Does one size fit all in the Euro Area? Some counterfactual evidence

The economic crises that occurred in Europe since 2007 have left member countries of the Economic and Monetary Union (EMU) in very heterogeneous economic conditions. While some members of the EMU experienced modest growth and high employment in the second half of the last decade, others remained in the process of recovery, suffering from unprecedented levels of unemployment. Clearly this experience is at odds with the goals of the EMU in general and the Euro Area in particular.¹

A popular view on the economic events in the EMU during the last two decades is that, by construction of the EMU, structural heterogeneity, limited scope for fiscal policy and a union-wide monetary policy have amplified the effects of adverse shocks and lead to sub-optimal macroeconomic performance. An EMU country that adopted the Euro has chosen a monetary regime where monetary policy is delegated to the ECB (see, e.g., Ball 2010). As a direct consequence, a Euro Area country can no longer offset country-specific shocks by a country-specific monetary policy. Moreover, the transmission of area-wide shocks may be heterogeneous due to structural differences among member countries. In consequence, ECB’s monetary policy is believed to be sub-optimal: ‘one size must fit all’ (Issing 2001) rather than ‘one size fits all’ (Issing 2005).

Consider the suggestive evidence in Fig. 1 below.² Panels 1a and 1b compare the unconditional variances of inflation deviations from an estimated target and the output gap for Euro Area countries and non-Euro OECD countries over three periods: beginning of the Great Moderation until inception of the Euro, start of the Euro until the period of Issing’s (2005) judgment roughly before the beginning of the Great Recession, and, the crisis period since then. The panels suggest that, according to these key indicators of macroeconomic performance, non-Euro OECD countries have been more successful in reducing inflation and output variability after the start of the Euro. Moreover, they have been more successful in stabilizing both inflation and output variability during the most recent period. However, this kind of evidence leaves many questions unanswered. Do these results depend on different shocks hitting the two country groups? Are they uniform across Euro countries, time or policy changes? In this paper, we address these questions through an empirical set-up drawing from both counterfactual analysis and the analysis of productive processes.

There is already a literature that tries to establish empirically if there exists a ‘one size fits all’ monetary policy for Euro Area countries. Most of these papers focus on the transmission of monetary policy shocks before and after the introduction of the Euro in 1999, with somewhat ambiguous results. On the one hand, studies such as Peersman and Smets (2003), and Cecioni and Neri (2011) at the Euro Area level and Peersman (2004) in a cross-country set up do not find asymmetric effects due to monetary policy across Euro Area countries. On the other hand, Barigozzi et al. (2014), Georgiadis (2015) and Burriel and Galesi (2018), in a cross-country empirical framework, show that the common monetary framework produces asymmetric effects driven by structural differences among Euro Area countries.³ Moreover, Ball (2010) finds that the Euro adoption had no significant effects on indicators of macroeconomic performance such as the level or variability of inflation or GDP. Nevertheless, the focus of Ball (2010) is on the effects of adopting inflation targeting (IT). In fact, the bulk of the empirical literature that quantifies the effect of a change in the monetary regime on macroeconomic performance focuses on IT. Two key themes in this literature stand out: first, this literature quantifies the effect of a change in the monetary regime on the moment of a single variable, e.g., the variability of inflation or GDP; second, a key challenge in this literature is endogeneity, as it is unanimously recognized that the choice of IT is affected by initial conditions.

