price setting behavior in Slovakia using a large micro-level dataset underlying the Slovak CPI.

Recently, inflation persistence at the aggregate level for the EU new members has also been

studied by Franta et al. (2007).

Therefore, a novel contribution of this study lies in exploring inflation persistence at the

disaggregate level in the Czech Republic using rich data collected by the Czech Statistical Office,

which cover about a thousand product categories over 1994–2005 (accounting also for structural 41 Notice that in general the forecasting ability of persistence-weighted measures of inflation may depend on the monetary regime and the degree of inflation persistence. For a discussion, see Smith (2004, 2005).


breaks). Furthermore, our study goes beyond a simple statistical description of the data and

makes an attempt to identify the determinants of inflation persistence. Of particular interest is the

examination of the so-called “services inflation persistence puzzle”, namely that more labor

intensive categories such as services often exhibit smaller persistence as compared to goods (see,

for example, Altissimo, Mojon and Zaffaroni, 2007; Clark, 2006; Coricelli and Horvath, 2006).

Finally, we construct “persistence-weighted” core inflation in line with Cutler (2001) and propose

a “persistence expenditure-weighted” core inflation measure that combines information on the

persistence of an individual product and its weight in the CPI basket, with the objective of

assessing its predictive performance (ability to capture inflation trends) compared to other

alternative approaches for core inflation measurement.

The paper is organized as follows. After this introduction to the subject and overview of the key

literature, the second section describes how inflation persistence is measured in practice,

formulates the research hypotheses and explains the estimation methodology. The third section

presents the data set used in the study. The fourth section provides the results. The last section

concludes and draws policy implications. An Appendix with additional results and sensitivity

checking follows.

3.2 Estimating inflation persistence

The literature generally applies two statistical approaches to estimating inflation persistence –

parametric and non-parametric. The parametric approach is more extensively applied in empirical

studies (Cecchetti and Debelle, 2006; Clark, 2006; Levin and Piger, 2004; Levin, Natalucci and

Piger, 2004). As advocated by Andrews and Chen (1994), the best scalar measure of persistence is

the sum of autoregressive coefficients in the dynamic equation for inflation:

επαπ µtjt

K

jjt

++=−

=∑

1

, (1)


where π t stands for the yearly inflation rate, µ and α j

are parameters, and ε t is the white-

noise disturbance. The lag length K is determined based on information criteria. Typically, ∑=

K

jj

1α

is interpreted as the measure of inflation persistence. Specification (1) may be labeled as naïve,

because it does not account for potential structural breaks. A number of recent studies apply

various tests for structural breaks (e.g. Cecchetti and Debelle, 2006; Levin and Piger, 2004).

A non-parametric approach has been recently put forward by Marquez (2004). This approach

builds on the idea that less persistent inflation is more likely to cross the long-run mean of the

inflation rate (or possibly the time-varying mean). Consequently, inflation persistence, ϕ , is

measured as Tn−= 1ϕ , where n is the number of times inflation crosses its equilibrium

value and T is the number of observations. Dias and Marquez (2005) derive the finite sample and

asymptotic properties of this non-parametric measure. They also conduct Monte Carlo

simulations and find that the bias of the estimate of persistence based on the non-parametric

approach is smaller for any sample size, as compared to the parametric measure from equation

(1). In addition, they argue that the non-parametric measure is more robust to structural breaks.

Nevertheless, the properties of this measure are investigated only for covariance stationary

processes.

Despite the potential attractiveness of the approaches described above, in our case we find that

most individual inflation rates follow an I(1) process (even if we control for structural breaks).

For such a case, the properties of the non-parametric approach have not been investigated yet.

Analogously, in the case of a parametric measure – e.g. the sum of autoregressive coefficients – it

is well known that non-stationarity of the variables would result in spurious regression.


Therefore, we do not report these measures and propose a different measure of the persistence

of inflation.42

Given the non-stationarity of inflation series, we opt for an examination of the degree of inflation

persistence using the complementary unit root and stationarity tests. Specifically, we use the

augmented Dickey-Fuller test (Dickey and Fuller, 1981), Phillips-Perron test (Phillips and Perron,

1988) and KPSS test (Kwiatkowski et al., 1992). Given that our data come from a former

transition country, we test the robustness of the results by carrying out a unit root test with a

structural break (Saikkonen and Lütkepohl, 2002, and Lanne et al., 2002, labeled as the LLS test

hereinafter).

For the Augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) unit root tests, the probability

of rejecting the null hypothesis of a unit root will be reported. The probability can vary from 0 to

1. Higher values thus correspond to more persistence. For example, a probability higher than

0.10 means that the null of a unit root cannot be rejected at the 10% significance level. For the

KPSS stationarity test, the t-statistic will be reported: higher t-statistic values increase the

probability of rejecting the null hypothesis of stationarity and hence characterize more

persistence in the underlying series.

The number of lags in the aforementioned tests for each product is determined according to the

Akaike information criterion. We address the sensitivity of the results by estimating persistence

first for the full sample and then for the restricted sample, i.e. using data only after the

introduction of inflation targeting in 1998.

42 A straightforward application of the non-parametric method to our data does not bring any meaningful insight: the degree of persistence across all sectors is found to be very similar.


Next, we also run a unit root test with a structural break. Given a relatively short time series, we

test for only one structural break on an unknown date (Lanne et al., 2002). As we find that most

of the time series exhibit a structural break around 1998–1999 (shortly after the adoption of

inflation targeting), we decided to employ a unit root test where we impose the break (captured

by the shift dummy) in 1998:1.43 The rationale for imposing the break is to ensure that we subject

each time series to the identical testing procedure and consequently to allow cross-sectional

comparability of our results. We take the t-value from this test as the measure of the persistence

of the series, with a more negative value indicating less persistence (increasing the probability of

rejecting the null hypothesis of a unit root process).

Furthermore, one can put forward a critique that p-values might not generally serve as a universal

measure for the degree of inflation persistence.44 Therefore, we also measure persistence by

simply running the aforementioned stationarity and unit root tests and examining whether we can

reject the corresponding null hypothesis at a reasonable level of significance.45 We then use the

following coding to assess the degree of persistence: 1 if the series is found to contain a unit root,

and 0 if the series is stationary. Subsequently, we calculate the share of unit root processes for

particular sectors. As a result, this exercise provides an additional sensitivity check of our results.

Obviously, the drawback of this measure is that it is not possible to evaluate the extent of

aggregation bias.

It is also vital to note that we use year-on-year inflation rates, for the following reasons. Other

possibilities, such as using month-on-month and quarter-on-quarter changes in the price level, are

associated with seasonality, which may contaminate the true extent of persistence. In addition, 43 Therefore, we estimate the LLS test only for our full sample (1995–2005) and do not estimate the test for the restricted sample (1998–2005, i.e. the inflation targeting period), as we do for the ADF, PP and KPSS tests. 44 Given that p-values are affected by the standard errors of the estimated coefficients, the distribution of p-values is also influenced by the sample size. Hence, p-values cannot be used to compare persistence in, for example, very short versus large samples. Since in our case the sample size is the same for all products (about 100 observations), p-values can be informative in characterizing the non-stationarity properties of the underlying series. 45 More specifically, we use the 5% and 10% significance levels.


these two aforementioned changes are typically not monitored by economic agents such as

households or unions. Most importantly, central banks set their inflation targets in year-on-year

changes in the price level. In addition, Aron and Muellbauer (2006) claim that year-on-year

inflation rates also capture the dynamics of month-on-month inflation.46

3.3 Data

The Czech Statistical Office included 1,022 narrowly defined products in the consumer basket

between 1994 and 2005 on a monthly frequency. Nevertheless, prices of many products were not

tracked over the whole sample period. Typically, the whole consumer basket includes about 700

products on any given date. As a result, we were able to identify 412 individual products for

which the price indexes are available for the whole period spanning from 1994:M1 to 2005:M12.

The selected 412 products represent 64% of the CPI basket for 2005.

As a benchmark, we construct sample inflation as a weighted average of 412 individual price

indices (year-on-year percentage changes). Figure 1 shows the official CPI inflation and our

sample inflation over 1995–2005 at monthly frequency. The high similarity between the two

series suggests that our sample of 412 products is fairly representative in terms of inflation

dynamics. On average, annual CPI inflation in the Czech Republic was about 4.3% over the

period 1994–2005. Prior to 1998, inflation fluctuated around 10%, while successful disinflation

policy resulted in average inflation of around 3% during 1999–2005.

46 Nevertheless, for the purposes of sensitivity checking, we replicate our analysis on month-on-month inflation rates (the results are available upon request). We find that in such case inflation exhibits less persistence compared to the yearly base. A similar observation was pointed out by Altissimo, Ehrmann and Smets (2006): the same series is found to be less persistent if considered in quarter-on-quarter changes compared to year-on-year changes.


Figure 5. Official CPI inflation and sample inflation, 1995–2005

-2

0

2

4

6

8

10

12

14

1619

95/0

1

1996

/01

1997

/01

1998

/01

1999

/01

2000

/01

2001

/01

2002

/01

2003

/01

2004

/01

2005

/01

Sample Official

To facilitate interpretation, the individual 412 products are further grouped into several broader

categories according to their characteristics (in line with the Czech National Bank internal

classification of products for reporting sectoral inflation rates). These are: tradables, non-

tradables, durables, regulated goods and services, non-regulated services, raw goods and

processed goods. Products are also classified by the statistical office into 12 main categories

according to the classification of individual consumption by purpose (COICOP). These

categories are food and non-alcoholic beverages; alcoholic beverages and tobacco; clothing and

footwear; housing, water, gas, and electricity; furnishings and maintenance of the house; health

care expenses; transport; communications; leisure and culture; education; hotels, cafés, and

restaurants; and miscellaneous goods and services.


3.4 Results

In the first part, we perform product-specific estimates of inflation persistence using the unit root

(ADF, PP, LLS) and stationarity (KPSS) tests. Then we examine the effect of aggregation on

inflation persistence and analyze whether inflation persistence changes over time. The second

part is devoted to an assessment of the determinants of inflation persistence. Finally, we evaluate

the predictive ability of persistence-weighted core inflation.

4.1 Inflation persistence estimates

The overall distribution of inflation persistence across product categories is summarized in Figure

2 below. The degree of persistence is depicted on the horizontal axis, while the vertical axis

displays the kernel density. Several stylized facts follow from Figure 2.

All three tests suggest that aggregate inflation exhibits significantly higher persistence than the

average inflation persistence as measured at the disaggregate level for the whole sample as well as

for the 1998–2005 sub-period47 (the results of Altissimo, Mojon and Zaffaroni, 2007, and Clark,

2006, for example, also indicate this discrepancy). Generally, there are two possible explanations

for this phenomenon. First, Granger (1980) showed that cross-sectional aggregation of (even

simple) time series may result in complex, often more persistent processes (i.e. aggregation bias).

Typically, the aggregation bias is likely to be greater when there is large heterogeneity in the

product-level inflation persistence. As a result, the estimated persistence of aggregate inflation

may change due to changes in sectoral heterogeneity. Second, it may also reflect the fact that

idiosyncratic shocks vanish due to aggregation. Next, we assess the robustness of these findings

47 The results are valid regardless of whether the sample aggregate inflation is constructed using the mean, weighted mean or median. The gap between aggregate inflation and the average inflation across the disaggregated components is different from zero at the 1% significance level, as suggested by the t-test. However, this significance may be overestimated since the conventional t-test is applied to the test statistics, not to the raw data.


by also running an LLS unit root test with a structural break (Saikkonen and Lütkepohl, 2002,

and Lanne et al., 2002). The break is captured by the shift dummy in 1998:M1. The results from

this test confirm the presence of aggregation bias (see Figure A.2 in the Appendix).

One can also observe a noticeable reduction in overall CPI inflation persistence for the sub-

period 1998–2005 (i.e. the inflation targeting period), while the sample aggregate inflation

persistence has decreased rather marginally (see the lower part of Figure 2). We find that it was

the persistence of tradables (especially durable goods) inflation rather than that of non-tradables

that declined after the adoption of inflation targeting.

Similar evidence of aggregation bias is observed when comparing inflation persistence for the

aggregate CPI and nine sectors (see Table 1 and Table 2; note that the results are obtained by

aggregating the product-specific estimates). Overall, the results in Table 1 and 2 seem to indicate

that inflation persistence in the Czech Republic is higher compared to the euro area members.

While for the Western European countries there are relatively few cases of I(1) processes at

sectoral and even aggregate levels (European Central Bank, 2005), and while the results of

stationarity and unit root tests are often inconclusive48 (Gadea and Mayoral, 2006), the results for

the Czech Republic are much more clear-cut. Czech inflation follows a unit root process for most

of the sectors. On the other hand, Franta et al. (2007) find that aggregate inflation persistence in

the new EU member states tends to be lower than in the euro area when allowing for the time-

varying inflation target.

48 In other words, Gadea and Mayoral find that many sectoral inflation series are fractionally integrated, i.e. follow a process between I(0) and I(1).


Figure 2. Distribution of inflation persistence across 412 products and aggregation bias

ADF 1995–2005 PP 1995–2005 KPSS 1995–2005

ADF 1998–2005 PP 1998–2005 KPSS 1998–2005

Notes: Vertical bold lines denote the persistence of aggregate CPI inflation; simple vertical lines represent the mean of disaggregate inflation persistence. The horizontal axis characterizes the level of inflation persistence (higher values mean more persistence). For all the measures of persistence displayed, higher values mean more persistent inflation. For the ADF and PP unit root tests, the probability of rejecting the null hypothesis of a unit root is reported. The probability can vary from 0 to 1. Higher values correspond to more persistence. For example, a probability higher than 0.10 means that the null of a unit root cannot be rejected at the 10% significance level. For the KPSS stationarity test, the t-statistic is reported. Higher t-statistic values increase the probability of rejecting the null hypothesis of stationarity and hence characterize more persistence in the underlying series.

Moreover, in the Czech case the results of the unit root and stationarity tests are quite similar at

the sectoral level (the test performance at the product level is assessed in the next paragraph). For

example, considering the period from 1995 to 2005 (Table 1), the results of the unit root and

stationarity tests give the same picture: 8 out of the 9 sectors exhibit a unit root process at the

10% significance level; raw goods (line 8) are the only sector which is stationary at the 10% level,

as supported by both the unit root (ADF/PP) and stationarity (KPSS) tests. This similarity

between unit root tests and stationarity tests gives support for I(1) behavior of sectoral inflation

rates. Note that these results are obtained assuming no trend in inflation. The incorporation of a


time trend in the inflation dynamics or accounting for a time-varying inflation target could be

further investigated.

In terms of ranking the persistence across sectors, we find that raw goods consistently exhibit the

smallest inflation persistence. On the other hand, durables inflation seems to be the most inertial.

Interestingly, services and regulated products do not display greater persistence. This finding is

also robust to our alternative indicator of inflation persistence – the share of unit roots. The

attendant results are available in Table A.1 in the Appendix.

Table 1. Inflation persistence, yearly inflation, 1995–2005 (132 obs.)

Measures of persistence Sector No. of products

Sample weights ADF PP KPSS LLS

Tradables 311 0.59 0.31 (0.29) 0.31 (0.27) 0.69** (0.39) -2.35 (1.12) Non-tradables 101 0.41 0.24 (0.21) 0.22 (0.20) 0.55** (0.30) -2.32 (1.03) Services 96 0.40 0.24 (0.21) 0.22 (0.20) 0.56** (0.30) -2.30 (1.05) Non-reg.serv. 74 0.30 0.24 (0.21) 0.21 (0.19) 0.56** (0.30) -2.32 (1.00) Regulated 27 0.11 0.23 (0.21) 0.24 (0.20) 0.53** (0.28) -2.32 (1.13) Durables 164 0.21 0.44 (0.29) 0.43 (0.28) 0.90*** (0.34) -1.86 (0.92) Non-durables 152 0.39 0.16 (0.20) 0.18 (0.18) 0.46* (0.31) -2.88** (1.05) Raw goods 42 0.11 0.07 (0.13) 0.09 (0.11) 0.24 (0.19) -3.43** (1.13) Processed goods 370 0.89 0.32 (0.28) 0.31 (0.26) 0.71** (0.36) -2.22 (1.02) Total prod. level 412 1.00 0.29 (0.28) 0.29 (0.26) 0.66** (0.38) -2.35 (1.09) Aggr. inflation 1 1 0.48 0.49 1.03*** -1.80

Notes: The pairs (tradables, non-tradables) and (raw goods, processed goods) make up a total of 412 products. Durables do not include regulated prices, while processed goods do. For all the measures of persistence displayed, higher values mean more persistent inflation. For the ADF and PP unit root tests, the probability of rejecting the null hypothesis of a unit root is reported. The probability can vary from 0 to 1. Higher values correspond to more persistence. For example, a probability higher than 0.10 means that the null of a unit root cannot be rejected at the 10% significance level. Standard deviations are shown in parentheses. For the KPSS stationarity test, the t-statistic is reported. Higher t-statistic values increase the probability of rejecting the null hypothesis of stationarity and hence characterize more persistence in the underlying series. *, **, and *** denote the 10%, 5% and 1% asymptotical significance levels for rejection of the stationarity hypothesis. Standard deviations are shown in parentheses. For the LLS (Lanne et al., 2002) unit root test in the presence of a structural break, the t-statistic is reported. More negative t-statistic values increase the probability of rejecting the null hypothesis of a unit root and thus characterize less persistence in the underlying series. *, **, and *** denote the 10%, 5% and 1% asymptotical significance levels for rejection of the unit root hypothesis.


Table 2. Inflation persistence, yearly inflation, 1998–2005 (96 obs.)


Sample weights ADF PP KPSS

Tradables 311 0.59 0.21 (0.21) 0.23 (0.19) 0.52** (0.35) Non-tradables 101 0.41 0.23 (0.19) 0.22 (0.17) 0.46* (0.28) Services 96 0.40 0.24 (0.19) 0.22 (0.17) 0.47** (0.29) Non-reg. serv. 74 0.30 0.27 (0.19) 0.25 (0.16) 0.46** (0.27) Regulated 27 0.11 0.12 (0.17) 0.14 (0.16) 0.47* (0.31) Durables 164 0.21 0.24 (0.24) 0.26 (0.23) 0.70** (0.32) Non-durables 152 0.39 0.16 (0.15) 0.20 (0.14) 0.31 (0.25) Raw goods 42 0.11 0.12 (0.14) 0.15 (0.13) 0.16 (0.12) Processed goods 370 0.89 0.22 (0.21) 0.24 (0.19) 0.54** (0.33) Total prod. level 412 1.00 0.21 (0.20) 0.23 (0.19) 0.50** (0.33) Aggr. inflation 1 1 0.26 0.27 0.63**

Notes: As for Table 1.

In addition, our results suggest that inflation persistence has decreased in the post-1998 period,

i.e. since inflation targeting was adopted. Vega and Winkelried (2005) find that inflation targeting

helps in reducing the volatility of inflation; however, the effect on inflation persistence is rather

ambiguous. On the other hand, the results of Levin et al. (2004) indicate that inflation targeters

indeed exhibit smaller inflation persistence. Likewise, Yigit (2007) documents that the adoption

of an inflation target provides a coordinating effect on the inflation expectations of economic

agents and therefore puts downward pressure on inflation persistence.

In this regard, while we find that there are 314 categories out of 412 for which we cannot reject

the null of a unit root based on the ADF test in the 1995–2005 sample at the 5% significance

level, there are 256 such categories in 1998–2005 (note that for the PP test the figures are 339

and 322 categories, respectively). In the case of the KPSS test, we reject the null of stationarity at

the 5% significance level for 269 categories over 1995–2005 and 207 categories for 1998–2005.

