Chapter

4 Exchange Rate Pass-Through in the Euro Area

Author(s):
Charalambos Tsangarides, Carlo Cottarelli, Gian Milesi-Ferretti, and Atish Ghosh
Published Date:
September 2008
Share
  • ShareShare
Show Summary Details

Since August 2007, financial market turmoil—precipitated by the U.S. subprime mortgage crisis—has disrupted money and credit markets in the United States, as well as Europe. Against concerns of a “credit crunch” and a sharper slowdown in the U.S. economy, the U.S. dollar has come under pressure, falling in value against many currencies, notably the euro. With Europe’s single currency touching new all–time highs against the dollar, European concerns about the economic impact of recent euro appreciation have figured prominently. In assessing the likely macroeconomic consequences, a key determining factor is the nature of exchange rate pass–through in the euro area.

More difficult to assess, however, are the behavioral features that underlie the nature and extent of pass–through and the responsiveness of trade flows to the exchange rate. The relevance of exchange rate pass–through for inflation is straightforward. But these issues are also important in determining the strength of expenditure–switching effects from relative price signals (see Obstfeld, 2002, and Engel, 2002, for recent reviews of these issues). Incomplete pass–through, for example, could delay or diminish the response in external variables and produce a certain degree of “exchange rate disconnect” (see Krugman, 1989, and Devereux and Engel, 2002). Thus, the degree and behavior of pass–through in the euro area is a key input for assessing the likely economic impact of such pass–through on growth and inflation.

Several economic explanations have been advanced to account for incomplete exchange rate pass–through, a feature that has strong empirical support for a large number of economies, including the euro area (see Goldberg and Knetter, 1997, for a survey; see Kieler, 2001, Hiifner and Schroder, 2002, Anderton, 2003, and Hahn, 2003, for euro area estimates). With nominal rigidity and local currency pricing, destination prices can change very little in the face of exchange rate variation (see Betts and Devereux, 1996, 2000, and Devereux and Engel, 2002). With pricing–to–market behavior, segmented markets allow firms to stabilize their destination prices (via changing markups) to preserve foreign market share. In the presence of local distribution costs, firms may also face offsetting factors when the exchange rate changes, leading to international price discrimination and incomplete pass–through (see Corsetti and Dedola, 2002, and Choudhri, Faruqee, and Hakura, 2002). These factors can help account for the difference in responses between “first–stage” pass–through (for example, in import prices) and “second–stage” pass–through (for example, in consumer prices). Moreover, these considerations have been shown to have important implications for optimal monetary and exchange rate policies (see, for example, Corsetti and Pesenti, 2002, and Devereux and Engel, 2003).

This chapter examines exchange rate pass–through and its behavioral determinants in the euro area. The methodology proceeds in two parts. First, the empirical analysis follows a vector autoregression (VAR) approach, in which the time–series behavior of the euro exchange rate and a system of euro area prices are examined. Specifically, the empirical analysis investigates exchange rate pass–through into a set of prices along the pricing chain. Second, the impulse response functions (IRFs) from the VAR estimation are used to calibrate in a new open economy macroeconomics model the key behavioral parameters that can help reproduce the pattern of pass–through and external adjustment in the euro area.

The use of a VAR approach to examine exchange rate pass–through has several advantages compared with single–equation methods. Previous studies typically have focused on pass–through into a single price (for example, import or consumer prices) without further distinguishing between the types of underlying exchange rate shocks (for example, permanent or transitory) that may be occurring. By investigating exchange rate pass–through into a set of prices along the pricing chain, the VAR analysis characterizes not only absolute but relative pass–through in upstream and downstream prices. Second, the VAR methodology potentially allows one to identify specific structural shocks affecting the system. In this case, a structural exchange rate shock is identified through a Cholesky decomposition of innovations, in which exchange rate fluctuations at higher frequencies are assumed to be largely driven by asset market—rather than goods market—disturbances.

Using this identification scheme, one can map the empirical results into a well–defined shock in an economic model of incomplete pass–through. Specifically, the estimated IRFs from an exchange rate shock in the VAR are matched to the relative response patterns generated by the corresponding asset market disturbance in the analytical pass–through model. By minimizing the distance between these sets of IRFs, one can identify key structural parameters that underlie the overall pattern of pass–through. The key parameters are (1) the pricing behavior of firms (that is, the extent of local currency pricing), (2) the degree of nominal rigidity, and (3) the extent of local distribution and trading costs generating international price discrimination. A comparison of these behavioral parameters across the major industrial countries can then be made.

Empirical Estimates

The empirical analysis focuses on euro area prices and the exchange rate at a monthly frequency. The time span covers the period from 1990 through 2002. All series are expressed in logarithms. For the euro exchange rate s, the nominal effective series is defined for the European Central Bank’s “narrow” group of partner countries.1 For factor prices (that is, wage earnings w), the series derive from quarterly data on nominal compensation per employee, extrapolated to a monthly frequency.2 For trade prices, import prices pm and export prices px are based on unit values in extra–area manufacturing trade.3 Producer prices py are based on the producer price index for manufacturing, excluding construction and energy. Consumer prices pc are based on the core consumer price index—that is, excluding energy and unprocessed foods—from the harmonized index of consumer prices, although the implications with regard to headline inflation are also discussed.4 The exclusion of energy prices is motivated primarily by the standard finding that their pass–through behavior differs from that of other goods; by eliminating them, we facilitate a more seamless transition to the pass–through model later that emphasizes differentiated goods with imperfect substitutability (see, for example, Campa and Goldberg, 2002).

The use of extra–area trade data helps avoid potential pitfalls from inferring pass–through for the euro area from estimates for individual member countries. To the extent that intra–area trade systematically differs from extra–area trade with respect to pass–through behavior, aggregating country estimates to generate an area–wide measure could suffer from a “fallacy of composition.” Hiifner and Schroder (2002), for example, construct pass–through estimates for euro area consumer prices by summing over individual country estimates, based on each country’s weight in the area–wide harmonized index of consumer prices. The analysis is then forced to make some “correction” of the estimates because of the presence of intra–area trade.

Before the VAR estimation of euro area pass–through was undertaken, some preliminary tests of the data were conducted. Unit root and stationarity tests indicate that these nominal variables are nonstationary in levels but stationary in first differences, suggesting that they are integrated of order one or I(1)series.5 Furthermore, residual–based cointegration tests do not find evidence of cointegration among the variables (see Appendix 4.1). Given potential nonstationarity and lack of cointegration in the data in levels, estimating the VAR in first differences is appropriate.

VAR Methodology

The VAR approach examines the joint historical time–series behavior of the euro exchange rate and a system of euro area prices. Specifically, the reduced–form VAR(p) can be written as follows:

where Y = [Δs Δw Δpm Δpx Δpy Δpc]′; c is a vector of deterministic terms (that is, monthly time dummies); A is a matrix polynomial of degree p in the lag operator L; and u is the (6 x 1) vector of reduced–form residuals with variance–covariance matrix Ω. The exchange rate is placed first in the order of variables, reflecting the presumption that exchange rate innovations at monthly frequency are primarily driven by exogenous asset market disturbances.6 For prices, the ordering after the exchange rate is motivated by the pricing chain, from factor input prices to trade prices to wholesale producer prices and retail consumer prices. The ordering among price variables after the exchange rate does not matter for the subsequent analysis of the exchange rate shock.