Conceptually, the focus on a single variable does not seem fully appropriate. Measurement and comparison of macroeconomic performance in a theoretical IT framework is routinely based on loss functions that involve inflation and output variability. Independently of whether one assumes optimal monetary policy or a simple Taylor (1993) rule, a central bank faces a long-run tradeoff between inflation and output variability. Moreover, the variability in these endogenous variables is jointly determined by structural supply shocks that move inflation and output in opposite directions, and also by structural demand shocks.⁴ Hence from the standpoint of such a framework, evidence based on the variability of inflation or output in isolation appears problematic. In case such research finds lower inflation variability for Euro Area countries compared to other countries, this might simply imply that the Euro Area countries are located on a different position of the inflation output variability tradeoff, but do not face an improved tradeoff due to the Euro.⁵

Against this background, this paper seeks to examine the claim of whether the Euro Area countries would have faced a more favorable inflation output variability tradeoff without the Euro. To this end we propose a novel empirical research design that is coherent with the bulk of theoretical IT frameworks and tackles the endogeneity issue that comes along with a monetary regime change such as the adoption of the Euro. Furthermore, we take advantage of a more extensive dataset than that available to previous work to explore possible heterogeneities in the responses to monetary regimes across countries and time.

Our research design involves several steps. First, we build a panel data set with observations on the unconditional variance of inflation deviations from target and of the output gap for twenty OECD countries over the sample period 1985 to 2019. We also estimate the variance of the structural supply and demand shocks for each country by the help of a Structural Vector Autoregression (SVAR). Second, as a clear novelty compared to the existing empirical literature on IT, we jointly condition the tradeoff between inflation and output gap variability on the variability of structural shocks using a set-up taken from the quantitative analysis of production processes (see, e.g., Kumbhakar 2012, 2013). In brief, we interpret the variability of inflation and output gap of each country as jointly determined inputs and the variability of structural supply and demand shocks of each country as exogenous outputs, or, more generally, as shifters. Third, in order to establish whether Euro adopters, on average, have been worse off by adopting this monetary regime, we first utilize a difference-in-differences (DiD) approach. However, as discussed in the IT literature (see, e.g., Ball 2010), the choice of adopting the Euro might have been affected by initial conditions and therefore be subject to endogeneity. This can be interpreted as a violation of the parallel trends assumption between the treated (Euro Area countries after the adoption of the Euro) and the control group (the countries taken to construct the counterfactual), which is required by the DiD approach. In consequence, the estimates obtained via the DiD approach may be biased. Therefore, we test for the parallel trends assumption and also consider the lagged dependent variable (LDV) model (for a detailed discussion see Angrist and Pischke 2009). The latter requires less stringent identification assumptions and controls for potential endogeneity of the Euro adoption.

We find that adopting the Euro worsened the macroeconomic performance of Euro countries on average. More precisely, when we account for the possibility that the effects of the Euro may vary over time, we find that the adoption of the Euro on average worsened macroeconomic performance only in the periods of the Financial Crisis and the European Sovereign Debt Crisis. Furthermore, for the Euro Area as a whole the detrimental effect of the Euro ceases after Mario Draghi’s announcement about ‘whatever it takes to preserve the Euro’ and the ECB’s enactment of more intense and additional unconventional policies such as the outright monetary transactions (OMTs), the targeted longer-term refinancing operations (TLTROs) and the expanded asset purchase program (EAPP). Therefore we interpret our findings as evidence that these measures have been effective in reducing inflation and output variability for the Euro Area as a whole. These measures may have credibly signaled that the ECB was going to act as ‘buyer of last resort’ (Acharya et al. 2017), i.e., what De Grauwe (2012) describes as a lender of last resort in the government bond markets.

Disaggregating the analysis across country groups shows that the detrimental effect of the Euro is more severe in peripheral countries. In addition, while this effect of the Euro becomes insignificant for the core of the Euro Area in the period of the above mentioned policy interventions, it remains significant for the peripheral countries until 2016Q1. These findings suggest that structural differences among Euro Area countries are a key element of the detrimental effect of the Euro and that monetary policy in the Euro Area during the crises period was best characterized as a ‘one size must fit all’ policy. The findings are consistent with Burriel and Galesi (2018) who find that the effects of ECB’s unconventional monetary policy measures on Euro Area countries are heterogeneous and related to Barigozzi et al. (2014) and Georgiadis (2015) who provide evidence for asymmetric effects at the country level to common monetary policy shocks in the Euro Area.

