Asymmetries in the transmission of oil price shocks to inflation in the eurozone

Asymmetries in the transmission of oil price shocks to inflation in the eurozone* Antonio J. Garzon a * , Luis A. Hierro a a Department of Economics and Economic History, University of Seville, Seville, Spain. Abstract Theory predicts that the level of inflation in an economy and the trend in the prevailing inflation rate may affect how oil price shocks, through their influence on firms’ expectations and price-setting behaviors, are transmitted to inflation. However, empirical evidence regarding this relationship is limited and calls for further exploration. Using data for 12 eurozone countries over the period from 1999 to 2020, we analyze how the inflation environment in which an oil price shock occurs influences its transmission. We find that the inflation environment is a determinant in the way oil supply shocks and oil-specific demand shocks are transmitted to inflation, with positive shocks displaying higher transmission in high inflation environments. Furthermore, transmission of shocks to core inflation, which represents an indirect effect, only occurs in high inflation environments. These findings highlight the need to consider the inflation environment in order to define appropriate monetary policies in response to inflationary pressures caused by oil price shocks. Keywords: oil price shocks; inflation; state-dependent; asymmetry JEL codes: E31; E37; Q43 * We thank the Editor (Sushanta Mallick), Associate Editor, Copy-Editor and three anonymous referees for their useful comments and suggestions, which have contributed enormously toward improving our paper. This work has been funded by the VI-PP of the University of Seville. * Corresponding author. Department of Economics and Economic History, University of Seville, 1, Avda. Ramon y Cajal, 41018, Seville, Spain. Email address: [email protected] 1. Introduction The price of oil as well as fluctuations in that price can significantly affect the evolution of a number of macroeconomic variables. The oil price shocks of the 1970s are a clear example of this (Hamilton, 1983; Mork, 1989; Hamilton, 1996); after those events, controlling inflation became a principal economic policy objective, particularly for monetary policy, and oil prices became a fundamental factor in analyzing economic conditions. Currently, most central banks have long-term inflation targets based on a core inflation measure that excludes an energy component. However, medium-term inflation targets are conditioned by the transmission of oil prices to consumer prices. Oil price shocks affect inflation through two different channels as follows: directly, as fuel prices are part of the energy component of the so-called headline inflation price index and indirectly, whereby oil price shocks alter production costs and generate second-round effects on wages and incomes that are indexed to changes in consumer prices (Peersman and Van Robays, 2009; Alvarez et al., 2011). Oil price changes are one of the main factors that influence the variability of the headline inflation rate in oil-importing countries, including the United States (US) (Blanchard and Gali, 2007) and regions such as the eurozone (Alvarez et al., 2011). Analyzing how oil price shocks are transmitted to inflation is of special relevance, particularly in a context in which cyclical fluctuations in the inflation rate are mainly determined by international factors such as commodity prices (Forbes, 2019), among which oil is key. In recent decades, inflation has trended downward in developed economies, especially after the global financial crisis, as shown in Figure 1 for eurozone inflation. In a climate of low inflation, analyzing how cost-push shocks such as oil price shocks are transmitted to consumer prices is of major importance. Theoretically, when the inflation trend is low, the degree to which cost shocks are transmitted to inflation is lower than in a high inflationary environment for a number of reasons. In an economy with staggered price settings and monopolistic competition, cost shocks in a low inflationary environment are perceived by firms as transitory such that firms are less likely to transmit these costs through their prices (Taylor, 2000). Following this line of thought, if the frequency with which firms update their prices is endogenous to the inflation environment such that the frequency of price updates increases when the inflation trend is higher (Devereux and Yetman, 2010), the transmission of cost shocks to inflation will be greater in an environment of high-trend inflation as more firms modify their prices more frequently, passing on these higher costs through their prices. Another possible source of non-linearity in the transmission of oil price shocks to inflation is the existence of an asymmetry that depends on the direction of the shock; in other words, the impact of positive shocks (oil price increases) and negative shocks may differ. This asymmetric transmission may arise from different channels. One such channel is via the direct impact of oil price shocks on fuel prices, as fuels represent an important share of the basket of goods and services in the eurozone’s Harmonized Index of Consumer Prices (HICP) and, therefore, directly affect headline inflation. This is the so-called “rockets and feathers” effect in the retail fuel market (Bacon, 1991) such that positive changes in the price of oil are transmitted to fuel prices to a greater degree than negative changes. This result has several theoretical explanations: the existence of market power in the retail fuel market because of its small number of competitors and/or high search costs (Borenstein, 1991; Borenstein et al., 3 1997); the asymmetric response of consumers to variations in fuel prices (Balke et al., 1998); or inventory management and accounting techniques used (Balke et al., 1998). Moreover, oil is an input in the production process of many firms and, hence, its price variation affects their marginal costs. In this sense, asymmetry may arise from the existence of menu costs. In an economy with positively trending inflation and monopolistic competition where firms set prices for several periods and face menu costs when changing their prices, firms adjust their prices more frequently in the face of increases in marginal costs than after decreases in those costs (Tsiddon, 1993; Ball and Mankiw, 1994). This implies that positive oil price shocks will transmit to inflation to a greater degree than negative shocks. Further, higher the inflation trend, the greater the asymmetry. Fig. 1. Evolution of eurozone average headline inflation and oil price changes (1999.01–2020.01). The black line represents the average year-on-year inflation rate of the 12 original eurozone countries. The dashed line represents the linear trend of the inflation rate. The red line represents the year-on-year change of oil prices, proxied by the US imported refiner acquisition cost of crude oil. Another source of this asymmetry is the differing impact of positive and negative oil price shocks on economic