These results suggest that inflation persistence may be somewhat lower after the adoption of

inflation targeting in 1998; however, this should be taken with caution, as the power of the tests


may decrease for the shorter sample. Table A.1 presents the detailed results on the (both simple

and consumption-weighted) share of unit root processes, including the LLS test.

We also find that the estimated inflation persistence falls when we control for structural breaks.

This is evident from comparing the ADF and LLS results. The construction of the LLS test

implies that it is essentially the ADF test “adjusted” for the structural break. The results

presented in Table A.1 indicate that the share of unit root processes is indeed smaller for the LLS

test as compared to the ADF test. The results thus comply with Levin and Piger (2004).

At the individual product level, the link between the various tests is illustrated in Figure A1 in the

Appendix. The correlation of the LLS test with the ADF, PP and KPSS tests stands at 0.76, 0.75

and 0.5, respectively. The P-values of the ADF and PP tests are closely related: the corresponding

correlation coefficient is 0.94 for 1995–2005 and 0.87 for 1998–2005. The correlation between

the unit-root tests and the KPSS test for stationarity is fairly high (0.63 and 0.67, respectively) for

1995–2005, and much lower (0.31 and 0.31, respectively) for 1998–2005.

Such a difference over the two periods is likely to be due to the following reasons. First, as the

number of observations decreases the tests lose their power to reject the null hypothesis – that of

an I(1) process for the ADF/PP tests, and of an I(0) process in the case of the KPPS. Second, as

inflation itself has decreased over time, it becomes more difficult to distinguish whether the series

follow an I(0) or I(1) process; the series may become fractionally integrated, as is the case for

disaggregate inflation in West European countries (see Gadea and Mayoral, 2006). In other

words, the growing differences between the unit root and stationarity tests may capture the effect

of structural changes in the Czech Republic and give further indirect support for our supposition

that inflation persistence decreased after the adoption of inflation targeting.


4.2 Explaining inflation persistence

Once the disaggregate estimates of inflation persistence are obtained, we test them for any

significant determinants. In particular, we analyze the ability of product characteristics to explain

the cross-sectional variation in persistence across 412 individual products. In addition, we analyze

the so-called “service inflation persistence puzzle”: several studies have revealed that (labor-

intensive) services, which are typically not subject to international competition, surprisingly

display smaller persistence than goods (see, for example, Altissimo, Mojon and Zaffaroni, 2007;

Clark, 2006; and Coricelli and Horvath, 2006). Thus, our results will add a piece of evidence on

this “service inflation persistence puzzle”. More generally, we analyze the implications of the

degree of competition for inflation persistence.

One hypothesis to explain the variation in inflation persistence is that it differs across sectors.

Concerning the sectoral categories, raw goods indeed demonstrate the lowest inflation

persistence (and the lowest dispersion) among the nine sectors considered. Non-durables have

the second-lowest persistence and dispersion of inflation. Apart from aggregate inflation, the

sector with the highest inflation persistence (and also dispersion) is durables, followed by

processed goods and tradables.

It is interesting to note that services are typically non-tradable and more labor-intensive, i.e. their

prices are likely to be set in a less competitive environment than that for goods. Naturally, the

incentives for price revision for services should then be weaker and thus the convergence to

frictionless equilibrium slower. Consequently, one would expect services prices to display greater

inertia. However, our results – like the empirical evidence – do not support this reasoning. We

find that inflation in services exhibits lower persistence, although for the post-1998 period this

difference diminishes and becomes sensitive to the choice of test. Similarly, Clark (2006) for the

U.S. as well as Coricelli and Horvath (2006) for Slovakia report smaller inflation persistence in


services than for manufacturing using micro level data. Lunnemann and Matha (2004) find that in

about 5 out of 15 EU countries the persistence in services inflation is smaller than the persistence

of the overall HICP.

In this regard, Coricelli and Horvath (2006) put forward an explanation for the finding that

services inflation is often found to exhibit smaller persistence than goods. Typically, it is assumed

that higher competition increases the incentives for price revisions and the market has a tendency

to adjust faster. On the other hand, Calvo (2000) shows that a greater degree of competition may

increase the inertia rather than decrease it. This is because when markets are highly competitive, it

is more likely that individual prices will not diverge far from the average (firms “follow the

pack”)49, otherwise the firm would be pushed out of the market. In other words, the degree of

strategic complementarity among price setters increases with higher competition and individual

pricing decisions will be more affected by the average pricing strategy in the market.

Consequently, greater competition reduces price dispersion; however, it does not have to

decrease persistence.

Price dispersion can be interpreted as a measure of market competition.50 Consequently, this

allows us to test the aforementioned supposition that the degree of competition may indeed be

positively related to inflation persistence. We measure price dispersion as the standard deviation

of price indexes within an individual COICOP category normalized to one in the initial period.

The resulting COICOP-specific measure of price dispersion is obtained by averaging the

standard deviations over time.

49 Note also that deviation from the price of competitors has been found to be one of the most important obstacles to price adjustment in surveys of euro area firms (see Fabiani et al., 2006). 50 A number of recent empirical studies find a negative relationship between price dispersion and the degree of market competition [Baye et al. (2004), Caglayan et al. (2007), Gerardi and Shapiro (2007), Leiter and Warin (2007)].


First, simple pair-wise correlations are illustrated in Table 3. Particularly strong correlations are

detected for the categories of durables and raw goods. We also find a significantly negative

correlation between our measure of price dispersion and inflation persistence. This is robust to

the measure of inflation persistence as well as the sample period.

Table 3 – Correlation matrix – Inflation persistence and product characteristics

1995–2005 1998–2005 ADF PP KPSS LLS ADF PP KPSS Price dispersion -0.25 -0.28 -0.32 -0.18 -0.08 -0.09 -0.27 Durables 0.44 0.45 0.53 0.36 0.13 0.12 0.47 Goods 0.10 0.14 0.14 -0.01 -0.08 0.01 0.05 Non-durables -0.37 -0.33 -0.42 -0.37 -0.20 -0.11 -0.43 Non-tradables -0.11 -0.16 -0.16 0 0.06 -0.02 -0.07 Processed goods 0.28 0.27 0.37 0.34 0.08 0.07 0.34 Raw goods -0.28 -0.27 -0.37 -0.34 -0.08 -0.07 -0.34 Regulated products -0.05 -0.06 -0.09 0 -0.12 -0.14 -0.08 Services -0.11 -0.13 -0.11 0.02 0.05 -0.01 -0.05 Services – non-regulated -0.10 -0.12 -0.08 0.02 0.13 0.09 -0.02 Tradables 0.1 0.16 0.16 0 -0.06 0.02 0.07

Note: Correlation coefficients greater than 0.08 in absolute terms are significant at the 5% level.

Next, we present our results on the determinants of inflation persistence using here the KPSS

test-based estimates of persistence in Table 4. The results suggest that greater price dispersion, a

measure of competition, is associated with smaller inflation persistence, implying that

competition is not conducive to reducing persistence. This finding holds for both our estimation

periods (the full sample, 1995–2005, and the inflation targeting-restricted sample, 1998–2005),

when controlling for product characteristics and altering our estimation technique (OLS vs.

GMM), and, on top of that, is largely unaffected by the measure of persistence (see Tables A.2,

A.3 and A.4 for the results based on targeting-restricted ADF, PP and LLS test-based estimates

of persistence). In addition, we present a logit estimation of the inflation persistence

determinants, which further confirms our findings. Our dependent variable is coded one if the


product inflation is found to follow an I(1) process at the 10% significance level51, and zero

otherwise. The results are available in Table A.5 in the Appendix.

We report both the OLS and GMM estimates to check the robustness of the results. While OLS

may be subject to endogeneity bias, it is known that GMM may give biased results for a smaller

sample. Next, we also control for product characteristics (two products with high correlation

with inflation persistence) and present the results for two sample periods. The Appendix also

contains Table A.6, where we study the impact of product characteristics on inflation persistence.

We find that raw goods as well as non-durables exhibit smaller inflation persistence. There is

some evidence that inflation in the services sector exhibits smaller persistence.

Table 4 – Determinants of inflation persistence

1995–2005 1998–2005 KPSS KPSS KPSS KPSS KPSS KPSS Price dispersion -1.25*** -10.4*** -2.57*** -0.91*** -9.23*** -1.71*** (0.18) (3.85) (0.18) (0.17) (3.53) (0.53) Non-durables -0.17** -0.17*** (0.08) (0.06) Raw goods -0.31*** -0.24*** (0.10) (0.07) Adj. R-squared 0.11 --- --- 0.07 --- --- Estimation method OLS GMM GMM OLS GMM GMM Sargan test (p-value) --- 0.2 (0.15) 0.4 (0.40) --- 1.5 (0.23) 0.9 (0.33) Observations 412 412 412 412 412 412

Note: Heteroscedasticity robust standard errors are shown in parentheses. The list of instruments for price dispersion is as follows: non-regulated services, non-durables, raw goods and regulated prices dummies. ***, **, and * denote significance at 1%, 5%, and 10%, respectively. P-value in brackets for the Sargan (overidentifying restrictions) test.

To further support our results that competition is likely to be negatively related to inflation

persistence, we present the determinants of price dispersion. Here we expect that non-

tradables/services, as they are typically not subject to international competition, will exhibit

51 The 5% significance level was used as the cut-off point for coding the dependent variable as well. The results remained largely unaffected.


greater price dispersion. Controlling for other product characteristics, the results in Table 5

indicate that the degree of non-tradability of a product, as captured by the services dummy, is

positively linked to price dispersion (see also Crucini et al., 2005).

Table 5 – Determinants of price dispersion

Price dispersion Services – non-regulated 0.06*** 0.09*** 0.07*** 0.09*** 0.11*** (0.01) (0.01) (0.01) (0.01) (0.01) Non-durables 0.07*** 0.07*** 0.08*** (0.01) (0.01) (0.01) Raw goods 0.03*** -0.001*** -0.001 (0.01) (0.001) (0.001) Regulated 0.17*** (0.02) Adj. R-squared 0.06 0.15 0.06 0.15 0.15 Estimation method OLS OLS OLS OLS OLS Observations 412 412 412 412 412

Note: Heteroscedasticity robust standard errors are shown in parentheses. ***, **, and * denote significance at 1%, 5%, and 10%, respectively.

4.3 Predictive ability of persistence-weighted core inflation

In order to improve inflation forecasts, a number of core inflation measures have been developed

to capture underlying inflation trends. Generally, the measures remove or reweight the most

volatile categories of inflation, such as energy prices. Smith (2004) notes that core inflation

measures typically exploit cross-sectional information, while time-series information has been

much less noted. In line with this, we construct a measure of core inflation, coretI , based on

product-level inflation rate persistence, giving a greater weight to categories exhibiting greater

persistence, and examine its predictive ability by comparison with other measures of core

inflation as well as various inflation forecasts.

The underlying idea is that the more persistent components of headline inflation may do a good

job in capturing inflation trends. In this context, Cutler (2001) finds that in the case of U.K. data,


persistence-weighted core inflation outperforms other core inflation measures. Cutler (2001)

argues that the exclusion of certain products from the basket in the construction of core inflation

can be arbitrary, and what is more, she finds that certain non-seasonal food prices (food prices

are typically excluded from core inflation) exhibit relatively persistent inflation and thus their

behavior may provide additional information for capturing trends in inflation series.

Our persistence-weighted core inflation, PWcoret

,π , is based on Cutler (2001) and is constructed as

follows:

iti

iPWcore

t p ,

413

1

, ∆=∑=

θπ

where iθ denotes the i-th product inflation persistence (normalized such that 1413

1=∑

=iiθ ) and

itp ,∆ is the i-th product yearly inflation rate at time t. As an alternative indicator, we combine

information on the persistence of an individual product, iθ , and the weight of that product in

the CPI basket in the following way,

iti

iPEWcore

t p ,

413

1

, ∆=∑=

ξπ

where iξ is the simple average of iθ – the individual inflation persistence – and iw is the sample

weight of the i-th product in the CPI basket, where iθ and iw are normalized such that

1413

1=∑

=iiθ and 1

413

1=∑

=iiw . Consequently, we label PEWcore

t,π as the persistence expenditure-

weighted core inflation.

We undertake a simple exercise here to evaluate the predictive ability of persistence-weighted

core inflation vis-à-vis other (core) inflation measures. Specifically, we compare it with net

inflation, median net inflation (the median net individual inflation rate), and so-called adjusted


inflation (net inflation excluding food, beverages and tobacco) over the horizons of 6, 12 and 18

months. The mean square error (MSE) will be used to measure the forecast quality:

( )2

1

,/1 ∑=

+ Π−Π=T

t

iCOREt

CPIhtTMSE ,

where T is the number of observations, h is the time horizon in months and iCOREt

,Π is the

selected core inflation measure.

Figure 3 depicts the predictive ability of the aforementioned core inflation measures. Here we

used the persistence measure based on the ADF test on the 1995–2005 data.52 The results

indicate that adjusted inflation exhibits the smallest MSE and thus is the best predictor of the

core inflation measures considered. Net inflation, median net inflation and persistence-weighted

core inflation, PWcoret

,π , do not perform particularly well. Current inflation and persistence-

weighted core inflation, PWcoret

,π , are relatively good predictors of inflation 6 months ahead, but

their predictive ability worsens substantially over longer time periods.

Figure 3. Predictive ability of core inflation measures, 1995–2005

02468

1012141618

persistenceweighted

pers. exp.weighted

net median net adjusted current

6 months 12 months 18 months

Note: The mean square error is plotted on the vertical axis. 52 The results based on other persistence measures (the PP, KPPS and LLS test-based measures for the full and restricted samples) are similar and available upon request.


3.5 Conclusions

In this paper, we have presented evidence on disaggregate inflation persistence in the Czech

Republic, exploring data from 412 individual narrowly defined products and 9 broader sectors

from 1995:M1 to 2005:M12. The results suggest that inflation persistence decreased after the

adoption of inflation targeting. A somewhat similar observation of falling rather than rising

inflation persistence in the euro area countries over the past decade is reported by the

Eurosystem Inflation Persistence Network (IPN).53 However, inflation persistence in the Czech

Republic still remains relatively high compared to that in the euro area countries.

The results unambiguously point to the presence of aggregation bias, that is, aggregate inflation is

more persistent than the mean of its underlying disaggregated components. This result is robust

to the choice of disaggregation level (412 components or 9 sectors) and weighting scheme

(simple mean, median, or weighted mean), to the choice of estimation technique (unit root ADF,

PP, LLS, or stationarity KPSS tests), and to the choice of period (full sample versus post-1998

inflation targeting period).

We identify that the sectoral structure explains the estimated variation in inflation persistence to a

certain extent. In particular, products belonging to the raw goods category exhibit smaller than

sample average persistence, while durables have higher than average persistence. Concerning the

“services inflation persistence puzzle”, there is evidence that (labor-intensive) services are

characterized by smaller persistence than goods for our 1995–2005 sample. However, the results

are sensitive to the choice of estimation technique and period, i.e. using a shorter sample over

53 A summary of the IPN’s findings is provided by Altissimo, Ehrmann and Smets (2006).


1998–2005 we do not find robust differences in terms of the persistence of goods and services.

Nevertheless, the regression results show that the services dummy is negatively associated with

inflation persistence.

We find that competition is not conducive to reducing inflation persistence. Price dispersion, as a

proxy for the degree of competition, is negatively related to inflation persistence. This finding

confirms the results of Calvo (2000), who shows that as the level of competition increases, the

firm’s pricing strategy is influenced more by the average pricing strategy in the market. The costs

of charging a different price for identical products increase with higher competition. As a result,

there can be a more inertial response to shocks in a more competitive environment.

Lastly, we construct a persistence-weighted core inflation measure and evaluate its predictive

ability by comparison with other available measures of core inflation over the period 1995–2005.

Generally, we find that adjusted inflation (headline inflation excluding regulated prices, fuel and

food prices and changes in indirect taxes) is the best predictor of future inflation trends in our set

of core inflation measures over the horizons of 6, 12 and 18 months. Our proposed measure –

persistence expenditure-weighted core inflation – may be viewed as an equally good predictor as

adjusted inflation for the 6-month horizon, but its predictive ability worsens over longer time

periods.


References

Altissimo, F., Ehrmann, M. and F. Smets, 2006. Inflation Persistence and Price Setting Behaviour in the Euro Area – A Summary of the IPN Evidence. European Central Bank Occasional Paper, No. 46. Altissimo, F., Mojon, B. and P. Zaffaroni, 2007. Fast Micro and Slow Macro: Can Aggregation Explain the Persistence of Inflation? European Central Bank Working Paper, No. 729. Andrews, D., and W. K. Chen, 1994. Approximately Median-Unbiased Estimation of Autoregressive Models. Journal of Business and Economic Statistics, 12, pp. 187–204. Angeloni, I., Aucremanne, L., Ehrmann, M., Galí, J., Levin, A. and F. Smets, 2006. New Evidence on Inflation Persistence and Price Stickiness in the Euro Area: Implications for Macro Modeling, Journal of the European Economic Association, 4, pp. 562–574. Aron, J. and J. Muellbauer, 2006. A Framework for Forecasting the Components of the Consumer Price Index: Application to South Africa, paper presented at the 21st Annual Congress of the European Economic Association, August 25, 2006. Baye, M. R., Morgan, J. and P. Scholten (2004): Price Dispersion in the Small and in the Large: Evidence from an Internet Price Comparison Site, Journal of Industrial Economics, 52, pp. 463–496. Caglayan, M., Filiztekin, A. and M.T. Rauh (2007): Inflation, Price Dispersion and Market Structure, European Economic Review, forthcoming. Calvo, G., 2000. Notes on Price Stickiness: With Special Reference to Liability Dollarization and Credibility. University of Maryland, mimeo, available at http://www.bsos.umd.edu/econ/ciecrp.htm. Calvo, G., 1983. Staggered Prices in a Utility Maximizing Framework, Journal of Monetary Economics, 12, pp. 383–398. Calvo, G., Celasun, O. and M. Kumhof, 2002. A Theory of Rational Inflationary Inertia, in: P. Aghion, R. Frydman, J. Stiglitz and M. Woodford (eds.), Knowledge, Information and Expectations in Modern Macroeconomics: In Honor of Edmund S. Phelps. Princeton: Princeton University Press. Cecchetti, S. and G. Debelle, 2006. Has the Inflation Process Changed? Third BIS Annual Conference, Economic Policy, pp. 311–352. Clark, T., 2006. Disaggregate Evidence on the Persistence in Consumer Price Inflation, Journal of Applied Econometrics, 21, pp. 563–587. Coricelli, F. and R. Horváth, 2006. Price Setting Behaviour: Micro Evidence on Slovakia, Centre for Economic Policy Research, CEPR Discussion Paper No. 5445. Cournede, B., Janovskaia, A. and P. van den Noord, 2005. Sources of Inflation Persistence in the Euro Area, OECD Economics Department Working Papers, No. 435.