Our study is also related to the literature that uses a Taylor (1979) curve to evaluate macroeconomic performance. Cecchetti et al. (2006) evaluate macroeconomic performance for single countries, based on a comparison between two different subsamples of the radial distance of actual unconditional variances from the optimal variances implied by the Taylor (1979) curve. Mishkin and Schmidt-Hebbel (2007) extend the approach used by Cecchetti et al. (2006) to a multi-country level, utilizing a dynamic panel with fixed effects estimated through GMM in order to infer on the macroeconomic implications of IT. However, as illustrated by Angrist and Pischke (2009), identification in a panel with lagged variables and fixed effects is problematic when the policy is endogenous to initial conditions. Olson and Enders (2012) have also made use of a Taylor (1979) curve framework, but use a different metric to measure the distance between observed and optimal variances compared to Cecchetti et al. (2006). Furthermore their analysis is conducted exclusively for the US.

Unlike this literature, our research design does not require explicit assumptions on whether monetary policy in the examined countries is best described by optimal monetary policy or by a simple Taylor (1993) rule. Rather the opposite, our framework encompasses both the inflation output variability tradeoffs implied by optimal monetary policies and by simple Taylor (1993) rules. Besides, in our empirical analysis we explicitly link the jointly determined variability of inflation and output gap to exogenous supply and demand shocks.

Note that there is also a large literature that interprets the Taylor (1979) curve as a policy menu. According to this view, a central bank can choose output variability relative to inflation variability on this tradeoff. The central bank trades a reduction of output variability for an increase in inflation variability (see, e.g., Chatterjee 2002; Olson and Enders 2012). Our paper estimates actual and counterfactual inflation output variability tradeoffs for the Euro Area to examine whether Euro Area countries would have faced lower output and inflation variability without the Euro.

The remainder of the paper is organized as follows. In Sect. 2 we outline the theoretical framework on which we base our empirical strategy. Section 3 describes the empirical implementation and the data in use. Section 4 presents the main results based on the DiD, while Sect. 5 contains our extensive robustness analyses. Section 7 concludes.

We start out by briefly elaborating the theoretical inflation output variability tradeoff in the context of the New Keynesian model. We take the latter as a benchmark for measuring and comparing macroeconomic performance and argue that the inflation output variability tradeoff exists for optimal discretionary monetary policy as well as monetary policy described by a Taylor (1993) rule. Then, we develop a theory-based empirical framework to estimate the inflation output variability tradeoffs in economies independent of any assumption about the type of monetary policy.

Frameworks for measuring and comparing macroeconomic performance in theory are routinely based on loss functions. A popular approach is to consider ad hoc period loss functions such aswhere \(\pi ^{2}_{t}\) denotes the deviation of inflation from an inflation target and \(x_{t}\) denotes the deviation of the output gap from steady state. Parameter \(\omega _{x}\) captures the central bank’s preference for output gap relative to inflation stabilization. Moreover, assume that the aggregate economy is best approximated by a standard New Keynesian model under the rational expectations hypothesis.⁶ Under optimal discretionary monetary policy (as elaborated in Clarida et al. 1999), the central bank minimizes (1) subject to the aggregate economy in each period. One can show that the minimum state variable solution of this model then implies the following long-run relationships in unconditional varianceswhere \(e_{t}\) is an exogenous supply disturbance assumed to be \(e_{t} \sim \text {iid}(0,\sigma ^{2}_{e})\).⁷ In short, (2) to (3) show that both the optimal variances of inflation and output gap depend on the supply shock variance. Moreover, one can verify that the larger the central banks’ preference for output gap stabilization, \(\omega _{x}\), the lower \(\sigma ^{2}_{x,*}\) and the larger \(\sigma ^{2}_{\pi ,*}\).