activity. It has been argued that positive shocks have a greater effect on economic activity than negative shocks. This asymmetric response may be caused by one or more factors, including: (1) rigidities in the labor market associated with worker reallocation costs in the most energy-intensive sectors, which amplifies the recessionary effect generated by oil price increases (Hamilton, 1988; Davis and Haltiwanger, 2001); (2) the asymmetric response of monetary policy to positive and negative shocks given that since the 1970s, central banks have shown a clear tendency to have a greater response to positive oil shocks than negative shocks (Bohi, 1991; Bernanke et al., 1997); and (3) the uncertainty and financial stress that amplifies positive shocks and dampens negative shocks, either because of their impact on future investment decisions (Bernanke, 1983; Pindyck, 1991) or their effect on precautionary savings (Edelstein and Kilian, 2009). As the effect of oil price shocks on inflation differs depending on the source of the shock (Barsky and Kilian, 2004), it is important to differentiate between different types of shocks. Following Kilian (2009), we distinguish three main types of shocks: (1) oil supply shocks caused by exogenous changes in oil production, whereby a decline in oil production translates to an increase in oil prices and vice versa; (2) oil-specific demand shocks, which reflect -100 -50 0 50 100 150 -1 0 1 2 3 4 5 1999M01 2000M01 2001M01 2002M01 2003M01 2004M01 2005M01 2006M01 2007M01 2008M01 2009M01 2010M01 2011M01 2012M01 2013M01 2014M01 2015M01 2016M01 2017M01 2018M01 2019M01 2020M01 Oil price changes (%) Headline inflation rate (%) Month 4 changes in oil demand caused by speculation, precautionary moves against the risk of future supply disruptions, or changes in firms’ oil needs because of changes in their technologies or production processes; and (3) aggregate demand shocks, which represent changes in oil demand caused by fluctuations in global economic activity. To the best of our knowledge, no study has comprehensively analyzed the effect of oil price shocks on inflation by jointly considering the inflation environment in which such shocks occur and the possible existence of asymmetric response. This study aims to fill this gap by examining how the inflation environment affects the transmission of oil price shocks to inflation as well as the existence of asymmetries, and how they vary depending on the inflation environment in which the shock occurs. In this study, we analyze the transmission of oil price shocks to inflation using a panel of the 12 initial eurozone countries. We analyze the transmission of these shocks to headline inflation, which includes energy prices, as well as core inflation. This allows us to study the transmission of shocks to prices that are not directly affected by fluctuations in oil prices, thereby allowing us to distinguish indirect effects. Our methodology proceeds using a twostage approach. In the first stage, we use an SVAR model of the oil market (Kilian, 2009) to identify the three types of shocks. In the second stage, we estimate the effect of the previously identified shocks on inflation, using local projection methods (Jordà, 2005). This method affords greater flexibility for estimating the impulse response function than is provided by traditional VAR analyses and is particularly useful for estimating nonlinear relationships and for the two-stage approach as it allows us to directly include the exogenous shocks we identify. To account for the effect of the inflation environment, we employ a smooth transition state-dependent local projection model (Auerbach and Gorodnichenko, 2012) using standardized past inflation as the state variable on which the transmission of oil price shocks depends. Further, to analyze the existence of asymmetric effects based on the direction of the shocks, we extend our state-dependent model by distinguishing between positive shocks (which increase oil prices) and negative shocks (which cause oil prices to fall) and obtain different impulse response functions for positive and negative shocks. The main contribution of this work is that, for the first time, the influence of the current inflation environment on the transmission of oil price shocks to inflation, and on asymmetries between positive and negative shocks is jointly analyzed empirically. This allows us to identify asymmetric and conditional behaviors that should be useful in designing the European Central Bank’s monetary policy with respect to its economic pillar (ECB, 2011). To date, only Sekine (2020) has studied the existence of non-linearity in the transmission of oil price changes to inflation as a function of the inflation environment. Our work extends that study in several ways. First, we distinguish the sources of oil price shocks as their transmission mechanisms to inflation differ substantially. Second, Sekine (2020) analyzes the effect using a static approach in the framework of an augmented Phillips curve, whereas we analyze the transmission of oil price shocks from a dynamic perspective by estimating impulse response functions. This allows us to obtain a more comprehensive view of the transmission at different time horizons. Finally, we also consider the presence of asymmetric responses in the transmission of oil shocks conditioned on the economy’s current inflation environment. 5 Our main findings are that oil supply shocks as well as oil-specific demand shocks differ depending on the inflation environment in which they occur. Headline inflation shows a greater response to positive shocks in high inflation environments, while negative shocks are transmitted to a greater degree in a low inflation environment. The transmission of aggregate demand shocks depends to a greater extent on the direction of the shock such that negative shocks that reflect depressed global economic activity show a high and persistent transmission to inflation, triggering a significant deflationary effect, while positive shocks show very low transmission, especially in low inflation environments. Finally, the effect of these shocks on core inflation, which reflects indirect effects, is only significant in high inflation environments, while transmission is muted in low inflation environments. These findings imply that when determining monetary policy response to oil price shocks, it is important to consider the inflation environment in which these shocks occur as their transmission to inflation can vary considerably depending on that environment. This study is related to three strands of literature. First, it is linked to the literature on statedependence in the transmission of cost shocks. Although there is ample empirical evidence that the transmission of exchange rate shocks depends on the inflation environment (Choudhri and Hakura, 2006; Junttila and Korhonen, 2012; Shintani et al., 2013; Cheikh and Louhichi 2016; López-Villavicencio and Mignon, 2017; Cheikh and Zaied, 2020), evidence on the transmission of oil price shocks is scarce. Only Sekine (2020) analyzes the influence of the inflation