The idea of the shock dependency of the pass-through emerged from the understanding that exchange rate movements can have different effects on prices. The underlying reasons for shock dependence of exchange rate pass-through (ERPT) are many. For example, price-setters may react differently to exchange rate movements triggered by different economic shocks. Reduced-form estimates using macro data can yield a useful rule of thumb for import prices, although better estimates of pricing equations can be obtained from micro data when an instrument for identification is available. Highly complex mechanisms, as well as the limited availability and lack of richness of micro datasets on consumer prices as compared to those on trade prices, make it difficult to rely on rules of thumb to assess the impact of the exchange rate on consumer prices. While a simple reduced-form estimate of ERPT may be informative for import prices, a more shock-dependent approach to assessing pass-through should also be considered for consumer prices and its components.
This assumption that the exchange rate is governed by exogenous shocks only to itself has already been challenged in theoretical contributions (Corsetti and Dedola 2008), but only recently applied in empirical frameworks.1 This topic has become especially relevant among policymakers following the contribution by Forbes et al. (2018), who study this for the UK (also described in their VoxEU column here).
We look at this new measure – the shock-dependent ERPT – and interpret this not as a ‘transmission’ of exchange rate changes, but rather as a ‘co-movement’ of prices and exchange rates given different shocks.2 Both the regular ERPT and the shock-dependent measure are useful to assess the relationship between exchange rates and prices, and they complement each other.
In this column, we aim to shed a light on shock-dependent ERPT for the euro area and its member states (Comunale 2020), drawing comparison and looking at the robust results across available structural empirical frameworks (here Bayesian VARs) and fully-fledged structural models (DSGEs) for the euro area.
The BVARs considered are as in Comunale and Kunovac (2017), based on the identification applied in Forbes et al. (2018),3 as in an updated version of Conti et al. (2017) and from Leiva et al. (2020). The main DSGE model is the one by the National Bank of Belgium (NBB) as described in De Walque et al. (2017). We use a common dataset across models, covering the period from 1999Q1 to 2019Q2. The identification schemes across the available BVARs are highly comparable and almost completely match (up to some unrestricted responses) those in the structural model from De Walque et al. (2017).4
Shock-dependent ERPT to consumer prices
Here we look at four common shocks: domestic demand, domestic supply, monetary policy, and exogenous exchange rate shocks (to broad nominal effective exchange rate, NEER, or bilateral euro/US dollar). Moreover, different global shocks are also taken into account. Monetary policy is included in both its forms, conventional as well as unconventional via shadow rates,5 and taken for the euro area domestically, and for the US, or in relative terms, depending on the model.
We find that for the euro area different domestic and global shocks can be associated with widely different pass-throughs. On impact, the results are similar across models (see Table 1). At longer horizons, the magnitude increases in fully-fledged structural models. The highest value is experienced in case of monetary policy shocks (either domestic or relative to the US). The shock-dependent ERPT for the exogenous exchange rate shock is particularly relevant for import prices, while smaller than the other values for consumer prices. We also find a relatively large value following domestic demand, even if with an opposite sign as expected.6 This may have important consequences for how we assess the pass-through when the euro exchange rate movement is predominantly driven by demand shocks.
Table 1 Empirical estimates for shock-dependent ERPT to consumer prices in the euro area
Impact after Q1-Q12 of 1% depreciation in exchange rate (NEER or euro/US dollar
Note: Impact after Q1-Q12 of 1% depreciation in exchange rate (NEER or euro/US dollar). Forbes refers to Forbes at el. (2018) identification, CK is Comunale and Kunovac (2017), BdE is Leiva et al. (2020), BdI is Conti et al. (2017). Forbes, BdI and CK use NEER-38, BdE instead includes EUR/USD. Monetary policy is taken in relative terms with respect to the US in CK and BdE. BdI treats US monetary policy as an exogenous variable. Data points until 2019Q1 (or Q2). Median of the DSGEs is obtained from different euro area models described in Ortega and Osbat (eds.) (2020). The model by the Bank of Finland is not included in the results for the supply shock.
Source: Comunale (2020).
Exchange rate historical decomposition
We also look at the historical shock decomposition of the exchange rates, which is, in a way, the ‘narrative’ that a given model provides about the economic forces driving the economy. It turns out that establishing robustness of such narratives is much harder than establishing the shape of impulse responses and the shock-dependent ERPTs if we do not restrict our analysis to one specific period with a clearer narrative (An et al. 2020).
For this purpose, we use the period just before and following the announcement of the ECB’s Asset Purchase Programme (APP), i.e. from 2014Q1 to 2016Q2, to look at whether the models agree qualitatively on the cumulative contribution of shocks to cumulative changes in exchange rates. The exchange rates started appreciating after 2014Q3 (euro/US dollar) or 2014Q4 (nominal effective exchange rate). In the DSGE model, the exchange rate shock can anticipate depreciation one quarter ahead, partially dampened by the appreciation due to demand shocks. All the empirical models find that the overall demand shocks explain a substantial part of change in the exchange rate, while the monetary policy shocks play a minor role. In Comunale and Kunovac’s (2017) model, global shocks also seem to matter.
Results for euro area member states
It is difficult to find a robust characterisation across models of the configuration of shocks that drive the exchange rates and prices. The modelling challenges increase when looking at individual countries, because exchange rate and monetary policy shocks are mostly common to the whole euro area in the considered time span and cannot be fully identified using data from only one country. This is the reason why we provide a local projection exercise à la Jordà (2005) with common euro area shocks, identified in the euro area-specific models and extrapolated and used as regressors (as in Lane and Stracca 2018).