Likewise, the economic structure impacts the shape of the Taylor Curve. Particularly the degree of price stickiness is a crucial determinant of the slope of the New Keynesian Phillips curve and therefore of the tradeoff. For given \(\omega _{x}\), if prices are more sticky, this has two effects: first, while the output gap variance declines, the inflation variance increases. Second, if the Taylor Curve is interpreted as a policy menu, the tradeoff improves in the sense that the rate of exchange at which a central bank can trade output gap variance for a unit of inflation variance declines. Thus, a central bank has to tolerate less additional inflation variance to reduce output gap variance by the same amount. The same holds for simple policy and given \(\phi _{\pi }\) rule (4) discussed below.

Figure 2a depicts this concept for the case of the US. The variation of \(\omega _{x}\) allows one to depict the Taylor (1979) curve, \(\textrm{F}_{\mathrm {USA,*}}\), which can be thought of as an efficient frontier. The idea is that country-specific supply shocks hit an economy and, given the structure of the economy, create a domestic tradeoff between inflation and output variability. In theory, a domestic central bank, e.g., the Federal Reserve (Fed), can conduct optimal policy and locate the economy on the tradeoff, \({\textrm{Fed}}_{\textrm{optimal}}\) in Fig. 2a.

When it comes to measuring the macroeconomic performance, in practice, the efficient frontier can be estimated, for instance, via a parsimonious reduced form VAR with a supply and demand equation, including reduced form shocks. Such a frontier states an approximation of optimal monetary policy.⁸ Actual observed variability in inflation and the output gap will routinely indicate that the economy is to the right of an estimated efficient inflation output variability tradeoff, \({\textrm{Fed}}_{\textrm{actual}}\) in Fig. 2a. Therefore a central bank’s monetary policy can be classified as sub-optimal.⁹

However, what if monetary policy in a country is not appropriately described by optimal monetary policy, but may be better approximated by a simple Taylor (1993) rule? The latter is a flexible way of describing monetary policies in theory. For instance, it can also involve terms for observed monetary policy inertia, feedback to real economic activity or exchange rates. Therefore, such rules may be a more suitable description of monetary policy for many countries.

As we discuss next, under such a Taylor (1993) rule, there is still an inflation output variability tradeoff. However, this tradeoff is neither optimal, nor is it captured by the approach pursued in Cecchetti et al. (2006) and related studies, which explicitly assume optimal monetary policy. For instance, consider the simple interest rate rule¹⁰The minimum state variable solution under policy (4) then implies the following long-run relationships in unconditional variancesIn contrast to (2) and (3), also demand shocks affect the unconditional variances of inflation and output gap in (5) to (6). The reason for this is, that unlike in the case of optimal policy, a simple rule does not necessarily offset the effects of demand shocks. In addition, the smaller the central bank’s coefficient on inflation, \(\phi _{\pi } \in (1,\infty ]\), the lower \(\sigma ^{2}_{x}\) and the larger \(\sigma ^{2}_{\pi }\).¹¹ Thus, there exists an inflation output variability tradeoff, although the latter is based on the simple interest rate rule (4). The challenge is then to develop an empirical framework that is flexible enough to encompass both the tradeoffs implied by optimal and simple monetary policy.

In this paper, we propose an empirical framework to tackle this challenge. We assume that a inflation output variability tradeoff exists independently of the specific monetary policy in a certain country. Coming back to the example of the Fed in Fig. 2, actual variances observed for the USA may be the result of optimal or sub-optimal monetary policy, but a tradeoff exists at any rate, see Fig. 2b. We solely assume that, consistent with the above theory, an exogenous supply shock shifts output and inflation in opposite directions and determines both the variability of inflation and output. Moreover, a stronger central bank preference for inflation stabilization, i.e., lower \(\omega _{x}\), or, a higher coefficient on inflation in the interest rate rule, \(\phi _{\pi }\), implies a higher variability of output and a lower variability of inflation. Using observations for more countries at different points in time, our framework allows us to fit a convex curve as depicted in Fig. 3, which is shifted by changes in the variance of supply shocks.