To recover the underlying exchange rate shock, the Cholesky decomposition of the matrix Ω is used to produce orthogonalized innovations ε. These disturbance terms are expressed in terms of the reduced–form VAR innovations as follows:

where C is the unique lower triangular Cholesky matrix with ones along its principal diagonal. Because the exchange rate appears first in the VAR, the recursive structure in Equation (4.2) imposes the assumption that orthogonalized innovations to the exchange rate depend only on the residuals from the exchange rate equation and not on those from the other equations. This identification allows for a simple correspondence between the VAR estimates and a well–defined shock in the model described later. For prices, the corresponding disturbance term will represent a mix of shocks, including the structural exchange rate shock.7

An alternative identification scheme would place the exchange rate last (or near last) in the VAR. This ordering is motivated by the view that prices (and quantities) are predetermined in the very short run and thus cannot respond to an exchange rate shock, whereas the exchange rate can respond to various shocks (see, for example, Peersman and Smets, 2001). Although this observation may be valid for many prices, it may not be appropriate for others. More to the point, the ordering imposes a specific pass–through pattern in the estimates and, ultimately, certain behavioral features in the model that the current analysis seeks to investigate. Nevertheless, results of sensitivity analysis for reorderings of the VAR are reported later.

The implications of the identifying restriction can be further understood from the structural representation of the VAR:

where F(L) is a matrix polynomial of degree p + 1, k is a transformation of deterministic terms (that is, Ck = c), and ε is the vector of structural shocks. One can show that the identification scheme based on the Cholesky decomposition introduces the following restriction: in the first equation (that is, for the exchange rate Δs), the coefficients on contemporaneous price changes Δw, Δpm, Δpx, Δpy, and Δpc are equal to zero. Granger causality and block exogeneity tests find that (lagged) euro area prices have no predictive value for the euro exchange rate.8 The VAR identification scheme takes this result one step further, assuming that concurrent price innovations also do not help explain exchange rate innovations.

The economic justification for this identifying assumption can be understood as follows. Exchange rates—especially at higher frequencies—are essentially driven by asset market rather than goods market disturbances. In the presence of noise traders, short–run exchange rate fluctuations may have little to do with economic fundamentals, including the price variables considered here (see, for example, Jeanne and Rose, 2002, and Devereux and Engel, 2002). The econometric assumption, though, is not necessarily incompatible with a more “fundamentalist” view. Under the asset market view of exchange rate determination, the predictable component is usually deemed small, and short–run changes in the exchange rate are likely to be dominated by “news” (see Mussa, 1984). In the universal presence of reporting lags, however, data releases on price indices, when they are made available, typically offer very little new information to move exchange rate markets. Furthermore, the empirical justification for this identification scheme is well known, given the overwhelming failure of exchange rate models to outperform a simple random walk over short horizons.9

Based on the reduced–form estimates of the VAR and the Cholesky decomposition to identify structural shocks, accumulated IRFs for a unit exchange rate shock are displayed in Figure 4.1.10 The figure shows that an exchange rate shock has only a small effect on prices in the euro area. Prices tend to be predetermined or very sticky (in local currency) initially in response to a depreciation in the euro effective exchange rate. Over time, the degree of exchange rate pass–through generally rises, although minimally so in factor and retail prices. Wholesale producer prices tend to rise more than retail consumer prices, but the greatest response is in trade prices. Twelve to eighteen months after the shock, export prices show almost half the response of import prices, which move in proportion to the exchange rate, suggesting full pass–through. Consequently, the (manufacturing) terms of trade for the euro area tend to worsen or decline in response to an exchange rate depreciation.11Choudhri, Faruqee, and Hakura (2005) report similar relative pass-through findings in a quarterly VAR for G–7 countries that further includes interest rates to account for monetary policy shocks.12

Normalizing the price responses in Figure 4.1 by the path of the exchange rate response, Table 4.1 shows dynamic pass–through elasticities for euro area prices. After 18 months, pass–through rates in export and import prices are about one–half and one, respectively. Wage pass–through is relatively very small at 5 percent. Pass–through in wholesale prices is nearly 20 percent, significantly exceeding the pass–through in retail prices.13 Using headline rather than core inflation (or including longer lags) raises the degree of pass–through in consumer prices to near 10 percent at 18 months, but the pattern of relative pass–through remains intact.

Figure 4.1.Effects of Exchange Rate Shock on Euro Area Prices

Source: International Monetary Fund staff estimates.

Note: Figure shows impulse response path in (log) prices in response to unit shock in the (log) exchange rate.

Full pass–through in euro area import prices over time may be a somewhat surprising result.14 Alternative specifications (for example, with coefficient restrictions or VAR reorderings) tend to yield smaller but still high pass-through elasticities of between 0.7 and 1. In a recent paper that also examines extra–area manufacturing trade prices, Anderton (2003) also finds generally high (albeit not full) pass–through of between 0.5 and 0.7, based on a single–equation approach.15 Given the identification of near–permanent exchange rate shocks here, it is not surprising that import price pass–through in this VAR analysis lies somewhat above those single–equation estimates.

Table 4.1.Euro Area Pass-Through Elasticities(Ratio of percent change in prices to percent change in exchange rate)
t = 1t = 6t = 12t = 18
CPI0.000.010.020.02
PPI0.000.040.110.17
Wage–0.020.040.070.05
px0.020.180.310.45
pm0.030.420.811.17
Source: IMF staff estimates.Note: Based on impulse response functions from six–variable VAR estimated on monthly data from 1990 through 2002. CPI refers to consumer price index; PPI refers to producer price index; Wage refers to nominal compensation per employee; px and pm refer to export and import prices, respectively.
Source: IMF staff estimates.Note: Based on impulse response functions from six–variable VAR estimated on monthly data from 1990 through 2002. CPI refers to consumer price index; PPI refers to producer price index; Wage refers to nominal compensation per employee; px and pm refer to export and import prices, respectively.

Cross–Country Evidence

Before relating the empirical findings to the theoretical model, it is useful to compare the pass–through results for the euro area to those for the other major industrial economies. Repeating the VAR exercise for the United States, Japan, the United Kingdom, and Canada produces the pass–through coefficients for trade prices shown in Table 4.2.

As is evident from the table, pass–through is incomplete, particularly in the short run. For the United States, pass–through in trade prices at the time of the shock is near zero, quite similar to the response of the euro area. But whereas pass–through rises significantly over time in the euro area, the increase over time is much smaller for the United States. For the other countries, pass–through to import and export prices is higher on impact than for the United States and the euro area. For Canada and Japan, estimated import price pass–through is about 60–70 percent initially and remains around those levels over time; for the United Kingdom, import price pass–through eventually reaches 60 percent, but is initially half that. Pass–through in export prices is eventually around 50 percent for these three countries, similar to that for the euro area.

Sensitivity Analysis

As is well known, impulse responses from a VAR can be sensitive to the ordering of variables. In the six–variable VAR system, there are 6! = 720 possible orderings under the Cholesky decomposition used to identify structural shocks. In general, when the reduced–form residuals from the VAR do not display high cross–correlations, the order of factorization makes little difference.16 Otherwise, the results can be sensitive to the choice of ordering. Since the focus here is solely on the effects of an exchange rate shock, suborderings among the price variables—given the exchange rate’s order in the VAR—do not matter. Thus, the relevant reorderings to consider surround the placement of the exchange rate in the VAR and the combinations of price variables selected to appear either before or after the exchange rate, given its order in the VAR.17

Table 4.2.Pass-Through Elasticities in Trade Prices: International Comparisons(Ratio of percent change in prices to percent change in exchange rate)
t = 1t = 6t = 12t = 18
Import prices
Euro area0.030.420.811.17
United States0.060.150.180.30
Japan0.610.560.570.57
United Kingdom0.280.580.570.60
Canada0.680.540.620.68
Export prices
Euro area0.020.180.310.45
United States0.000.000.060.12
Japan0.620.500.480.47
United Kingdom0.160.470.460.50
Canada0.350.190.350.44
Source: IMF staff estimates.Note: Based on impulse response functions from six–variable VAR estimated on monthly data from 1990 through 2002. For euro area, import and export prices are based on unit values in manufacturing trade. For others, import and export prices are based on unit values in total trade.
Source: IMF staff estimates.Note: Based on impulse response functions from six–variable VAR estimated on monthly data from 1990 through 2002. For euro area, import and export prices are based on unit values in manufacturing trade. For others, import and export prices are based on unit values in total trade.