Crucini, M., Telmer, C. I. and M. Zachariadis, 2005. Understanding European Real Exchange Rates, American Economic Review, pp. 724–738. Cutler, J., 2001. Core Inflation in the UK, External MPC Unit Discussion Paper No. 3, Bank of England. Dias, D. and C. Marquez, 2005. Using Mean Reversion as a Measure of Persistence, European Central Bank Working Paper, No. 450. Dickey, D. A. and W. A. Fuller, 1981. Likelihood Ratio Statistics for Autoregressive Time Series with a Unit Root, Econometrica, 49, pp. 1057–1072. Erceg, C. and A. Levin, 2003. Imperfect Credibility and Inflation Persistence, Journal of Monetary Economics, 50(4), May, pp. 915–944. European Commission, 2003. The EU Economy: 2003 Review, Brussels. Fabiani, S., Druant, M., Hernando, I., Kwapil, C., Landau, B., Loupias, C., Martins, F., Matha, T., Sabbatini, R. and A. Stockman, 2006. What Firms’ Surveys Tell Us about Price-Setting Behavior in the Euro Area, International Journal of Central Banking, 2(3), pp. 3–47. Franta, M., Saxa, B. and K. Šmídková, 2007. Inflation Persistence: Euro Area and New Member States, European Central Bank, Working Paper, No. 810. Fuhrer, J. C. and G. Moore, 1995. Inflation Persistence, Quarterly Journal of Economics, February, pp. 127–159. Gadea, M. D. and L. Mayoral, 2006. The Persistence of Inflation in OECD Countries: A Fractionally Integrated Approach, International Journal of Central Banking, 2(1), pp. 52–103. Galí, J. and M. Gertler, 1999. Inflation Dynamics: A Structural Econometric Analysis, Journal of Monetary Economics, 44(2), pp. 195–222. Gerardi, K. and A.H. Shapiro (2007): The Effects of Competition on Price Dispersion in the Airline Industry: A Panel Analysis, Federal Reserve Bank of Boston Working Paper, No. 07/2007. Granger, C., 1980. Long Memory Models and Aggregation of Dynamic Models, Journal of Econometrics, pp. 227–238. Jonas, J. and F. Mishkin, 2003. Inflation Targeting in Transition Countries: Experience and Prospects, National Bureau for Economic Research, WP No. 9667. Konieczny, J. and A. Skrzypacz, 2005. The Behaviour of Price Dispersion in a Natural Experiment, Journal of Monetary Economics, pp. 621–632. Kotlán, V. and D. Navrátil, 2003. Inflation Targeting as a Stabilization Tool: Its Design and Performance in the Czech Republic. Finance a uver – Czech Journal of Economics and Finance, 53, pp. 220–242.


Kwiatkowski, D., Phillips, P. C. B., Schmidt P. and Y. Shin, 1992. Testing the Null of Stationarity against the Alternative of Unit Root, Journal of Econometrics, 54, pp. 159–178. Lanne, M., Lütkepohl, H. and P. Saikkonen, 2002. Comparison of Unit Root Tests for Time Series with Level Shifts, Journal of Time Series Analysis, 23, pp. 667–685. Leiter, D.B. and T. Warin (2007): An Empirical Study of Price Dispersion in Homogenous Goods Markets, Middlebury College Economics Discussion Paper, No. 10/2007. Levin, A., Natalucci, F. and J. Piger, 2004. Explicit Inflation Objectives and Macroeconomic Outcomes, European Central Bank Working Paper, No. 383. Levin, A. and J. Piger, 2004. Is Inflation Persistence Intrinsic in Industrial Countries? The Federal Reserve Bank of Saint Louis, Working Paper, No. 023E. Lunnemann, P. and T. Matha, 2005. Regulated and Services’ Prices and Inflation Persistence, European Central Bank Working Paper, No. 466. Mankiw, N. and R. Reis, 2002. Sticky Information Versus Sticky Prices: A Proposal to Replace the New Keynesian Phillips Curve, Quarterly Journal of Economics, 117(4), pp. 1295–1328. Marquez, C., 2004. Inflation Persistence: Facts or Artefacts? European Central Bank Working Paper, No. 371. Nimark, K., 2005. Calvo Pricing and Imperfect Common Knowledge – A Forward Looking Model of Rational Inflation Inertia, European Central Bank Working Paper, No. 474.

O’Reilly, G. and K. Whelan, 2005. Has Euro-Area Inflation Persistence Changed Over Time? Review of Economics and Statistics, 87(4), pp. 807–820. Orphanides, A. and J. Williams, 2003. Imperfect Knowledge, Inflation Expectations, and Monetary Policy, in: M. Woodford (ed.), Inflation Targeting, Chicago: University of Chicago Press. Phillips, P. C. B. and P. Perron, 1988. Testing for a Unit Root in Time Series Regression, Biometrika, 75, pp. 335–346. Ratfai, A., 2006. Linking Individual and Aggregate Price Changes, Journal of Money, Credit and Banking, 38(8), pp. 2199–2224.

Saikkonen, P. and H. Lütkepohl, 2002. Testing for a Unit Root in a Time Series with a Level Shift at Unknown Time, Econometric Theory, 9, pp. 343–362. Sargent, T., 1999. The Conquest of American Inflation, Princeton, Princeton University Press. Smith, J. K., 2005. Inflation Targeting and Core Inflation, Canadian Journal of Economics 38(3), pp. 1018–1036. Smith, J. K., 2004. Better Measures of Core Inflation? Lafayette College Working Paper Series (September).


Taylor, J., 1980. Aggregate Dynamics and Staggered Contracts, Journal of Political Economy, 88, pp. 1–23. Vega, M. and D. Winkelried, 2005. Inflation Targeting and Inflation Behavior: A Successful Story? International Journal of Central Banking, pp. 153–175.

Yigit, T., 2007. Inflation Targeting: An Indirect Approach to Assess the Direct Impact, Bilkent University Working Paper, No. 6/2007. Zaffaroni, P., 2004. Contemporaneous Aggregation of Linear Dynamic Models in Large Economies, Journal of Econometrics, 120, pp. 75–102.


APPENDIX Figure A.1 – Link between ADF, PP, KPSS and LLS tests (based on 412 product groups)

ADF vs PP, 1995-2005

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0.00 0.20 0.40 0.60 0.80 1.0

ADF vs PP, 1998-2005

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0.00 0.20 0.40 0.60 0.80 1.00

LLS vs. ADF, 1995-2005

-8

-7

-6

-5

-4

-3

-2

-1

0

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.

ADF vs KPSS, 1995-2005

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

0.00 0.20 0.40 0.60 0.80 1.0

ADF vs KPSS, 1998-2005

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

0.00 0.20 0.40 0.60 0.80 1.0

LLS vs. PP, 1995-2005

-8

-7

-6

-5

-4

-3

-2

-1

0

0.00 0.20 0.40 0.60 0.80 1.0

PP vs KPSS, 1995-2005

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

0.00 0.20 0.40 0.60 0.80 1.0

PP vs KPSS, 1998-2005

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

0.00 0.20 0.40 0.60 0.80 1.0

LLS vs. KPSS, 1995-2005

-8

-7

-6

-5

-4

-3

-2

-1

0

0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.

1995-2005corr(adf,pp) 0.94corr(adf,kpss) 0.63corr(pp,kpss) 0.67

1998-2005corr(adf,pp) 0.87corr(adf,kpss) 0.31corr(pp,kpss) 0.31

1995-2005corr(lls,adf) 0.76corr(lls,pp) 0.76corr(lls,kpss) 0.50

Notes: For the ADF and PP tests, the probability of rejecting the null hypothesis of a unit root is employed. The probability can vary from 0 to 1. Higher values correspond to more persistence. For example, a probability higher than 0.10 means that the null of a unit root cannot be rejected at the 10% significance level. For the KPSS stationarity test, the t-statistic is used (shown on the vertical axes). Higher t-statistic values increase the probability of rejecting the null hypothesis of stationarity and hence characterize more persistence in the underlying series. LLS test stands for the Lanne et al. (2002) unit root test with a structural break; the t-statistic is used in the Figure.


Figure A.2 – Distribution of inflation persistence across 412 products and aggregation

bias; Results from Lanne et al. (2002) unit root test with structural break

Notes: The vertical bold line denotes the persistence of aggregate CPI inflation; the simple vertical line represents the mean of the disaggregate inflation persistence. The horizontal axis characterizes the level of inflation persistence (more negative values mean more persistence). Thus, the results are indicative of aggregation bias.


Table A.1 – Inflation persistence, Share of unit root processes

1995–2005 10% significance level Share of unit roots Share of unit roots (weighted)

no_prod sample_w ADF PP KPPS LLS ADF_w PP_w KPPS_w LLS_w Tradables 311 0.59 0.66 0.72 0.73 0.63 0.62 0.71 0.67 0.63 Non_tradables 101 0.41 0.68 0.67 0.76 0.65 0.83 0.79 0.90 0.76 Services 96 0.40 0.70 0.66 0.77 0.67 0.83 0.79 0.90 0.76 Non_regul_serv 74 0.30 0.70 0.66 0.77 0.68 0.81 0.74 0.90 0.72 Regulated 27 0.11 0.63 0.70 0.74 0.59 0.89 0.94 0.91 0.85 Durables 164 0.21 0.85 0.86 0.90 0.82 0.89 0.89 0.80 0.86 Non_durables 152 0.39 0.45 0.57 0.55 0.42 0.47 0.61 0.60 0.51 Raw_goods 42 0.11 0.21 0.29 0.26 0.21 0.29 0.31 0.13 0.33 Processed 370 0.89 0.72 0.75 0.79 0.68 0.75 0.79 0.84 0.73 Total_prod_level 412 1.00 0.67 0.71 0.74 0.64 0.70 0.74 0.76 0.68

5% significance level Share of unit roots Share of unit roots (weighted) no_prod sample_w ADF PP KPPS LLS ADF_w PP_w KPPS_w LLS_w Tradables 311 0.59 0.76 0.82 0.67 0.73 0.76 0.84 0.62 0.75 Non_tradables 101 0.41 0.76 0.80 0.60 0.76 0.86 0.88 0.54 0.82 Services 96 0.40 0.77 0.79 0.60 0.77 0.87 0.88 0.54 0.82 Non_regul_serv 74 0.30 0.74 0.78 0.61 0.76 0.82 0.85 0.59 0.76 Regulated 27 0.11 0.81 0.85 0.59 0.78 0.97 0.98 0.40 0.97 Durables 164 0.21 0.90 0.92 0.87 0.86 0.92 0.94 0.78 0.89 Non_durables 152 0.39 0.59 0.72 0.45 0.59 0.67 0.78 0.53 0.67 Raw_goods 42 0.11 0.36 0.50 0.14 0.38 0.43 0.62 0.09 0.44 Processed 370 0.89 0.80 0.85 0.71 0.78 0.85 0.88 0.65 0.82 Total_prod_level 412 1.00 0.76 0.82 0.65 0.74 0.80 0.86 0.59 0.78

1998–2005 10% significance level Share of unit roots Share of unit roots (weighted)

no_prod sample_w ADF PP KPPS ADF_w PP_w KPPS_w Tradables 311 0.59 0.59 0.70 0.60 0.62 0.73 0.55 Non_tradables 101 0.41 0.69 0.77 0.53 0.82 0.84 0.72 Services 96 0.40 0.73 0.78 0.55 0.83 0.84 0.73 Non_regul_serv 74 0.30 0.82 0.88 0.54 0.94 0.95 0.68 Regulated 27 0.11 0.33 0.48 0.52 0.52 0.56 0.84 Durables 164 0.21 0.63 0.66 0.81 0.62 0.64 0.73 Non_durables 152 0.39 0.54 0.73 0.36 0.61 0.79 0.45 Raw_goods 42 0.11 0.38 0.57 0.07 0.45 0.70 0.02 Processed 370 0.89 0.65 0.73 0.64 0.73 0.79 0.69 Total_prod_level 412 1.00 0.62 0.72 0.58 0.70 0.78 0.62

5% significance level Share of unit roots Share of unit roots (weighted) no_prod sample_w ADF PP KPPS ADF_w PP_w KPPS_w Tradables 311 0.59 0.70 0.77 0.52 0.73 0.82 0.47 Non_tradables 101 0.41 0.77 0.79 0.44 0.84 0.85 0.70 Services 96 0.40 0.79 0.80 0.46 0.84 0.85 0.70 Non_regul_serv 74 0.30 0.91 0.91 0.46 0.96 0.96 0.66 Regulated 27 0.11 0.41 0.48 0.37 0.53 0.56 0.80 Durables 164 0.21 0.71 0.74 0.75 0.70 0.74 0.69 Non_durables 152 0.39 0.68 0.80 0.26 0.74 0.86 0.35 Raw_goods 42 0.11 0.52 0.67 0.05 0.55 0.75 0.01 Processed 370 0.89 0.74 0.79 0.55 0.80 0.84 0.63 Total_prod_level 412 1.00 0.72 0.78 0.50 0.77 0.83 0.56


Table A.2 – Determinants of inflation persistence, ADF test

1995–2005 1998–2005 ADF ADF ADF ADF ADF ADF Price dispersion -0.73*** -6.66*** -1.63*** -0.17* -2.08** -0.58*** (0.14) (2.49) (0.46) (0.1) (1.04) (0.16) Non-durables -0.10** -0.05* (0.05) (0.02) Raw goods -0.16*** -0.002 (0.04) (0.04) Adj. R-squared 0.07 --- --- 0.01 --- --- Estimation method OLS GMM GMM OLS GMM GMM Sargan test (p-value) --- 1.8 (0.19) 1.5 (0.22) --- 0.1 (0.7) 5.2 (0.02) Observations 412 412 412 412 412 412

Note: Heteroscedasticity robust standard errors are shown in parentheses. ***, **, and * denote significance at 1%, 5%, and 10%, respectively. P-value in brackets for the Sargan (overidentifying restrictions) test. The list of instruments for price dispersion is as follows: non-regulated services, non-durables, raw goods and regulated prices dummies.

Table A.3 – Determinants of inflation persistence, PP test

1995–2005 1998–2005 PP PP PP PP PP PP Price dispersion -0.73*** -5.66*** -1.51*** -0.17* -0.87*** -0.49*** (0.14) (2.13) (0.42) (0.1) (0.30) (0.18) Non-durables -0.08* -0.01 (0.04) (0.02) Raw goods -0.16*** -0.04 (0.04) (0.03) Adj. R-squared 0.08 --- --- 0.01 --- --- Estimation method OLS GMM GMM OLS GMM GMM Sargan test (p-value) --- 1.9 (0.17) 5.5 (0.02) --- 0.4 (0.82) 5.5 (0.02) Observations 412 412 412 412 412 412

Note: Heteroscedasticity robust standard errors are shown in parentheses. ***, **, and * denote significance at 1%, 5%, and 10%, respectively. P-value in brackets for the Sargan (overidentifying restrictions) test. The list of instruments for price dispersion is as follows: non-regulated services, non-durables, raw goods and regulated prices dummies.


Table A.4 – Determinants of inflation persistence, LLS test

1995–2005 LLS LLS LLS Price dispersion -1.99*** -27.7*** -2.69** (0.49) (11.1) (1.24) Non-durables -0.57*** (0.15) Raw goods -0.84*** (0.21) Adj. R-squared 0.03 --- --- Estimation method OLS GMM GMM Sargan test (p-value) --- 2.5 (0.11) 0.6 (0.46)Observations 412 412 412

Note: Heteroscedasticity robust standard errors are shown in parentheses. ***, **, and * denote significance at 1%, 5%, and 10%, respectively. P-value in brackets for the Sargan (overidentifying restrictions) test. The LLS test is a unit root test with a structural break on an unknown date. The test was carried out only for the full sample, 1995–2005; see the main text for explanations. The list of instruments for price dispersion is as follows: non-regulated services, non-durables, raw goods and regulated prices dummies.

Table A.5 – Determinants of inflation persistence, Logit estimates

1995–2005 1998–2005 ADF PP KPSS LLS ADF PP KPSS Price dispersion -3.68*** -2.90** -2.59** -3.45*** -1.37 -0.97 -3.11*** (1.16) (1.15) (1.17) (1.12) (1.11) (1.16) (1.06) Non-durables -1.13*** -0.49* -1.06*** -1.07*** -0.26 0.41 -1.04*** (0.24) (0.25) (0.26) (0.24) (0.23) (0.26) (0.24) Raw goods -1.64*** -1.74*** -1.82*** -1.48*** -0.91*** -0.95*** -2.61*** (0.41) (0.37) (0.42) (0.40) (0.57) (0.37) (0.63) Pseudo R-squared 0.15 0.10 0.15 0.13 0.03 0.06 0.16 Estimation method Logit Logit Logit Logit Logit Logit Logit Observations 412 412 412 412 412 412 412

Note: Heteroscedasticity robust standard errors are shown in parentheses. ***, **, and * denote significance at 1%, 5%, and 10%, respectively. The LLS test is a unit root test with a structural break on an unknown date. The test was carried out only for the full sample, 1995–2005; see the main text for explanations.


Table A.6 – Determinants of inflation persistence, Product characteristics

1995–2005 1998–2005 ADF PP KPSS LLS ADF PP KPSS Non-durables -0.24*** -0.20*** -0.34*** -0.79*** -0.08*** -0.04** -0.32*** (0.03) (0.03) (0.03) (0.11) (0.02) (0.02) (0.03) Raw goods -0.16*** -0.16*** -0.34*** -0.86*** -0.01 -0.02 -0.24*** (0.02) (0.02) (0.05) (0.16) (0.03) (0.03) (0.04) Services – nonregulated -0.21*** -0.20*** -0.30*** -0.45*** 0.02 0.01 -0.21*** (0.03) (0.03) (0.04) (0.14) (0.03) (0.03) (0.04) Regulated -0.17*** 0.17*** -0.33*** -0.38*** -0.011*** -0.11*** -0.25*** (0.04) (0.03) (0.06) (0.13) (0.03) (0.03) (0.05) Adj. R-squared 0.24 0.22 0.33 0.21 0.06 0.04 0.29 Estimation method OLS OLS OLS OLS OLS OLS OLS Observations 412 412 412 412 412 412 412

Note: Heteroscedasticity robust standard errors are shown in parentheses. ***, **, and * denote significance at 1%, 5%, and 10%, respectively.