environment on the transmission of oil prices to inflation. Focusing on the US, he finds a higher transmission of oil price fluctuations during periods of high inflation. Nevertheless, that study takes a static perspective and does not distinguish among sources of asymmetries. This work is also related to the literature addressing the existence of asymmetry in the effects of oil price shocks. Although there is a vast body of research on the asymmetric effect of oil price shocks on economic activity, empirical studies of the asymmetric effects of oil price shocks on inflation are scarce, and the results are not conclusive. In the US, positive shocks have been shown to have a greater effect on inflation than negative shocks (Balke et al., 2002), especially when the shocks are larger than usual (An et al., 2014). López-Villavicencio and Pourroy (2019) find that in countries where the central bank does not follow an inflation targeting strategy, positive oil price shocks also exhibit a greater degree of transmission to inflation than negative shocks, although this asymmetry does not hold in countries with inflation targeting. For the eurozone, studies have found that negative shocks have a greater degree of transmission and persistence than positive ones (Evgenidis, 2018) in periods of financial stress. Donayre and Wilmot (2016) found that in Canada, the transmission of negative shocks is higher than positive shocks, especially during periods of low economic growth. Furthermore, our work is also related to the literature addressing the importance of the source of oil price shocks in their transmission to other variables related to the oil market and domestic variables. Studies show that the degree of transmission of oil price shocks to macroeconomic variables, including inflation, depends on whether those price shocks are caused by supply or demand shocks in the case of the US (Kilian, 2009), the United Kingdom (UK) (Lorusso and Pieroni, 2018), and the eurozone as a whole (Peersman and Van Robays, 2009; Herwartz and Plödt, 2016; Enders and Enders, 2017). 6 In contrast to the works cited above, our study is the first to jointly consider the influence of the inflation environment on the transmission of oil price shocks, while distinguishing shocks according to their source and sign. The rest of this study is organized as follows: section 2 describes our methodology and dataset; section 3 presents the empirical results, and section 4 tests the robustness of our findings. In section 5, we discuss the results and their implications and then offer conclusions in section 6. 2. Methodology Our methodology consists of a two-stage approach as proposed by Kilian (2009), which has been extensively used to analyze the effect of oil price shocks on different macroeconomic variables (Kilian et al., 2009; Habib et al., 2016; Lorusso and Pieroni, 2018; Jibril et al., 2020). In the first stage, we identify oil price shocks through a structural model of the global oil market, following the methodology in Kilian (2009). In the second stage, we estimate the transmission of those shocks to inflation in the eurozone using local projection methods (Jordà, 2005). An alternative strategy would be to estimate an SVAR model of the oil market extended to include the macroeconomic variables to be studied. However, Kilian and Zhou (2020) point out that to correctly estimate a structural model of the oil market, it is necessary to use a long time series, such as one spanning from 1974 to present. When macroeconomic data are available for a shorter period of time, as is the case of the eurozone that has only been in existence since 1999, the two-stage approach is more suitable than the extended SVAR approach to correctly identify oil price shocks. Furthermore, Hamilton and Herrera (2004) and Kilian (2009) highlight the importance of including a large number of lags in the model to adequately capture the dynamics of the oil market and how oil price shocks are transmitted to the economy, which also supports our decision to use a two-stage approach, given the length of our time series data. 2.1. Identifying structural oil price shocks To identify oil price shocks, we estimate an SVAR model with recursive identification as developed by Kilian (2009), which has been widely used to study the effect of oil price shocks on other macroeconomic variables (Kilian et al., 2009; Kang and Ratti, 2013; Habib et al., 2016; Enders and Enders, 2017; Kang et al., 2017; Degiannakis et al., 2018; Lorusso and Pieroni, 2018; Jibril et al., 2020) 1 . Thus, we identify three types of shocks: oil supply, oilspecific demand, and aggregate demand shocks by estimating the following trivariate SVAR model: 𝐴0𝑌𝑡= 𝛼+ ∑𝐴𝑖𝑌𝑡−𝑖+𝜀𝑡 24 𝑖=1 (1) 1 An alternative option is to identify oil price shocks using an SVAR model with sign restrictions (Kilian and Murphy, 2012; Kilian and Murphy, 2014; Baumeister and Hamilton, 2019). In section 4, we test the robustness of the results using oil price shocks identified via sign restrictions. 7 where 𝑌𝑡 is a 3x1 vector of endogenous variables, namely, ∆𝑝𝑟𝑜𝑑𝑡, representing the percentage change in oil production, 𝑟𝑒𝑎𝑡 is the real economic activity index developed by Kilian (2009) representing fluctuations in global economic activity, and 𝑟𝑝𝑜𝑡 is the real refiner acquisition cost of imported crude oil expressed in log form. 𝜀𝑡 is a vector of structural innovations that are serially and mutually uncorrelated. To obtain 𝜀𝑡 from the reduced-form errors, we assume that 𝐴0−1 has a recursive structure such that we can retrieve the structural innovations 𝜀𝑡 from the VAR reduced-form errors, 𝑒𝑡: 𝑒𝑡= (𝑒𝑡∆𝑝𝑟𝑜𝑑 𝑒𝑡𝑟𝑒𝑎 𝑒𝑡𝑟𝑝𝑜 )=(𝑎11 0 0 𝑎21 𝑎22 0 𝑎31 𝑎32 𝑎33)(𝜀𝑡𝑜𝑖𝑙 𝑠𝑢𝑝𝑝𝑙𝑦 𝑠ℎ𝑜𝑐𝑘 𝜀𝑡𝑎𝑔𝑔𝑟𝑒𝑔𝑎𝑡𝑒 𝑑𝑒𝑚𝑎𝑛𝑑 𝑠ℎ𝑜𝑐𝑘 𝜀𝑡𝑜𝑖𝑙−𝑠𝑝𝑒𝑐𝑖𝑓𝑖𝑐 𝑑𝑒𝑚𝑎𝑛𝑑 𝑠ℎ𝑜𝑐𝑘) (2) To identify the three structural shocks, we use a recursive structure based on the following assumptions: - First, to identify oil supply shocks, we assume that oil production is inelastic in the short run. This implies that production does not react contemporaneously to changes in global economic activity or the price of oil. - To identify aggregate demand shocks, we assume that global economic activity can be affected in the same month by oil production changes but not by fluctuations in oil prices. In other words, global economic activity only reacts to oil price fluctuations after a certain delay. We, therefore, identify aggregate demand shocks as innovations in the economic activity index that are not explained by oil supply shocks. - Finally, in the case of oil-specific demand shocks, we assume that oil prices react contemporaneously to both oil production changes and fluctuations in global economic activity. In this way, we identify oil-specific demand shocks as oil price innovations that are not explained by oil supply shocks and aggregate demand shocks. After identifying these structural oil shocks 𝜀𝑡, we construct three variables corresponding to oil supply, oil-specific demand, and aggregate demand shocks for the period from January 1999 to January 2020, which we use to study the transmission of oil shocks to inflation in the eurozone. 