Changes taken from this subset of possibilities, however, can be shown to have very little impact on the results for the euro area (and the United States); the results are robust to reorderings of the VAR. Take the extreme cases. Compared with the case in which the exchange rate appears first in the VAR, placing the exchange rate last in the VAR produces very similar pass–through elasticities for the euro area (see Table 4.A.5).18 Similarly, placing different combinations of prices before or after the exchange rate does not materially affect the results. For the other countries, the results largely obtain so long as the trade prices appear after the exchange rate in the VAR. Otherwise, import and export prices are predetermined (that is, have zero response on impact) by construction, which, as shown in Table 4.2, does not reflect their behavior under less restrictive assumptions. Given the objective of investigating (without prejudice) the pass–through behavior in these key prices, making fewer (possibly erroneous) data restrictions a priori is clearly desirable.

Model of Incomplete Pass–Through

This section describes the analytical framework used to interpret the VAR evidence in terms of underlying economic behavior. The stylized model generates incomplete pass–through by drawing on many of the common themes found in the new open economy macroeconomics paradigm, following the seminal work of Obstfeld and Rogoff (1995) (see Lane, 2001, for a survey). A detailed derivation of the microfounded model can be found in Choudhri, Faruqee, and Hakura (2005). A sketch of the model follows.

Imperfect competition characterizes the production and allocation of two differentiated goods—a traded intermediate good and a nontraded final (consumption) good. One differentiated primary factor (labor) and intermediate inputs enter the production of traded goods. A domestic retail sector relying on labor services is needed to transform intermediate goods into final consumption goods in each country. This structure gives rise to a pricing chain with five price indices: retail consumer prices, wholesale producer prices, import and export prices, and factor prices (that is, wages).

Beginning with the demand side and consumer preferences from the home country’s perspective, expected lifetime utility of a household, indexed by l∈[0,1], is assumed to be

where Cτ(l) and Lτ(l) are the household’s consumption basket and labor supply, respectively. The consumption basket, reflecting household preferences, is a constant–elasticity–of–substitution (CES) aggregate over individual varieties, indexed by c∈[0,1], of the final good and is given by

Each consumption good variety, meanwhile, is produced according to the following technology:

where Qt(c) is an index of both home and foreign varieties of the intermediate good, and NCt(c) is a bundle of differentiated labor services. This technology combines traded goods with nontraded labor services to produce consumption services.

The technology for producing a home variety, indexed by y, of the intermediate good is given by

where Zt(y) and NYt(y) are composites of differentiated intermediate and labor inputs. From these two equations, note that intermediate goods (indexed by Q) are used directly in the production of the final consumer good, and others (indexed by Z) are used in the production of other intermediate goods.

Specifically, each home intermediate variety is allocated across the four sectors summarized in the following equation:

where YQHt(y) and YZHt(y) denote the amounts of the home variety y used to meet domestic final and intermediate demands, respectively, and YQMt*(y)andYZMt*(y) denote the exports of the home variety y used to satisfy foreign final and intermediate demands, respectively.

Using YQMt(y*)andYZMt(y*) to denote the corresponding amounts of an imported foreign variety, indexed by y∈[0,1], the intermediate CES input bundles in the home country are defined as

Here, the elasticity of substitution σ between domestic and foreign baskets of the intermediate good is allowed to be different from the elasticity Ω between varieties within each bundle. Similar equations would apply to the foreign country.

Imported varieties (in both countries) are also assumed to go through distribution channels before their use, and the distribution process requires local labor services. Following Corsetti and Dedola (2002), one unit of an imported variety y* in the home country requires δ units of the labor service bundle, NMt(y*), so that19

Firms that face these local trading or distribution costs δ have an incentive for pricing to market or price discrimination across local and export markets.20 With this dependence on local currency wages, changes in the exchange rate lead, through a cost or supply mechanism, to incomplete pass-through in trade prices and the pricing-to–market behavior described by Krugman (1987).21 The introduction of pricing to market along these lines helps limit the effects of import price pass–through in retail prices and helps generate greater overall persistence in the degree of incomplete pass–through over time.22

Given this structure, household labor supply employed in the production of intermediate and final goods and distribution services,

and aggregate labor service bundles in the three activities are combined as follows:23

Marginal costs for final and intermediate goods derive from the employment of labor services and the costs of intermediate goods. From the production technologies in Equations (4.6) and (4.7), these are

where PQt and PZt are the familiar price indices associated with the CES bundles for intermediate inputs Qt and Zt in Equations (4.9) and (4.10), respectively, and Wt is the aggregate wage.

Given the CES demands for each variety derived (ultimately) from household preferences, monopolistically competitive firms set price as a markup over marginal costs to maximize profits. But prices and wages are not fully flexible. Instead, they are updated only infrequently based on Calvo-type (1983) adjustment.24 Specifically, the (fixed) probability that a firm will leave its price unchanged at a particular point in time is equal to π. Correspondingly, the average interval over which prices remain fixed is π/(1 - π).25 This framework generates staggered price–setting behavior in the economy.

With nominal rigidity in an open economy, the question arises as to whether prices are sticky in terms of domestic or foreign currency. The two limiting cases are referred to as (1) local currency pricing (LCP), in which all prices are rigid in the destination or buyer’s currency; and (2) producer currency pricing (PCP), in which all prices are rigid in the origin or seller’s currency. Allowing for a hybrid case, both types of pricing behavior are nested in the model. Specifically, the parameters Φ and Φ* denote the share of domestic and foreign firms, respectively, following PCP behavior. If Φ = 1, there is strict PCP; Φ = 0 represents strict LCP.

In the PCP case, a home producer would set its home export price PXt(y) at the point of origin; under LCP, the same firm would set the foreign import price PMt*(y) at the destination, where PMt*(y)=Pχt(y)/St and St is the nominal exchange rate. Using superscripts P and L to denote price setting under each type of behavior, the values χYtPandχχtP represent (under PCP) the domestic and export contract prices, respectively, under Calvo-type adjustment chosen by firms at time t to maximize the present discounted value of profits:

where DRt,τ is the stochastic discount rate (consistent with preferences), πτ–t is the probability that a price set at t will remain in place at τ, and the expressions within square brackets represent the domestic and foreign demands for the home variety, conditional on the price set at t.26 The analogous objective functions under LCP (and for retail firms) can be similarly derived.

Solving Equation (4.17) and its analog for retailers and for foreign firms—and under both PCP and LCP—yields the following central (log–linearized) pricing equations for aggregate producer, consumer, import, and export prices, respectively:

where lowercase letters denote logarithms of variables. In the presence of Calvo-type staggered price adjustment, aggregate price indices display inertial dynamics depending on the parameter related to the probability of price (non)adjustment π In the import and export price–setting equations, the behavior is also influenced strongly by the shares Φ, Φ* of LCP versus PCP firms. The underlying contracted prices (or components) entering these expressions are given by

The last equation (and that for its counterpart, xD,t*) reflects the role of distribution costs in the setting of trade prices, where ψCD is a weight related to the importance of unit distribution costs δ relative to marginal production costs for imports.27 Along with a similar inertial equation for wages, Equations (4.18)(4.21) represent the major behavioral price–setting equations in the model, and θ = { Φ,Φ*,π,δ} is the vector of key structural parameters used to help map model dynamics into those generated by the VAR with respect to exchange rate pass–through.

