Journal of Macroeconomics 32 (2010) 1118–1130

Contents lists available at ScienceDirect

Journal of Macroeconomics
journal homepage: www.elsevier.com/locate/jmacro

Understanding changes in exchange rate pass-through

Yelena F. Takhtamanova *
Federal Reserve Bank of San Francisco, USA

a r t i c l e i n f o a b s t r a c t

Article history: Recent research suggests that there has been a decline in the extent to which firms ‘‘pass-
Received 4 August 2008 through” changes in exchange rates to prices. This paper provides further evidence in sup-
Accepted 13 April 2010 port of this claim. Additionally, it proposes an explanation for this phenomenon. The paper
Available online 24 April 2010
then presents empirical evidence of a structural break during the 1990s in the relationship
between the real exchange rate and CPI inflation for a set of fourteen OECD countries. It is
JEL classification: suggested that the recent reduction in the real exchange rate pass-through can in part be
E31
attributed to the low inflationary environment of the 1990s.
E52
F41
Ó 2010 Elsevier Inc. All rights reserved.

Keywords:
Pass-through
Inflation
Exchange rate
Monopolistic competition
Staggered price setting

1. Introduction

In open economies, exchange rate fluctuations affect the behavior of inflation. This makes the exchange rate pass-through
(defined as the effect of exchange rate changes on domestic inflation) an important consideration with respect to monetary
policy. Several authors claim that even considerably large changes in exchange rates during the 1990s had a surprisingly
weak effect on the domestic inflation in some small open economies.1 This puzzlingly weak effect of the exchange rate on
domestic inflation can possibly be explained by the presence of other forces; for instance, deviations of output from its natural
level. Nevertheless, it has been suggested that the recently observed reduction in the exchange rate pass-through is too large
and pervasive to be explained by special factors (Taylor, 1999).
The literature on exchange rate pass-through is voluminous.2 It is also important to note that there is no uniform definition
of the term ‘‘pass-through”. Much of the existing research focuses on the relationship between movements in nominal exchange
rates and import prices. A smaller, but equally important strand of the literature, concentrates on the macroeconomic exchange
rate pass-through to aggregate price indices (Bacchetta and van Wincoop, 2003; Campa and Goldberg, 2005; Gagnon and Ihrig,
2004). This paper too focuses on the relationship between aggregate prices and inflation.

* Address: 101 Market Street, MS 101, San Francisco, CA 94105, USA.

E-mail address: Yelena.Takhtamanova@sf.frb.org
1
Cunningham and Haldane (2000) show evidence of a reduced pass-through in the UK, Sweden and Brazil. Garcia and Restrepo (2001) discuss the low pass-
through in Chile. McCarthy (2007) studies exchange rate pass-through in nine developed countries, and concludes that exchange rates ‘‘have not assumed a
bigger role in domestic consumer price inflation in recent years, and may even have had a smaller role.”
2
For a survey of the literature on the exchange rate pass-through to import prices, see Goldberg and Knetter (1997). Examples of studies done on the pass-
through to domestic producer and consumer prices are Woo (1984), Feinberg (1986, 1989), McCarthy (2000).

0164-0704/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jmacro.2010.04.004
 Y.F. Takhtamanova / Journal of Macroeconomics 32 (2010) 1118–1130 1119

This analysis differs from other work showing a decline in exchange-rate pass-through as it combines three important
features. First, it systematically uses cross-country evidence, which is more informative than single country studies. Second,
by utilizing a Phillips curve framework, it allows to separate changes in exchange rate pass-through from other forces influ-
encing inflation behavior. Lastly it focuses on the link between real exchange rate and inflation, which is a key effect in
macroeconomics.
The remainder of this paper is organized as follows. The next section discusses the empirical approach. I introduce
the open economy Phillips curve to be estimated and discuss the two parameters of focus for this project: the
impact effect of real exchange rate changes on inflation (short-run pass-through) and the total effect (long-run pass-
through). I discuss possible determinants of exchange rate pass-through (which include the degree of openness in the
economy, the fraction of flexible-price firms in the economy, the credibility of a central bank, and the monopoly power
of firms).
As previously mentioned, the sample consists of fourteen OECD countries. The data come from the Organization for Eco-
nomic Co-operation and Development (OECD) databases. To allow for the correlation of residuals across countries, a seem-
ingly unrelated regression model is estimated. Average Phillips curve coefficients and average dynamic inflation responses to
a real exchange rate shock over time are then compared. This methodology differs from those previously utilized.3 The
advantage of using this technique is that it allows us to reach some general conclusions about the recent change in the nature
of the relationship between the exchange rate and the CPI inflation, rather than concentrating on individual country results
(which are not informative due to the fact that only short samples are available for each individual country).
Section 3 describes the results of the empirical investigation. Two crucial points are made. Firstly, that, for the set of coun-
tries investigated as a group, there is evidence indicating that a reduction in short- and long-run exchange rate pass-through
indeed took place during the 1990s. Secondly, the change in the long-run relationship between the exchange rate and infla-
tion was caused by a reduction in the fraction of flexible-price firms in the economy. Section 4 concludes the paper.

2. Empirical approach

In this section, the empirical model used in this study is presented. An explanation of the proposed estimation method is
then given. Finally, the data used is described.

2.1. Empirical model

The purpose of this study is to analyze the effect of exchange rate changes on domestic inflation. Exchange rate changes
affect domestic inflation through prices of imports.4 Indeed, domestic inflation is the change in the overall cost of the goods
and services bought by a typical consumer. These goods include those produced abroad and sold domestically (imports), and
prices of such goods fluctuate with real exchange rates. In particular, domestic currency depreciation tends to lead to an in-
crease in the prices of imports.
As is well documented in the import price pass-through literature, the relationship between import prices and exchange
rate needs not be one-to-one. First, the strength of this relationship depends on the monopoly power of firms.5 Also, it de-
pends on the fraction of the firms that adjust their prices every period (I refer to such firms as flexible-price firms).6 All other
things equal, a higher share of flexible-price firms would lead to a larger exchange rate pass-through (as more firms would react
to an exchange rate shock in the period when it occurs).
While the literature on the relationship between import prices and exchange rates is vast, less attention has been paid to
the relationship between consumer price inflation and exchange rates. In addition to the factors just discussed, this relation-
ship also depends on the share of imports in the typical consumer basket (I refer to this as the degree of openness in the
economy). All other things being equal, higher share of imports would lead to a larger response of aggregate inflation to real
exchange rate fluctuations.
Exchange rates are not the only force affecting inflation. For instance, inflation is affected by the degree of slack in the
economy. This is an important point to remember when analyzing exchange rate pass-through – failure to control for the
degree of slack in the economy might lead a researcher to mistake disinflationary effects of recessions for episodes of de-
clines in pass-through (or inflationary effects of booms for episodes of increases in pass-through). Thus, an open economy
Phillips curve appears to be a model well suited for pass-through analysis.
I estimate an open economy Phillips curve of the following form7:

3
Previously utilized estimation techniques include ordinary least squares, panel data, vector autoregression, cointegration analysis and error correction
models, as well as state-space models (Garcia and Restrepo, 2001).
4
As this focus of this paper is my empirical work, I only present intuition behind the relationship between variables of interest. A formal model and detailed
derivation of all the results can be found in Takhtamanova (2008).
5
It can be shown (Takhtamanova, 2008) that the degree to which firms pass changes in exchange rate onto inflation depends on elasticities of demand and
cost functions faced by the good producing firms. Changes in these elasticities can be interpreted as changes in the monopoly power of the firm.
6
These firms differ from the so-called fixed prices firms who update their prices less frequently and set their prices equal to the expected optimal price.
7
See Takhtamanova (2008) for a formal derivation of an open economy Phillips curve that serves as a base to the empirical model presented here.
1120 Y.F. Takhtamanova / Journal of Macroeconomics 32 (2010) 1118–1130

X
4 X
4 X
4 X
4
pjt ¼ cj0 þ cpji pjti þ cyji y
~t þ cqji Dqjti þ czji Dzti þ et : ð1Þ
i¼1 i¼1 i¼1 i¼1

In the above equation, p denotes inflation, y ~ is the measure of slack in the economy (this is commonly measured by out-
put gap – the deviation of actual real GDP from potential), q denotes real exchange rate, z denotes supply shock variables, j
indexes countries, and cpi , cyi , cqi , and czi are constants. All variables are in logs.
This model is consistent with the existing open economy Phillips curve literature. As in Ball (1999), Svensson (1998), Gor-
don (1997) and Staiger et al. (1997), this model allows for inflation inertia. Mankiw and Reis (2002) point out that a Phillips
curve must allow for inflation inertia if it has any hope of explaining the data. As in Ball (1999), Svensson (1998), and Razin
and Yuen (2001), the inflation rate is influenced by the real exchange rate. In particular, a real appreciation of domestic cur-
rency puts a downward pressure on the inflation rate. Unlike some of the existing open economy Phillips models, the one
employed here does not assume a complete pass-through (i.e. the coefficient on the real exchange rate growth rate is not
set to 1).
The initial impact of exchange rate changes on inflation is measured by cq1 (I refer to it as the short-run exchange rate
pass-through). Also, it is possible to calculate the total impact of exchange rate on inflation (this is typically referred to as
the long-run exchange rate pass-through). The long-run pass-through coefficient can be derived from (1) as8:
P4 q
i¼4 c ji
uLR
i ¼ P4 p : ð2Þ
1  i1 ci;j

2.2. Estimation methodology

Instead of estimating an open economy Phillips curve given by Eq. (1) for each country in the sample, I estimate the equa-
tions as a system. There appears to be no justification for restricting the coefficients in the system in any way. There is no
reason to assume that the Phillips curve coefficients are the same for all the countries in the sample. It seems plausible, how-
ever, to allow for a correlation of the residuals across countries. Since the equations in the system are linked only by their
disturbances, the resulting model is a seemingly unrelated regression model (often referred to as SUR).
Though the covariance matrix of the disturbances is unknown, feasible generalized least squares (FGLSs) estimators are
devised. The system is estimated in two steps. The least squares residuals are used to estimate consistently the elements of
the variance–covariance matrix of the residuals. With such estimate in hand, the FGLS can then proceed as usual to estimate
the system (Greene, 1997).
The procedure generates open economy Phillips curve estimates for all the countries in the sample. While the individual
country coefficient estimates are presented, the discussion focuses on average coefficients across countries. The main reason
for utilizing such an approach is that the goal of this paper is to address very recent changes in the structure of the open
economy Phillips curve. As a result, only short samples for each individual country are available. With such limited data,
individual country estimation results are not especially informative. On the other hand, the average coefficient estimates
are more precise. This approach also allows for some general conclusions to be made about the structural break in the open
economy Phillips curve during the 1990s. To be specific, the average Phillips curve is

X
4 X
4 X
4 X
4
pt ¼ c0 þ cpi pti þ cyi y
~t þ cqi Dqti þ czi Dzti ; ð3Þ
i¼1 i¼1 i¼1 i¼1

P P
where c0 ¼ 1n nj¼1 cxj0 , cxi ¼ 1n nj¼1 cxij , x = p, y, q or z.
The average short-term exchange rate pass-through is given by cq1 (for consistency of notation, I will refer to it as uSR),
while the average long-run pass-through is given by
P4 q
i¼4 c i
uLR ¼ P4 p ; ð4Þ
1  i1 ci

where cqi and cpi are average Phillips curve coefficients as shown in Eq. (3).
Since the long-run pass-through is a nonlinear combination of estimated coefficients, the computations of standard errors
is worth a special discussion. These standard errors are obtained using the delta method. STATA now allows one to compute
delta-method standard error using the ‘‘nlcom” command. For those without access to software that allows for computations
of the standard error for a nonlinear combination of underlying model parameters Papke and Wooldridge (2004) propose a
method for obtaining the delta-method standard error that can be used by those who have software supporting only basic
regression analysis.

8
See derivation of the long-run multiplier in a textbook discussion of the distribution of autoregressive distributed lag model (for instance, see Greene, 1997
or Wickens and Breusch, 1988).
 Y.F. Takhtamanova / Journal of Macroeconomics 32 (2010) 1118–1130 1121

2.3. Data

This study considers 1980–2007 time period and uses data from 14 OECD countries: Australia, Canada, Denmark, Finland,
France, Ireland, Italy, Japan, Netherlands, New Zealand, Norway, Sweden, the United Kingdom, and the United States. The
sample begins with the first quarter of 1980 in order to exclude the Great Inflation of the 1970s. The countries were chosen
based on data availability (I chose only those countries for which data for all variables of interest was available starting in
1980). The data are quarterly.
The series employed are those for the rate of inflation, the output gap, the real exchange rate growth rate, and the growth
rate for the prices of food and energy. Data comes from OECD, and Appendix A provides details on the specific databases
used. I compute the rate of inflation as the annualized growth rate of the Consumer Price Index (i.e. inflation is defined
as pt ¼ 400  ðln P t  ln P t1 Þ; where P is CPI). Output gap is the difference between actual and potential gross domestic prod-
uct (GDP) as a per cent of potential GDP (I use OECD estimates of potential output). Supply shock variables are the growth
rate in the relative prices of food and energy (computed as 100  ðln P t  ln Pt1 Þ  100  ðln Pct  ln Pct1 Þ, where P is CPI and Pc
is core CPI).
For the real exchange rate, I use two alternative series. The first one is the real effective exchange rate series computed by
the OECD. The effective exchange rate index is a chain-linked index. Percentage changes in the index are calculated by com-
paring the change in the index based on consumer prices for the country concerned (expressed in US dollars at market ex-
change rates) to a weighted average of changes in its competitors’ indices (also expressed in US dollars), using the weighting
matrix of the current year. The indices of real effective exchange rates are then calculated from a starting period by cumu-
lating percentage changes. This gives a set of real effective exchange rates based on moving weights.
To check the robustness of my estimation results with respect to the choice of the real exchange rate data, an alternative
EN P
measure of the real exchange rate is also computed. The real exchange rate is defined as Q t ¼ tP t , where P is price level of the
t
* N
country in question (measured by CPI), P is the US CPI, and E is the nominal exchange rate between the currency of the
country under consideration and the US dollar. In other words, I compute the real exchange rate between the currency of
the country in question and the US dollar. In comparison to the real effective exchange rate series, this is a narrower measure
of the real exchange rate, which makes it perhaps less appealing theoretically. On the other hand, given that no measure of
real exchange rate is perfect (due to the ambiguity about the best use of deflators, currencies to go into the index and the
weights given to those currencies), it seems useful to do a robustness check with a simple yet transparent measure.
The study compares the relationship between inflation and the real exchange rate growth over two decades: the 1980s
and the 1990s. Two sub-samples are used – the first runs from 1980:1 to 1989:4, and the second from 1990:1 to 2007. Sum-
mary statistics for CPI inflation, output gap and the real exchange rate growth rate are summarized in Table 1. For all the
countries in the sample, average inflation in the first sub-period was higher than in the later period. With the exception
of Netherlands, the difference between the means is significant at a 1% level. Moreover, the later sub-period was also marked
by lower inflation volatility (with the exception of Sweden and Australia, variance of inflation was significantly higher in the
first period). For most countries, average gap was negative (indicating that actual output was below potential) for both peri-
ods (output gap is positive only for Finland in period 1). In terms of output gap and real exchange rate variability, the results
are mixed. For Canada, Denmark, Finland, Italy, Netherlands, UK and the US, variance of output gap was significantly lower in
the second period. For Ireland and New Zealand, the opposite is true. Real exchange rate growth rate series exhibited