2.2. Estimation of the transmission of oil price shocks to inflation 2.2.1. Linear model To analyze the transmission of oil price shocks to inflation, we use headline inflation and, alternatively, core inflation as dependent variables in an effort to pinpoint the indirect effects generated by those shocks. We estimate the following equation using local projection methods, following other studies such as Sekine and Tsuruga (2018) and Caselli and Roitman (2019): 8 𝑝𝑖,𝑡+ℎ−𝑝𝑖,𝑡−1= 𝛼𝑖,ℎ+∑𝜇𝑙ℎ(𝑝𝑖,𝑡−𝑖−𝑝𝑖,𝑡−1−𝑖) 12 𝑙=1 + 𝛽ℎ𝑆ℎ𝑜𝑐𝑘𝑎,𝑡 +∑𝜃𝑙ℎ 12 𝑙=0 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑖,𝑡−𝑙+𝜖𝑖,𝑡+ℎ (3) where 𝑝𝑖,𝑡 represents the price index, in log form, for country i = 1, 2,…,12 on date t. 𝑆ℎ𝑜𝑐𝑘𝑎,𝑡 refers to the three structural oil price shocks identified previously, where a = 1, 2, 3 refers to oil supply, oil-specific demand, and aggregate demand shocks, respectively, whose impacts on inflation we estimate. The estimates for each type of shock are the same for every country in the panel. 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑖,𝑡−𝑙 is a set of control variables that includes the percentage change in the industrial production index for each country and the euro/dollar exchange rate, including its contemporaneous value and its value at 12 lags. We also include 12 lags of the inflation rate to control for persistence in inflation (Sekine and Tsuruga, 2018). Finally, 𝜖𝑖,𝑡+ℎ represents the error term. To obtain the impulse response function, we estimate equation (3) for each horizon h and obtain 𝛽ℎ, a coefficient that shows the response of the dependent variable in h to an exogenous shock in t. Therefore, the impulse response function is obtained by adding the coefficients for the entire estimation horizon, represented as 𝐼𝑅𝐹(ℎ)= 𝛽ℎ for h = 1, 2,…, H. We choose a time horizon of 24 months. To estimate equation (3), we use the Least Squares Dummy Variable Estimator with the Newey West Heteroskedasticity and Autocorrelation- Consistent covariance matrix. 2.2.2. Estimation of the transmission of oil price shocks based on the inflation environment The local projection method can be extended to estimate impulse response functions in statedependent models. Specifically, following Auerbach and Gorodnichenko (2012), the local projection method can be estimated as a smooth transition model using a logistic function as the transition between different states. To test whether the inflation environment affects the transmission of oil price shocks to inflation, we estimate the following equation: 𝑝𝑖,𝑡+ℎ−𝑝𝑖,𝑡−1= 𝛼𝑖,ℎ+(1−𝐹(𝑧𝑖,𝑡−1)[∑𝜇𝐻,𝑙 ℎ(𝑝𝑖,𝑡−𝑖−𝑝𝑖,𝑡−1−𝑖) 12 𝑙=1 + 𝛽𝐻ℎ𝑆ℎ𝑜𝑐𝑘𝑎,𝑡+∑𝜃𝐻,𝑙 ℎ 12 𝑙=0 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑖,𝑡−𝑙] +(𝐹(𝑧𝑖,𝑡−1)[∑𝜇𝐿,𝑙 ℎ(𝑝𝑖,𝑡−𝑖−𝑝𝑖,𝑡−1−𝑖) 12 𝑙=1 + 𝛽𝐿ℎ𝑆ℎ𝑜𝑐𝑘𝑎,𝑡+∑𝜃𝐿,𝑙 ℎ 12 𝑙=0 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑖,𝑡−𝑙]+𝜖𝑖,𝑡+ℎ (4) where F(.) is the transition function represented by the following logistic function: 𝐹(𝑧𝑖,𝑡−1)= exp (−𝛾𝑧𝑖,𝑡−1) 1+𝑒𝑥𝑝 (−𝛾𝑧𝑖,𝑡−1) with 𝛾>0, where 𝑧𝑖,𝑡−1 is the transition variable. Here, we include the standardized past inflation rate as the state variable, as proposed by Sekine and Tsuruga (2018). Specifically, we use core inflation in our baseline model as we think it is a better indicator of the inflation environment, given that it excludes the most volatile components of the HICP. The state variable is represented as follows: 𝑧𝑖,𝑡−1= 𝜋𝑖,𝑡−1−𝜋𝑖 𝜎𝑖  (5) 9 where 𝜋𝑖,𝑡−1 is the lagged inflation rate of country i, 𝜋𝑖 is country i’s average inflation rate during the period analyzed, and 𝜎𝑖  is the standard deviation of that inflation rate. 𝐹(𝑧𝑖,𝑡−1) denotes the probability of being in a low inflation state. When 𝑧𝑖,𝑡−1→ ∞, then 𝐹(𝑧𝑖,𝑡−1)→ 0, and the impulse response function would be represented by 𝛽𝐻ℎ, where the subscript H denotes a high inflation environment. In contrast, when 𝑧𝑖,𝑡−1→ −∞, 𝐹(𝑧𝑖,𝑡−1)→1, the impulse response function is given by 𝛽𝐿ℎ where the subscript L refers to a low inflation environment. The parameter 𝛾 sets the speed of the transition between states. If 𝛾=0, the model becomes the linear model as 𝐹(𝑧𝑖,𝑡−1) is always ½. On the other hand, if 𝛾 is very high, the transition between states is immediate, and the model would approximate a discrete transition model with a threshold dummy variable. Following Tenreyro and Thwaites (2016), we select 𝛾=3, which allows an intermediate speed for regime transitions. However, in section 4 we show that our results hold when we use different values of 𝛾. In general terms, the impulse response function is the weighted sum of 𝛽𝐻ℎ and 𝛽𝐿ℎ, where the weight depends on the value of 𝐹(𝑧𝑖,𝑡−1) and is obtained as follows: 𝐼𝑅𝐹(ℎ)= (1 − 𝐹(𝑧𝑖,𝑡−1))𝛽𝐻ℎ+ (𝐹(𝑧𝑖,𝑡−1))𝛽𝐿ℎ. In the extremes, a low inflation environment where 𝐹(𝑧𝑖,𝑡−1) takes the value 1, the impulse response function is given by 𝐼𝑅𝐹𝐿(ℎ) = 𝛽𝐿ℎ and in a high inflation environment, 𝐹(𝑧𝑖,𝑡−1) takes the value 0 such that the impulse response function is given by 𝐼𝑅𝐹𝐻(ℎ)= 𝛽𝐻ℎ. 2.2.3. Including asymmetry in the transmission of oil price shocks To analyze the asymmetric transmission of positive and negative shocks, we extend the model represented in equation (4) by substituting the variable 𝑆ℎ𝑜𝑐𝑘𝑎,𝑡 for the following variables: 𝑆ℎ𝑜𝑐𝑘𝑎,𝑡 +={𝑆ℎ𝑜𝑐𝑘𝑡 𝑖𝑓 𝑆ℎ𝑜𝑐𝑘𝑎,𝑡>0 0 𝑆ℎ𝑜𝑐𝑘𝑎,𝑡≤0, 𝑆ℎ𝑜𝑐𝑘𝑡−={𝑆ℎ𝑜𝑐𝑘𝑡 𝑖𝑓 𝑆ℎ𝑜𝑐𝑘𝑎,𝑡≤0 0 𝑖𝑓 𝑆ℎ𝑜𝑐𝑘𝑎,𝑡>0 (6) where 𝑆ℎ𝑜𝑐𝑘𝑎,𝑡 + refers to positive oil shocks, that is, those that trigger an increase in oil prices, and 𝑆ℎ𝑜𝑐𝑘𝑎,𝑡 − refers to negative shocks that reduce oil prices. We introduce both variables simultaneously to avoid any possible bias effects that might be induced by truncated variables. Using this method, we can analyze whether there is any asymmetry in the transmission of these shocks depending on their direction and whether these differences vary depending on the inflation environment for the economy at the time of the shock. 2.3. Data We use monthly data for the period from January 1999 to January 2020 for a panel of the 12 initial countries of the eurozone (Austria, Belgium, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal, and Spain). Our study period ranges from the creation of the euro to the latest data available. The HICP and the HICP excluding food and energy (HICPex) are obtained from Eurostat. The euro/dollar exchange rates are obtained from the International Monetary Fund’s International Financial Statistics 2 database, 2 In the case of Greece for the period from January 1999 to December 2000, we convert the drachma/dollar exchange rate to euro/dollar by multiplying the former by the fixed drachma/euro exchange rate set during the country’s adoption of the euro. 16 Fig. 6. Cumulative response of core inflation to the three types of oil shocks in a low inflation environment (column 1) and a high inflation environment (column 2). Notes: Figures show the cumulative effect on inflation of a one standard deviation shock. The black line denotes the response of inflation to a positive shock. The red line denotes the response of inflation to a negative shock. Shaded areas denote 90% confidence intervals. Aggregate demand shocks show a different pattern, compared to the two other shocks as in both