To close the model, monetary policy is determined by an interest rate rule, and uncovered interest rate parity is assumed to hold (up to an error term):

where ξ represents an exchange rate shock, perhaps reflecting the impact of noise traders.28

Benchmarks for parameters held fixed in the model follow the calibration in Choudhri, Faruqee, and Hakura (2005) and are generally based on average estimated values for the major industrial countries—including euro area members Germany, France, and Italy.29 Monetary policy, for example, is specified by an interest rate rule targeting expected inflation in consumer prices and allowing for interest rate smoothing, with the corresponding parameter weights based on the average estimates for the (non-U.S.) G–7 countries.

Matching Theory and Empirics

To match the empirical impulse response patterns, an asset market shock ξ to uncovered interest rate parity in the model is considered. The scale and persistence of ξ is calibrated to reproduce the VAR’s accumulated impulse response path for the exchange rate from a structural exchange rate shock εs.30 Simulating the model’s price responses to this uncovered interest rate parity shock generates the analytical impulse response functions. To find the optimal structural parameter values in the model consistent with the empirical evidence, the following loss function is minimized: 31

where IR(P,ξ;θ) denotes the vector of simulated impulse responses for the price vector P = {w, px, pm, pc, py} from a ξ shock in the model conditional on the key set of structural parameters θ = {Φ, Φ*, π, δ}.32IRVAR(Ps) denotes the corresponding vector of accumulated impulse responses to a structural exchange rate shock εs in the VAR. The time horizon for the impulse response paths is 18 months.

Figure 4.2 displays the combinations of LCP and PCP behavior in domestic and foreign firms that best replicate the price responses—particularly for trade prices—derived from the VAR estimates for each of the major industrial economies. The vertical axis measures the extent of PCP behavior Φ* in foreign firms (that is, those producing home imports), and the horizontal axis measures the extent of PCP behavior Φ in domestic firms (that is, producing home exports).

In the figure, the origin represents local currency pricing at home and abroad, whereas the opposing corner represents the case of universal producer currency pricing. The 45–degree line connecting them reflects symmetric combinations of local and producer currency pricing. In terms of symmetric outcomes, the midpoint on the 45–degree line tends to outperform either corner or extreme empirically, as shown by Choudhri, Faruqee, and Hakura (2002). The intuition is as follows: exclusive reliance on LCP behavior in both countries produces stable import prices at the destination, which generally squares well with the data, at the expense of unstable export prices at the point of origin, which does not. The reverse is true in the case of PCP. As is evident from the VAR results in Table 4.2, both export and import prices in a given currency show incomplete pass–through. Consequently, a mix of the LCP and PCP behaviors fares better than either behavior exclusively.

Figure 4.2.Pricing Behavior of Domestic and Foreign Firms

Figure 4.2 indicates, however, that the opposing diagonal to the one that is shown is even more relevant empirically in describing the pricing behavior of traded goods across countries. Specifically, asymmetric pricing behavior between domestic and foreign firms appears more consistent with the data. Allowing for LCP to a greater extent in certain trade prices and PCP to a greater extent in others tends to be more consistent with the data than requiring equal degrees of the two. Specifically, the smaller, more open economies and Japan gravitate toward pricing in foreign currencies (that is, LCP in their exports and PCP in their imports), whereas the larger, more insular economies of the United States and the euro area gravitate toward pricing in their domestic currencies (that is, LCP in their imports and PCP in their exports). As further supporting evidence, the implicit currency pricing behavior suggested by Figure 4.2 appears very consistent with independent data on invoice currencies for these countries.33

For the United Kingdom, pricing behavior is consistent with an outward orientation. Domestic firms exporting to foreign markets tend toward LCP and set and stabilize prices in terms of foreign or destination currencies, safeguarding export market share. Import prices also tend to be rigid in terms of foreign currencies, as foreign exporting firms are less sensitive to fluctuations in their destination currency prices. The situation in Canada is similar to that in the United Kingdom (albeit closer to the symmetric or equal–weighting case), with foreign currencies playing a large role in price–setting behavior of traded goods.

Despite its relative economic size and modest trade openness, Japan exhibits considerable outward orientation in its pricing behavior. From Table 4.2, recall that Japanese import and export prices show considerable pass–through on impact akin to that in the smaller, more open economies considered here. These results suggest that, in the model, Japanese exporting firms predominantly engage in local currency pricing, whereas foreign firms largely engage in producer currency pricing. This finding accords with the conventional wisdom regarding the behavior of Japanese and foreign (predominantly U.S.) exporting firms described in Giovannini (1988), Marston (1990), and others.34

For the larger, more insular economies, excluding Japan, both domestic and foreign firms tend to price in the dominant currency associated with the large economy. For foreign firms exporting to the United States and to the euro area, this translates into destination or local currency pricing in dollars and euros, respectively. For U.S. and euro area firms exporting abroad, this translates into origin or producer currency pricing. For the United States, this result is well established in the literature.

The similarity between the United States and the euro area is somewhat striking. Although it is generally accepted that U.S. firms exporting to the euro area (and elsewhere) primarily invoice in dollars and engage in dollar currency pricing, less is known about the pricing behavior of euro area firms.35 The results here suggest that the euro area already behaves in some measure like the United States with respect to the predominant role of domestic currency pricing for traded goods. The finding is largely driven by the similarly low degree of pass-through on impact in U.S. and area–wide import and export prices shown in Table 4.2. The use of manufacturing trade prices for the euro area accentuates this similarity, although using total extra–area trade prices (which show higher pass–through) would alter the relative placements shown in Figure 4.2 only slightly.

Some important differences between the United States and the euro area are worth highlighting. Specifically, the dynamic response of prices suggests significantly higher pass–through over time (particularly in import prices) for the euro area. Interestingly, this translates, in the model, into a higher estimate for the Calvo parameter π, reflecting greater nominal flexibility.36 Imposing a similar π value would shift the euro area in Figure 4.2 toward the other countries and away from the United States in terms of the mix of LCP versus PCP firms, although the euro area would remain more closely aligned with United States.

In all countries, exchange rate pass–through in consumer prices is significantly lower than that in import prices. In the model, this wedge between intermediate and final goods is captured by local distribution, marketing, and trading costs. Consequently, implicit estimates of these costs are fairly substantial. Country estimates of distribution costs—measured as a share of final goods price—are shown in Table 4.3. The estimates are expressed as broad ranges, reflecting the fact that certain combinations of distribution costs and degree of nominal rigidity produce similar fits of the model. The overall range is from 45 percent for the United States to 70 percent for Japan.37Burstein, Neves, and Rebelo (2003) estimate a 40 percent share for distribution costs in the United States.

Table 4.4 summarizes the “optimized” structural parameters that minimize the distance between the IRFs from the model and each country’s VAR. The table also provides an R2 value, based on the minimized loss or sum of squared deviations between model and VAR normalized by the total sum of squares for prices from the VAR impulse response trajectories. The fit of the pass–through model is generally very good overall, albeit less so for the United States. In the case of the euro area, the impulse responses from the model compared with those of the VAR, including 90 percent confidence intervals, are shown in Figure 4.3.

Table 4.3.Implicit Estimates of Local “Distribution” Costs(In percent of final goods price)
United StatesEuro AreaJapanUnited KingdomCanada
45–6555–6565–7060–6550–65
Source: IMF staff estimates.Note: Estimate ranges are based on minimum distance between calibrated model and country VAR impulse response functions for alternative degrees of price rigidity.
Source: IMF staff estimates.Note: Estimate ranges are based on minimum distance between calibrated model and country VAR impulse response functions for alternative degrees of price rigidity.