Our empirical strategy builds on tools developed in the quantitative production analysis. We use a specification based on a translog transformation function (TTF). Kumbhakar (2012, 2013) shows that an input-oriented TTF can be used to model the determination of one or more endogenous production inputs, for exogenous production outputs, and technology. Here we use the input-oriented TTF to model the joint determination of the endogenous variances of inflation and output gap (i.e., the two inputs in the TTF framework), for a given variance of an exogenous structural supply shock (i.e., a single output or shifter in the TTF framework) and a given monetary policy. In this way, the macroeconomic performance of different countries can be gauged controlling for country-specific supply shocks. As we show in Appendix A, we can obtain the following empirical specificationThe above function is normalized with respect to \(\sigma ^{2}_{x}\), but exactly the same econometric results would be obtained by normalizing on \(\sigma ^{2}_{\pi }\). Furthermore, provided that shifters are exogenous, the presence of \(\sigma ^{2}_{x}\) (in the \(\sigma ^{2}_{\pi }-\sigma ^{2}_{x}\)-ratio) among the regressors does not make OLS estimates inconsistent (see Kumbhakar 2012, 2013, for a formal treatment of this issue). Estimating (7) allows one to test whether a tradeoff between the variability of output and inflation actually exists in the data. Note that conditionally on the existence of this tradeoff, estimation of (7) uses the statistical information on macroeconomic performance more efficiently than the usual estimates based on either inflation or output gap variability alone. The reason is that in this set-up one can use the variability of the output gap (respectively inflation) to model the variability of inflation (respectively output gap).

Our goal is to estimate the inflation output variability tradeoff for a number of countries \(i=1,\ldots ,N\) over time \(t=1,\ldots ,T\) based on (7). However, the empirical implementation of (7) is not obvious. In principle, as said above, one can consistently estimate the inflation output variability tradeoff by a two-way fixed-effect model, i.e.,where we assume \(v_{i,t} = \alpha _{i} + \delta _{t} + \varepsilon _{i,t}\). \(\varepsilon _{i,t}\) is a stochastic error term, \(\alpha _{i}\) a fixed effect aimed at capturing unobserved time invariant country factors and \(\delta _{t}\) can be thought of as a flexible (nonlinear) time trend, i.e., a common unobserved factor (shock) affecting all countries by the same amount (for further details see Smith and Fuertes 2016).

Yet, some major challenges emerge with regard to the estimation of (8). First, consistent with theoretical inflation output variability tradeoff, we require observations of the variances of inflation deviation from target, of the output gap, and of the structural supply shock. This in turn implies some de-trending of the inflation and output data and the derivation of structural supply and demand shocks for each country. We estimate the latter by SVARs. Moreover, we want to examine the effect of the Euro monetary policy on macroeconomic performance of Euro Area countries after they have adopted the Euro relative to a comparable set of countries without the Euro. Hence, we need to develop an identification strategy for the effect of the Euro on macroeconomic performance. We address these issues below.¹²

The first three columns of Table 2 present the baseline results for the DiD specification (9). First, notice that the coefficient for the ratio of the variability of inflation to output gap, \({\hat{\beta }}_{2}\), is highly significant in all columns, in line with the considerations developed at the end of Sect. 2. Next, the coefficient for the non-linear term, \({\hat{\beta }}_{2,2}\), is not significantly different from zero. Yet, this is not necessarily evidence against a convex inflation output variability tradeoff for the Euro Area as a whole, as well as for the core and the periphery.¹⁸ Second, the coefficient \({\hat{\alpha }}_{e}\) shows that the variance of the supply shock has a highly significant impact on the location of the tradeoff in all the Euro Area, the core and the periphery. It has a negative sign, \({\hat{\alpha }}_{e} < 0\), therefore, the larger the variance of the supply shock, the larger the variance of inflation and the output gap.¹⁹ Finally, demand shocks appear to have some impact on the location of the tradeoff in all the Euro Area, the core and the periphery. In sum, the significant coefficient estimates are evidence for the existence of an inflation output variability tradeoff for the Euro Area consistent with the theoretical tradeoff discussed in Sect. 2.