states, a persistent drop in inflation is seen after a negative shock, while positive shocks have a minor effect and even show a negative cumulative effect in low inflationary environments. Regarding transmission to core inflation, in an environment of low inflation, transmission is insignificant, while in high inflation environments, negative shocks trigger a significant drop in core inflation, and positive shocks also eventually produce a negative cumulative effect. Contrary to what we observe for the other types of shocks, these results are consistent with Evgenidis (2018) who found that in periods of high financial stress, negative oil price shocks have a greater impact on inflation than positive shocks in the eurozone and Donayre and Wilmot (2016) who show that negative shocks have a greater impact on inflation than positive shocks in periods of low economic growth. -1.5 -1 -0.5 0 0.5 1 1.5 0 2 4 6 8 10 12 14 16 18 20 22 24 HICPex Months Oil supply shock (low inflation environment) -1.5 -1 -0.5 0 0.5 1 1.5 0246810 12 14 16 18 20 22 24 HICPex Months Oil supply shock (high inflation environment) -1.5 -1 -0.5 0 0.5 1 1.5 0 2 4 6 8 10 12 14 16 18 20 22 24 HICPex Months Oil-specific demand shock (low inflation environment) -1.5 -1 -0.5 0 0.5 1 1.5 0246810 12 14 16 18 20 22 24 HICPex Months Oil-specific demand shock (high inflation environment) -1.5 -1 -0.5 0 0.5 1 1.5 0 2 4 6 8 10 12 14 16 18 20 22 24 HICPex Months Aggregate demand shock (low inflation environment) -1.5 -1 -0.5 0 0.5 1 1.5 0246810 12 14 16 18 20 22 24 HICPex Months Aggregate demand shock (high inflation environment) 17 State Sign Positive Negative Difference by shock Shock Oil supply shock Oilspecific demand shock Agg. demand shock Oil supply shock Oilspecific demand shock Agg. demand shock Oil supply shock Oilspecific demand shock Agg. demand shock Low inflation Max impact (%) 0.07 0.13 0.05 0.18 0.1 0.14 −0.12 0.03 −0.09 Impact after 12 months (%) −0.16 0.06 0.01 0.06 0.03 0.1 −0.22 0.03 −0.09 Impact after 24 months (%) −0.22 0.11 0.05 0.18 0.08 0.13 −0.41 0.02 −0.08 High inflation Max impact (%) 0.36 0.36 −0.02 −0.04 −0.01 0.43 0.4 0.37 −0.45 Impact after 12 months (%) 0.31 0.11 −0.17 −0.27 −0.11 0.15 0.57 0.22 −0.32 Impact after 24 months (%) 0.15 0.36 −0.44 −0.39 −0.38 0.43 0.54 0.74 −0.86 Difference by state Max impact (%) 0.29 0.22 −0.07 −0.22 −0.11 0.29 Impact after 12 months (%) 0.46 0.05 −0.18 −0.33 −0.14 0.05 Impact after 24 months (%) 0.37 0.25 −0.48 −0.58 −0.47 0.3 Notes: The cumulative impact represents the % change in the HICPex generated by a one standard deviation shock, estimated using the model in equation (4) and including the shocks as defined in equation (6). Table 2. Summary of the transmission of structural oil price shocks to core inflation. Further, it is important to emphasize the differences in the response of inflation to oil-specific demand shocks versus aggregate demand shocks. In a high inflation environment, negative aggregate demand shocks trigger a persistent drop in inflation, while oil-specific demand shocks, after an initial drop in inflation, cause price levels to increase such that the cumulative impact is higher inflation in the long run. One possible explanation for this is the transmission mechanisms of these shocks to the output. A negative aggregate demand shock reduces the price of oil and triggers a decline in global economic activity; this can be translated into a decline in domestic economic activity in the eurozone. On the other hand, a negative oilspecific demand shock, perhaps caused by precautionary or speculative actions, sparks the opposite response in terms of production (Kilian, 2009; Peersman and Van Robays, 2009); in other words, a lower cost of oil has a positive effect on production, so that, in an environment of high inflation, that translates into greater pressures on prices. The long-term effect is an increase in both headline and core inflation. 4. Robustness In this section, we test the robustness of the results obtained thus far to the following: (i) the method used to identify exogenous oil price shocks; (ii) the choice of the speed of transition between regimes, determined by the parameter 𝛾; (iii) the choice of the state variable 𝑧𝑖,𝑡−1 used to define the inflationary environment; and (iv) the method used to define the transition between states or inflation environments. 4.1. Alternative identification of oil price shocks In the baseline model, we identify oil price shocks using an exclusion restriction with a recursive structure, following the identification strategy in Kilian (2009). However, in studies of the oil market, oil price shocks are often identified via sign restrictions. To determine whether our results are robust to the identification method, we follow the strategy in Kilian 18 and Murphy (2014) who distinguish between flow supply, flow demand, and speculative demand shocks. To do so, we extend the model in Kilian (2009) by adding a variable that represents changes in OECD crude oil inventories 4 . Following Kilian and Murphy (2014), we use sign restrictions in the contemporary response of the variables to these different types of shocks, as shown in Table 3. We impose two additional restrictions. First, we include dynamic restrictions on the response of oil production, global economic activity, and the price of oil to maintain the sign imposed on the contemporary response for 12 months after a flow supply shock. Second, the impact price elasticity of oil supply is restricted to be less than or equal to 0.25, while the impact price elasticity of oil demand is restricted to be between 0 and −0.80 (see Kilian and Murphy, 2014). Flow supply shock Flow demand shock Speculative demand shock Oil production - + + Real activity - + - Price of oil + + + Inventories + Source: Kilian and Murphy (2014). Notes: All shocks are normalized to imply an increase in the price of oil. Table 3. Sign restrictions on the contemporary impact responses in the SVAR model. The results, shown in Appendix B, are in line with those obtained in the baseline model. Specifically, flow demand shocks are equivalent to aggregate demand shocks, while speculative demand shocks are equivalent to oil-specific demand shocks. This demonstrates that the conclusions obtained from our baseline results are robust to the method of identifying oil price shocks. 4.2. Speed of regime switching In our baseline model, the parameter that determines the speed of transition between states is set to 𝛾=3, following Tenreyro and Thwaites (2016). To test the sensitivity of our results to the value of this parameter, we re-estimate the model in equation (4), setting 𝛾 to 1.5 and 6 and, thereby, allowing the speed of transition to be lower and higher than in the baseline model, respectively. Results are provided in Appendix C and show that in both scenarios, the results are in line with those found in the baseline model. 4.3. State variable In the baseline model, we employ standardized core inflation as the state variable. The choice of core inflation is motivated by the fact that it better represents the true inflation environment than headline inflation as it is not exposed to the volatility of energy and food prices. 