Local currency pricing and pricing-to–market behavior notwithstanding, expenditure–switching effects still operate in these economies, although short–run trade elasticities may be small.38 Based on the calibrated model, an exchange rate depreciation can be shown to improve the trade balance overall, once both volume and pass–through effects are factored in.39 Given its generally small effects on the trade balance, however, the exchange rate adjustment required to achieve a moderate degree of external adjustment may be substantial in the absence of adjustment in other variables.

Conclusions

Incomplete exchange rate pass–through in prices is a well–documented empirical regularity for many economies, including the euro area. A better understanding of the economic behavior underlying limited pass–through is an important consideration for investigating the implications of currency fluctuations and the role of monetary and exchange rate policy. The analysis here has sought to examine pass–through in a set of euro area prices along the pricing chain using a VAR approach to identify the effects of an exogenous exchange rate shock. Mapping the effects of this structural shock into an analytical framework helps identify behavioral features that could help account for the nature of incomplete pass–through in the euro area. On the basis of the analysis, a number of conclusions can be drawn.

Table 4.4.Summary of Behavioral Parameter Estimates
United StatesEuro AreaJapanUnited KingdomCanada
Φ1.000.940.240.400.66
Φ*0.040.180.800.680.78
π0.050.100.050.050.10
δ0.650.600.650.650.60
R20.610.910.980.910.88
Source: IMF staff estimates.Note: Table presents optimized values based on minimum distance between calibrated model and country VAR impulse response functions. R2 values are based on sum of squared deviations between model and VAR impulse responses normalized by total sum of squares from VAR impulse responses over 18 months. Values for δ represent local distribution costs expressed in percent of final goods price.
Source: IMF staff estimates.Note: Table presents optimized values based on minimum distance between calibrated model and country VAR impulse response functions. R2 values are based on sum of squared deviations between model and VAR impulse responses normalized by total sum of squares from VAR impulse responses over 18 months. Values for δ represent local distribution costs expressed in percent of final goods price.

First, short-run pass–through is low in the euro area for a wide range of prices. An exchange rate shock has nearly a zero impact on factor and trade prices, and on wholesale and retail prices, much as is the case with the United States.

Second, pass–through tends to rise over time in the euro area, although the extent of wage and consumer price pass–through remains comparatively small. Pass–through in euro area producer and export prices is somewhat higher, but the highest degree of pass–though (near unity) is reflected in import prices. The differences in relative pass–through in import prices and consumer prices, for example, suggest that the roles of the retail sector and local distribution costs are important for price determination.

Third, the pattern of pass–through in trade prices suggests a fair degree of asymmetry with respect to the pricing behavior of domestic and foreign firms operating in the euro area. Specifically, the impulse response patterns suggest a high degree of local currency pricing in import prices and producer currency pricing in export prices. For the euro area (and the United States), this suggests that these traded goods (that is, both imports and exports) are priced predominantly in euros (and dollars). For Japan and smaller, more open economies, the behavior of pass–through is more consistent with importers’ and exporters’ operating significantly in foreign currencies.

Finally, local currency pricing and pricing–to–market behavior notwithstanding, expenditure–switching effects still operate in all these economies, although short–run trade elasticities can be small. Nevertheless, the model suggests that an exchange rate depreciation improves the trade balance, once both volume and pass–through effects are factored in. Given the exchange rate’s generally small effects on the trade balance, however, the exchange rate adjustment required to achieve a moderate degree of external adjustment may, in some circumstances, be substantial in the absence of adjustment in other variables.

Figure 4.3.Empirical and Theoretical Impulse Response Paths for the Euro Area

Note: Shaded areas show 90 percent confidence intervals based on VAR estimates.

Appendix 4.1 Stationarity, Causality, and Sensitivity Analysis

Unit Root and Stationarity Tests

Test statistics based on Phillips and Perron’s (1998) nonparametric unit root test and the stationarity test of Kwiatkowski, Phillips, Schmidt, and Shin (KPSS; 1992) are shown in Table 4.A.1. The alternative hypothesis with the Phillips-Perron test and the null hypothesis with the KPSS test is trend stationarity. The tests suggest that the levels (first differences) of these variables are nonstationary (stationary). The borderline cases, in which the tests give conflicting answers, are CPI inflation (for which tests reject a unit root and stationarity, respectively) and import price inflation (for which tests fail to reject a unit root and stationarity, respectively).

Table 4.A.1.Unit Root and Stationarity Tests(Euro area monthly data, 1990-2002)
VariablePhillips–Perron Zt TestKPSS ητ TestOrder of Integration
NEER–2.340.34**I(1)
Δ NEER–8.46**0.07I(0)
CPI–2.890.74**
Δ CPI–10.87**0.36**I(0) or
PPI–1.660.32**I(1)
Δ PPI–5.86**0.09I(0)
px–0.920.27**I(1)
Δ px–3.29*0.10I(0)
pm–2.380.15**I(1)
Δ pm–2.610.11I(0) or I(1)
Wage–2.460.63**I(1)
Δ Wage–6.56**0.07I(0)
Source: IMF staff estimates.Note: NEER refers to nominal effective exchange rate. The Phillips-Perron (1998) unit root test includes deterministic time trend under the alternative. The Kwiatkowski and others (KPSS; 1992) stationarity test includes deterministic time trend under the null.

significant at the 10 percent level;

significant at the 5 percent level.

Source: IMF staff estimates.Note: NEER refers to nominal effective exchange rate. The Phillips-Perron (1998) unit root test includes deterministic time trend under the alternative. The Kwiatkowski and others (KPSS; 1992) stationarity test includes deterministic time trend under the null.

significant at the 10 percent level;

significant at the 5 percent level.

Table 4.A.2.Cointegration Tests(Euro area monthly data, 1990-2002)
Number of RegressorsDemeanedDemeaned and Detrended
n = 5115.10120.43
(225.23)(284.01)
Source: IMF staff estimates.Note: The table reports the multivariate trace statistic or Pz test statistic based on Phillips and Ouliaris (1990); 10 percent critical value given in parentheses.
Source: IMF staff estimates.Note: The table reports the multivariate trace statistic or Pz test statistic based on Phillips and Ouliaris (1990); 10 percent critical value given in parentheses.

Cointegration Tests

Given that the unit root and stationarity tests suggest that the log levels of the exchange rate and various price measures are nonstationary, cointe–gration tests were conducted to examine whether a linear combination of these variables is stationary. Table 4.A.2 reports the results of the Phillips and Ouliaris (1990) residual–based test for cointegration with window size = 2. Other window sizes produce similar results. The tests fail to reject the null of no cointegration.

Granger Causality Tests

Granger causality tests were conducted to examine whether changes in euro area exchange rates and prices have predictive content for each other. Simple bivariate tests indicate that the nominal effective exchange rate Granger–causes (that is, helps predict) several price measures, but prices fail to Granger–cause exchange rates. The exchange rate is found to be a significant predictor for trade prices and has predictive content for wages at shorter lag length (see Table 4.A.3).

Table 4.A.3.Bivariate Granger Causality Tests(Euro area monthly data, 1990-2002)
Price MeasureLagsH0: Exchange Rate Does Not Cause PricesLagsH0: Prices Do Not Cause Exchange Rate
CPI120.6620.96
PPI140.9920.87
px130.0620.27
pm130.0020.70
Wage1140.5420.47
Source: IMF staff estimates.Note: Reported numbers are p–values on the relevant exclusion restriction (F–test). Lag length selection based on Akaike information criterion.

At one lag, the exchange rate Granger-causes wages (p = 0.04).