Table A.7 – Detailed product-specific results

Products Units ADF95 ADF98 KP95 KP98 LLS PP95 PP98 WeightsBread, white 1 kg 0.07 0.36 0.41 0.19 -2.08 0.24 0.20 113.43 Bread, whole meal 1 kg 0.30 0.30 0.17 0.40 -2.36 0.17 0.26 94.57 Baguettes (white) 1 kg 0.01 0.33 0.31 0.11 -2.31 0.16 0.13 14.48 Pastry, cake 1 kg 0.19 0.08 0.49 0.09 -2.49 0.28 0.22 19.72 Puff pastry 1 kg 0.40 0.13 0.62 0.07 -1.70 0.39 0.12 5.78 Sponge cake 1 kg 0.47 0.13 0.99 0.38 -1.62 0.41 0.12 6.96 Biscuit dry 1 kg 0.16 0.41 0.49 0.65 -2.38 0.33 0.35 20.20 Biscuit with filling 1 kg 0.00 0.00 0.78 0.33 -3.42 0.00 0.00 20.20 Waffles 1 kg 0.03 0.01 0.79 0.70 -3.42 0.02 0.00 20.20 Wheat flour (impalpable powder)

1 kg 0.11 0.01 0.08 0.19 -2.35 0.15 0.34 7.91

Wheat flour 1 kg 0.13 0.38 0.09 0.22 -2.51 0.24 0.46 13.79 Spaghetti, without eggs 1 kg 0.11 0.46 0.19 0.56 -1.84 0.25 0.28 4.55 Pasta, with eggs 1 kg 0.28 0.39 0.18 0.55 -2.02 0.33 0.35 11.93 Bread dumpling, powder 1 kg 0.01 0.01 0.26 0.09 -3.82 0.10 0.13 8.47 Pudding (powder) 10 pcs 0.46 0.00 0.86 0.23 -2.02 0.34 0.00 5.24 Rice, long-grain 1 kg 0.06 0.32 0.16 0.45 -2.88 0.10 0.26 13.25 Center loin roast 1 kg 0.08 0.27 0.17 0.08 -2.47 0.08 0.32 38.69 Boneless sirloin roast 1 kg 0.04 0.26 0.18 0.08 -2.86 0.04 0.31 38.39 Sirloin chop 1 kg 0.04 0.23 0.21 0.08 -2.74 0.06 0.31 24.34 Boneless blade roast 1 kg 0.07 0.30 0.27 0.08 -2.70 0.06 0.31 24.97 Belly-pork 1 kg 0.07 0.25 0.09 0.09 -2.71 0.09 0.30 21.86 Boneless rump roast 1 kg 0.00 0.00 0.49 0.13 -6.01 0.00 0.07 36.27 Boneless shoulder pot-roast 1 kg 0.00 0.05 0.58 0.14 -5.17 0.00 0.13 22.12 Fore shank 1 kg 0.00 0.08 0.37 0.13 -4.90 0.00 0.14 10.68 Minced meat 1 kg 0.00 0.27 0.34 0.08 -3.79 0.01 0.33 14.67 Liver, Pork 1 kg 0.00 0.16 0.13 0.09 -3.53 0.01 0.30 14.49 Rabbit 1 kg 0.13 0.42 0.73 0.36 -2.61 0.26 0.26 5.03 Veal leg 1 kg 0.02 0.19 0.82 0.52 -4.62 0.01 0.04 2.52


Products Units ADF95 ADF98 KP95 KP98 LLS PP95 PP98 WeightsSmall sausage 1 kg 0.04 0.29 0.14 0.09 -3.04 0.05 0.30 26.16 Sausage 1 kg 0.06 0.06 0.16 0.11 -2.71 0.06 0.25 26.16 Salami, Gothaj 1 kg 0.05 0.13 0.38 0.13 -2.46 0.06 0.27 26.16 Ring of Lyoner sausage 1 kg 0.06 0.17 0.15 0.13 -2.79 0.06 0.28 26.16 Salami (ham) 1 kg 0.01 0.00 0.31 0.13 -3.45 0.02 0.36 26.16 Sausage (pepper) 1 kg 0.01 0.02 0.54 0.08 -3.90 0.05 0.27 24.37 Salami, Polican 1 kg 0.09 0.01 0.24 0.14 -2.62 0.10 0.41 24.37 Ham (pork) 1 kg 0.01 0.21 0.19 0.11 -3.06 0.01 0.31 15.44 Sliced bacon 1 kg 0.01 0.11 0.68 0.17 -3.37 0.03 0.31 14.95 Liver pâté 1 kg 0.00 0.34 0.53 0.10 -4.28 0.02 0.25 9.62 Sausage (pork) 1 kg 0.09 0.25 0.59 0.08 -3.44 0.08 0.21 9.16 Sausage (poultry) 1 kg 0.08 0.06 0.56 0.11 -3.01 0.06 0.28 17.87 Luncheon meat 1 kg 0.21 0.32 0.52 0.13 -2.59 0.19 0.32 21.28 Beef (canned meat) 1 kg 0.00 0.18 0.52 0.10 -4.87 0.00 0.37 15.69 Chicken 1 kg 0.01 0.07 0.06 0.06 -3.16 0.09 0.25 67.93 Duck, without heart, liver and gizzard

1 kg 0.03 0.06 0.38 0.17 -3.34 0.11 0.13 7.47

Carp chilled, frozen 1 kg 0.72 0.27 0.77 0.73 -1.16 0.58 0.26 22.72 Salted herring 125 g 0.06 0.28 0.17 0.57 -3.06 0.09 0.28 19.53 Fresh chicken eggs 10 pcs 0.05 0.00 0.08 0.07 -3.34 0.06 0.16 47.16 Milk pasteurized (fat content 1.5%)

1 l 0.18 0.11 0.61 0.12 -2.00 0.28 0.26 22.25

Milk, long life (fat content 1.5%)

1 l 0.04 0.05 0.16 0.06 -3.12 0.07 0.14 66.77

Milk condensed, not sweetened 500 g 0.69 0.30 0.67 0.23 -1.53 0.45 0.20 5.46 Powdered milk, for babies 400 g 0.05 0.04 0.81 0.54 -2.41 0.19 0.15 9.08 Mellow cheese (Ermine) 1 kg 0.59 0.05 1.06 0.55 -1.63 0.55 0.06 12.21 Processed cheese (not flavored) 1 kg 0.71 0.18 0.97 0.25 -2.29 0.63 0.24 39.56 Cottage cheese (LUCINA) 1 kg 0.28 0.08 0.90 0.48 -1.57 0.32 0.18 5.33 Fermented milk products, liquid 1 l 0.56 0.12 0.86 0.14 -2.42 0.46 0.11 10.58 Cream, sweet 1 l 0.43 0.22 0.79 0.20 -1.93 0.38 0.20 28.88 Natural yoghurt, fat content low 150 g 0.62 0.31 0.66 0.18 -1.87 0.54 0.24 44.65 Fruit yoghurt 150 g 0.24 0.08 0.94 0.18 -2.12 0.23 0.08 66.97 Ice-cream 1 l 0.26 0.12 0.25 0.11 -2.19 0.17 0.40 24.86 Butter, unsalted 1 kg 0.10 0.29 0.17 0.32 -2.21 0.18 0.37 39.91 Pure lard 1 kg 0.00 0.27 0.08 0.09 -3.45 0.00 0.29 6.44 Olive oil 1 l 0.16 0.57 0.41 0.33 -2.36 0.30 0.58 2.19 Sunflower oil 1 l 0.07 0.07 0.69 0.53 -2.38 0.07 0.35 16.28 Margarine, type Hera 1 kg 0.08 0.12 0.59 0.29 -1.96 0.13 0.17 14.67 Margarine, type Planta 1 kg 0.27 0.59 0.94 0.80 -2.53 0.23 0.54 5.70 Fresh apples 1 kg 0.01 0.01 0.19 0.08 -3.10 0.05 0.10 32.28 Fresh peaches, nectarines 1 kg 0.00 0.00 0.17 0.25 -3.51 0.00 0.00 18.70 Fresh grapes 1 kg 0.00 0.00 0.15 0.08 -3.71 0.01 0.00 21.22 Fresh water melon 1 kg 0.02 0.06 0.10 0.16 -3.33 0.00 0.00 11.09 Fresh oranges 1 kg 0.00 0.00 0.22 0.13 -3.98 0.02 0.04 41.74 Fresh lemons 1 kg 0.00 0.02 0.20 0.15 -3.73 0.01 0.01 11.08 Fresh bananas 1 kg 0.00 0.04 0.20 0.24 -3.54 0.00 0.00 41.40 Fresh kiwis 1 kg 0.00 0.00 0.08 0.14 -3.16 0.02 0.08 6.21 Dried raisins 1 kg 0.09 0.07 0.11 0.29 -2.63 0.21 0.23 6.72 Potatoes 1 kg 0.00 0.02 0.22 0.11 -6.69 0.01 0.01 30.29 Frozen chipped potatoes 1 kg 0.00 0.00 0.09 0.28 -4.15 0.15 0.03 12.75


Products Units ADF95 ADF98 KP95 KP98 LLS PP95 PP98 WeightsPotato dumpling (powder) 1 kg 0.00 0.36 0.77 0.39 -2.98 0.46 0.32 3.00 Fresh white cabbage 1 kg 0.00 0.00 0.08 0.08 -3.75 0.05 0.08 8.28 Fresh cucumber 1 kg 0.00 0.00 0.17 0.11 -4.74 0.00 0.00 15.18 Fresh green peppers 1 kg 0.00 0.00 0.14 0.14 -3.93 0.00 0.00 22.46 Fresh tomatoes 1 kg 0.00 0.00 0.05 0.06 -5.35 0.00 0.00 23.74 Fresh cauliflower 1 kg 0.00 0.00 0.16 0.17 -4.34 0.00 0.00 10.04 Fresh carrots 1 kg 0.00 0.00 0.09 0.08 -4.53 0.01 0.02 5.42 Fresh celeriac 1 kg 0.00 0.00 0.12 0.17 -3.99 0.03 0.12 7.62 Fresh cultivated mushrooms 1 kg 0.33 0.10 0.41 0.14 -1.99 0.30 0.08 3.90 Garlic (dry) 1 kg 0.10 0.12 0.49 0.15 -2.29 0.10 0.28 10.32 Cabbage, jar 1 kg 0.03 0.15 0.20 0.47 -2.98 0.02 0.23 6.07 Pickled gherkins 1 kg 0.00 0.06 0.36 0.09 -4.65 0.00 0.26 4.97 Dried lentils 1 kg 0.10 0.27 0.14 0.19 -2.71 0.16 0.35 4.20 Jam, strawberry 1 kg 0.00 0.37 0.48 0.77 -6.16 0.07 0.41 0.40 Granulated sugar 1 kg 0.03 0.04 0.08 0.11 -3.04 0.17 0.15 31.13 Confectioner’s sugar 1 kg 0.11 0.20 0.13 0.15 -3.09 0.19 0.27 8.68 Chocolate, milk 100 g 0.02 0.13 0.97 0.57 -2.18 0.03 0.17 34.97 Chocolate dessert 250 g 0.24 0.13 1.04 0.41 -2.27 0.24 0.17 32.68 Chocolate bar 100 g 0.15 0.04 0.13 0.10 -2.52 0.00 0.02 13.23 Fruit drops 100 g 0.29 0.05 1.16 0.73 -1.86 0.35 0.07 10.18 Chewing gum 1 pack 0.05 0.53 0.30 0.26 -2.78 0.04 0.30 10.39 Cake from egg yolk 10 pcs 0.54 0.32 0.91 0.17 -2.55 0.55 0.28 16.81 Sherbet 1 l 0.35 0.24 0.20 0.68 -1.81 0.29 0.24 10.92 Honey 1 kg 0.29 0.00 0.33 0.12 -2.01 0.39 0.00 0.97 Meat extract 100 g 0.05 0.22 0.43 0.67 -3.07 0.06 0.34 15.35 Delicious salad 1 kg 0.00 0.07 0.64 0.23 -3.70 0.00 0.11 9.31 Table salt 1 kg 0.59 0.14 0.70 0.29 -1.70 0.49 0.16 13.29 Black pepper 100 g 0.03 0.03 0.47 0.36 -3.02 0.17 0.23 9.39 Tomato ketchup 1 kg 0.20 0.13 0.14 0.11 -3.13 0.11 0.13 11.48 Mustard 1 kg 0.37 0.37 0.22 0.32 -1.89 0.43 0.52 13.05 Yeast 1 kg 0.27 0.00 0.35 0.30 -1.94 0.34 0.00 10.78 Roust coffee beans 100 g 0.00 0.57 0.29 0.18 -6.60 0.00 0.38 20.56 Instant coffee 100 g 0.00 0.00 0.70 0.55 -5.40 0.00 0.22 22.11 Black tea bags 100 g 0.08 0.00 0.49 0.71 -2.77 0.24 0.01 19.39 Green tea bags 100 g 0.17 0.00 0.37 0.54 -1.93 0.25 0.00 9.44 Coffee substitutes 100 g 0.21 0.01 0.58 0.36 -1.74 0.44 0.00 8.85 Fruity syrup 1 kg 0.07 0.15 0.19 0.85 -1.98 0.39 0.15 24.41 Orange juice 1 l 0.14 0.06 0.22 0.18 -2.08 0.15 0.06 27.11 Spring water 1 l 0.05 0.19 0.28 0.15 -2.80 0.01 0.07 43.21 Mineral water (fizzy) 1 l 0.10 0.03 0.87 0.21 -2.66 0.09 0.00 59.40 Coca-cola (Pepsi-cola) 1 l 0.00 0.00 0.08 0.15 -4.04 0.03 0.00 11.80 Rum (domestic production) 1 l 0.20 0.29 0.07 0.24 -3.00 0.06 0.10 39.33 Vodka (fine) 1 l 0.05 0.12 0.41 0.45 -3.50 0.07 0.19 31.05 Fernet stock (liqueur) 1 l 0.54 0.54 0.14 0.24 -1.64 0.20 0.30 70.74 Becher’s (Carlsbad) liqueur 1 l 0.15 0.13 0.28 0.28 -2.32 0.25 0.19 42.80 Scotch whisky 1 l 0.61 0.32 0.67 0.85 -1.25 0.44 0.32 13.89 White wine (high quality) 1 l 0.61 0.00 0.98 0.41 -1.99 0.61 0.00 74.11 Red wine (high quality) 1 l 0.78 0.06 0.91 0.43 -1.51 0.68 0.23 69.88 Sparkling wine semi-dry 0,75 l 0.03 0.01 0.53 0.06 -3.09 0.16 0.07 38.17 Italian vermouth 1 l 0.30 0.09 0.96 0.38 -2.31 0.32 0.11 11.76 Bottled light beer 0,5 l 0.27 0.14 0.86 0.67 -1.72 0.36 0.19 232.41


Products Units ADF95 ADF98 KP95 KP98 LLS PP95 PP98 WeightsBottled light lager 0,5 l 0.08 0.15 0.80 0.82 -3.07 0.23 0.29 57.37

PETRA (filter tipped cigarettes) 1

package 0.10 0.35 0.82 0.56 -2.50 0.14 0.44 221.14 SPARTA LIGHT (filter tipped cigarettes)

1 package 0.11 0.27 0.71 0.56 -2.62 0.17 0.40 147.43

START (filter tipped cigarettes) 1

package 0.10 0.22 1.01 0.77 -2.48 0.16 0.40 73.71 MARLBORO (filter tipped cigarettes)

1 package 0.01 0.09 0.76 0.50 -3.45 0.03 0.25 81.47

Clothing materials for business suit (wool)

1 m2 0.46 0.48 0.74 0.56 -2.09 0.43 0.49 5.11

Briefs (for men) 1 pc 0.66 0.09 0.81 0.96 -1.20 0.40 0.09 13.02 Men’s pajamas (cotton) 1 pc 0.95 0.59 1.20 1.03 -0.62 0.92 0.57 5.87 Men’s shirt (classic) 1 pc 0.92 0.19 1.06 0.71 -0.62 0.87 0.20 20.53 Men’s waistcoat 1 pc 0.96 0.92 1.31 1.02 -0.33 0.95 0.88 10.65 Men’s sweatshirt 1 pc 0.82 0.12 1.22 1.02 -0.75 0.88 0.21 17.20 Panty made of cotton (ladies underwear)

1 pc 0.68 0.21 0.96 1.07 -0.68 0.68 0.14 13.22

Bra 1 pc 0.89 0.01 1.13 0.80 -0.93 0.80 0.01 19.68 Nightdress 1 pc 0.81 0.39 1.28 1.08 -0.74 0.83 0.39 7.38 Swimsuit 1 pc 0.57 0.53 1.28 1.00 -1.63 0.55 0.51 8.21 Ladies pullover – long-sleeved 1 pc 0.94 0.67 1.32 1.02 -0.87 0.93 0.63 16.23 Ladies tracksuit 1 pc 0.91 0.32 1.29 1.11 -0.15 0.91 0.30 5.83 Panty made of cotton (girl’s underwear)

1 pc 0.62 0.68 0.70 0.97 -0.97 0.55 0.69 5.44

Children’s pajamas (cotton) 1 pc 0.83 0.25 1.04 1.01 -0.77 0.73 0.24 6.02 Tracksuit 1 pc 0.90 0.61 1.29 0.99 -0.56 0.88 0.58 14.49 Children’s sweatshirt (cotton) 1 pc 0.91 0.74 1.32 1.18 -1.07 0.91 0.75 13.69 Men’s suit 1 pc 0.81 0.04 1.22 1.01 -0.84 0.80 0.04 7.87 Men’s jacket (for summer) 1 pc 0.79 0.58 1.28 1.23 -1.11 0.80 0.59 6.93 Men’s jacket (for winter) 1 pc 0.65 0.03 1.01 0.32 -1.52 0.65 0.03 18.67 Men’s trousers 1 pc 0.80 0.01 1.20 1.10 -1.33 0.78 0.00 21.25 Men’s jacket (leather) 1 pc 0.50 0.46 1.14 0.92 -2.28 0.51 0.45 5.42 Ladies overcoat 1 pc 0.05 0.26 1.23 0.88 -1.70 0.05 0.25 8.75 Ladies winter coat 1 pc 0.50 0.13 1.10 0.55 -1.41 0.52 0.13 17.42 Ladies windcheater (for winter) 1 pc 0.71 0.12 1.17 0.63 -1.35 0.76 0.14 18.41 Two-piece suit 1 pc 0.65 0.01 1.21 0.80 -1.14 0.43 0.04 22.86 Ladies jacket (for summer) 1 pc 0.11 0.01 1.27 0.92 -1.77 0.08 0.01 13.40 Ladies trousers (wool) 1 pc 0.93 0.36 1.17 0.77 -0.52 0.93 0.36 17.34 Dress (for summer) 1 pc 0.32 0.03 1.10 0.61 -3.05 0.31 0.03 21.72 Smock 1 pc 0.44 0.08 1.32 1.16 -1.21 0.43 0.06 31.14 Skirt 1 pc 0.14 0.02 1.35 1.13 -1.27 0.14 0.01 13.21 Dress (for girls) 1 pc 0.78 0.89 1.32 1.22 -0.53 0.83 0.88 5.73 Children’s trousers (cotton) 1 pc 0.66 0.84 1.31 1.21 -1.89 0.67 0.86 18.22 Men’s socks (cotton) 1 pair 0.73 0.09 1.05 0.48 -1.16 0.77 0.13 12.90 Ladies socks (cotton) 1 pair 0.54 0.08 0.78 0.15 -1.57 0.49 0.07 6.61 Ladies tights 1 pc 0.44 0.42 0.88 0.79 -1.55 0.52 0.41 14.86 Children’s tights 1 pc 0.76 0.03 0.94 0.78 -0.99 0.77 0.02 5.38 Ladies neckerchief 1 pc 0.00 0.45 0.99 1.01 -4.23 0.00 0.45 4.53 Handkerchief 1 pc 0.47 0.34 0.65 0.29 -1.61 0.46 0.34 1.29 Men’s leather gloves 1 pair 0.58 0.08 0.60 0.30 -1.76 0.54 0.03 5.25 Thread, sewing 500 m 0.02 0.05 0.36 0.64 -3.30 0.00 0.50 1.24 Knitting yarn 100 g 0.20 0.13 0.46 0.73 -2.10 0.18 0.35 2.84


Products Units ADF95 ADF98 KP95 KP98 LLS PP95 PP98 WeightsZip fastener 1 pc 0.26 0.11 1.05 0.62 -3.10 0.26 0.24 2.61 Cleaning of overcoat 1 pc 0.35 0.14 0.67 0.08 -2.44 0.25 0.11 8.80 Shortening or elongation of coat 1 repair 0.05 0.12 1.00 0.32 -3.17 0.04 0.10 4.34 Men’s footwear suitable for everyday (leather)

1 pair 0.49 0.15 1.31 1.00 -1.57 0.50 0.16 21.31

Men’s footwear suitable for summer (leather)