Table 1
Summary statistics.

Country Inflation Output gap Exch. rate growth

Period 1 Period 2 Period 1 Period 2 Period 1 Period 2
l r l r l r l r l r l r
Australia 7.97 3.04 2.61 2.53 1.23 2.00 1.24 2.18 0.17 4.84 0.03 3.22
Canada 6.10 3.19 2.15 2.24 0.76 2.99 0.75 2.15 0.27 1.51 0.17 2.31
Denmark 6.31 3.71 2.01 1.57 0.26 2.37 0.43 1.31 0.02 1.95 0.07 1.29
Finland 6.83 3.55 1.80 2.07 1.09 2.24 3.00 4.76 0.52 1.81 0.50 2.43
France 6.74 4.22 1.76 1.28 1.44 1.51 0.86 1.39 0.19 1.70 0.03 1.10
Ireland 8.36 7.08 2.93 2.17 2.04 2.37 0.46 3.27 0.29 2.49 0.16 2.11
Italy 10.04 5.48 3.24 1.75 0.93 2.16 0.66 1.48 0.49 1.42 0.14 2.41
Japan 2.36 3.23 0.45 2.31 1.12 1.80 0.05 2.25 0.84 4.61 0.19 4.55
Netherlands 2.68 3.05 2.31 1.43 1.08 1.85 0.02 2.03 0.26 1.81 0.09 1.48
New Zealand 10.80 6.22 2.15 1.78 0.56 1.56 0.48 2.32 0.12 4.68 0.14 3.02
Norway 7.89 3.87 2.13 2.52 1.65 2.35 0.90 3.13 0.27 1.59 0.07 2.04
Sweden 7.45 3.95 2.31 3.84 0.71 2.38 1.95 2.42 0.11 2.80 0.31 2.83
UK 6.36 5.01 2.43 3.15 2.12 3.57 0.70 1.45 0.15 4.23 0.17 2.64
US 5.05 3.49 2.79 1.70 1.22 2.58 0.65 1.34 0.07 3.20 0.06 2.09
Average 6.78 2.22 0.92 0.79 0.13 0.05

Period 1 refers to 1980:Q1–1989:Q4. Period 2 refers to 1990:Q1–2007:Q2.

l denotes mean, r denotes standard deviation.
1122 Y.F. Takhtamanova / Journal of Macroeconomics 32 (2010) 1118–1130

significantly higher volatility in the first sub-period for Australia, Denmark, France, New Zealand, UK and the US and signif-
icantly higher volatility in the second sub-period for Canada, Finland and Italy.

3. Results

This section presents the evidence supporting the claim that a structural break in the relationship between inflation and
real exchange rate occurred during the 1990s. Some explanations for this phenomenon are then investigated.

3.1. Evidence of a decline in the real exchange rate pass-through

As discussed in Section 2.3, the system is estimated for two sub-periods – 1980:1–1989:4 and 1990:1–2007:2. For each
sub-period, the system is estimated twice; once for each real exchange rate series. Table 2 shows the sums of estimated coef-
ficients on inflation, output gap, and real exchange rate (Table 2a shows results obtained with OECD real exchange rate and
Table 2b shows results obtained with the self-constructed real exchange rate). By observing these estimation results, one can
reach a general conclusion that inflation responsiveness to changes in any of the dependent variables has declined during the
1990s. More than half the countries in the sample experienced a decline in inflation persistence (measured by the sum of the
coefficient on the lagged inflation terms). Regardless of the real exchange rate series used in estimation, the decline in infla-
tion persistence was statistically significant for five countries out of the sample. Most of the countries also experienced a
decline in responsiveness to exchange rate shocks (measures by the sum of coefficients on the real exchange rate terms).

Table 2a
Sums of Estimated Coefficients (Real Exchange Rate Data Source: OECD). Standard errors in parenthesis.
P4 pp P4 y P4 q P4
pt ¼ c0 þ i¼1 c i ti þ i¼0 c i ðyti  yN
ti Þ þ i¼0 c i Dqti þ z
i¼1 c i Dzti þ et ;

Estimates of the Sums of coefficients

P3 p P3 y P3 q
i¼1 c i i¼0 c i i¼0 c i

Period 1 Period 2 Difference Period 1 Period 2 Difference Period 1 Period 2 Difference