4 Since data on OECD crude oil inventories are not available, Kilian and Murphy (2014) calculated the ratio of oil and petroleum product inventories between the OECD and the US and multiplied US crude oil inventories by this ratio. 19 However, other studies (Sekine and Tsuruga, 2018; Cheikh and Zaied, 2020; Sekine, 2020) have used headline inflation as the state variable. In order to check the robustness of the model to this specification, we substitute core inflation with headline inflation as the state variable. In general, these results (shown in Appendix D) do not differ from those obtained in the baseline model, allowing us to conclude that the baseline results are robust to the choice of the state variable. 4.4. Definition of the transition between states In this section, we test the robustness of our baseline findings to an alternative approach to determining the economy’s current inflation environment (high or low). To do so, we extend the linear model to a threshold state-dependent model (Ahmed and Cassou, 2016; Ramey and Zubairy, 2018; Ahmed and Cassou, 2021), represented in the following equation: 𝑝𝑖,𝑡+ℎ−𝑝𝑖,𝑡−1= 𝛼𝑖,ℎ+𝐷𝑖,𝑡−1[∑𝜇𝐻,𝑙 ℎ(𝑝𝑖,𝑡−𝑖−𝑝𝑖,𝑡−1−𝑖) 12 𝑙=1 + 𝛽𝐻ℎ𝑆ℎ𝑜𝑐𝑘𝑎,𝑡+∑𝜃𝐻,𝑙 ℎ 12 𝑙=0 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑖,𝑡−𝑙] +(1−𝐷𝑖,𝑡−1)[∑𝜇𝐿,𝑙 ℎ(𝑝𝑖,𝑡−𝑖−𝑝𝑖,𝑡−1−𝑖) 12 𝑙=1 + 𝛽𝐿ℎ𝑆ℎ𝑜𝑐𝑘𝑎,𝑡+∑𝜃𝐿,𝑙 ℎ 12 𝑙=0 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑖,𝑡−𝑙]+𝜖𝑖,𝑡+ℎ (7) where 𝐷𝑖,𝑡−1∈ {0,1} is a dummy variable that represents the inflation environment and is defined as follows: 𝐷𝑖,𝑡−1={1 𝑖𝑓 𝜋𝑖,𝑡−1>𝜋𝑖 0 𝑖𝑓 𝜋𝑖,𝑡−1≤𝜋𝑖 (8) where 𝜋𝑖,𝑡−1 is the lagged core inflation of country i, and 𝜋𝑖 is the average core inflation over the period studied. Therefore, when the lagged core inflation is higher than the country’s average over the period, 𝐷𝑖,𝑡−1 takes a value of 1, indicating that the economy is in a high inflation environment; when the lagged core inflation is below the country’s average, the economy is in a low inflation environment, with 𝐷𝑖,𝑡−1 taking a value of 0. This method differs from the approach used in equation (4) as it does not assume a smooth transition between inflation regimes or environments but rather that the transition occurs when the state variable exceeds the threshold value. The results, displayed in Appendix E, do not show any significant differences with respect to the baseline model, confirming the robustness of our findings. 5. Discussion of the results The results presented in this study show that when analyzing the effect of oil price shocks on inflation, it is essential to consider both the inflationary environment in which they occur and the sign of the shocks, as the consequences of the shocks differ depending on these factors. When we analyze oil price shocks originating from the oil market itself, that is, oil supply and oil-specific demand shocks, the results are in line with theoretical predictions. In high inflation environments, positive shocks, that is, those that increase the price of oil, are transmitted to inflation to a greater extent than negative shocks, producing indirect effects reflected in an increase in core inflation. This relationship is consistent with the theoretical model in Taylor (2000), which argues that in an environment of higher inflation, firms have 20 expectations of higher future inflation and greater persistence of oil price shocks and, thus, transmit these higher costs to their prices to a greater degree. Furthermore, our results show that the degree of transmission of negative price shocks is greater in low inflation environments, while in high inflation environments, that transmission is very low. This is in line with the theoretical prediction in Ball and Mankiw (1994) who show that the frequency of price increases because changes in costs grows with higher trend inflation, while the frequency of price reductions decline, leading to greater asymmetry. These results with respect to the frequency of price changes have been empirically supported at the microeconomic level for Mexico (Gagnon, 2009), Argentina (Alvarez et al., 2019), the eurozone, and the US (Dhyne et al., 2006). Moreover, our results for aggregate demand shocks show a reaction pattern for both headline and core inflation that differs from the pattern found for the other type of shocks. In both low and high inflation environments, we find a high degree of asymmetry between positive and negative shocks. Slumps in global economic activity trigger a persistent decline in consumer prices, while increases in global economic activity have a very low level of transmission to inflation and even show a negative cumulative impact in a low inflation environment. One possible explanation for the reduced impact of positive aggregate demand shocks is a greater impact of monetary policies that respond to demand pressure in the economy, which would be reflected in the negative transmission seen in core inflation in our estimates and offset the inflationary effect of the energy component. Another possible explanation could be an asymmetric impact of global economic activity on domestic economies within the eurozone, whereby an increase in global economic activity does not necessarily result in higher domestic economic activity, whereas a downturn in global economic activity triggers a decline in domestic demand and, therefore, a drop in consumer prices in the eurozone. Other empirical studies have found evidence of negative asymmetry, such as Evgenidis (2018) who shows that in periods of high financial stress, negative shocks have a greater deflationary effect than the inflationary effect of positive shocks in the eurozone. Donayre and Wilmot (2016) found that negative oil price shocks in Canada have a greater transmission to inflation than positive shocks, especially in periods of low economic growth. Our results imply that periods of declining global economic activity, such as the global financial crisis of 2008 or the decline in economic activity caused by the slowdown in China’s economy in 2014–2015, would have triggered greater deflationary pressure than the inflationary pressures generated during periods of booming global economic activity, such as during the period from 2006 to 2007. Another important finding is the pattern of transmission of oil price shocks to core inflation. Our results show that in a low inflation environment, the transmission is negligible for all types of shocks, while transmission is significant in high inflation environments, with an inflationary effect from shocks originating in the oil market itself. These findings are relevant for monetary policy, as in a context of low inflation such as the one the eurozone has experienced since the global financial crisis, the inflationary effects of oil price shocks are small and transitory and do not trigger the indirect effects that force a central bank to tighten monetary policies to counter inflationary pressures. These considerations should be considered by central bank authorities, especially given that the trade-off between controlling inflation or stabilizing output that oil price shocks represent is reduced in an environment of low inflation. 