1 pair 0.77 0.09 1.30 1.11 -1.10 0.83 0.06 8.09

Men’s footwear suitable for winter (leather)

1 pair 0.77 0.28 1.20 0.82 -1.24 0.73 0.27 13.15

Ladies footwear suitable for everyday (leather)

1 pair 0.83 0.04 1.27 1.07 -0.77 0.85 0.06 33.56

Ladies footwear suitable for summer (leather)

1 pair 0.76 0.03 1.29 1.08 -1.23 0.79 0.01 23.05

Ladies footwear for home wear (textile)

1 pair 0.66 0.02 1.12 0.58 -1.72 0.67 0.02 6.47

Children’s footwear suitable for everyday (leather)

1 pair 0.83 0.18 1.30 1.18 -0.61 0.88 0.17 6.25

Children’s footwear suitable for summer (leather)

1 pair 0.16 0.10 1.36 1.20 -2.20 0.41 0.12 6.07

Children’s footwear for leisure wear (leather)

1 pair 0.75 0.17 1.20 0.73 -1.31 0.72 0.17 7.17

Children’s footwear for leisure wear (textile)

1 pair 0.34 0.02 0.83 0.14 -2.19 0.37 0.02 7.17

Children’s footwear for home wear (textile)

1 pair 0.22 0.00 0.85 0.26 -3.49 0.18 0.00 4.60

Children’s footwear suitable for winter (plastic)

1 pair 0.50 0.19 0.98 0.28 -1.88 0.36 0.19 5.84

Repair of ladies heel (replace old with new heels promptly)

1 pair 0.56 0.02 0.80 0.92 -1.61 0.52 0.01 100.55

Actual rentals paid by tenants, first category – 3 rooms, rent regulated by the government

monthly

0.52 0.03 0.80 0.91 -1.62 0.51 0.01 118.25 Actual rentals paid by tenants, first category – 4 rooms, rent regulated by the government

monthly

0.53 0.03 0.80 0.91 -1.58 0.52 0.01 48.57 Actual rentals paid by tenants, second category – 2 rooms, rent regulated by the government

monthly

0.46 0.04 0.74 0.87 -1.71 0.45 0.02 20.07 Actual rentals paid by tenants, first category – 2 rooms, cooperative flat

monthly


monthly


monthly

0.18 0.44 0.39 0.89 -1.36 0.15 0.40 26.76 Imputed rentals of owner-occupied flat – 2 rooms

monthly 0.24 0.49 0.61 1.01 -0.91 0.20 0.40 157.25

Imputed rentals of owner-occupied flat – 3 rooms

monthly 0.38 0.62 0.41 0.93 -0.53 0.32 0.60 393.47

Imputed rentals of owner-occupied flat – 4 rooms

monthly 0.40 0.66 0.48 0.66 -1.33 0.29 0.59 717.89

Tiles 1 m2 0.62 0.55 1.05 0.88 -1.43 0.54 0.49 17.69 Washbasin 1 pc 0.44 0.00 0.69 0.79 -1.28 0.43 0.00 13.54 Mixer tap 1 pc 0.38 0.08 0.60 0.54 -1.37 0.46 0.08 15.06 Decorator 1 m2 0.47 0.18 0.55 0.33 -1.60 0.37 0.18 13.15


Products Units ADF95 ADF98 KP95 KP98 LLS PP95 PP98 WeightsPainter 1 m2 0.52 0.08 1.03 0.49 -1.64 0.53 0.10 13.61 Tiler 1 m2 0.02 0.08 0.59 0.29 -3.24 0.03 0.07 26.89 Heating engineer 1 hour 0.47 0.01 0.87 0.50 -0.93 0.52 0.00 11.05 Paraffin oil 1 l 0.01 0.01 0.20 0.10 -3.62 0.07 0.10 0.40 Black coal 100 kg 0.02 0.38 0.73 0.66 -3.97 0.02 0.41 8.16 Brown coal 100 kg 0.37 0.43 0.73 0.69 -1.85 0.37 0.45 32.96 Briquettes (made from brown coal)

100 kg 0.31 0.29 1.05 0.63 -1.84 0.35 0.31 3.13

Coke 100 kg 0.02 0.09 0.10 0.15 -2.97 0.08 0.25 3.37 Firewood 100 kg 0.26 0.01 0.86 0.38 -2.68 0.25 0.01 5.54 Heat for fuel and preparation of hot water

1 GJ 0.21 0.20 0.45 0.53 -1.54 0.18 0.20 523.14

Upholstered chair 1 pc 0.53 0.40 1.26 1.01 -2.33 0.51 0.39 14.13 Kitchen dining table 1 pc 0.00 0.00 1.05 0.79 -4.85 0.00 0.00 10.83 Wardrobe 1 pc 0.22 0.03 0.34 0.20 -2.08 0.16 0.03 26.31 Studio couch 1 pc 0.03 0.23 0.88 0.19 -3.21 0.05 0.18 32.78 Kitchen unit 1 set 0.00 0.13 1.01 1.19 -4.87 0.00 0.13 27.91 Wall system 1 set 0.33 0.48 1.14 0.72 -1.65 0.34 0.50 26.11 Table in the garden 1 pc 0.68 0.86 0.57 0.32 -1.82 0.35 0.74 4.69 Table lamp 1 pc 0.10 0.01 1.02 0.44 -2.35 0.11 0.05 31.21 Woven carpet 1 m2 0.21 0.00 1.10 0.84 -1.63 0.27 0.00 32.15 Tufted carpet 1 m2 0.16 0.00 0.87 0.10 -3.85 0.22 0.00 8.75 Upholstered armchair repair 1 repair 0.26 0.00 0.59 0.64 -2.13 0.28 0.00 5.92 Quilt 1 pc 0.08 0.13 0.71 0.76 -3.02 0.17 0.13 8.69 Blanket (synthetic fiber) 1 pc 0.13 0.12 0.64 0.51 -1.55 0.47 0.36 8.69 Decorative textile made of cotton

1 m2 0.46 0.01 1.11 0.51 -1.85 0.61 0.01 14.97

Knitted synthetic curtains 1 m2 0.37 0.26 0.57 0.79 -1.02 0.29 0.26 12.84 Bed linen (not crape) 1 set 0.24 0.00 0.47 0.67 -3.23 0.16 0.00 11.88 Bed linen (crape) 1 set 0.70 0.65 0.76 0.50 -1.36 0.43 0.48 11.88 Bed sheet made of cotton 1 pc 0.11 0.57 0.73 0.63 -2.04 0.27 0.49 6.79 Terry towel 1 pc 0.80 0.19 1.12 0.94 -0.89 0.66 0.44 7.34 Dishcloth 1 pc 0.80 0.11 1.08 0.63 -1.76 0.61 0.11 4.12 Refrigerator 1 pc 0.21 0.00 1.11 0.97 -2.26 0.60 0.00 8.09 Freezer 1 pc 0.21 0.00 0.75 0.73 -1.77 0.40 0.00 9.58 Washing machine 1 pc 0.82 0.02 1.25 1.18 -0.92 0.78 0.02 52.97 Dishwasher 1 pc 0.75 0.07 0.96 1.11 -1.02 0.75 0.09 17.93 Electric range (with a grill) 1 pc 0.97 0.83 1.21 1.12 -0.17 0.95 0.81 7.90 Microwave oven 1 pc 0.41 0.02 0.92 0.72 -1.72 0.47 0.04 20.84 Electric boiler 1 pc 0.59 0.23 0.60 0.73 -1.47 0.51 0.23 8.10 Vacuum cleaner 1 pc 0.85 0.06 1.22 0.88 -0.95 0.80 0.07 26.04 Sewing machine 1 pc 0.73 0.48 0.76 0.78 -1.05 0.69 0.44 3.10 Electric hand-held beater 1 pc 0.74 0.65 1.23 1.12 -1.23 0.83 0.76 8.44 Electric deep fryer 1 pc 0.54 0.57 0.25 0.59 -1.49 0.39 0.57 5.18 Iron 1 pc 0.28 0.03 1.03 1.07 -2.02 0.59 0.04 4.32 Repair of a refrigerator 1 repair 0.19 0.05 1.14 0.79 -2.10 0.07 0.12 14.01 Repair of a washing machine 1 repair 0.80 0.21 1.19 0.90 -0.92 0.74 0.29 16.37 Fireproof bowl 1 pc 0.77 0.67 1.18 0.83 -1.52 0.73 0.62 11.23 Mug (porcelain) 1 pc 0.81 0.01 1.04 0.81 -1.19 0.74 0.02 5.80 Plate (porcelain) 1 pc 0.76 0.60 1.13 0.86 -1.12 0.72 0.57 7.47 Cup and saucer (pottery) 1 pc 0.70 0.00 0.83 0.96 -1.54 0.57 0.00 5.30 Vase 1 pc 0.51 0.22 1.00 0.58 -2.12 0.49 0.21 13.66


Products Units ADF95 ADF98 KP95 KP98 LLS PP95 PP98 WeightsFrying pan 1 pc 0.08 0.45 1.20 0.85 -1.96 0.08 0.42 5.11 Cutlery 6 pcs 0.14 0.00 0.29 0.20 -2.88 0.04 0.00 3.30 Kitchen knife 1 pc 0.13 0.30 0.29 0.41 -2.68 0.10 0.30 3.71 Soup ladle 1 pc 0.17 0.12 0.93 0.20 -2.54 0.17 0.10 5.17 Mixing/wooden spoon 1 pc 0.26 0.26 0.75 0.69 -1.95 0.10 0.28 5.17 Kitchen scales 1 pc 0.54 0.39 1.05 0.90 -1.61 0.44 0.63 3.86 Bucket 1 pc 0.48 0.44 0.91 1.07 -1.98 0.22 0.51 5.19 Ironing board 1 pc 0.34 0.42 1.24 0.94 -1.28 0.44 0.46 3.27 Lawn mower (type: rotary mower, electric)

1 pc 0.47 0.00 1.03 0.91 -1.83 0.34 0.00 21.20

Electric drill 1 pc 0.36 0.01 0.71 0.08 -2.10 0.23 0.11 10.59 Screwdriver 1 pc 0.52 0.12 1.16 1.03 -1.35 0.51 0.06 8.33 Lawn rake (with wooden handle) 1 pc 0.14 0.10 0.86 0.48 -1.53 0.01 0.10 5.89 Rocker switch 1 pc 0.52 0.12 1.23 1.16 -1.00 0.71 0.10 3.47 Light bulb 1 pc 0.08 0.01 0.74 0.20 -2.71 0.15 0.03 8.95 AA battery 1.5 V 1 pc 0.01 0.04 0.89 0.66 -3.41 0.01 0.04 8.95 Nails 1 kg 0.03 0.04 0.23 0.40 -3.19 0.13 0.31 8.97 Detergent 1 kg 0.21 0.14 0.73 1.01 -2.28 0.28 0.38 66.95 Anticalcareous for washing machine, powder

1 kg 0.31 0.07 0.13 0.49 -2.30 0.27 0.09 3.65

Dish washing liquid 1 l 0.10 0.17 0.57 0.50 -2.58 0.24 0.35 13.59 Liquid scourer 1 l 0.20 0.30 0.28 0.41 -2.54 0.18 0.35 15.31 Furniture polish 1 l 0.57 0.22 0.94 0.95 -1.28 0.56 0.22 1.55 Broom 1 pc 0.00 0.51 0.91 0.80 -2.71 0.00 0.34 7.82 Insecticide 200 ml 0.06 0.22 1.14 0.86 -3.03 0.06 0.22 1.74 Adhesive 50 ml 0.22 0.26 0.43 0.36 -1.79 0.14 0.49 4.97 Paper napkin 100 pcs 0.03 0.35 0.49 0.22 -3.57 0.26 0.17 7.56 Plastic bag 50 pcs 0.00 0.48 0.31 0.64 -4.93 0.01 0.52 5.22 Aluminum foil 1 m2 0.51 0.14 0.60 0.21 -1.69 0.38 0.10 3.47 Scissors 1 pc 0.33 0.27 1.14 0.68 -1.86 0.50 0.33 1.43 Carpet cleaning 1 m2 0.13 0.02 0.13 0.35 -2.32 0.05 0.23 5.16 Laundry 1 amount 0.55 0.27 0.50 0.16 -1.80 0.35 0.20 4.34 ACYLPYRIN 10 pcs 0.12 0.06 0.56 0.34 -2.37 0.13 0.28 5.24 ATARALGIN 20 pcs 0.01 0.00 0.63 0.31 -2.98 0.13 0.00 5.78 CELASKON 250 30 pcs 0.06 0.08 0.13 0.09 -2.91 0.18 0.30 7.86 B KOMPLEX FORTE 20 pcs 0.01 0.04 0.24 0.36 -2.85 0.10 0.33 7.86 Chamomile 50 g 0.06 0.14 0.40 0.12 -2.65 0.16 0.27 5.93 Medical thermometer 1 pc 0.33 0.02 0.51 0.17 -1.84 0.18 0.01 1.35 Medical examination at the request of a patient

1 service0.44 0.00 0.27 0.74 -1.99 0.20 0.00 4.26

plastic surgery – eyelids 1 service 0.04 0.14 0.37 0.51 -3.04 0.03 0.10 10.00 Partly removable tooth replacement

1 pc 0.06 0.00 0.50 0.23 -2.71 0.12 0.00 13.79

Eye refraction 1 service 0.04 0.17 0.78 0.47 -2.49 0.06 0.17 12.97 Tire casing (bike) 1 pc 0.12 0.40 1.03 0.70 -2.19 0.12 0.43 10.31 Tire (radial) 165 R 13 1 pc 0.65 0.45 1.28 1.00 -1.22 0.80 0.45 22.00 Battery L1 12V 1 pc 0.54 0.67 0.25 0.19 -2.02 0.26 0.45 4.98 Petrol 95 1 l 0.20 0.34 0.07 0.10 -2.46 0.10 0.21 193.24 Petrol Super 98 1 l 0.31 0.46 0.07 0.12 -2.16 0.12 0.26 38.39 Diesel for car 1 l 0.16 0.47 0.07 0.11 -2.41 0.16 0.31 29.01 Engine oil 1 l 0.10 0.12 0.36 0.68 -1.81 0.36 0.50 8.84 Centering of rear wheel (bicycle) 1 repair 0.16 0.47 0.46 0.49 -2.30 0.27 0.35 7.83


Products Units ADF95 ADF98 KP95 KP98 LLS PP95 PP98 Weights

Charge for driving licenses course

fee 0.05 0.00 0.53 0.17 -1.77 0.30 0.10 32.29 Parking charge for cars 1 hour 0.13 0.57 0.50 0.39 -2.25 0.07 0.43 7.16

Motorway tax disc annual

fee 0.30 0.11 0.69 0.41 -2.12 0.36 0.23 15.32 Individual fare in public urban transport by bus

1 ticket 0.38 0.30 0.55 0.34 -2.11 0.34 0.13 1.71

Payments for the delivery of a letter inland

1 pc 0.02 0.02 0.10 0.09 -3.35 0.01 0.02 10.87

Payments for the delivery of a parcel inland

1 pc 0.06 0.03 0.47 0.14 -3.35 0.04 0.03 1.76

Installation costs of private telephone equipment

1 pc 0.00 0.00 0.31 0.30 -6.59 0.00 0.00 2.49

Television set – color 1 pc 0.80 0.35 1.30 1.18 -0.78 0.83 0.38 12.73 Hi-fi music centre 1 pc 0.92 0.38 1.06 0.82 -0.83 0.87 0.36 4.69 Film for color prints (36 pictures)

1 pc 0.25 0.48 1.07 0.89 -1.31 0.25 0.45 12.79

Repair of color TV set 1 repair 0.82 0.80 0.91 1.13 -0.83 0.80 0.75 26.25 Guitar (not electric and not for children)

1 pc 0.49 0.00 1.02 0.34 -1.76 0.52 0.00 10.95

Doll (from PVC) 1 pc 0.43 0.56 1.04 0.51 -1.56 0.42 0.56 5.49 Toy car (with an electric cell) 1 pc 0.89 0.79 1.29 1.14 -0.85 0.84 0.79 6.99 Building set (type Lego) 1 pc 0.77 0.44 1.24 1.08 -1.67 0.75 0.47 10.38 Soft toy 1 pc 0.48 0.10 1.07 0.52 -1.77 0.48 0.10 4.36 Inflatable ball 1 pc 0.50 0.02 1.01 0.53 -1.76 0.51 0.02 2.89 Baby carriage (toy) 1 pc 0.67 0.04 1.06 0.39 -1.54 0.67 0.04 1.64 Ball (for volleyball) 1 pc 0.07 0.38 1.02 0.28 -2.41 0.07 0.25 8.83 Tent 1 pc 0.30 0.88 1.15 0.83 -1.05 0.29 0.81 7.61 Rucksack 1 pc 0.22 0.29 0.72 0.97 -2.29 0.41 0.43 5.18 Carnation 1 pc 0.00 0.10 0.61 0.28 -2.94 0.00 0.11 11.22 Rose 1 pc 0.11 0.00 1.02 0.49 -2.43 0.15 0.01 11.22 Pot plants (type African violet) 1 pc 0.63 0.11 0.94 0.37 -1.56 0.51 0.09 5.76 Artificial flower 1 pc 0.01 0.21 0.67 0.76 -2.66 0.02 0.22 8.97 Outdoor plant – garden bush (rose)

1 pc 0.51 0.23 1.01 0.36 -1.85 0.51 0.20 4.22

Dog-food, dried 500 g 0.19 0.47 0.31 0.22 -2.24 0.21 0.43 38.77 Veterinary service 1service 0.17 0.22 0.12 0.14 -2.68 0.07 0.18 8.11 Ticket, ski lift 1 pc 0.14 0.18 0.13 0.19 -2.95 0.11 0.23 16.36 Ticket, aerobics centre or fitness centre

1 hour 0.43 0.23 0.83 0.68 -1.83 0.41 0.24 9.50

Swimming pool, indoor 1 ticket 0.21 0.21 0.61 0.71 -1.52 0.28 0.24 9.09 Ticket, football game average 0.32 0.41 0.42 0.89 -1.90 0.30 0.41 3.05 Charge for dancing lessons (adolescent people)

course fee 0.02 0.16 1.01 0.89 -4.04 0.02 0.13 3.92

Ticket, cinema average 0.77 0.05 0.76 0.99 -0.96 0.74 0.02 10.41 Ticket, theatre average 0.26 0.00 0.77 0.63 -2.49 0.15 0.00 16.48 Ticket, concert average 0.01 0.10 0.37 0.60 -2.52 0.01 0.24 7.67 Lending fee, video cassette 24 hours 0.24 0.29 0.11 0.24 -2.49 0.15 0.22 5.89 Blow-up of a color picture 10 pcs 0.19 0.35 0.37 0.33 -2.51 0.13 0.30 21.91 Developing color film 36 prints 1 pc 0.14 0.27 0.15 0.09 -2.73 0.07 0.18 5.56 License for radio – monthly monthly 0.09 0.01 0.43 0.33 -2.87 0.06 0.01 30.66 License for television – monthly monthly 0.16 0.03 0.16 0.29 -2.61 0.12 0.03 68.37 Children’s book (aged 9 years or less)

average 0.10 0.02 0.46 0.31 -2.11 0.15 0.02 13.25


Products Units ADF95 ADF98 KP95 KP98 LLS PP95 PP98 WeightsBelles letters by domestic author average 0.16 0.00 0.59 0.39 -2.37 0.08 0.00 11.66 Belles letters by worldwide-known author

average 0.15 0.10 0.17 0.17 -2.38 0.15 0.10 23.31

Daily newspaper MLADA FRONTA DNES

monthly 0.13 0.04 0.41 0.17 -2.77 0.14 0.05 12.55

Daily newspaper, tabloid – BLESK

monthly 0.12 0.12 0.33 0.55 -3.96 0.10 0.10 18.48

Daily newspaper – PRAVO monthly 0.03 0.03 0.62 0.45 -2.83 0.03 0.03 13.68 Daily newspaper – LIDOVE NOVINY

monthly 0.48 0.28 0.44 0.11 -1.84 0.30 0.21 12.15

Picture postcard 10 pcs 0.14 0.00 0.32 0.22 -2.70 0.11 0.00 4.72 Desk calendar 1 pc 0.55 0.20 0.81 0.76 -1.71 0.55 0.21 5.96 Domestic recreation – stay in the mountains