Australia 0.10 0.26 0.36 1.93*** 0.16 1.77*** 0.44 0.22 0.22
(0.16) (0.16) (0.23) (0.19) (0.14) (0.24) -(4.49) (0.14) (4.49)
Canada 1.24*** 0.09 1.33*** 0.20** 0.39*** 0.19 0.42** 0.32* 0.10
(0.15) (0.22) (0.26) (0.08) (0.12) (0.15) (0.18) (0.16) (0.24)
Denmark 1.29** 0.16 1.45*** 0.77 0.46*** 0.31 1.67** 0.28** 1.39*
(0.51) (0.18) (0.54) (0.58) (0.12) (0.60) (0.72) (0.14) (0.73)
Finland 0.78*** 0.58 0.20 0.42*** 0.07 0.35** 0.68** 0.26** 0.42
(0.11) (0.58) (0.59) (0.16) (0.05) (0.17) (0.27) (0.12) (0.30)
France 0.73*** 0.30* 0.43** 0.75*** 0.21 0.54 0.50 0.23 0.74
(0.12) (0.16) (0.20) (0.26) (0.21) (0.34) (0.40) (0.23) (0.46)
Ireland 0.58*** 0.59*** 0.01 0.82** 0.16** 0.66 0.72 0.58*** 0.14
(0.14) (0.17) (0.22) (0.40) (0.06) (0.41) (0.53) (0.14) (0.55)
Italy 0.57*** 0.85*** 0.28** 0.37** 0.08 0.30* 0.79 0.19*** 0.59
(0.10) (0.06) (0.12) (0.15) (0.07) (0.16) -(0.79) (0.06) (0.79)
Japan 0.23 0.23 0.46 0.26 0.36** 0.10 0.23** 0.12* 0.11
(0.41) (0.27) (0.49) (0.25) (0.15) (0.30) (0.11) (0.06) (0.13)
Netherlands 1.31*** 0.21 1.10*** 0.42*** 0.36*** 0.06 1.36*** 0.36*** 0.99***
(0.22) (0.21) (0.31) (0.16) (0.07) (0.17) (0.26) (0.13) (0.29)
New Zealand 0.83*** 0.35*** 0.47*** 1.61*** 0.10 1.51*** 1.20*** 0.03 1.17***
(0.12) (0.12) (0.17) (0.46) (0.11) (0.48) (0.28) (0.13) (0.31)
Norway 0.73*** 0.22 0.51 0.63*** 0.05 0.58*** 0.42 0.09 0.50
(0.18) (0.28) (0.33) (0.20) (0.09) (0.21) -(0.98) (0.23) (1.01)
Sweden 0.74*** 0.61*** 0.13 0.27 0.23 0.04 0.43 0.27 0.17
(0.27) (0.14) (0.30) (0.25) (0.17) (0.30) (0.29) (0.22) (0.36)
UK 0.78*** 0.78*** 0.00 0.38* 0.34*** 0.03 0.50** 0.09 0.60***
(0.21) (0.10) (0.24) (0.22) (0.09) (0.24) (0.21) (0.08) (0.22)
US 0.86*** 0.65*** 0.21 0.10 0.22* 0.12 0.15 0.28** 0.13
(0.13) (0.17) (0.22) (0.10) (0.12) (0.16) 0.14 0.13 (0.19)

Period 1 refers to 1980:Q1–1989:Q4. Period 2 refers to 1990:Q1–2007:Q2.

*
Indicate that the sum of coefficients is significantly different from zero at the 10% level.
**
Indicate that the sum of coefficients is significantly different from zero at the 5%.
***
Indicate that the sum of coefficients is significantly different from zero at the 1% level.
 Y.F. Takhtamanova / Journal of Macroeconomics 32 (2010) 1118–1130 1123

Table 2b
Sums of estimated coefficients (real exchange rate data source: self-constructed). Standard errors in parenthesis.
P3 pp P3 y P3 q P3
pt ¼ c0 þ i¼1 c i ti þ i¼0 c i ðyti  yN
ti Þ þ i¼0 c i Dqti þ z
i¼1 c i Dzti þ et ;

Estimates of the Sums of coefficients

P3 p P3 y P3 q
i¼1 c i i¼0 c i i¼0 c i

Period 1 Period 2 Difference Period 1 Period 2 Difference Period 1 Period 2 Difference

Australia 0.01 0.40** 0.38* 1.42*** 0.28** 1.14*** 0.99*** 0.14 0.84***
(0.15) (0.16) (0.22) (0.18) (0.14) (0.23) (0.21) (0.19) (0.29)
Canada 1.04*** 0.06 1.10*** 0.17* 0.32** 0.15 0.06 0.01 0.05
(0.16) (0.22) (0.27) (0.09) (0.13) (0.15) (0.07) (0.07) (0.11)
Denmark 0.27 0.18 0.10 1.48** 0.51*** 1.99*** 0.51 0.10 **
0.61*
(0.51) (0.18) (0.54) (0.71) (0.12) (0.72) (0.32) (0.04) (0.32)
Finland 0.50*** 0.47*** 0.03 0.32** 0.08* 0.24 0.28** 0.13 **
0.15
(0.16) (0.13) (0.21) (0.17) (0.04) (0.17) (0.11) (0.06) (0.13)
France 0.70*** 0.19 0.51*** 0.47** 0.20** 0.27 0.33** 0.07 0.41***
(0.10) (0.17) (0.20) (0.22) (0.09) (0.24) (0.15) (0.05) (0.15)
Ireland 0.37** 0.36** 0.01 1.64*** 0.18*** 1.45*** 0.11 0.22 ***
0.12
(0.17) (0.17) (0.24) (0.37) (0.06) (0.37) (0.23) (0.07) (0.24)
Italy 0.54*** 0.86*** 0.33*** 0.31** 0.02 0.29** 0.26** 0.00 0.26**
(0.10) (0.06) (0.11) (0.13) (0.07) (0.15) (0.11) (0.03) (0.11)
Japan 0.05 0.24 0.19 0.37 0.35** 0.02 0.14* 0.09 0.04
(0.37) (0.27) (0.46) (0.23) (0.15) (0.28) (0.08) (0.06) (0.10)
Netherlands 1.28*** 0.07 1.20*** 0.75*** 0.35*** 0.40** 0.42*** 0.14 ***
0.29***
(0.23) (0.22) (0.32) (0.17) (0.08) (0.18) (0.08) (0.05) (0.09)
New Zealand 0.61*** 0.32*** 0.29* 1.89*** 0.19 1.70*** 0.42** 0.02 0.41**
(0.12) (0.12) (0.17) (0.51) (0.09)** (0.52) (0.17) (0.08) (0.19)
Norway 0.69*** 0.03 0.66* 0.03 0.06 0.09 0.63*** 0.10 0.73***
(0.22) (0.27) (0.34) (0.31) (0.08) (0.32) (0.22) (0.11) (0.24)
Sweden 0.35* 0.56*** 0.20 0.08 0.18 0.26 0.38*** 0.04 0.34*
(0.20) (0.13) (0.24) (0.19) (0.16) (0.25) (0.13) (0.11) (0.17)
UK 0.53*** 0.75*** 0.21 0.39* 0.34*** 0.04 0.40*** 0.05 0.35**
(0.20) (0.10) (0.23) (0.23) (0.10) (0.25) (0.12) (0.07) (0.14)

Period 1 refers to 1980:Q1–1989:Q4. Period 2 refers to 1990:Q1–2007:Q2.

*
Indicate that the sum of coefficients is significantly different from zero at the 10% level.
**
Indicate that the sum of coefficients is significantly different from zero at the 5% level.
***
Indicate that the sum of coefficients is significantly different from zero at the 1% level.