21 By way of an example, in July 2008, just prior to the start of the global financial crisis, the ECB raised interest rates by 25 basis points in reaction to a rise in the price of oil that was attributed to increased oil demand. Monetary policy was consistent with what was predictable according to a linear model. However, this policy action would not have been consistent with the results of a model that accounts for the current inflation environment and the direction of the shock. This illustrates the practical implications for pursuing a detailed analysis of the transmission of oil price shocks to inflation in determining effective monetary policy. 6. Conclusions This study analyzes the transmission of oil price shocks to inflation using a state-dependent model. Distinguishing between three types of oil price shocks (oil supply, oil-specific demand, and aggregate demand shocks), the model conditions the response of inflation to the inflation environment in which shocks occur and the direction of the shocks themselves, producing a comprehensive study of the transmission of oil price shocks to inflation. The results show that the inflation environment in which shocks take place is a key determinant when analyzing whether and how those shocks are transmitted to inflation, as it affects both the degree of transmission and the asymmetry between positive and negative shocks. The transmission of oil price shocks differs depending on the current inflation environment and on the direction of the shocks, and those differences vary among the different types of shocks. The shocks generated in the oil market itself, that is, oil supply and oil-specific demand shocks, show a greater inflationary effect in periods of high inflation, while negative shocks have a higher deflationary impact in a low inflation environment, in line with the results predicted in theoretical models (Ball and Mankiw, 1994; Taylor, 2000). However, aggregate demand shocks, that is, those caused by variations in global economic activity, show an asymmetry that holds in both states; negative shocks generate a persistent decline in inflation, whereas positive shocks are barely transmitted. This last finding implies that the deflationary risk posed by negative aggregate demand shocks is higher than the inflationary risk of positive aggregate demand shocks. This is particularly relevant in low inflation environments, where the risk of deflation is higher and monetary policy must react in order to prevent the economy from sinking into a deflationary spiral. In high inflation environments, these findings are less relevant for monetary policy as the inflationary risk of positive shocks is low, while the decline in inflation caused by negative aggregate demand shocks can offset the inflationary pressures that emerge in a high inflation environment, thus allowing a central bank to accommodate these shocks and avoid tightening its monetary policy. Another important finding is that the transmission of oil price shocks to core inflation, which represents the indirect effects of oil price shocks, is negligible in a low inflation environment but significant in a high inflation environment. This implies that monetary policy authorities, whose inflation target focuses on the medium-term dynamics of core inflation (ECB, 2011) should not be concerned by the inflationary effect of these shocks when the macroeconomic environment is characterized by low inflation. In contrast, in a high inflation environment, the transmission of shocks, especially those originating from the oil market itself, are substantially transmitted to core inflation; in this situation, positive shocks generate inflationary pressures that the central bank must take into account when establishing its monetary policy. 22 The main implication of these results is that when deciding how to react to oil price shocks, a central bank must consider the current inflation environment to design an optimal monetary policy as the extent to which inflation reacts to oil price shocks depends to a great extent on the inflation setting in which the shocks occur. Further, our results show certain counterintuitive behaviors that call for future research on their possible causes. This is especially relevant in the case of aggregate demand shocks as inflation shows little response to positive shocks, contrary to what is expected by economic theory. However, this finding highlights the importance of taking into account all the factors that interact when analyzing oil price shocks, such as the inflation environment as well as the source and direction of the shock itself. 23 References Ahmed, M. I., & Cassou, S. P. (2016). Does consumer confidence affect durable goods spending during bad and good economic times equally? Journal of Macroeconomics, 50, 86- 97. https://doi.org/10.1016/j.jmacro.2016.08.008. Ahmed, M. I., & Cassou, S. P. (2021). Asymmetries in the effects of unemployment expectation shocks as monetary policy shifts with economic conditions. Economic Modelling, 100, 105502. https://doi.org/10.1016/j.econmod.2021.105502. Alvarez, F., Beraja, M., Gonzalez-Rozada, M., & Neumeyer, P. A. (2019). From hyperinflation to stable prices: Argentina’s evidence on menu cost models. The Quarterly Journal of Economics, 134(1), 451-505. https://doi.org/10.1093/qje/qjy022. Álvarez, L. J., Hurtado, S., Sánchez, I., & Thomas, C. (2011). The impact of oil price changes on Spanish and euro area consumer price inflation. Economic Modelling, 28(1-2), 422-431. https://doi.org/10.1016/j.econmod.2010.08.006. An, L., Jin, X., & Ren, X. (2014). Are the macroeconomic effects of oil price shock symmetric?: A Factor-Augmented Vector Autoregressive approach. Energy Economics, 45, 217-228. https://doi.org/10.1016/j.eneco.2014.06.003. Auerbach, A. J., & Gorodnichenko, Y. (2012). Measuring the output responses to fiscal policy. American Economic Journal: Economic Policy, 4(2), 1-27. https://doi.org/10.1257/pol.4.2.1. Bacon, R. W. (1991). Rockets and feathers: The asymmetric speed of adjustment of UK retail gasoline prices to cost changes. Energy Economics, 13(3), 211-218. https://doi.org/10.1016/0140-9883(91)90022-R. Balke, N. S., Brown, S. P., & Yucel, M. K. (1998). Crude oil and gasoline prices: An asymmetric relationship? Economic and Financial Policy Review- Federal Reserve Bank of Dallas, (Q 1), 2-11. Balke, N. S., Brown, S. P., & Yucel, M. K. (2002). Oil price shocks and the US economy: Where Does the Asymmetry Originate? The Energy Journal, 23(3), 27-52. https://doi.org/10.5547/ISSN0195-6574-EJ-Vol23-No3-2. Ball, L., & Mankiw, N. G. (1994). Asymmetric price adjustment and economic fluctuations. The Economic Journal, 104(423), 247-261. https://doi.org/10.2307/2234746. Barsky, R. B., & Kilian, L. (2004). Oil and the macroeconomy since the 1970s. Journal of Economic Perspectives, 18(4), 115-134. https://doi.org/10.1257/0895330042632708. Baumeister, C., & Hamilton, J. D. (2019). Structural interpretation of vector autoregressions with incomplete identification: Revisiting the role of oil supply and demand shocks. American Economic Review, 109(5), 1873-1910. https://doi.org/10.1257/aer.20151569. Bernanke, B. S. (1983). Irreversibility, uncertainty, and cyclical investment. The Quarterly Journal of Economics, 98(1), 85-106. https://doi.org/10.2307/1885568. Bernanke, B. S., Gertler, M., Watson, M., Sims, C. A., & Friedman, B. M. (1997). Systematic monetary policy and the effects of oil price shocks. Brookings Papers on Economic Activity, 1997(1), 91-157. https://doi.org/10.2307/2534702. Blanchard, O. J., & Gali, J. (2007). The macroeconomic effects of oil shocks: Why are the 2000s so different from the 1970s? (No. w13368). National Bureau of Economic Research. https://doi.org/10.3386/w13368. 