Source: IMF staff estimates.Note: Reported numbers are p–values on the relevant exclusion restriction (F–test). Lag length selection based on Akaike information criterion.

At one lag, the exchange rate Granger-causes wages (p = 0.04).

Table 4.A.4.Block Exogeneity Tests(Euro area monthly data, 1990-2002)
Lag LengthH0: A2 = 0H0: B1 = 0
p = 732.42 (0.59)50.48 (0.04)
Source: IMF staff estimates.Note: Test statistic is based on likelihood ratio test with degrees–of–freedom correction and is distributed as χ2(5p). Significance levels are given in parentheses.
Source: IMF staff estimates.Note: Test statistic is based on likelihood ratio test with degrees–of–freedom correction and is distributed as χ2(5p). Significance levels are given in parentheses.

Block Exogeneity Tests

To generalize Granger causality tests to a multivariate context, consider the following partitioned VAR(p) system:

where c1 and c2 represent a constant term and monthly time dummies, P represents the (5 x 1) vector of price variables, X1 and X2 are (p x 1) and (5p x 1) vectors of lagged changes in exchange rates and prices, respectively, with conformable matrices of autoregressive coefficients A1, A2, B1, and B2. Block exogeneity test results are shown in Table 4.A.4.

Sensitivity Analysis

As shown in Table 4.A.5, the implied pass–through elasticities from the two VARs (that is, those with the exchange rate first and last in the ordering) are fairly similar. On impact, euro area prices are essentially predetermined in response to the exchange rate shock in the two VARs. It should be stressed, however, that the exchange rate shock has zero contemporaneous effect on prices (that is, prices are exactly predetermined) by construction in the latter VAR, based on the ordering of variables. In the case of the euro area (and the United States) this restriction appears valid empirically. Hence, the reordering makes little difference. For other industrial countries, however, this is generally not true for trade prices, as shown by Table 4.2, and should thus not be imposed a priori. So long as the exchange rate appears before trade prices in the VAR for these countries, alternative impulse response functions will closely match those reported in the text.

Table 4.A.5.Euro Area Pass-Through Elasticities Across VAR Reorderings
t = 1t = 6t = 12t = 18
a. Exchange rate first in VAR
CPI0.000.010.020.02
PPI0.000.040.110.17
Wage–0.020.040.070.05
px0.020.180.310.45
pm0.030.420.811.17
b. Exchange rate last in VAR
CPI00.020.020.03
PPI00.050.110.18
Wage00.040.080.05
px00.100.200.30
pm00.300.630.97
Source: IMF staff estimates.Note: Table reports percentage change in prices divided by percentage change in exchange rate in response to unit exchange rate shock at different time intervals, in terms of months after the shock.
Source: IMF staff estimates.Note: Table reports percentage change in prices divided by percentage change in exchange rate in response to unit exchange rate shock at different time intervals, in terms of months after the shock.
References

    AndertonRobert2003“Extra-Euro Area Manufacturing Import Prices and Exchange Rate Pass–Through,”ECB Working Paper No. 219 (Frankfurt: European Central Bank).

    • Search Google Scholar
    • Export Citation

    BekxPeter1998“The Implications of the Introduction of the Euro for Non–EU Countries,”Euro Papers No. 26 (Brussels: European Commission).

    • Search Google Scholar
    • Export Citation

    BerginPaul and RobertC. Feenstra2001“Pricing–to–Market, Staggered Contracts, and Real Exchange Rate Persistence,”Journal of International EconomicsVol. 54 (August) pp. 33359.

    • Search Google Scholar
    • Export Citation

    BettsCaroline and MichaelB. Devereux1996“The Exchange Rate in a Model of Pricing–to–Market,”European Economic ReviewVol. 40 (April) pp. 100721.

    • Search Google Scholar
    • Export Citation

    BettsCaroline and MichaelB. Devereux2000“Exchange Rate Dynamics in a Model of Pricing–to–Market,”Journal of International EconomicsVol. 50 (February) pp. 21544.

    • Search Google Scholar
    • Export Citation

    BursteinArielT.JoaoC. Neves and SergioRebelo2003“Distribution Costs and Real Exchange Rate Dynamics During Exchange–Rate–Based Stabilizations,”Journal of Monetary EconomicsVol. 50 (September) pp. 1189214.

    • Search Google Scholar
    • Export Citation

    CalvoGuillermoA.1983“Staggered Prices in a Utility-Maximizing Framework,”Journal of Monetary EconomicsVol. 12 (September) pp. 38398.

    • Search Google Scholar
    • Export Citation

    CampaJoseM. and LindaS. Goldberg2002“Exchange Rate Pass-Through into Import Prices: A Macro or Micro Phenomenon?”Staff Reports No. 149 (New York: Federal Reserve Bank of New York).

    • Search Google Scholar
    • Export Citation

    CheungY.M.Chinn and A.Pascual2002“Empirical Exchange Rate Models of the Nineties: Are Any Fit to Survive?”NBER Working Paper No. 9393 (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation

    ChoudhriEhsanU.HamidFaruqee and DaliaS. Hakura2002“Explaining the Exchange Rate Pass–Through in Different Prices,”IMF Working Paper 02/224 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    ChoudhriEhsanU.HamidFaruqee and DaliaS. Hakura2005“Explaining the Exchange Rate Pass-Through in Different Prices,”Journal of International EconomicsVol. 65 (March) pp. 34974.

    • Search Google Scholar
    • Export Citation

    ChoudhriEhsanU. and DaliaS. Hakura2001“Exchange Rate Pass-Through to Domestic Prices: Does the Inflationary Environment Matter?”IMF Working Paper 01/194 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    CorsettiGiancarlo and LucaDedola2002“Macroeconomics of International Price Discrimination,”ECB Working Paper No. 176 (Frankfurt: European Central Bank).

    • Search Google Scholar
    • Export Citation

    CorsettiGiancarlo and PaoloPesenti2002“Self–Validating Optimum Currency Areas,”NBER Working Paper No. 8783 (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation

    DevereuxMichael and CharlesEngel2002“Exchange Rate Pass-Through, Exchange Rate Volatility, and Exchange Rate Disconnect,”Journal of Monetary EconomicsVol. 49 (July) pp. 91340.

    • Search Google Scholar
    • Export Citation

    DevereuxMichael and CharlesEngel2003“Monetary Policy in the Open Economy Revisited: Price Setting and Exchange–Rate Flexibility,”Review of Economic StudiesVol. 70 (October) pp. 76583.

    • Search Google Scholar
    • Export Citation

    DevereuxMichael and CharlesEngelCedricTille1999“Exchange Rate Pass-Through and the Welfare Effects of the Euro,”NBER Working Paper No. 7382 (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation

    DominguezKathryn1999“The Role of the Yen,” in International Capital Flowsed. by MartinFeldstein (Chicago: University of Chicago Press), pp. 13371.

    • Search Google Scholar
    • Export Citation

    EngelCharles2002“Expenditure Switching and Exchange Rate Policy,” in NBER Macroeconomics Annual 2002ed. by MarkGertler and KennethRogoff (Cambridge, Massachusetts: MIT Press).

    • Search Google Scholar
    • Export Citation

    FaruqeeHamid1995“Pricing to Market and the Real Exchange Rate,”IMF Staff PapersVol. 42 (December) pp. 85581.

    FaruqeeHamid2003“Exchange Rate Pass-Through and External Adjustment in the Euro Area,” in Euro Area Policies: Selected IssuesIMF Country Report No. 03/298 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    FlathDavid2003“Regulation, Distribution Efficiency, and Retail Density,”NBER Working Paper No. 9450 (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation

    FloodRobertP. and AndrewK. Rose1995“Fixing Exchange Rates, A Virtual Quest for Fundamentals,”Journal of Monetary EconomicsVol. 36 (August) pp. 337.