These declines were statistically significant for four countries in the sample when the OECD real exchange rate series are
used (Table 2a) and for eight countries in the sample when self-constructed real exchange rate series are used (Table 2b).
Turning to parameters of focus of this research project, Tables 3 and 4 show estimates of country specific exchange rate
pass-through coefficients (short-term and long-term, respectively). Table 3 suggests that most countries in the sample expe-
rienced a decline in short-run pass-through (the coefficient on the first lag of the exchange rate in the estimated equation).
For about half of the countries in the sample, this decline was actually statistically significant.9
Table 4 summarizes estimates of long-run exchange rate pass-through coefficients (given by Eq. (2)) for each country in
the sample.10 As shown in the table, although estimation results for many countries suggest a decline in the exchange rate
pass-through, the change is rarely statistically significant (this is true for either exchange rate series).
As can be seen from Tables 2, 3 and 4, individual country coefficients tend to be estimated rather imprecisely. Therefore, I
consider the average Phillips curve coefficient estimates to be more informative than individual country coefficient estimates
(see Section 3 for discussion). Average Phillips curve coefficient estimates are presented in Table 5. The first panel of Table 5
shows estimation results obtained with OECD real effective exchange rate data, and the second one presents estimation re-
sults obtained with self-constructed real exchange rate data. The table reports the sum of coefficients on inflation (denoted
with p in the table), output gap (denoted with y), and the real exchange rate growth rate (Dq). The last two columns in the

9
To be precise, this is true for seven countries (out of 14) when OECD real exchange rate measure is used, and for six countries (out of 13) when the self-
constructed real exchange rate measure is used.
10
The standard errors are obtained using the delta method. STATA now allows one to compute delta-method standard error using the ‘‘nlcom” command. For
those without access to software that allows for computations of the standard error for a nonlinear combination of underlying model parameters Papke and
Wooldridge (2004) propose a method for obtaining the delta-method standard error that can be used by those who have software supporting only basic
regression analysis.
1124 Y.F. Takhtamanova / Journal of Macroeconomics 32 (2010) 1118–1130

Table 3a
Estimates of short-run real exchange rate pass-through coefficients real exchange rate data source: OECD. Standard errors in parenthesis.

Period 1 Period 2 Difference

Australia 0.20*** 0.0.03 0.18*
(0.05) (0.08) (0.10)
Canada 0.54*** 0.04 0.50***
(0.14) (0.11) (0.18)
Denmark 0.68** 0.20*** 0.47
(0.34) (0.07) (0.35)
Finland 0.30** 0.19*** 0.10
(0.14) (0.07) (0.15)
France 0.04 0.01 0.03
(0.18) (0.09) (0.20)
Ireland 0.49*** 0.26*** 0.23
(0.18) (0.09) (0.20)
Italy 0.27 0.10*** 0.18
(0.20) (0.04) (0.20)
Japan 0.18** 0.00 0.18**
(0.08) (0.04) (0.09)
Netherlands 0.46*** 0.10 0.36**
(0.15) (0.07) (0.16)
New Zealand 0.56*** 0.07 0.49***
(0.09) (0.06) (0.11)
Norway 0.96*** 0.09 1.05***
(0.29) (0.13) (0.31)
Sweden 0.26** 0.25* 0.01
(0.11) (0.13) (0.17)
UK 0.21 0.06 0.26*
(0.13) (0.05) (0.14)
US 0.03 0.06 0.04
(0.09) (0.06) (0.11)
Average 0.37*** 0.08*** 0.29***
(0.04) 0.02 (0.05)

Period 1 refers to 1980:Q1–1989:Q4. Period 2 refers to 1990:Q1–2007:Q2.

*
Designate that long-run pass-through coefficient or the difference between coefficients is significantly different from zero at the 10% level.
**
Designate that long-run pass-through coefficient or the difference between coefficients is significantly different from zero at the 5% level.
***
Designate that long-run pass-through coefficient or the difference between coefficients is significantly different from zero at the 1% level.

table show the short-run and long-run pass-through coefficient for the average Phillips curve (denoted with uSR and uLR,
respectively; recall that the average short-run pass-through is just the coefficient on the first lag of inflation in the average
Phillips curve, and the expression for the average long-run pass-through coefficient is given by Eq. (4) in the text).
As shown in Table 5, for the countries as a group, both inflation persistence (measured by the sum of coefficients on infla-
tion lags) and the link between real exchange rate growth rate and inflation (measured by the sum of coefficients on the real
exchange rate) declined significantly between the two periods. This result is consistent with other researchers’ findings of a
reduction in inflation persistence over time (Ball and Sheridan, 2003; Sheridan, 2001). Recall that while a decline in the per-
sistence of inflation does not change the impact effect of real exchange rate shocks on inflation (short-run pass-through), it
does reduce the long-term effect. Thus, it is no surprise that estimation results show a significant decline in the long-run
pass-through coefficient. In addition, we also see a significant decline in the short-run exchange rate pass-through. These
conclusions are robust with respect to real exchange rate measure choice.
When OECD real effective exchange rate is used, the estimation results suggest that a 1% real appreciation of domestic
currency (increase in real exchange rate) lead to a 0.37% increase in inflation at first (see the short-run pas through coeffi-
cient for period 1) and the cumulative effect on inflation amounts to 2.44% decline (see the long-run pass-through coefficient
in period 2). With self-constructed real exchange rate, a 1% appreciation of domestic currency (increase in the real exchange
rate) is estimated to lead to a 0.1% decline in inflation at first and the total effect of 0.45% decline in inflation. The short- and
long-run pass-through coefficients for the second sub-period are not significantly different from zero, indicating no signif-
icant effect of real exchange rate changes on inflation during the later period (for either exchange rate series).

3.2. The determinants of real exchange rate pass-through: time series cross-section results

In addition to documenting the significant changes in the relationship between real exchange rate and inflation, I also
investigate the reasons behind these changes. One can think of several possible determinants for the magnitude of the real
 Y.F. Takhtamanova / Journal of Macroeconomics 32 (2010) 1118–1130 1125

Table 3b
Estimates of short-run real exchange rate pass-through coefficients real exchange rate data source: self-constructed. Standard errors in parenthesis.

Period 1 Period 2 Difference

Australia 0.13 0.06 0.18
(0.21) (0.13) (0.25)
Canada 0.03 0.05 0.08
(0.04) (0.05) (0.07)
Denmark 0.01 0.04 0.05
(0.11) (0.02) (0.11)
Finland 0.16*** 0.05 0.12*
(0.06) (0.03) (0.07)
France 0.16*** 0.02 0.14**
(0.06) (0.03) (0.06)
Ireland 0.13 0.11** 0.02
(0.09) (0.05) (0.10)
Italy 0.03 0.01 0.02
(0.05) (0.01) (0.05)
Japan 0.11* 0.00 0.11
(0.06) (0.04) (0.07)
Netherlands 0.16*** 0.04 0.11**
(0.05) (0.03) (0.05)
New Zealand 0.41*** 0.00 0.41***
(0.08) (0.05) (0.10)
Norway 0.18 0.02 0.20
(0.11) (0.06) (0.13)
Sweden 0.30*** 0.09 0.20**
(0.07) (0.08) (0.10)
UK 0.22*** 0.04 0.26***
(0.08) (0.04) (0.09)
Average 0.10*** 0.03* 0.07**
(0.03) 0.02 (0.03)

Period 1 refers to 1980:Q1–1989:Q4. Period 2 refers to 1990:Q1–2007:Q2.