1 person 0.22 0.01 0.94 0.34 -2.22 0.21 0.01 63.51

Spain 1 person 0.03 0.16 0.15 0.21 -3.18 0.03 0.11 65.08

Italy 4

persons 0.01 0.17 0.20 0.26 -3.07 0.00 0.00 20.62 School-fees at nursery school monthly 0.48 0.57 1.15 0.99 -1.10 0.56 0.57 12.73 Tuition at private secondary school

monthly 0.46 0.52 0.96 0.77 -1.69 0.46 0.50 7.83

School-fees at higher level than secondary school

yearly 0.06 0.06 0.19 0.75 -3.05 0.06 0.08 2.97

Examination fee for entrance to university

fee 0.77 0.45 1.07 1.19 -0.97 0.77 0.45 1.81

Language teaching 1 hour 0.02 0.24 1.00 0.56 -3.59 0.01 0.22 15.20 School fees at art school (lower level)

yearly 0.17 0.12 1.03 0.54 -2.26 0.17 0.12 13.25

After-school care centre fee monthly 0.50 0.83 0.49 0.29 -1.75 0.29 0.80 2.09 Thick soup 0,33 l 0.02 0.04 0.65 0.21 -3.73 0.17 0.04 4.45 Meat soup 0,33 l 0.22 0.25 0.70 0.11 -2.45 0.13 0.16 5.56 Roast sirloin in cream sauce 100 g 0.02 0.30 0.63 0.18 -4.21 0.02 0.30 27.05 Beef goulash 100 g 0.00 0.13 0.71 0.14 -4.92 0.00 0.13 34.49 Pork roast 100 g 0.03 0.34 0.49 0.28 -4.01 0.02 0.26 92.53 Schnitzel 100 g 0.02 0.45 0.33 0.48 -3.16 0.03 0.45 82.67 Pepper with minced meat filling 100 g 0.00 0.19 0.67 0.18 -4.53 0.02 0.18 24.98 Cheese deep fried in breadcrumbs

100 g 0.20 0.34 0.18 0.24 -2.34 0.15 0.27 10.32

Dumplings (side dish) 160 g 0.07 0.40 0.65 0.17 -2.82 0.08 0.27 36.30 Sliced ham – starter 70 g 0.00 0.37 0.22 0.49 -5.03 0.02 0.27 13.11 Pancake – warm dessert 100 g 0.13 0.35 0.10 0.16 -2.41 0.07 0.20 6.46 Coffee 1 portion 0.00 0.39 0.57 0.13 -4.53 0.02 0.20 18.38 Coke (Pepsi Cola) in a restaurant 0,2 l 0.04 0.25 0.18 0.34 -2.94 0.03 0.18 14.68 Draught beer – light 0,5 l 0.04 0.14 0.55 0.27 -2.78 0.10 0.21 47.21 Draught beer – light (lager) 0,5 l 0.19 0.17 0.13 0.37 -2.62 0.21 0.26 16.79 Light beer (lager) 0,5 l 0.02 0.08 0.13 0.09 -2.64 0.02 0.12 2.31 White wine 0,2 l 0.27 0.29 0.36 0.41 -1.75 0.36 0.29 10.19 Red wine 0,2 l 0.31 0.22 0.47 0.34 -1.63 0.33 0.25 7.45 Inland rum – dark 0,05 l 0.17 0.32 0.44 0.31 -2.01 0.13 0.35 1.57 Spirit, brandy – FERNET STOCK

0,05 l 0.34 0.28 0.59 0.35 -2.13 0.24 0.35 5.25

Spirit, Becher’s (Carlsbad) liqueur

0,05 l 0.28 0.27 0.69 0.41 -2.09 0.37 0.37 2.05


Products Units ADF95 ADF98 KP95 KP98 LLS PP95 PP98 WeightsA two or three-course meal (lunch or supper) in canteens

1 menu 0.10 0.45 0.36 0.18 -3.12 0.08 0.26 236.03

Lunch in canteens in schools – pupils aged 7–10 years

1 menu 0.30 0.44 0.61 0.52 -1.84 0.18 0.32 50.20

Lunch in canteens in schools – pupils aged 11–14 years

1 menu 0.35 0.47 0.66 0.53 -1.89 0.21 0.30 58.71

Lunch in canteens in (secondary) schools – students aged 15 years or more

1 menu

0.10 0.45 0.66 0.51 -2.04 0.18 0.30 33.58 Lunch in canteens in universities 1 menu 0.09 0.18 0.13 0.27 -2.36 0.17 0.33 13.84 Hotel **** 1 night 0.28 0.27 1.09 0.91 -1.00 0.41 0.45 2.22 Hotel *** 1 night 0.43 0.19 1.17 0.85 -1.59 0.15 0.12 8.94 Hostel 1 night 0.38 0.30 0.64 0.55 -1.67 0.35 0.32 3.97 Cottage 1 night 0.18 0.26 0.92 0.66 -1.70 0.28 0.26 7.40 Accommodation services of universities

monthly 0.86 0.79 0.15 0.30 -0.98 0.79 0.68 7.38

Barber 1 service 0.11 0.00 0.46 0.15 -1.91 0.23 0.15 13.19 Hairdresser (for ladies) 1 service 0.31 0.34 0.52 0.15 -1.89 0.15 0.30 71.49 Deep complexion clearing incl. face pack

1 service0.79 0.42 0.99 0.44 -1.30 0.61 0.32 21.11

Hair dryer 1 pc 0.65 0.00 0.98 0.56 -2.13 0.49 0.01 6.16 Electric razor 1 pc 0.35 0.48 1.04 0.92 -2.05 0.26 0.41 7.40 Toilet soap 100 g 0.69 0.49 1.14 0.99 -1.28 0.64 0.47 26.92 Toothpaste 75 ml 0.39 0.44 0.96 0.99 -1.30 0.29 0.42 27.87 Toilet paper 1 pc 0.01 0.25 0.18 0.52 -2.78 0.04 0.31 27.29 Toothbrush 1 pc 0.45 0.00 0.91 0.40 -3.10 0.35 0.00 10.87 Non-electrical razor 1 pc 0.17 0.19 0.51 0.65 -2.68 0.21 0.42 7.96 Hair shampoo 250 ml 0.79 0.43 1.10 0.87 -0.98 0.64 0.18 19.95 Cream NIVEA 150 ml 0.12 0.12 0.61 0.89 -2.33 0.12 0.14 30.73 Deodorant 100 g 0.29 0.50 0.95 0.69 -2.63 0.50 0.30 20.21 Lipstick 1 pc 0.63 0.46 0.81 1.08 -1.22 0.60 0.46 22.51 Ladies wrist watch 1 pc 0.08 0.07 0.58 0.61 -2.81 0.06 0.08 15.21 Wedding ring (gold) 1 pc 0.00 0.03 0.16 0.19 -3.46 0.01 0.03 20.48 Electronic wall clock 1 pc 0.21 0.22 0.97 0.50 -4.38 0.23 0.22 8.08 Ladies umbrella 1 pc 0.75 0.15 0.81 0.75 -1.04 0.55 0.38 5.27 Pram 1 pc 0.79 0.03 1.03 0.61 -1.02 0.64 0.32 1.53 Accommodation in old people’s home

monthly 0.26 0.03 1.01 0.78 -2.17 0.35 0.03 63.31

Cremation fee 0.12 0.22 0.15 0.16 NA 0.09 0.19 1.50 Registration fee for a dog fee 0.12 0.21 0.45 0.35 -2.61 0.09 0.17 3.78

Note: Sample weight multiplied by 100.


APPENDIX 2

In Appendix 2, we re-estimate Table 1 and 2 using different time coverage (e.g. 1995-1997 pre-

targeting period and 1999-2001 targeting period) to assess the supposed fall in inflation

persistence further. Therefore, we keep the sample size identical. Obviously, the drawback is that

the sample size is rather small.

The results presented in Table A.8 and A.9 show that the estimated degree of inflation

persistence falls for almost all categories after the adoption of inflation targeting in 1998, albeit

the standard error of estimates is large and thus, the results should be interpreted with caution.

Table A.8. Inflation persistence, yearly inflation, 1995–1997 (36 obs.)



Tradables 311 0.59 0.42 (0.31) 0.46 (0.30) 0.41* (0.17) Non-tradables 101 0.41 0.45 (0.28) 0.49 (0.26) 0.42* (0.19) Services 96 0.40 0.45 (0.28) 0.48 (0.27) 0.43* (0.19) Non-reg. serv. 74 0.30 0.44 (0.27) 0.54 (0.27) 0.44* (0.17) Regulated 27 0.11 0.48 (0.31) 0.47 (0.26) 0.36* (0.22) Durables 164 0.21 0.46 (0.31) 0.46 (0.30) 0.42* (0.17) Non-durables 152 0.39 0.37 (0.30) 0.45 (0.29) 0.41* (0.17) Raw goods 42 0.11 0.30 (0.32) 0.37 (0.32) 0.42* (0.16) Processed goods 370 0.89 0.44 (0.30) 0.48 (0.29) 0.41* (0.18) Total prod. level 412 1.00 0.43 (0.31) 0.47 (0.29) 0.41* (0.17) Aggr. inflation 1 1 0.49 0.32 0.14

Notes: The pairs (tradables, non-tradables) and (raw goods, processed goods) make up a total of 412 products. Durables do not include regulated prices, while processed goods do. For all the measures of persistence displayed, higher values mean more persistent inflation. For the ADF and PP unit root tests, the probability of rejecting the null hypothesis of a unit root is reported. The probability can vary from 0 to 1. Higher values correspond to more persistence. For example, a probability higher than 0.10 means that the null of a unit root cannot be rejected at the 10% significance level. Standard deviations are shown in parentheses. For the KPSS stationarity test, the t-statistic is reported. Higher t-statistic values increase the probability of rejecting the null hypothesis of stationarity and hence characterize more persistence in the underlying series. *, **, and *** denote the 10%, 5% and 1% asymptotical significance levels for rejection of the stationarity hypothesis. Standard deviations are shown in parentheses. For the LLS (Lanne et al., 2002) unit root test in the presence of a structural break, the t-statistic is reported. More negative t-statistic values increase the probability of rejecting the null hypothesis of a unit root and thus characterize less


persistence in the underlying series. *, **, and *** denote the 10%, 5% and 1% asymptotical significance levels for rejection of the unit root hypothesis.

Table A.9. Inflation persistence, yearly inflation, 1999–2001 (36 obs.)



Tradables 311 0.59 0.36 (0.29) 0.40 (0.30) 0.40* (0.17) Non-tradables 101 0.41 0.22 (0.28) 0.24 (0.28) 0.36* (0.18) Services 96 0.40 0.22 (0.28) 0.24 (0.28) 0.37* (0.18) Non-reg. serv. 74 0.30 0.19 (0.27) 0.20 (0.26) 0.36* (0.16) Regulated 27 0.11 0.32 (0.27) 0.35 (0.29) 0.37* (0.22) Durables 164 0.21 0.32 (0.29) 0.34 (0.30) 0.39* (0.17) Non-durables 152 0.39 0.36 (0.31) 0.42 (0.31) 0.39* (0.17) Raw goods 42 0.11 0.40 (0.26) 0.51 (0.27) 0.38* (0.17) Processed goods 370 0.89 0.32 (0.30) 0.35 (0.30) 0.39* (0.18) Total prod. level 412 1.00 0.33 (0.30) 0.37 (0.30) 0.39* (0.17) Aggr. inflation 1 1 0.001 0.07 0.54**

Notes: See Table A.8.

The Effects of Monetary Policy - 126 -

4 The Effects of Monetary Policy in the Czech Republic: An

Empirical Study *

4.1 Introduction

Understanding the transmission of monetary policy to inflation and other real economic variables

is of key importance if central bankers are to conduct monetary policy effectively. Not

surprisingly, there is extensive theoretical as well as empirical literature studying the effects of

monetary policy shocks on real economy aggregates and prices. For a small open economy such

as the Czech Republic, it is vital to analyze monetary policy transmission for several reasons.

First, there is somewhat mixed evidence regarding monetary policy transmission, as many studies

estimate standard vector autoregression (VAR) models mixing data from two distinct policy

regimes, i.e., from the fixed exchange rate regime under which the Czech National Bank

conducted its policy until May 1997, and from the inflation targeting regime that was adopted in

January 1998.54 Not surprisingly, the identification of monetary policy shocks then becomes

somewhat cumbersome and all these studies exhibit the price puzzle (see our Table 1 in the

results section).

* We thank Oxana Babetskaia-Kukharchuk, Sophocles Brissimis, Martin Cincibuch, Michal Franta, Paolo Giordani, Jan Hanousek, Jaromír Hurník, Marek Jarocinski, Michal Kejak, Evžen Kočenda, Francesco Lippi, Benoit Mojon, Michal Skořepa, Wadim Strielkowski, Petr Zemčík, and seminar participants at the Austrian National Bank, the Czech National Bank, Charles University, and the International Trade and Finance Association Annual Conference for helpful comments and discussions. All remaining errors are entirely our own. The views expressed in this paper are not necessarily those of the Czech National Bank. This paper was supported by Czech National Bank Research Project No. A3/07.

54 See Coats, Laxton, and Rose (2003) and Kotlán and Navrátil (2003) for an overview of Czech monetary policy.


Therefore, it is worthwhile to update previous results reflecting the monetary policy regime

changes, to utilize a wider range of econometric techniques and, on top of that, to incorporate

real-time and forward-looking variables into the VAR analysis. To our knowledge, real-time data

has not been applied to study monetary transmission in the Czech Republic. This is in a sense

paradoxical, as an important feature of monetary policy conduct is that it is based on the

information set available at the time of policy-making. This implies that using ex-post revised

data (note that these are typically more precise, but are not available at the time of monetary

policy action) may contaminate the estimated effects of monetary policy (Croushore and Evans,

2006). The revisions are typical for output data.55

There is also no empirical evidence about monetary policy effects on sectoral prices. This is

striking, because tradable prices in a small open economy may be driven to a large extent by

international factors that domestic monetary policy is unlikely to affect. Our prior assumption is

that as non-tradable prices are typically less exposed to international competition and more labor-

intensive, the reaction of non-tradable prices is likely to be more persistent (see e.g. Barro, 1972,

and Martin, 1993, for models relating the degree of competition to price rigidity).56

In this paper, we examine the effects of monetary policy within the vector autoregression (VAR),

structural VAR (SVAR), and factor-augmented VAR (FAVAR) frameworks during the inflation

targeting period in the Czech Republic. More specifically, we focus on assessing the persistence

and magnitude of monetary policy shocks on output (including the real-time output gap), prices

55 We therefore utilize the real-time estimates of the output gap available from the Czech National Bank (CNB). Using the central bank output gap is advantageous for monetary policy shock identification, as the central bank conducts its policy based on its estimate of the degree of economic activity, not the estimates of other institutions or individuals. Note that price indices are not revised ex post by the Czech Statistical Office. An additional rationale for using the output gap is that in an environment of changing potential growth of the economy, as is the case in our sample, actual GDP growth does not necessarily give an accurate picture about the degree of economic activity. 56 The negative link between the degree of competition and price rigidity is also documented empirically using microeconomic data at the price-setter level by Alvarez and Hernando (2006) for the euro area and Coricelli and Horvath (2006) for Slovakia.


(at both the aggregate and sectoral level) and the exchange rate, controlling for a standard set of

factors.

The paper is organized as follows. Section 2 discusses the related literature. The data are

presented in section 3. Section 4 is focused on identification issues. Section 5 contains our results

on the effects of monetary policy. We present our conclusions in section 6, and an appendix

follows.

4.2 Related VAR Literature

Vector autoregressions (VARs), as introduced by Sims (1980), are considered to be benchmarks

in econometric modeling of monetary policy transmission. It has been argued that this class of

models provides a certain mix between a mere “data-driven” approach and an approach

coherently based on economic theory (see Fry and Pagan, 2005, on the applications of VARs for

macroeconomic research). In terms of monetary policy analysis, the VAR methodology has been

further developed among others by Gerlach and Smets (1995); Leeper, Sims, and Zha (1998); and

Christiano, Eichenbaum, and Evans (1999). The last-mentioned study provides a detailed review

of the literature on this topic in the United States. Similarly, there has been extensive research

undertaken in Europe to study various aspects of monetary transmission in the euro area

countries (see Angeloni, Kashyap, and Mojon, 2003). The research on monetary transmission in

the euro area either focuses on euro area-wide analysis (Peersman and Smets, 2001) or studies

specific countries in detail (Mojon and Peersman, 2001).

The economic theory suggests that output and prices should temporarily fall after a monetary

contraction. Nevertheless, as regards prices, a number of papers document that, on the contrary,

prices rise after a monetary contraction. This effect has been labeled as the “price puzzle.” The


literature typically argues that the price puzzle is a consequence of some model misspecification

(Brissimis and Magginas, 2006, and Giordani, 2004). Meanwhile, Barth and Ramey (2001) suggest

that a fall in both prices and output would indicate that monetary policy affects the economy

mainly through the demand channel. On the other hand, falling output and rising price levels

would point to the prevalence of the supply or cost channel.57

In addition, the literature examines the effect of monetary policy on exchange rate behavior.

Generally, an immediate exchange rate appreciation after a monetary tightening and then a

gradual depreciation of the domestic currency is expected according to uncovered interest rate

parity. However, the empirical evidence is again somewhat mixed. Some authors find a rather

persistent appreciation of the domestic currency (“delayed overshooting”, Eichenbaum and

Evans, 1995), while others report that the exchange rate actually depreciates with a monetary

contraction and provide explanations for the so-called exchange rate puzzle (Kim and Roubini,

2000).

A number of approaches to dealing with model misspecification related to monetary policy shock

identification have been stressed in the literature. For example, Brissimis and Magginas (2006)

show that by adding forward-looking variables such as federal funds futures to a standard VAR

specification, one is able to obtain responses to monetary policy that are consistent with the

theory. The rationale for including federal funds futures is that they contain market expectations

about future monetary policy action (this expectation element may also be found in commodity

prices or money, to a certain extent).

57 If a firm has to borrow to finance its production, interest rates enter its cost function. Consequently, a monetary policy tightening increases the firm’s costs, to which the firm may react by increasing the price of the products it sells. In consequence, this argument suggests that the price puzzle does not have to be caused by model misspecification. In general, see Coricelli et al. (2006) for more specific explanations of the price puzzle.


In addition, Croushore and Evans (2006) emphasize the role of data revisions for monetary

policy shock identification. Monetary policy makers react to the information set available at the

time they make their decision, and it is often the case that GDP data are revised afterwards. As a

result, using ex-post GDP data series may contaminate the estimated monetary policy effects.