    • Search Google Scholar
    • Export Citation

    GagnonJoseph and JaneIhrig2001“Monetary Policy and Exchange Rate Pass-Through,”International Finance Discussion Papers No. 704 (revised March2002) (Washington: Board of Governors of the Federal Reserve System).

    • Search Google Scholar
    • Export Citation

    GiovanniniAlberto1988“Exchange Rates and Traded Goods Prices,”Journal of International EconomicsVol. 24 (February) pp. 4568.

    • Search Google Scholar
    • Export Citation

    GoldbergP. and M.Knetter1997“Goods Prices and Exchange Rates: What Have We Learned?”Journal of Economic LiteratureVol. 35 (September) pp. 124372.

    • Search Google Scholar
    • Export Citation

    HahnElke2003“Pass–Through of External Shocks to Euro Area Inflation,”ECB Working Paper No. 243 (Frankfurt: European Central Bank).

    • Search Google Scholar
    • Export Citation

    HamiltonJamesD.1994Time Series Analysis (Princeton, New Jersey: Princeton University Press).

    HerzbergValerieGeorgeKapetanios and SimonPrice2003“Import Prices and Exchange Rate Pass–Through: Theory and Evidence from the United Kingdom,”Bank of England Working Paper No. 182 (London: Bank of England).

    • Search Google Scholar
    • Export Citation

    HiifnerFelixP. and MichaelSchroder2002“Exchange Rate Pass-Through to Consumer Prices: A European Perspective,”ZEW Discussion Paper No. 02–20 (Mannheim, Germany: Center for European Economic Research).

    • Search Google Scholar
    • Export Citation

    IsardPeterHamidFaruqeeRussellKincaid and MartinFetherstonPeter2001Methodology for Current Account and Exchange Rate AssessmentOccasional Paper No. 209 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    JeanneOlivier and AndrewK. Rose2002“Noise Trading and Exchange Rate Regimes,”Quarterly Journal of EconomicsVol. 117 (May) pp. 53769.

    • Search Google Scholar
    • Export Citation

    KasaKenneth1992“Adjustment Costs and Pricing-to-Market Theory and Evidence,”Journal of International EconomicsVol. 32 (February) pp. 130.

    • Search Google Scholar
    • Export Citation

    KielerMads2001“Tracking the Pass-Through of External Shocks to Euro Area Inflation,” in Monetary and Exchange Rate Policies of the Euro Area—Selected Issuesprepared byA.JaegerK.RossZ.KontolemisM.Kieler and G.MeredithIMF Country Report No. 01/201 (Washington: International Monetary Fund), pp. 5267.

    • Search Google Scholar
    • Export Citation

    KollmannRobert2001“The Exchange Rate in a Dynamic-Optimizing Business Cycle Model with Nominal Rigidities: A Quantitative Investigation,”Journal of International EconomicsVol. 55 (December) pp. 24362.

    • Search Google Scholar
    • Export Citation

    KrugmanPaul1987“Pricing to Market When the Exchange Rate Changes,” in Real–Financial Linkages Among Open Economicsed. by SvenW. Arndt and J. DavidRichardson (Cambridge, Massachusetts: MIT Press), pp. 4970.

    • Search Google Scholar
    • Export Citation

    KrugmanPaul1989Exchange-Rate Instability (Cambridge, Massachusetts: MIT Press).

    KwiatkowskiDenisPeterPhillipsPeterSchmidt and YongcheolShin1992“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 EconometricsVol. 54 (October-December) pp. 15978.

    • Search Google Scholar
    • Export Citation

    LanePhili2001“The New Open Economy Macroeconomics: A Survey,”Journal of International EconomicsVol. 54 (December) pp. 23566.

    • Search Google Scholar
    • Export Citation

    MarstonRichard1990“Pricing to Market in Japanese Manufacturing,”Journal of International EconomicsVol. 29 (November) pp. 21736.

    • Search Google Scholar
    • Export Citation

    McCarthyJonathan2000“Pass–Through of Exchange Rates and Import Prices to Domestic Inflation in Some Industrialized Economies,”Staff Reports No. 111 (New York: Federal Reserve Bank of New York).

    • Search Google Scholar
    • Export Citation

    MeeseRichard and KennethRogoff1983“Empirical Exchange Rate Models of the Seventies: Do They Fit out of Sample?”Journal of International EconomicsVol. 14 (February) pp. 324.

    • Search Google Scholar
    • Export Citation

    MussaMichael1984“The Theory of Exchange Rate Determination,” in Exchange Rate Theory and Practiceed. by JohnF. O. Bilson and RichardC. Marston (Chicago: University of Chicago Press), pp. 1378.

    • Search Google Scholar
    • Export Citation

    ObstfeldMaurice2002“Exchange Rates and Adjustment: Perspectives from the New Open Economy Macroeconomics,”NBER Working Paper No. 9118 (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation

    ObstfeldMauriceKennethRogoff1995“Exchange Rate Dynamics Redux,”Journal of Political EconomyVol. 103 (June) pp. 62460.

    ObstfeldMaurice2000“The Six Major Puzzles in International Macroeconomics: Is There a Common Cause?”NBER Working Paper No. 7777 (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation

    PeersmanGert and FrankSmets2001“The Monetary Transmission Mechanism in the Euro Area: More Evidence from VAR Analysis,”ECB Working Paper No. 91 (Frankfurt: European Central Bank).

    • Search Google Scholar
    • Export Citation

    PhillipsP.C.B. and S.Ouliaris1990“Asymptotic Properties of Residual Based Tests for Cointegration,”EconometricaVol. 58 (January) pp. 16593.

    • Search Google Scholar
    • Export Citation

    PhillipsP.C.B. and P.Perron1998“Testing for a Unit Root in Time Series Regressions,”BiometrikaVol. 75 (June) pp. 33546.

    SarnoL. and M.Taylor2002The Economics of Exchange Rates (Cambridge: Cambridge University Press).

    SmetsFrank and RafWouters2002“Openness, Imperfect Exchange Rate Pass-Through and Monetary Policy,”Journal of Monetary EconomicsVol. 49 (July) pp. 94781.

    • Search Google Scholar
    • Export Citation

Special thanks are due to Susanna Mursula for her outstanding research assistance and to Bankim Chada, Ehsan Choudhri, Dalia Hakura, Albert Jaeger, and Alessandro Rebucci for helpful discussions and comments.

1

Partner countries are Australia, Canada, Denmark, Hong Kong SAR, Japan, Norway, Singapore, Republic of Korea, Sweden, Switzerland, the United Kingdom, and the United States.

2

Price and wage data are from Eurostat and are not seasonally adjusted. The area–wide measures reflect aggregates of the 11 euro area members through December 2000. Thereafter, the chained series include Greece as the 12th member country after it adopted the euro.

3

Manufactured goods include Sections 5–8 of the Standard International Trade Classification. The use of unit values (that is, aggregate, implicit deflators) is not ideal, but data limitations on direct trade prices make it necessary.

4

For Canada, Japan, the United Kingdom, and the United States, all corresponding monthly series were drawn from the IMF’s International Financial Statistics, except that for wages, which was obtained from the Organization for Economic Cooperation and Development’s Analytical Database (ADB).

5

The differenced series for euro area consumer and import prices were found to be borderline nonstationary (see Appendix 4.1). But as is well known, unit root and stationarity tests have low power, making it difficult to distinguish between stationary and unit root processes in finite samples.

6

The identification scheme largely follows Choudhri, Faruqee, and Hakura (2005). That analysis also includes the interest rate in the VAR in order to further distinguish between the effects of interest rate and exchange rate shocks on prices. McCarthy (2000) and Hahn (2003) also use a Cholesky decomposition to examine pass–through based on a somewhat different model.