*
Designate that long-run pass-through coefficient or the difference between coefficients is significantly different from zero at the 10% level.
**
Designate that long-run pass-through coefficient or the difference between coefficients is significantly different from zero at the 5% level.
***
Designate that long-run pass-through coefficient or the difference between coefficients is significantly different from zero at the 1% level.

exchange rate pass-through to aggregate inflation. For both short- and long-term exchange rate pass-through, the openness
of the economy (share of imports in average consumer’s basket), the share of flexible-price firms and the monopoly power of
firms come to mind (please see Takhtamanova (2008) for a formal derivation of the determinants of the real exchange rate
pass-through).
It is easy to see why a lower degree of openness in the economy would cause response of inflation to real exchange rate
fluctuations to be weaker. As previously discussed, changes in exchange rate are transmitted to aggregate inflation through
the prices of imports. A decline in the share of imports would, therefore, lower inflation responsiveness to real exchange rate
shocks. A decline in the fraction of flexible-price firms has a similar effect. Only flexible-price firms have the ability to incor-
porate information about current exchange rate shocks when setting prices for their goods. Thus, if the fraction of flexible-
price firms goes down, exchange rate shocks have less of an effect on aggregate inflation.
Also, a decline in the aggregate real exchange rate pass-through could be caused by a reduction of real exchange rate
pass-through at the level of the individual firms. Suppose each importing firm begins to respond less to real exchange rate
changes. Then aggregate response of prices to real exchange rate movements will also be dampened. The decline in an indi-
vidual firm’s responsiveness to real exchange rate fluctuations, in turn, can be caused by an increase in either the elasticity of
the demand function or the elasticity of the cost function. An increase in either elasticity can be interpreted as a loss of
monopoly power by the firm. Thus, a reduction in the responsiveness of an individual foreign firm’s prices to real exchange
rate shocks happens when the firm loses its monopoly power.
In addition to the three determinants described above, the long-term exchange rate pass-through is affected by infla-
tion persistence. A reduction in inflation persistence would cause a decline in the long-term effect of real exchange rate
fluctuations on inflation. Takhtamanova (2008) shows that inflation persistence, in turn, is affected by central bank cred-
ibility, degree of real rigidity (responsiveness of individual firm optimal price to total spending in the economy), and the
weight the central bank places on inflation (i.e. the relative importance of inflation to a central bank faced with several
goals).
1126 Y.F. Takhtamanova / Journal of Macroeconomics 32 (2010) 1118–1130

Table 4a
Estimates of long-run real exchange rate pass-through coefficients real exchange rate data source: OECD. Standard errors in parenthesis.

Period 1 Period 2 Difference

Australia 0.40*** 0.30 0.10
(0.10) (0.20) (0.22)
Canada 1.73 0.29* 2.02*
(1.10) (0.17) (1.11)
Denmark 5.68 0.24** 5.92
(8.61) (0.12) (8.61)
Finland 3.06 0.61** 2.45
(2.10) (0.26) (2.12)
France 1.89 0.33 2.22
(1.47) (0.24) (1.48)
Ireland 1.71 1.41* 0.30
(1.42) (0.75) (1.60)
Italy 1.81** 1.27** 0.54
(0.87) (0.55) (1.03)
Japan 0.19* 0.15 0.04
(0.11) (0.11) (0.15)
Netherlands 4.36 0.46** 4.82
(3.11) (0.20) (3.12)
New Zealand 6.87 0.05 6.82
(5.05) (0.05) (5.05)
Norway 1.52 0.10 1.62
-(1.52) (0.29) (1.55)
Sweden 1.65 0.68 0.97
(2.58) (0.65) (2.66)
UK 2.32 0.31 2.00
(2.83) (0.39) (2.86)
US 1.06 0.79* 0.27
(1.55) (0.46) (1.61)
Average 0.77 0.44*** 0.33
(0.85) 0.11 (0.86)

Period 1 refers to 1980:Q1–1989:Q4. Period 2 refers to 1990:Q1–2007:Q2.

*
Designate that long-run pass-through coefficient or the difference between coefficients is significantly different from zero at the 10% level.
**
Designate that long-run pass-through coefficient or the difference between coefficients is significantly different from zero at the 5% level.
***
Designate that long-run pass-through coefficient or the difference between coefficients is significantly different from zero at the 1% level.

This study focuses on the openness in the economy and the fraction of flexible-price firms as pass-through determi-
nants.11 It is necessary to use proxies for both, as the data for these variables are not available. The share of imports in GDP
is selected as a proxy for the share of imported goods in the CPI basket. Average inflation for the decade is used as a proxy
for the fraction of flexible-price firms in the economy. This choice of proxy is based on previous research on the frequency
of price adjustment. Ball et al. (1988) argue that, theoretically, in the presence of a higher average rate of inflation, firms adjust
prices more frequently to keep up with the rising price level. They then go onto show the empirical aggregate evidence support-
ing this theoretical claim. The microeconomic literature also shows evidence supporting a positive relationship between the
average inflation for a period and the fraction of flexible-price firms (Ceccetti, 1986). Hence, it is reasonable to conjecture that
there are more flexible-price firms in an economy where average inflation is higher.
In this second stage of the exercise, exchange rate pass-through (short-run or long run) becomes a dependent variable. As
the reader recalls, I already have short- and long-run pass-through coefficient estimates (shown in Tables 3 and 4). I also
calculate for each decade the average import share in GDP and the average inflation for each country. With the data in hand,
I then regress the pass-through coefficient on the two explanatory variables and also allow for a decade effect:

uxjt ¼ aj þ bsAVE
jt þ cpAVE
jt þ qdt þ ejt ; ð5Þ

where j is the index for countries, t is the index for decades, u stands for the pass-through coefficient, x is either SR (for short-
run pass-through) or LR (for long-run pass-through), sAVE denotes the average share of imports in GDP for the decade, pAVE is
the average inflation for the decade, and d is decade indicator (equal to 1 for the first sample period and to 0 otherwise).

11
Data on the elasticities of the demand and cost functions for all the countries in the sample are not readily available. Therefore, this project does not analyze
whether the monopoly power of firms changed between the two decades, and whether this change is the cause of the deterioration in the relationship between
aggregate inflation and the real exchange rate. Additionally, finding a measure for central bank credibility for every country in the sample proved unviable. One
suggestion for future research would be to work on constructing such a measure by relating central bank credibility to central bank independence.
 Y.F. Takhtamanova / Journal of Macroeconomics 32 (2010) 1118–1130 1127

Table 4b
Estimates of long-run real exchange rate pass-through coefficients real exchange rate data source: self-constructed. Standard errors in parenthesis.