For common exchange rate shocks, the impact on consumer prices is largest in some new member states, but there is a wide range of estimates across models. For core consumer prices, the coefficients are even smaller. This is mainly because of low pass-through to price of services.
For common monetary policy shocks (for the euro area only and relative to the US), the impact is larger in all the considered models compared to the other responses. For instance, euro area monetary policy plays a bigger role for consumer prices compared to exogenous exchange rate shocks, and especially so for some new member rates. The impact of relative monetary policy shocks on consumer prices is the largest in Estonia, Spain, Portugal, and Luxembourg (Figure 1), while seems smaller for the Netherlands. This heterogeneity that we find across member states is strongly in line with the recent literature (Lane and Stracca 2018). The responses of core consumer prices are very small and can be mostly attributed to the price of both services and non-energy industrial goods.
Figure 1 Relative monetary policy shock and impact on consumer prices
Note: Local projections. Median values with 95% bands. Results are across BVAR models for Q1. Monetary policy is taken in relative terms with respect to the monetary policy in the US.
Source: Comunale (2020)
Overall, we conclude that different domestic and global shocks are associated with widely different pass-throughs, and that country characteristics may matter. On impact, the shock-dependent ERPTs are similar across models. The highest value is experienced in the case of monetary policy shocks (either domestic or relative); this is in line with the literature. We do find a contribution for domestic demand and exogenous exchange rate shocks in the APP announcement period.
Looking at country-specific results, the impact of exchange rates on consumer prices is the largest in some new member states, such as the Baltic States and Malta. Generally speaking, monetary policy plays a bigger role for consumer prices, especially for some of the new member states and the euro area periphery. This heterogeneity within the euro area accords well with Lane and Stracca (2018). Moreover, accounting for time-variation (as in Leiva et al. 2020), exogenous shocks to exchange rate seem to have passed to consumer prices with more intensity after 2010, and this is particularly true for most of the old euro area countries.
Authors’ note: The views expressed in this column are those of the authors and do not necessarily represent those of the ECB, the Bank of Lithuania or the ESCB.
An, L, M A Wynne and R Zhang (2020), “Shock-Dependent Exchange Rate Pass-Through: Evidence Based on a Narrative Sign Approach”, Federal Reserve Bank of Dallas, Globalization Institute Working Paper 379.
Comunale, M and D Kunovac (2017), “Exchange rate pass-through in the euro area”, s.l.: European Central Bank WP n.2003/2017.
Comunale, M (2020), “Shock dependence of exchange rate pass-through: a comparative analysis of BVARs and DSGEs”, Bank of Lithuania, Working Paper Series, No. 75/2020.
Conti, A M, S Neri and A Nobili (2017), “Low inflation and monetary policy in the euro area”, s.l.: ECB Working Paper Series No.2005.
Corsetti, G, L Dedola and S Leduc (2008), “High exchange-rate volatility and low pass-through”, Journal of Monetary Economics 55: 1113-1128.
De Walque, G, T Lejeune, Y Rychalovska and R Wouters (2017), “An estimated two-country EA-US model with limited exchange rate pass-through”,Working Paper Research 317, National Bank of Belgium.
De Walque, G, T Lejeune, A Rannenberg and R Wouters (2019) “Low pass-through and high spillovers in NOEM: what does help and what does not”, s.l.: National Bank of Belgium Working Paper, forthcoming.
Ortega, E and C Osbat (eds.) (2020), “Exchange rate pass-through in the euro area and EU countries”, Occasional Paper Series, ECB, Frankfurt am Main, 2020, forthcoming.
Forbes, K, I Hjortsoe and T Nenova (2018), “The shocks matter: improving our estimates of exchange rate pass-through”, Journal of International Economics, Volume 115, pp. 255-275 (see also a 2016 Vox column here)
Ha, J, M Stocker and H Yilmazkuday (2020), “Inflation and Exchange Rate Pass-Through” Journal of International Money and Finance, Volume 105, July 2020, 102187.
Jordá, O (2005), “Estimation and Inference of Impulse Responses by Local Projections” The American Economic Review, pp. 161-182.
Krippner, L (2016), “Documentation for measures of monetary policy”, New Zealand’s central bank, mimeo.
Lane, P and L Stracca (2018), “Can appreciation be expansionary? Evidence from the euro area”, Economic Policy, April 2018: 225–264.
Leiva, D, J Martínez-Martín and E Ortega (2020) “Exchange Rate Shocks and Inflation Comovement in the Euro Area”, s.l.: European Central Bank WP n.2383/2020.
Shambaugh, J (2008), “A new look at pass-through”, Journal of International Money and Finance 27(6): 560-591.
1 Past empirical works on the shock-dependent ERPT include Shambaugh (2008), Forbes et al. (2018) for the UK, Comunale and Kunovac (2017) for the euro area, An et al. (2020) for Japan and Ha et al. (2020) for a panel of countries.
2 This is also called the Price-to-Exchange-Rate Ratio (PERR) in Ortega and Osbat (2020).
3 The identification in Forbes et al. (2018), for the UK, is applied here for the euro area case. As a result, it is important to note that the outcomes reported here are not directly taken from Forbes et al. (2018) but only their identification is used for the purpose.
4 For more details on the shock-dependent ERPT in this DSGE model, see De Walque (2019).
5 We make use of the series by Leo Krippner available here https://www.rbnz.govt.nz/research-and-publications/research-programme/ad… also Krippner (2016).
6 This is especially true for the BVARs, while the values are positive in the DSGEs medians after Q1. However there are differences across DSGE models, with some of them experiencing also negative values at later horizons (Ortega and Osbat 2020).