7

For the first variable (that is, the exchange rate), the orthogonalized disturbance term is given by ε1t= μ1t. For the jth variable (j > 1) in the VAR, the corresponding shock term is given by εjt = μjt–cj,1–εjt …–cj, j–1εjt, where cj, i correspond to the entries of the Cholesky matrix (see Hamilton, 1994).

8

The exchange rate, however, helps predict (at least) trade prices (see Appendix 4.1). Restricted VAR estimates (not reported) excluding lagged prices from the exchange rate equation yield results very similar to those reported here.

9

See the seminal paper by Meese and Rogoff (1983). More recently, see, for example, Flood and Rose (1995) and Cheung, Chinn, and Pascual (2002). For a review, see Sarno and Taylor (2002).

10

Given the large number of parameters in the VAR, a parsimonious lag structure is sought to conserve degrees of freedom. The lag length p is chosen by starting with given maximum lag length and sequentially testing the incremental significance of dropping an additional lag based on the likelihood ratio test. Confidence intervals for the IRFs derived from bootstrapping methods are shown in Figure 4.3.

11

Obstfeld and Rogoff (2000) provide evidence that the terms of trade decline with a currency depreciation—an observation that appears at odds with the implications of strict local currency pricing models. See Lane (2001) for a review.

12

The exchange rate path is typically found to be less persistent (that is, more mean–reverting) in the quarterly VAR when interest rates are included (and are first in the ordering), but the (absolute and relative) pass–through effects are more similar to those reported here.

13

Gagnon and Ihrig (2001) find low degrees of pass–through (around 5 percent) in consumer prices for 20 industrial countries and argue that pass–through has been declining.

14

These estimates should be taken as indicative, given the standard errors of the impulse response functions. Based on bootstrapped standard errors, the 90 percent confidence interval for import prices is the widest, suggesting pass–through at 18 months in the range (0.6, 1.4). The median value of this band suggests a slightly lower degree of import price pass–through (that is, 0.95 at 18 months); otherwise, the median and estimated IRFs closely coincide.

15

The time pattern is similar, though, with most of the price adjustment occurring within five quarters. Based on a quarterly VAR, Hahn (2003) reports similar findings on the degree and time pattern of pass–through in the euro area non–oil import deflator.

16

From the variance–covariance matrix, the correlations between residuals are less than 0.2, with the notable exceptions of those between the exchange rate and trade prices and those between trade prices themselves.

17

This reduces the number of relevant reorderings considerably. The possible cases can be enumerated as follows: Σx=055Cx=32, where nCr denotes “n choose r” combinations.

18

These results obtain despite the fact that the impulse response path for the exchange rate itself differs across the two orderings, displaying less persistence in the latter case, where the exchange rate shock depends recursively on the reduced–form residuals from all the price equations. The interpretation of the exchange rate shock becomes more difficult in this case, since it incorporates a mix of innovation terms.

19

The Leontief–type distribution process is needed only for moving intermediate goods across borders. For simplicity, the distribution cost for imports is the same whether they are used in the production of the intermediate or the final good. Note, however, that the consumption technology (Equation (4.6)) specifies a Cobb–Douglas process for transforming both home and foreign intermediates into a final good.

20

Investigating extra–area import prices, Anderton (2003) finds that foreign suppliers attach a significant weight to the pricing–to–market strategy in efforts to maintain market share. Herzberg, Kapetanios, and Price (2003) find pricing to market to be the dominant consideration for U.K. import prices. Kieler (2001) provides comparative estimates for several industrial countries.

21

See also Kasa (1992) and Faruqee (1995) for analyses that introduce market–specific costs as a way to generate incomplete pass–through and pricing–to–market behavior. An alternative approach focusing on varying markups and demand elasticities through trans–log (that is, non–CES) preferences can be found in Bergin and Feenstra (2001).

22

The home import price (in local currency terms) of a foreign variety, after distribution costs, is P˜Mt(y*) = PMt(y*) + δWt(where PMt(y*) is the price before distribution costs) and thus depends partly on the domestic wage W.

23

Individual labor demands facing households depend on relative wages and aggregate labor demand: Ltd(l)=(NCt+NYt+NMt)(Wt(l)/Wt)ɛ. See Choudhri, Faruqee, and Hakura(2005).

24

See Kollmann (2001) for a new open economy macroeconomics analysis with sticky wages and prices but without distribution costs.

25

Choudhri, Faruqee, and Hakura (2002) also examine the cases of wage and price flexibility (that is, π = 0) and find that these model variants generally fall short in explaining the empirical impulse responses.

26

The firms’ stochastic discount factor is consistent with that of households, based on the Euler condition in consumption:

where r is the rate of interest.

27

Specifically, one can derive

evaluated at steady-state values.

28

In a noise trader model, market participants can be subject to a stochastic bias in their expectations,

generating an uncovered interest rate parity or exchange rate shock in Equation (4.27).

29

Reflecting the higher (monthly) frequency of the data, the discount factor and interest rate are adjusted to produce annualized rates consistent with Choudhri, Faruqee, and Hakura (2005).

30

The exchange rate shocks from the VAR have permanent or highly persistent effects on the log level of the exchange rate. Consequently, the ξ, shocks in the model are constructed to be permanent or highly persistent as well.

31

The standard errors for the accumulated impulse response trajectories are similar, suggesting that minimizing the unweighted least squares should yield parameters broadly similar to the weighted least–squares approach used in Smets and Wouters (2002). The weighted approach would give less weight to import prices in favor of consumer prices.

32

The elasticity of substitution σ between traded goods, broadly defined, was also chosen to minimize the distance between IRFs and found to be fairly low (σ = 2), in line with Choudhri, Faruqee, and Hakura (2002), Smets and Wouters (2002), and the references cited therein. The elasticity of substitution 0 between specific varieties was also allowed to vary within a range (θ = 4 to 10), consistent with plausible markups.

33

Bekx (1998) reports the following 1995 percentages of import and export prices, respectively, that are invoiced in domestic currency: Germany (52, 75), Japan (23, 36), the United Kingdom (43, 62), and the United States (81, 92). Note that the percentages for Germany refer to deutsche marks rather than all euro area legacy currencies.

34

See Dominguez (1999) for a related discussion of the limited role of the yen as an international currency, particularly as an invoicing currency. In principle, the choice of invoicing currency is distinct from the issues of local versus producer currency pricing, but these issues have considerable overlap in practice.

35

Some have suggested that the international role of the euro in this regard will expand over time. See Bekx (1998) and Devereux, Engel, and Tille (1999).

36

The estimates for π typically range from 5 percent to 10 percent, suggesting an average duration between price changes of three to six quarters. The π estimates for the euro area (United States) are at the higher (lower) end of this range, suggesting shorter (longer) durations between price changes.

37

High estimates of implicit distribution costs in Japan may reflect aspects of that country’s peculiar distribution system—including restrictive government regulations, predominance of small stores, complex marketing channels, and pervasive constraints on vertical integration. See Flath (2003).

38

The classic Marshall–Lerner–Robinson condition requires that the sum of trade elasticities exceed unity for a depreciation to improve the trade balance under traditional pass–through assumptions (that is, zero and full pass–through in export and import prices, respectively). For example, Isard and others (2001) use elasticity benchmarks of 0.7 and 0.9 for export and import volumes. When pass–through is incomplete, however, the Marshall–Lerner–Robinson condition need not apply.

39

Using the calibrated model described here, Faruqee (2003) examines the trade impact of a uniform dollar decline on the external positions of the major industrial economies.

    Other Resources Citing This Publication