Period 1 Period 2 Difference

Australia 1.00*** 0.24 0.76
(0.31) (0.34) (0.46)
Canada 1.48 0.01 1.49
(6.09) (0.07) (6.09)
Denmark 0.40* 0.08** 0.48**
(0.21) (0.04) (0.21)
Finland 0.56*** 0.25** 0.31*
(0.15) (0.10) (0.18)
France 1.11** 0.09 1.20**
(0.53) (0.06) (0.53)
Ireland 0.17 0.35** 0.18
(0.33) (0.15) (0.37)
Italy 0.56*** 0.02 0.54*
(0.20) (0.26) (0.32)
Japan 0.14 0.12 0.02
(0.09) (0.10) (0.14)
Netherlands 1.54 0.15** 1.69
(1.30) (0.06) (1.30)
New Zealand 1.08* 0.02 1.06*
(0.56) (0.12) (0.57)
Norway 2.04 0.11 2.15
(1.74) (0.11) (1.74)
Sweden 0.59** 0.09 0.49
(0.28) (0.26) (0.38)
UK 0.85** 0.19 0.67
(0.41) (0.28) (0.49)
Average 0.05 0.12* 0.07
(0.49) 0.06 (0.49)

Period 1 refers to 1980:Q1–1989:Q4. Period 2 refers to 1990:Q1–2007:Q2.

*
Designate that long-run pass-through coefficient or the difference between coefficients is significantly different from zero at the 10% level.
**
Designate that long-run pass-through coefficient or the difference between coefficients is significantly different from zero at the 5% level.
***
Designate that long-run pass-through coefficient or the difference between coefficients is significantly different from zero at the 1% level.

Table 5a
Average Phillips curve estimates (real exchange rate data source: OECD). Standard errors in parenthesis.

Sums of coefficients on
p y Dq uSR uLR
*** *** *** ***
Period 1 0.72 0.63 0.68 0.37 2.44***
(0.06) (0.08) (0.10) (0.04) (0.73)
Period 2 0.38*** 0.24 0.19 0.08 0.31
(0.06) (0.04) (0.04) 0.02 0.08
Difference 0.34*** 0.39*** 0.49*** 0.29*** 2.13***
(0.09) (0.09) (0.10) (0.05) (0.73)

The model is estimated for each type of pass-through coefficients (short-run and long-run) and for each real exchange
measure (OECD and self-constructed). In each case, the Hausman test results suggest that a random effects model be esti-
mated. The estimation results are presented in Table 6. The first panel focuses on short-run pass-through. It shows little evi-
dence of a link between short-run pass-through and the proposed explanatory variable. When OECD real exchange rate is
used, we see that the openness of the economy has a statistically significant relationship with short-run pass-through.
However, the effect is rather small – a 1% point increase in import to GDP ratio causes a 0.01% decline in the short-run ex-
change rate pass-through. When self-constructed real exchange rate is used, there is no statistically significant relationship
between the short-run exchange rate pass-through. Given the small sample size and the imprecision of country-specific
short-run exchange rate pass-through estimates, the large standard errors on the coefficients are not surprising.
The results for long-run pass-through are more interesting. The table shows that the average inflation for the decade has a
statistically significant impact on the total response of inflation to real exchange rate fluctuations. For the estimation per-
formed with the real exchange rate series prepared by the OECD, a 1% point increase in the average inflation for the decade
1128 Y.F. Takhtamanova / Journal of Macroeconomics 32 (2010) 1118–1130

Table 5b
Average Phillips curve estimates (real exchange rate data source: self-constructed). Standard errors in parenthesis.

Sums of coefficients on
p y Dq uSR uLR
Period 1 0.49*** 0.47*** 0.20*** 0.10 0.40***
(0.06) (0.08) (0.05) (0.03) (0.11)
Period 2 0.31*** 0.24 0.07** 0.03 0.11
(0.06) (0.24) (0.03) (0.03) 0.05
Difference 0.18** 0.24 0.13** 0.07* 0.29**
(0.09) (0.25) (0.06) (0.04) (0.12)

Period 1 refers to 1980:Q1–1989:Q4. Period 2 refers to 1990:Q1–2007:Q2.

*
Designate that long-run pass-through coefficient or the difference between coefficients is significantly different from zero at the 10% level.
**
Designate that long-run pass-through coefficient or the difference between coefficients is significantly different from zero at the 5% level.
***
Designate that long-run pass-through coefficient or the difference between coefficients is significantly different from zero at the 1% level.

Table 6a
Determinants of short-run real exchange rate pass-through. Standard errors in parenthesis.

Measures of exchange rate

OECD Self-constructed
Constant 0.27 0.03
(0.25) (0.07)
Decade indicator 0.22* 0.05
(0.11) (0.08)
Average Inflation 0.02 0.01
(0.02) (0.01)
Import Share in GDP 0.01*** 0.00
(0.00) (0.00)
*
Designate that the coefficient is significantly different from zero at the 10% level.
**
Designate that the coefficient is significantly different from zero at the 5% level.
***
Designate that the coefficient is significantly different from zero at the 1% level.

Table 6b
Determinants of long-run real exchange rate pass-through. Standard errors in parenthesis.

Measures of exchange rate

OECD Self-constructed
Constant 7.48*** 0.21
(2.57) (0.42)
Decade Indicator 3.46*** 0.97**
(1.16) (0.49)
Average Inflation 0.81*** 0.18**
(0.21) (0.08)
Import Share in GDP 0.02 0.01
(0.03) (0.01)
*
Designate that the coefficient is significantly different from zero at the 10% level.
**
Designate that the coefficient is significantly different from zero at the 5% level.
***
Designate that the coefficient is significantly different from zero at the 1% level.

yields a 0.8% increase in the long-run exchange rate pass-through. Using the self-constructed real exchange rate data, a 1%
increase in the average inflation yields an additional 0.18% increase in the same parameter. Based on these observations, it is
possible to conclude that the low inflation environment in the 1990s led to a reduction in the fraction of flexible-price firms
in the economy. This reduction in turn contributed to a decline in the long-run pass-through of real exchange rate changes to
inflation. This conclusion stands irrespective of the choice of real exchange rate series.

3.3. How much do determinants of real exchange rate pass-through explain?

The final question addressed in this empirical study is how much of the decline in the long-run pass-through can be ex-
plained by the reduction in the frequency of price updating by firms.12 I begin with considering the estimate of exchange rate
pass-through obtained with OECD real effective exchange rate series. From Table 6b, I discovered that average inflation is a

12
I do not address short-run pass-through in this section, because no strong relationship between this variable and the two proposed determinants has been
established.
 Y.F. Takhtamanova / Journal of Macroeconomics 32 (2010) 1118–1130 1129

significant determinant of the long-run pass-through. Thus, it is useful to know how much of the drop in the estimated long-
run pass-through coefficient is explained by the change in average inflation. I do that by comparing the actual decline in the
long-run pass-through coefficients (0.33, as shown in Table 4a) to the one predicted by the change in average inflation. The
latter one (denoted with u ^DpAVE
^ LR ) is computed as c t , where c
^ is the estimated coefficient on average inflation (from Table 6b)
and DpAVE is the change in average inflation between decades (from Table 1). For the estimation results obtained with OECD
real exchange rate, this predicted drop in real long-run exchange rate pass-through of 3.7. Thus, in this case the average infla-
tion over-explains the drop in exchange-rate pass-through.
I reach the same conclusion when using estimates obtained with the self-constructed real exchange rate. The drop in long-
run exchange rate pass-through estimated this real exchange rate measures is 0.07 (see Table 4b). The predicted change is 0.8
(the product of the coefficient on average inflation shown in second column of Table 5b and the change in average inflation).