Also, monetary policy makers often tend to react to the output gap rather than GDP growth. In

addition, Giordani (2004) shows that using the output gap instead of GDP growth alleviates the

price puzzle. These concerns are especially appealing in our case. First, the CNB’s main

forecasting model (the so-called Quarterly Projection Model) indeed contains an output gap in its

reaction function (Coats et al., 2003). Second, GDP growth may still be useful as the measure of

the degree of economic activity if potential output growth is not changing much. However, in the

case of the Czech economy, it is estimated that potential output growth sharply increased from

some 2% in 1998 to around 5.5% in 2005 (Dybczak, Flek, Hájková, and Hurník, 2006).

Next, there has been a lot of research focusing on the sensitivity of the responses of aggregate

variables such as aggregate inflation and output to monetary policy within the VAR framework.

However, much less is known about the responses to monetary policy at the more disaggregated

level. Erceg and Levin (2006) find that the durable goods sector is more sensitive to interest rate

changes than the non-durable goods sector in the U.S. Based on this empirical finding, they

investigate the impact of monetary policy on these two industries and find, as expected, that

monetary policy effects are much stronger in the durable goods industry. Dedola and Lippi

(2005) study the responses to monetary policy of various industrial sectors for a number of

OECD countries. They find that the responses vary between sectors in terms of their magnitude

and persistence. This result is confirmed by Peersman and Smets (2005), who find a number of

significant differences between various industries in the euro area in terms of both the magnitude

of the output response as well as the asymmetry of the responses over the business cycle.


Bouakez et al. (2005) is one of the few studies that examine the impact of monetary policy on

disaggregate prices. Their results suggest that monetary transmission affects household

consumption in the construction and durable manufacturing sectors the most, but the impact of

a monetary policy shock vanishes relatively quickly. They also find significant differences between

the sectors’ inflation in terms of variance decomposition, volatility, and persistence. Bouakez et al.

(2005) find that the response of services inflation to monetary policy shocks is relatively

pronounced and also the most persistent. Boivin et al. (2007) study the effect of macroeconomic

fluctuations on disaggregate prices within the factor-augmented VAR framework. Among other

things, their results indicate that the degree of market power explains the diversity of the

responses of disaggregate prices to monetary policy shocks.

Several papers study the monetary policy effects for the Czech Republic within the VAR

framework.58 Using the sample period after the adoption of inflation targeting (1998–2004),

Hurník and Arnoštová (2005) find that prices respond with a peak around 5–6 quarters after a

shock, although there is some evidence for a price puzzle in the first two quarters after the shock.

Output falls after a monetary contraction, with a peak after one year or so. There is a delayed

overshooting of the exchange rate, as it depreciates only some 4 to 5 quarters after the monetary

policy innovation. Extending the sample back to 1994, when the fixed exchange rate regime was

in use, yields less satisfactory results, as it is obviously more difficult to identify monetary policy

shocks across two monetary policy regimes. In our paper, we use a similar, slightly extended time

horizon (after the adoption of inflation targeting), but we opt for monthly rather than quarterly

data. In addition, in our paper we include the real-time output gap in the benchmark

specification, as opposed to the ex-post revised GDP used by Hurník and Arnoštová (2005). The

effects of a monetary policy contraction estimated in our paper are largely in line with the

responses observed in more developed economies and countries in the Eurozone, in particular. 58 See Coricelli, Égert, and MacDonald (2006) for a survey of current findings on monetary policy transmission in Central and Eastern Europe, including those undertaken within the VAR framework.


Contrary to Hurník and Arnoštová (2005), we do not find evidence for a price puzzle in the

Czech Republic.

Next, there are a number of papers analyzing and comparing the effects of monetary policy in

groups of Central and Eastern European countries vis-à-vis other, more advanced economies

(Creel and Levasseur, 2005; Darvas, 2006; EFN59, 2004; Héricourt, 2005). Many studies find

evidence of price and/or exchange rate puzzles for the Czech Republic. As argued by Coricelli et

al. (2006), the price puzzle is generally avoided in studies that allow for changes in coefficients

and in papers employing more sophisticated identification schemes. As we argue below, the price

puzzle in these studies often arises because monetary policy regime changes are ignored. In our

paper we consider only the period after the change of monetary policy regime, characterized by

stable coefficients (as assessed by the estimation of recursive coefficients).

Among the studies that do not find evidence of a price puzzle, Jarocinski (2006) provides a

Bayesian VAR analysis of monetary policy effects in Western and Central Europe. Interestingly,

Jarocinski finds that monetary policy is more potent in Central Europe, despite a lower level of

financial development and smaller indebtedness. Regarding the Czech Republic, he uncovers that

there is a relatively strong appreciation of exchange rates as well as a larger price decline after a

monetary policy innovation, as compared to other Central European countries. Elbourne and

Haan (2006) study the interactions between the financial system and monetary transmission

within the structural VAR framework for a group of ten Central and Eastern European countries.

For the Czech Republic, they find a hump-shaped response of prices, an exchange rate

appreciation, and a fall in industrial production after a monetary policy innovation. Next,

financial structure is found to be of little importance for monetary transmission.

59 European Forecasting Network (2004)


4.3 Data

This section contains a description of our dataset. We restrict our sample to the data from 1998

onwards, i.e., since the inflation targeting framework was adopted by the Czech National Bank

(previously – until May 1997 – it had operated a fixed exchange rate regime). Our sample thus

spans from 1998:1 to 2006:5 at monthly frequency. While studies in this stream of literature often

employ quarterly data, given the length of our sample we decided to work at monthly frequency.

As a result, we have 101 observations. The source of our data is the CNB’s public database

ARAD (except for the output gap, which is only available internally within the CNB). The plots

of all the series are available in Appendix 1.

We use GDP, lgdpt, and the real-time output gap estimate, outputgaprealt, as measures of economic

activity.60 GDP is traditionally used for this kind of exercise, but Giordani (2004) suggests using

the output gap. In addition, by using the real-time output gap estimate we avoid the risk resulting

from the use of ex-post data, which are not available to central bankers at the time of monetary

policy formulation (Croushore and Evans, 2005). As GDP and the output gap are only available

at quarterly frequency, we interpolate these two using the quadratic-match average procedure.61

Note that all the other variables we use are not revised afterwards.

Next, we employ the net price index, lnett, (the net price index is the consumer price index

excluding regulated prices). For our disaggregate analysis, we employ the tradable price index,

tradablet, and the non-tradable price index, nontradablet. Note that the individual components

underlying the consumer price indices are grouped into tradables and non-tradables categories in

line with the internal CNB classification.

60 See Coats et al. (2003, chapter 5) on the construction of the output gap used by the CNB. The output gap is the difference between actual and potential output, where the latter is estimated by a multivariate filter, more specifically by the Kalman filter procedure, where the system of equations is in the state-space representation. 61 We admit that interpolation introduces information not available at the time of policy making.


Further, the nominal CZK/EUR exchange rate, lexratet, and the three-month interbank interest

rate (3M PRIBOR62), pribort, are used. To capture external developments, the 1-year EURIBOR,

euribort, and the commodity price index, lcommodityt, are utilized. The forward rate agreement rate

(9*12 FRA rate), frat, is used to bring in an additional forward-looking element. Given that there

are no futures or forwards in the Czech Republic that are directly linked to the monetary policy

rate (2W repo) as is the case in the U.S., we decided to use forwards on interbank rates, which are

very closely related to the policy rate. Finally, all data are in logs except interest rates and the real-

time output gap.

4.4 Identification

In this section, we discuss the VAR framework we adopt.63 The choice of variables for our VAR

model is largely motivated by an open economy New Keynesian model (see, for example, Gali

and Monacelli, 2005). The main equations of this class of models are aggregate demand, the

Phillips curve, the monetary policy rule, and uncovered interest rate parity.

We estimate two benchmark models and then undertake a sensitivity analysis. The difference

between these two benchmark models is that the first includes only the aggregate price index,

while the second distinguishes between the tradable and non-tradable price indices. The

specification of the first baseline model is the following:

ttptt uXLBYLAY ++= − )()( (1)

62 The actual monetary policy instrument of the CNB is the 2W repo rate. Since the repo rate is not changed continuously and is censored, we opt for the 3M PRIBOR, which is very closely linked to the 2W repo rate; its correlation stands at 0.998 in our sample. In addition, the 3M PRIBOR may capture central bank communication. See Horvath (2008) for a discussion related to the use of the monetary policy rate vs. the interbank market rate in the Czech Republic. 63 See Rudebusch (1998) and Sims (1998) on the discussion regarding appropriateness of VAR modelling for monetary policy.


where Yt and Xt represent endogenous and exogenous variables64, respectively. The data vectors

are: Yt = outputgaprealt, lnett, pribort, lexratet and Xt = euribort, lcommodityt, frat. For our second

benchmark specification: Yt = outputgaprealt, lnontradablet, tradablet, pribort, lexratet and Xt remains

the same.

The VAR specification in (1) represents a so-called reduced-form equation. In order to identify

the original shocks we can apply the recursiveness assumption by imposing restrictions on a

matrix linking the structural shocks to the reduced-form disturbances. The variables are ordered

in a specific way so as to represent the assumption that the monetary authorities choose the

interest rate taking into account the current level of prices and output (as in Mojon and

Peersman, 2001). In addition, the output gap and prices are assumed not to react immediately to

the monetary policy shock, but rather with a one-period lag. Mojon and Peersman (2001) follow a

recursive specification to analyze the impact of a monetary policy shock in some of the euro area

countries.

We analyze the sensitivity of our benchmark models first by using GDP instead of the output

gap, second by estimating a very parsimonious model without exogenous variables, and third by

estimating the baseline models by structural VAR instead of recursive VAR.

As regards the first sensitivity check, actual GDP data are used instead of the output gap. The

rationale for this exercise is that the output gap, as opposed to GDP, is unobservable. Our

second sensitivity check is motivated by degrees-of-freedom considerations. Here, we assume

that external shocks influence the Czech economy only via the exchange rate (i.e., B(L)=0).

Admittedly, this is a simplistic specification, but its main advantage is its limited number of

64 The inclusion of foreign variables that are considered exogenous is motivated by the need to control for foreign shocks and thus not to confuse domestic monetary shocks with the central bank’s responses to external developments (Jarocinski, 2006).


variables and thus its greater degree of freedom in comparison to our other models. As the third

robustness check, the two baseline models are estimated by structural VAR (SVAR). SVAR

represents an alternative identification scheme in order to recover the original residuals from the

reduced-form VAR. For structural VAR, we apply here the AB-model of Amisano and Giannini

(1997), which is defined as follows in a reduced form:

ttptt uXLBYLAY ++= − )()( ** (2),

tt BeAu 1−= , ( )Kt Ie ,0~ , where I is the identity matrix and K is the number of variables. A and

B are kk × matrices to be estimated. In the case of our first benchmark model, they are specified

as follows.

⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢

⎣

⎡

=

110

0010001

434241

3431

21

aaaaa

aA

⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢

⎣

⎡

=

44

33

22

11

000000000000

bb

bb

B

It follows from matrix A that a forward-looking monetary authority does not consider

contemporaneous prices while deciding on monetary policy (i.e., a32=0). However, monetary

authorities are likely to react to contemporaneous output (a31, as output can be regarded as an

excess demand pressure indicator) and exchange rate shocks (a34), which is a reasonable

assumption for small open economies according to Kim and Roubini (2000). More specifically,

exchange rate fluctuations influence the inflation forecast if they are deemed not to be transitory.

For our second benchmark model, in which we consider disaggregate prices (hence five

variables), matrices A and B look as follows:


⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢

⎣

⎡

=

1100

001000100001

54535251

4541

3231

21

aaaaaa

aaa

A

⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢

⎣

⎡

=

55

44

33

22

11

00000000000000000000

bb

bb

b

B

Following each VAR estimation, we perform stability checks in order to ensure the robustness of

our results (the results of these tests are available upon request). It is important to note that the

variables used in the VAR analysis do not need to be stationary. Sims (1980), among others,

argues against differencing even if the series contain a unit root. The main goal of the VAR

analysis is to analyze the co-movements in the data. What matters for the robustness of the VAR

results is the overall stationarity of the system (see Lütkepohl, 2006, for details).

A description of the FAVAR model is presented in Appendix 3.

4.5 Results

In this section, we discuss the estimated effects of Czech monetary policy within the

aforementioned specifications. The number of lags has been chosen according to the Schwartz

criterion and the parameter stability addressed by the CUSUM and CUSUM of squares tests and

the recursive coefficient estimation (the results are available upon request).

Figure 1 present our results regarding the effects of a contractionary monetary policy shock on

several economic variables of interest to a monetary authority. These figures contain the impulse

responses and the associated 95% confidence interval, which was bootstrapped using 1,000

replications according to the percentile method by Hall (1988).


Figure 1: Contractionary monetary policy shock, impulse responses

Notes: This figure shows the impulse responses to a one standard deviation contractionary monetary policy shock.

Time (on the horizontal axis) is measured in months.

We find that prices fall after a monetary tightening and bottom out after one year or so. This is in

line with the targeting horizon of the CNB, which is considered to be between 12 and 18

months. In terms of magnitude, our results show that a one Cholesky standard deviation of

interest rates (a 30 basis point monetary policy shock) decreases the log of prices by about 0.1%.65

Notably, there is essentially no evidence for a price puzzle.

The degree of economic activity, as measured by the output gap, falls after a contractionary

monetary policy shock, bottoming out after about four months (this, however, is not confirmed

in our sensitivity analysis, which identifies the bottom after about twelve months, which is more

sensible). The results indicate that a monetary shock of 30 basis points decreases the output gap

65 Several authors have raised the question of the accuracy of monetary policy shocks within VARs. See Boivin and Giannoni (2002) for a related discussion.


by about 5%. The responses of output and prices to a monetary shock show no support for the

cost channel of monetary policy.

Next, our results show a delayed overshooting in exchange rate behavior, i.e., a rather persistent

appreciation of the domestic currency after a monetary tightening (lasting typically about 6

months) and a gradual depreciation afterwards. However, it has to be pointed out that the

estimated confidence intervals are relatively wide, which brings some margins of uncertainty into

interpreting the results. Nevertheless, we can see that irrespective of specification and estimation

technique, the exchange rate depreciates over the longer term, which conforms to the uncovered

interest rate parity hypothesis (Kim and Roubini, 2000).

Figure 2: Contractionary monetary policy shock, impulse responses:

Tradable vs. non-tradable prices

Notes: This figure shows the impulse responses to a one standard deviation contractionary monetary policy shock.

Time (on the horizontal axis) is measured in months.


Figure 2 contains the estimates of the effect of monetary policy shocks on tradable and non-

tradable prices. Generally, tradable prices react faster than non-tradable prices to a monetary

contraction. While the bottom response of tradable prices is at one year or so (even 9–10

months), the bottom response of non-tradable prices occurs only after one and a half years. This

result matches the findings based on micro-level data (Alvarez and Hernando, 2006; Coricelli and

Horvath, 2006), which show that the frequency of non-tradable price changes is lower (and

negatively affected by the degree of competition); hence, a slower response to the monetary

policy shock is to be expected. On the other hand, the reaction of non-tradable prices is more

pronounced. A monetary shock of about 0.3% decreases tradable and non-tradable prices by

0.1% and 0.2%, respectively. In addition, the results in Figure 2 largely confirm the results of the

effect of monetary policy on output and the exchange rate from Figure 1.

Next, we analyze the sensitivity of our benchmark models, with all the results reported in

Appendix 2. First, we investigate how our results change when we include ex-post revised data

(GDP) instead of the real-time output gap in our data vector. Real-time variables are part of the

information set available at the time of policy-making, so by using these variables in the VAR

analysis we avoid the likely contamination of the results caused by data revisions.66 There is no

statistically significant reaction of GDP to the monetary shock and it seems that GDP does not

capture adequately the degree of demand pressures in an environment of sharply changing

potential output growth. Thus, our results stress the importance of using the real-time output gap

in the VAR specification, as it improves the precision of the empirical analysis.

Second, we estimate a very parsimonious model without exogenous variables, including the

forward-looking component. The rationale behind this is merely degrees-of-freedom

considerations. Interestingly, we find that a four-variable VAR is able to generate quite sensible 66 In general, the output gap should be a better measure of demand pressures (especially when potential output growth is changing), but one should keep in mind that it is unobservable and thus subject to greater uncertainty.


and precisely estimated impulse responses.67 This would suggest that economic agents during our

sample period form their expectations in a rather backward-looking manner. This is somewhat

surprising, but one has to consider the transition process of the Czech economy and the

corresponding greater uncertainty in economic development, which could make agents rely more

on current data than on forecasts.

Finally, we also estimate the benchmark models by structural VAR instead of recursive VAR, but

SVAR seems to provide little value added and typically generates impulse responses close to

those of VAR, but with much larger confidence intervals.

Next, we compare our results with other recent studies that analyze monetary policy shocks in

the Czech Republic within the VAR approach. The comparison is summarized in Table 1. Most

of the existing studies ignore the monetary policy regime change in the Czech Republic (the fixed

exchange rate regime until May 1997 and the adoption of inflation targeting in January 1998). In

consequence, it is not surprising that simple VAR methods have difficulty in identifying monetary

policy shocks across these two regimes; i.e., they do not deliver plausible results and all exhibit

the price puzzle (some of them even report a positive reaction of output to a monetary

tightening).68 This suggests that the price puzzle in these studies is associated with the monetary

policy regime change. This is further confirmed by two papers that employ data from the

inflation targeting period (Elbourne and Haan, 2006, and this paper), as their results do not

67 The output gap and prices fall after a contractionary monetary policy shock, bottoming out after about twelve months. The exchange rate first appreciates, but later depreciates significantly, in line with uncovered interest rate parity (see also Eichenbaum and Evans, 1995). The results on the reaction of tradable and non-tradable prices largely comply with the benchmark case, except that non-tradable prices reach their bottom response a bit later (about two years). 68 The exemption is Jarocinski (2006). His sample starts in June 1997, which is before the adoption of inflation targeting, but after the exchange rate turbulence and the abandonment of the fixed exchange rate regime. As a result, we code his sample in Table 1 as coming from a single monetary policy regime. Another approach to dealing with monetary policy changes is presented by Darvas (2005), who estimates a time-varying coefficient VAR. Indeed, his results suggest that the values of the estimated parameters change rather abruptly around the year 1997 and remain relatively stable afterwards. (This is confirmed in this study by the recursive estimation of the parameters. The results are available upon request.)


exhibit the price puzzle. Finally, the results in Table 1 indicate that the bottom responses of

output and prices seem to be at around 4 quarters, which is in line with our findings.