Most importantly, the coefficient for the dummy on Euro adoption, \({\hat{\beta }}_{{\mathcal {E}}}\), is significant for all the Euro Area, the core, and the periphery. A negative sign for \({\hat{\beta }}_{{\mathcal {E}}}\) means that, on average, Euro adopters have been worse off due to adopting this monetary regime.

We can provide some further interpretation for the coefficient estimates. For instance, for \({\hat{\beta }}_{{\mathcal {E}}} = -0.605\) follows that \(\exp ({\hat{\beta }}_{{\mathcal {E}}}) \approx 0.55\). The latter can be interpreted as the ratio of the Euro Area (post-Euro introduction) and control group transformation functions. Thus, the inverse of this ratio is \(\approx 1.83\) and means that, conditionally on the supply and demand shock variances, the Euro Area has a joint variance of inflation and output, which is around \(83\%\) larger than that of the control group. Likewise for the core and the periphery this variance was respectively \(75\%\) and \(94\%\) larger.

Finally, notice the p-values for the Ramsey (1969) Reset test. Under this test, rejection of the null provides clear evidence of misspecification, especially in the form of omitted variables (Godfrey and Orme 1994). However, the null, i.e., omitted variables being orthogonal with respect to the included variables, is never rejected. Therefore, we cannot find evidence that estimates of equation (9) suffer from misspecification.

In the next three columns of Table 2, we present results for the specification (11), where we include lags of the treatment, along the lines of Autor (2003). This exercise allows us to provide a more articulated economic interpretation of the basic finding of a detrimental effect of the Euro. Consider the fourth column in Table 2, which relates to the Euro Area as a whole. Compared to the previous results, three observations stand out. First, there is again significant and well-specified evidence in favor of an inflation output variability tradeoff consistent with the theoretical tradeoff developed in Sect. 2. Second, coefficients \({\hat{\beta }}_{{\mathcal {E}},0}\) and, \({\hat{\beta }}_{{\mathcal {E}},2}\), \({\hat{\beta }}_{{\mathcal {E}},3}\) and \({\hat{\beta }}_{{\mathcal {E}},4}\) relating to periods including the Financial Crisis, the European Sovereign Debt Crisis, and the period of Mario Draghi’s ‘whatever it takes’ and OMTs announcements and the EAPP announcement and implementation are significant, while \({\hat{\beta }}_{{\mathcal {E}},5}\), relating to the time after this policy turnaround, is insignificant.²⁰ We interpret these findings as evidence that the impact of the Euro adoption has changed over time. The inflation output variability tradeoff for the Euro Area countries is worse during the crisis period (relative to the control group) until the ECB’s 2012 policy turnaround, but not thereafter. The coefficient estimates can be interpreted as follows: \(\exp ({\hat{\beta }}_{{\mathcal {E}},0}) \approx 0.55\), \(\exp ({\hat{\beta }}_{{\mathcal {E}},2}) \approx 0.47\), \(\exp ({\hat{\beta }}_{{\mathcal {E}},3}) \approx 0.35\), and \(\exp ({\hat{\beta }}_{{\mathcal {E}},4}) \approx 0.57\) imply that the Euro Area had a joint variance of inflation and output, which is more than \(80\%\) (\(111\%\), \(187\%\), \(77\%\)) larger during 1999Q1 to 2002Q2 (2006Q1 to 2009Q2, 2009Q3 to 2012Q2, 2012Q3 to 2016Q1) compared to the control group.²¹

The fifth and sixth column in Table 2 relate to the analysis on the core and periphery. Their main features can be summed up by the following four remarks. First, throughout all specifications the estimated coefficients regarding the inflation output variability tradeoff are similar to the previous results. Second, again the Reset test does not give rise to concerns about omitted variables or specification problems. Third, the effect of the Euro on the core is less detrimental than for the Euro Area as a whole, except for period 5. There is again a detrimental effect of the Euro, but once we allow for heterogeneous effects of the treatment over time, the effect is observed in periods 8, but not in periods 9 and 10.