24 Bohi, D. R. (1991). On the macroeconomic effects of energy price shocks. Resources and Energy, 13(2), 145-162. https://doi.org/10.1016/0165-0572(91)90012-R. Borenstein, S. (1991). Selling costs and switching costs: Explaining retail gasoline margins. The RAND Journal of Economics, 354-369. https://doi.org/10.2307/2601052. Borenstein, S., Cameron, A. C., & Gilbert, R. (1997). Do gasoline prices respond asymmetrically to crude oil price changes? The Quarterly Journal of Economics, 112(1), 305- 339. https://doi.org/10.1162/003355397555118. Caselli, F. G., & Roitman, A. (2019). Nonlinear exchange‐rate pass‐through in emerging markets. International Finance, 22(3), 279-306. https://doi.org/10.1111/infi.12344. Cheikh, N. B., & Louhichi, W. (2016). Revisiting the role of inflation environment in exchange rate pass-through: A panel threshold approach. Economic Modelling, 52, 233-238. https://doi.org/10.1016/j.econmod.2014.11.004. Cheikh, N. B., & Zaied, Y. B. (2020). Revisiting the pass-through of exchange rate in the transition economies: New evidence from new EU member states. Journal of International Money and Finance, 100, 102093. https://doi.org/10.1016/j.jimonfin.2019.102093. Choudhri, E. U., & Hakura, D. S. (2006). Exchange rate pass-through to domestic prices: Does the inflationary environment matter? Journal of International Money and Finance, 25(4), 614-639. https://doi.org/10.1016/j.jimonfin.2005.11.009. Davis, S. J., & Haltiwanger, J. (2001). Sectoral job creation and destruction responses to oil price changes. Journal of Monetary Economics, 48(3), 465-512. https://doi.org/10.1016/S0304-3932(01)00086-1. Degiannakis, S., Filis, G., & Panagiotakopoulou, S. (2018). Oil price shocks and uncertainty: How stable is their relationship over time? Economic Modelling, 72, 42-53. https://doi.org/10.1016/j.econmod.2018.01.004. Devereux, M. B., & Yetman, J. (2010). Price adjustment and exchange rate passthrough. Journal of International Money and Finance, 29(1), 181-200. https://doi.org/10.1016/j.jimonfin.2008.12.009. Dhyne, E., Alvarez, L. J., Le Bihan, H., Veronese, G., Dias, D., Hoffmann, J., … & Vilmunen, J. (2006). Price changes in the euro area and the United States: Some facts from individual consumer price data. Journal of Economic Perspectives, 20(2), 171-192. https://doi.org/10.1257/jep.20.2.171. Donayre, L., & Wilmot, N. A. (2016). The asymmetric effects of oil price shocks on the Canadian economy. International Journal of Energy Economics and Policy, 6(2), 167-182. Edelstein, P., & Kilian, L. (2009). How sensitive are consumer expenditures to retail energy prices? Journal of Monetary Economics, 56(6), 766-779. https://doi.org/10.1016/j.jmoneco.2009.06.001. Enders, A., & Enders, Z. (2017). Second-round effects after oil-price shocks: Evidence for the Euro area and Germany. Economics Letters, 159, 208-213. https://doi.org/10.1016/j.econlet.2017.08.017. European Central Bank (ECB) (2011). The monetary policy of the ECB. Frankfurt, Germany: European Central Bank. Evgenidis, A. (2018). Do all oil price shocks have the same impact? Evidence from the euro area. Finance Research Letters, 26, 150-155. https://doi.org/10.1016/j.frl.2017.12.013. 25 Forbes, K. (2019). Inflation dynamics: Dead, dormant, or determined abroad? (No. w26496). National Bureau of Economic Research. https://doi.org/10.3386/w26496. Gagnon, E. (2009). Price setting during low and high inflation: Evidence from Mexico. The Quarterly Journal of Economics, 124(3), 1221-1263. https://doi.org/10.1162/qjec.2009.124.3.1221. Habib, M. M., Bützer, S., & Stracca, L. (2016). Global exchange rate configurations: Do oil shocks matter? IMF Economic Review, 64(3), 443-470. https://doi.org/10.1057/imfer.2016.9. Hamilton, J. D. (1983). Oil and the macroeconomy since World War II. Journal of Political Economy, 91(2), 228-248. https://www.journals.uchicago.edu/doi/abs/10.1086/261140. Hamilton, J. D. (1988). A neoclassical model of unemployment and the business cycle. Journal of Political Economy, 96(3), 593-617. https://www.journals.uchicago.edu/doi/abs/10.1086/261553. Hamilton, J. D. (1996). This is what happened to the oil price-macroeconomy relationship. Journal of Monetary Economics, 38(2), 215-220. https://doi.org/10.1016/S0304- 3932(96)01282-2. Hamilton, J. D., & Herrera, A. M. (2004). Comment: Oil shocks and aggregate macroeconomic behavior: The role of monetary policy. Journal of Money, Credit and Banking, 265-286. https://www.jstor.org/stable/3839020. Herwartz, H., & Plödt, M. (2016). The macroeconomic effects of oil price shocks: Evidence from a statistical identification approach. Journal of International Money and Finance, 61, 30-44. https://doi.org/10.1016/j.jimonfin.2015.11.001. Jibril, H., Chaudhuri, K., & Mohaddes, K. (2020). Asymmetric oil prices and trade imbalances: Does the source of the oil shock matter? Energy Policy, 137, 111100. https://doi.org/10.1016/j.enpol.2019.111100. Jordà, Ò. (2005). Estimation and inference of impulse responses by local projections. American Economic Review, 95(1), 161-182. https://doi.org/10.1257/0002828053828518. Junttila, J., & Korhonen, M. (2012). The role of inflation regime in the exchange rate passthrough to import prices. International Review of Economics & Finance, 24, 88-96. https://doi.org/10.1016/j.iref.2012.01.005. Kang, W., & Ratti, R. A. (2013). Structural oil price shocks and policy uncertainty. Economic Modelling, 35, 314-319. https://doi.org/10.1016/j.econmod.2013.07.025. Kang, W., de Gracia, F. P., & Ratti, R. A. (2017). Oil price shocks, policy uncertainty, and stock returns of oil and gas corporations. Journal of International Money and Finance, 70, 344-359. https://doi.org/10.1016/j.jimonfin.2016.10.003. Kilian, L. (2009). Not all oil price shocks are alike: Disentangling demand and supply shocks in the crude oil market. American Economic Review, 99(3), 1053-69. https://doi.org/10.1257/aer.99.3.1053. Kilian, L., & Murphy, D. P. (2012). Why agnostic sign restrictions are not enough: Understanding the dynamics of oil market VAR models. Journal of the European Economic Association, 10(5), 1166-1188. https://doi.org/10.1111/j.1542-4774.2012.01080.x. Kilian, L., & Murphy, D. P. (2014). The role of inventories and speculative trading in the global market for crude oil. Journal of Applied Econometrics, 29(3), 454-478. https://doi.org/10.1002/jae.2322.