Table 1: Comparison to other VAR studies on monetary transmission in the Czech

Republic

Sample period

Single monetary policy regime Estimation technique

Reaction of output to MP shock

Reaction of prices to MP shock

Bottom reaction of output and prices

EFN (2004) 1994–2003 No VAR (-), sig. (+), sig. 6Q/--- Ganev et al. (2004) 1995–2000 No VAR (+), n.a. (+), n.a. ---- Creel and Levasseur (2005) 1993–2004 No SVAR (+), sig. (+), sig. ---- Darvas (2005) 1993–2004 No TVC-SVAR (-), n.a. n.a. 4Q/n.a.Héricourt (2005) 1995–2004 No VAR (-), sig. (+), sig. 1Q/--- Hurník and Arnoštová (2005) 1994–2004 No VAR insig. insig. 8Q/6Q Elbourne and Haan (2006) 1998–2004 Yes SVAR (-), sig. (-), sig. 4Q/4Q Jarocinski (2006) 1997–2004 Yes Bayesian VAR (-), sig. (-), sig. 4Q/4Q Gavin and Kemme (2007) 1995–2006 No SVAR (-), sig. (+), sig. ---- Anzuini and Levy (2007) 1993–2002 No VAR, SVAR (-), sig. insig. 4Q/8Q This paper 1998–2006 Yes VAR, SVAR, FAVAR (-), sig. (-), sig. 3Q/4Q

Note: (-) and (+) denote, respectively, a statistically significant decline and increase of the variable after a monetary policy shock. The column “Single monetary policy regime” indicates whether the sample period of the study comes from a single monetary regime or spans different regimes (the fixed exchange rate regime until May 1997 and the inflation targeting regime adopted in January 1998). Abbreviations: TVC-SVAR – time-varying coefficient SVAR, Sig. – the reaction of the variable to a monetary policy shock is statistically significant at the 5% level, and Q – quarters. If the reaction of the variable to a monetary shock does not have the correct sign, the bottom reaction of the variable is not reported (denoted as “----” in the table; n.a. indicates that the corresponding estimates were not available in the original study).

4.6 Concluding Remarks

In this paper, we analyze the transmission of monetary policy shocks in the Czech Republic

within the VAR, SVAR, and FAVAR frameworks. In general, monetary transmission in the

Czech Republic seems to be similar, in terms of persistence of the responses of economic

variables to monetary shocks, to that in more developed countries, including the euro area (see

e.g. Mojon and Peersman, 2001).


All in all, subject to various sensitivity tests, we find that prices and output decline after a

monetary tightening, with the bottom response occurring after about one year. This finding

corresponds with the actual targeting horizon of the Czech National Bank.69 In addition, we

document that the reaction of tradable prices is faster than that of non-tradable prices. While the

maximum effect of a monetary shock on tradables can be seen after a year or so, it is at least a

year and a half for non-tradable prices. This result broadly confirms the microeconomic evidence

on the effect of competition on price rigidity (Alvarez and Hernando, 2006; Coricelli and

Horvath, 2006). We avoid a price puzzle within the system. Thus, our results support the notion

that the price puzzle is associated with model misspecification rather than with the actual

behavior of the economy. This is also supported in other VAR studies on monetary transmission

in the Czech Republic, as all studies estimating the effects of monetary policy across different

monetary policy regimes (i.e., the fixed exchange rate regime and inflation targeting regime mixed

together) exhibit the price puzzle.

Next, there is a rationale for using the real-time output gap estimate instead of current GDP

growth, as using the former results in much more precise estimates. The impulse responses of

GDP to an interest rate shock are less precisely estimated, and thus our findings point to the

importance of real-time data in monetary policy analysis. Finally, our results also indicate a

persistent appreciation of the domestic currency after a monetary tightening (“delayed

overshooting”, Eichenbaum and Evans, 1995), although the confidence intervals are in this case

rather wide, with a gradual depreciation afterwards.

69 However, note that the targeting horizon (i.e., the horizon minimizing the loss function of the monetary authority) and the horizon at which the monetary policy impact is the most profound are not identical concepts. See Strasky (2005) for details.


References Alvarez, L. and I. Hernando (2006): Competition and Price Adjustment in the Euro Area, Working Paper No 0629, Bank of Spain. Amisano, G. and C. Giannini (1997): Topics in Structural VAR Econometrics, 2nd ed., Springer-Verlag, Berlin. Angeloni, I., A. Kashyap, and B. Mojon (2003): Monetary Policy Transmission in the Euro Area, Cambridge University Press. Anzuini, A. and A. Levy (2007): Monetary Policy Shocks in the New EU Members: A VAR Approach, Applied Economics 39, pp. 1147–1161. Barro, R. (1972): A Theory of Monopolistic Price Adjustment, Review of Economic Studies 39, pp. 17–26. Barth, M. J. and V. A. Ramey (2001): The Cost Channel of Monetary Transmission, 2001 NBER Macroeconomics Annual, pp. l99–239. Bernanke, B., J. Boivin, and P. Eliasz (2005): Measuring the Effects of Monetary Policy: A Factor-Augmented Vector Autoregressive (FAVAR) Approach, Quarterly Journal of Economics 120, pp. 387–422. Boivin, J. and P. Giannoni (2002): Assessing Changes in the Monetary Transmission Mechanism: A VAR Approach, Federal Reserve Bank of San Francisco Economic Policy Review, 8/1, pp. 97–111. Boivin, J., P. Giannoni, and I. Mihov (2007): Sticky Prices and Monetary Policy: Evidence from Disaggregated U.S. Data, NBER Working Paper No 12824. Boivin, J. and S. Ng (2006): Are More Data Always Better for Factor Analysis?, Journal of Econometrics 132, pp. 169–194. Bouakez, H., E. Cardia, and F. Ruge-Murcia (2005): The Transmission of Monetary Policy in a Multi-Sector Economy, University of Montreal Working Paper, Cahier de recherche CIREQ, 20/2005. Brissimis, S. and N. Magginas (2006): Forward-Looking Information in VAR Models and the Prize Puzzle, Journal of Monetary Economics 53, pp. 1225–1234. Christiano, L., M. Eichenbaum, and C. Evans (1999): Monetary Policy Shocks: What Have We Learned and To What End?, in: Taylor, J., Woodford, M., eds., Handbook of Macroeconomics. North Holland, pp. 65–148. Coats, W., D. Laxton, and D. Rose (2003): The Czech National Bank’s Forecasting and Policy Analysis System, Czech National Bank (available at www.cnb.cz). Coricelli, F., B. Égert, and R. MacDonald (2006): Monetary Transmission Mechanism: Gliding on a Wind of Change, BOFIT Discussion Papers No 8/2006, Bank of Finland.


Coricelli, F. and R. Horváth (2006): Price Setting Behavior: Micro Evidence on Slovakia, CEPR Discussion Paper No 5445. Creel, J. and S. Levasseur (2005): Monetary Policy Transmission Mechanisms in the CEECs: How Important are the Differences with the Euro Area?, OFCE Working Paper No 2. Croushore, D. and C. Evans (2006): Data Revision and the Identification of Monetary Policy Shocks, Journal of Monetary Economics 53, pp. 1135–1160. Darvas, Z. (2005): Monetary Transmission in the New Members of the EU: Evidence from Time-Varying Coefficient Structural VARs, ECOMOD 2005 conference, mimeo. Dedola, L. and F. Lippi (2005): The Monetary Transmission Mechanism: Evidence from the Industries of Five OECD Countries, European Economic Review 49, pp. 1543–1569. Dybczak, K., V. Flek, D. Hájková, and J. Hurník (2006): Supply-Side Performance and Structure in the Czech Economy (1995–2005), Czech National Bank Working Paper No 4/2006. Eichenbaum, M. and C. Evans (1995): Some Empirical Evidence on the Effects of Monetary Policy Shocks on Exchange Rates, Quarterly Journal of Economics 110, pp. 975–1010. Elbourne, A. and J. de Haan (2006): Financial Structure and Monetary Policy Transmission in Transition Countries, Journal of Comparative Economics 34, pp. 1–23. Erceg, C. and A. Levin (2006): Optimal Monetary Policy with Durable Consumption Goods, Journal of Monetary Economics 53, pp. 1341–1359. European Forecasting Network (2004): Monetary Transmission in Acceding Countries, Annex, pp. 97–142. Fry, R. and A. Pagan (2005): Some Issues in Using VARs for Macroeconometric Research, CAMA Working Papers Series No 19/2005, The Australian National University. Gali, J. and T. Monacelli (2005): Monetary Policy and Exchange Rate Volatility in a Small Open Economy, Review of Economic Studies 72, pp. 707–734. Ganev, G., K. Molnár, K. Rybinski, and P. Wozniak (2002): Transmission Mechanism of Monetary Policy in Central and Eastern Europe, CASE Reports No. 52. Gavin, W. and D. Kemme (2007): Using Extraneous Information to Analyze Monetary Policy in Transition Economies, FRB of St. Louis Working Paper No. 2004-034B. Gerlach, S. and F. Smets (1995): The Monetary Transmission Mechanism: Evidence From the G-7 Countries, CEPR Discussion Paper No. 1219. Giordani, P. (2004): An Alternative Explanation of the Price Puzzle, Journal of Monetary Economics 51, pp. 1271–1296. Hall, P. (1988): On Symmetric Bootstrap Confidence Intervals, Journal of the Royal Statistical Society B50, pp. 35–45.


Héricourt, J. (2005): Monetary Policy Transmission in the CEECs: Revisited Results Using Alternative Econometrics, University of Paris 1, Mimeo. Horváth, R. (2008): Time-Varying Policy Neutral Rate in Real Time: A Predictor for Future Inflation?, Economic Modelling, forthcoming. Hurník, J. and K. Arnoštová (2005): The Monetary Transmission Mechanism in the Czech Republic: Evidence from VAR Analysis, Czech National Bank Working Paper Series No 4/2005. Jarocinski, M. (2006): Responses to Monetary Policy Shocks in the East and the West of Europe: A Comparison, Austrian National Bank Working Paper No. 124. Kim, S. and N. Roubini (2000): Exchange Rate Anomalies in the Industrial Countries: A Solution with a Structural VAR Approach, Journal of Monetary Economics 45, pp. 561–586. Kotlán, V. and D. Navrátil (2003): Inflation Targeting as a Stabilization Tool: Its Design and Performance in the Czech Republic, Finance a úvěr – Czech Journal of Economics and Finance 53, pp. 220–242. Leeper, E., C. Sims, and T. Zha (1998): What Does Monetary Policy Do?, Brookings Papers on Economic Activity 2, pp. 1–63. Lütkepohl, H. (2006): New Introduction to Multiple Time Series Analysis, 2nd ed., New York: Springer. Martin, C. (1993): Price Adjustment and Market Structure, Economics Letters 41, pp. 139–143. Mojon, B. and G. Peersman (2001): A VAR Description of the Effects of Monetary Policy in the Individual Countries of the Euro Area, ECB Working Paper No 92. Peersman, G. and F. Smets (2001): The Monetary Transmission Mechanism in the Euro Area: More Evidence from VAR Analysis, ECB Working Paper No 91. Peersman, G. and F. Smets (2005): The Industry Effects of Monetary Policy in the Euro Area, Economic Journal 115, pp. 319–342. Rudebusch, G. (1998): Do Measures of Monetary Policy in a VAR Make Sense?, International Economic Review, 39(4), pp. 907-931. Strasky, J. (2005): Optimal Forward-Looking Policy Rules in the Quarterly Projection Model of the Czech National Bank, Czech National Bank Research and Policy Note, No. 5/2005. Sims, C. (1980): Macroeconomics and Reality, Econometrica 48, pp. 1–48. Sims, C. (1998): Comment on Glenn Rudebusch's "Do Measures of Monetary Policy in a VAR Make Sense?, International Economic Review, pp. 933-941. Woodford, M. (2003): Interest and Prices: Foundations of a Theory of Monetary Policy, Princeton University Press.


Appendix 1

Figure 3: Time series 3M PRIBOR

1Y Euribor

Output gap – real time

Log of GDP

Log of net price index

Log of CZK/EUR exchange rate

Log of tradable price index Log of non-tradable price index


Forward rate agreements Log of commodity price index


Appendix 2 – Additional Results:

Impulse Responses to a Monetary Shock

Figure 4: GDP instead of output gap

Figure 5: GDP instead of output gap, sectoral prices


Figure 6: No exogenous variables

Figure 7: No exogenous variables, sectoral prices


Figure 8: SVAR

Figure 9: SVAR, sectoral prices


Appendix 3 – Factor-Augmented VAR

In this Appendix, we briefly document our attempt to study monetary policy effects within

factor-augmented VAR (FAVAR). However, as documented below, we find that the results

based on FAVAR are very sensitive and the confidence intervals for the impulse responses are

rather large.

We follow an approach developed by Bernanke et al. (2005).70 FAVAR can be represented in the

following form:

FY L

FY vt

t

t p

t pt

⎡

⎣⎢

⎤

⎦⎥ =

⎡

⎣⎢

⎤

⎦⎥ +

−

−Φ ( )

Vector Yt contains observable economic variables, whereas Ft represents unobserved factors

which provide additional economic information not fully captured by the Yt. We estimate these

unobservable factors using a principal components approach, which exploits the assumption that

information about the unobservable economic factors can be inferred from a large number of

economic time series Xt. Specifically, we can think of the unobservable factors in terms of

concepts such as “economic activity” or “investment climate.” They can be represented not by a

single economic variable, but rather by several time series of economic indicators.

The FAVAR methodology allows us not only to use a richer information set in the model

specification, but also to analyze the effects of a monetary policy shock on a greater number of

economic variables. There are two main approaches to estimating FAVAR: a two-step principal

components approach and a one-step approach that estimates (3) and a dynamic factor model

70 We followed the algorithm developed by Bernanke et al. (2005) to estimate FAVAR, which is available on the personal website of Jean Boivin: http://www2.gsb.columbia.edu/faculty/JBoivin/Personal/


jointly. As Bernanke et al. (2005) do not find any particular differences between these two

estimators in terms of inference, we opt for the computationally simpler two-step approach.71

In our FAVAR specification, Xt consists of a balanced panel of 40 series that have been

transformed in order to ensure their stationarity. The description of these series and their

transformations is included in Table 2. The data is at monthly frequency and spans the period

from February 1998 to May 2006. Following Bernanke et al. (2005), we assume that the monetary

policy instrument (the 3-month interest rate) is the only observable factor, hence the only

variable included in Yt. For identification purposes the monetary policy instrument is ordered

last, which implies that latent factors do not respond contemporaneously (within a month) to

innovations in monetary policy. As in Bernanke et al. (2005), we distinguish between “slow-

moving” and “fast-moving” variables. A “slow-moving” variable is assumed not to react

contemporaneously to shocks, while the “fast-moving” variables react instantaneously to changes

in monetary policy or economic conditions. The classification of the variables into these two

categories is included in Table 2.

In the first step of the two-step estimation, we can distinguish three stages. First, we use principal

component analysis to estimate the common factors Ct from all the variables in Xt. Second, after

dividing the series in Xt into slow- and fast-moving ones, we estimate the “slow-moving” factors

stF)

as the principal components of the “slow-moving” variables. Finally, we estimate the

following regression:

tetYstFst YbFbC ++=))

71 See Bernanke et al. (2005) for a more detailed discussion of principal component analysis and FAVAR.


Based on these estimates, tF)

is constructed as tYt YbC))

− . In the second step, we estimate the

VAR in tF)

and Yt, using a recursive assumption.

One caveat in our analysis is the fact that we have only 40 series available for principal

component analysis, as compared to the 120 used in Bernanke et al. (2005). While this may be

viewed as a weakness at first sight, it has been argued by Boivin and Ng (2006) that more series

do not necessarily ensure better data quality, due to cross-correlation of idiosyncratic errors. In

one of their tests, they show that the factors extracted from 40 pre-screened series may in some

cases yield better results as compared to using 147 series. Therefore, for the sake of size, 40

series, at least in general, should not pose a problem.

Our main results are given in Figure 9. Each panel shows the impulse responses of selected

macroeconomic variables to a monetary policy shock with 90% confidence intervals. The

FAVAR model in Figure 9 includes three principal factors, but the results were no different when

the number of factors was changed. In the benchmark specification we use one lag. The results

are highly sensitive to the numbers of lags used, with more lags resulting in highly improbable

results.

As a result, the FAVAR model does not appear to properly capture the developments in the

Czech economy. Most importantly, the confidence intervals are too large to infer the direction of

the impact of the monetary policy change on the macroeconomic variables. The exception is

actual GDP growth, lgdpt, which declines after monetary policy shock, as predicted by the theory.

There may be several reasons for the lack of significant results in the FAVAR estimation for the

Czech economy. One reason is likely to do with the relatively short span of the available data;


another may be data quality, as discussed by Boivin and Ng (2006). As it is at monthly frequency,

our dataset lacks variables related to consumption, housing starts and sales as well as real

inventories and therefore some important economic information may be missing.

Figure 9: FAVAR results

Note: Impulse responses with 90% confidence intervals are presented.

Table 2 – Data description

VAR DESCRIPTION TRANSFORMATION SOURCE

Real output and income var1* Industrial Production, Index number (sa) 3 IFS var10* Construction output, constant prices - % (sa) 1 ARAD

var11* Contracted construction work in enterprises with 20 employees or more - constant prices - (%) (sa) 1 ARAD

var20* Outputgap real - interpolated from quarterly values 2 ARAD var21* GDP - interpolated from quarterly values (sa) 3 ARAD var30 Total agricultural goods output (sa) 3 ARAD Employment and Hours var2* Industrial Employment (sa) 3 IFS var3* Unemployment Rate (sa) 1 Eurostat


var12* Registered job applicants, total (thousand persons, sa) 1 ARAD var13* Vacancies (thousand, sa) 3 ARAD var14* Newly registered job applicants (thousand persons, sa) 3 ARAD var15* Registered job applicants on unemployment benefit (thousand persons, sa) 3 ARAD Industry Sales

var6* Total sales revenues, Index sales in industry-constant price (corresponding period of preceding year=100, sa) 3 ARAD

var7* Mining and quarrying, Index sales in industry-constant price (corresponding period of preceding year=100, sa) 1 ARAD

var8* Manufacturing, Index sales in industry-constant price (corresponding period of preceding year=100, sa) 1 ARAD

var9* Electricity, gas and water supply, Index sales in industry-constant price (corresponding period of preceding year=100, sa) 1 ARAD

Exchange Rates var22 Foreign Exchange Rate (Czech Krown per Euro) 3 IFS var23 Foreign Exchange Rate (Czech Krown per U.S. $) 3 IFS Interest Rates var26 Treasury Bill Rate 3 IFS var27 Deposit Rate 3 IFS var28 Lending Rate 3 IFS var29 Government Bond Yield 3 IFS var31 1 day Interbank Rate PRIBOR (%) 1 ARAD var32 7 day Interbank Rate PRIBOR(%) 1 ARAD var33 14 day Interbank Rate PRIBOR(%) 1 ARAD var34 1 month Interbank Rate PRIBOR(%) 2 ARAD var35 2 month Interbank Rate PRIBOR (%) 2 ARAD var36 6 month Interbank Rate PRIBOR(%) 2 ARAD var37 9 month Interbank Rate PRIBOR (%) 2 ARAD var38 1 year Interbank Rate PRIBOR(%) 2 ARAD var41 3 month Interbank Rate PRIBOR(%); monetary policy instrument 1 ARAD Money Aggregates var24 Money (sa) 3 IFS var25 Money plus Quasi Money (sa) 3 IFS Price Indexes var16* Consumer Prices CPI (sa) 3 ARAD var17* Industrial Produces Prices (sa) 3 ARAD var18* Tradable prices (sa) 3 ARAD var19* Nontradable prices (sa) 3 ARAD

var40 Prague Stock Exchange Index PX50, Historical close, average of observations through period 3 IFS

Exports and Imports var4* Exports 3 IFS var5* Imports, FOB 3 IFS


All series were tested for unit root and when necessary were transformed to achieve stationarity. The transformation codes are: 1-no transformation, 2-first difference, and 3-first difference of logarithm. An asterisk (*) next to the mnemonic indicates a variable assumed to be “slow-moving” in the estimation.

phd dissertation horvath

Documents

federal reserve

filter tipped

global development

policy neutral

puts downward

european economic

slovak statistical

monetary policy