4. Conclusions

The empirical investigation for fourteen OECD countries confirms the suggestion made by other authors about the decline
in the exchange rate pass-through to aggregate inflation during the 1990s (for example, Cunnigham and Haldane, 2000; Gar-
cia and Restrepo, 2001; McCarthy, 1999). The paper looks at both short- and long-run exchange rate pass-through and sug-
gests a statistically significant decline in both.
This paper attempts to look for an explanation for the weakening in the relationship between the real exchange rate and
CPI inflation. It is hypothesized that the decline in the long-run exchange rate pass-through is in part caused by a reduction
in the fraction of flexible-price firms in the economy. That is, firms updated their prices less frequently during the 1990s as
compared to the previous decade, due to the 1990s unusually low inflation environment. This finding is consistent with the
conclusions of Taylor (1999), who claims that ‘‘the decline in the pass-through or pricing power is due to the low inflation
environment that has been recently achieved in many countries.”
The implication of the analysis presented in this paper is that the recently observed reduction in the real exchange rate
pass-through cannot be regarded as a permanent change. Should the inflation level and inflation persistence increase in the
future (as some fear might happen in the US in the near future) we would see an increase in the real exchange rate pass-
through. As Taylor (1999) points out, such an increase in real exchange rate pass-through could expedite inflationary pres-
sures again.

Acknowledgements

The author thanks Laurence Ball, Louis Maccini, Christopher Carroll, Carl Christ, Benjamin Dennis, Frederick Furlong,
Galina Hale, Thomas Lubik, Eva Sierminska, Gary Zimmerman, the participants in the Johns Hopkins University Macroeco-
nomics Lunch Seminar, 39th Annual Meetings of the Canadian Economics Association, 80th Annual Conferences of the Wes-
tern Economics Association International, and 39th Global Conference on Business and Economics for many useful
comments and discussions. The author is also grateful to Charles Notzon and Renee Courtois for excellent research assis-
tance. All errors are her own.

Appendix A. Data sources

Variable Source Description

Consumer OECD Main Economic Two series are used in this study. Consumer price index (CPI): all items
prices Indicators Database is used to measure inflation.
CPI ‘‘all items non-food non-energy” provides an indication about the
core inflation as used by the OECD
Output gap OECD Economic Outlook The difference between actual and potential GDP as a per cent of
Database (this study potential GDP. Potential GDP is defined in the Economic Outlook as the
utilized Economic Outlook level of output that an economy can produce at a constant inflation rate
No. 81)
Real effective OECD Main Economic Real effective exchange rates take account of price level differences
exchange rate Indicators (MEI) between trading partners. Movements in real effective exchange rates
provide an indication of the evolution of a country’s aggregate external
price competitiveness

References

Bacchetta, Philippe, van Wincoop, Eric., 2003. Why do consumer prices react less than import prices to exchange rates. Journal of the European Economic
Association 1 (2-3), 662–670.
Ball, Laurence, 1999. Monetary policy rules in an open economy. In: Taylor, John B. (Ed.), Monetary Policy Rules. University of Chicago Press, Chicago.
1130 Y.F. Takhtamanova / Journal of Macroeconomics 32 (2010) 1118–1130

Ball, Laurence, Mankiw, N. Gregory, Romer, David, 1988. The new Keynesian economics and the output-inflation trade-off, Brookings Papers on Economic
Activity, p. 1.
Ball, Laurence, Sheridan, Niamh, 2003. Does Inflation Targeting Matter, Johns Hopkins University, manuscript.
Campa, José Manuel, Goldberg, Linda S., 2005. Exchange-rate pass-through into import prices. Review of Economics and Statistics 87 (4), 679–690.
Ceccetti, Stephen G., 1986. The frequency of price adjustment: a study of the newsstand prices of magazines. Journal of Econometrics 31, 255–274.
Cunnigham, Alastair, Haldane, Andrew G., 2000. The Monetary Transmission Mechanism in the United Kingdom: Pass Through and Policy Rules, Central
Bank of Chile Working Papers, No. 83.
Feinberg, Robert M., 1986. The interaction of foreign exchange and market power effect on German domestic prices. Journal of Industrial Economics, 35.
Feinberg, Robert M., 1989. The effects of foreign exchange rate movements on US domestic prices. Review of Economics and Statistics, 71.
Gagnon, Joseph., Ihrig, Jane, 2004. Monetary policy and exchange rate pass-through. International Journal of Finance and Economics 9 (4), 315–338.
Garcia, Restrepo, 2001. Price Inflation and Exchange Rate Pass-Through in Chile, Central Bank of Chile Working Papers, No. 128.
Goldberg, Pinelopi, Knetter, Michael, 1997. Goods prices and exchange rates: what have we learned? Journal of Economic Literature, 35.
Gordon, Robert., 1997. Time-varying NAIRU and its implications for monetary policy. Journal of Economic Perspectives 11 (Winter), 32.
Greene, William H., 1997. Econometric Analysis, third ed. Prentice Hall International Editions.
Mankiw, N. Gregory, Reis, Ricardo, 2002. Sticky information versus sticky prices: a proposal to replace the new Keynesian curve. Quarterly Journal of
Economics 117 (4).
McCarthy, Jonathan, 1999. Pass-through of exchange rates and import prices to domestic inflation in some industrialized countries, Federal Reserve Bank of
New York Working Papers.
McCarthy, Jonathan, 2007. Pass-through of exchange rates and import prices to domestic inflation in some industrialized countries. Eastern Economic
Review 33, 511–537.
Papke, Leslie E., Wooldridge, Jeffrey M., 2004. A computation trick for delta-method standard errors. Economic Letters 86 (2005), 413–417.
Razin, Assaf, Yuen, Chi-Wa, 2001. The ‘New Keynesian’ Phillips Curve: Closed Economy vs. Open Economy, NBER Working Papers, No. 8313.
Sheridan, Niamh, 2001. Inflation Dynamics, Ph.D. Dissertation. Johns Hopkins University.
Staiger, Douglas, Stock, James, Watson, Mark, 1997. The NAIRU, unemployment, and monetary policy. Journal of Economic Perspectives 11 (Winter), 33–51.
Svensson, Lars, 1998. Open Economy Inflation Targeting, Stockholm University, manuscript.
Takhtamanova, Yelena F., 2008. Understanding changes in exchange rate pass-through. Federal Reserve Bank of San Franciso Working Paper 2008-13.
Taylor, John, 1999. Low inflation, pass-through, and the pricing power of firms. European Economic Review, 44.
Wickens, M.R., Breusch, T.S., 1988. Dynamic specification, the long-run and the estimation of transformed regression models. The Economic Journal 98
(390), 189-205. (Supplement: Conference Papers).
Woo, Wing T., 1984. Exchange rates and the prices of nonfood, nonfuel products, Brookings Papers on Economic Activity, p. 2.