However, and fourth, the coefficient estimates for the periphery reveal differences vis-à-vis the former results. From 2009Q3 onwards the detrimental effect of the Euro is more severe in the periphery as the respective coefficients are larger in absolute value. This is consistent with the asymmetric effects of shocks in the Euro Area as found in Barigozzi et al. (2014), Georgiadis (2015) and Burriel and Galesi (2018). More crucially, the detrimental effect of the Euro ceases in the periphery only in period 10. Our findings can be interpreted as evidence that the core has benefited significantly earlier from the monetary policy turnaround occurring in 2012.

The purpose of this section is threefold. First, we consider the LDV approach, which is an alternative identification strategy for the ATT. Second, we assess the robustness of our findings obtained via the DiD and LDV approach with regard to choice of the de-trending method. Third, we present an additional robustness analysis relating to the identification scheme in the estimation of the structural shocks.²²

Summing up, we consistently find a detrimental effect of the Euro on macroeconomic performance in period 8 (2009Q3 to 2012Q2), and, depending on the specification and filter also in periods 5 (1999Q1 to 2002Q2), 7 (2006Q1 to 2009Q2) and 9 (2012Q3 to 2016Q1). Moreover, we find that the detrimental effect of the Euro on macroeconomic performance during crises is more severe in peripheral countries and does already disappear in period 9 in the core, but not in the periphery.

It is worth emphasizing that these findings are conditional on controlling for country-specific supply and demand shocks in Euro Area and control group countries. Therefore, bad luck does not strike us as a reasonable explanation of our findings.²³

We surmise an interpretation of these results in terms of a monetary policy for the Euro Area that can be denoted a ‘one size must fit all’ (Issing 2001) monetary policy during the start of the Euro Area (period 5) and a ‘one size fits all’ policy in period 6 (2002Q3 to 2005Q4) as the stabilizing channels of a currency union become more effective (Issing 2005). However, during the subsequent periods of crises (periods 7, 8 and 9) the ECB’s policy can again be qualified as ‘one size must fit all’. Next, our finding that the detrimental effect of the Euro for the whole Euro Area ceases in period 10 has two implications. First, it suggests that the detrimental effect during periods 7, 8 and 9 is directly related to periods of crises and so is our ‘one size must fit all’ judgment. This is consistent with De Grauwe’s (2012) Eurozone fragility hypothesis. The ECB did not immediately react to solvency concerns regarding some peripheral Euro Area countries by signaling its willingness to act as ‘buyer of last resort’ on the market for bonds of European governments, although this would have been a natural policy for independent national central banks in these peripheral Euro Area countries. Second, we interpret our finding that the detrimental effect of the Euro ceases for the whole Euro Area in period 10 meaning that the ECB finally acted in such a way to make clear that it was willing to act as ‘buyer of last resort’. In turn, European sovereign debt markets calmed down, leading to the disappearance of the detrimental effect of the Euro on macroeconomic performance not only for the core, but also for the periphery. This narrative is also supported by empirical work on the effects of these announcements on European sovereign debt markets (see, e.g., De Grauwe and Ji 2013; Saka et al. 2015) and broadly consistent with the interpretation of events, monetary policies and evidence surveyed in Dell’Ariccia et al. (2018).

To gain further understanding on these issues, it is also interesting to consider, throughout periods 7–10, the evolution of the shadow rate developed in Wu and Xia (2016). We depict this rate for the US and the UK as examples of the control group on the one side and the Euro Area on the other side in Fig. 4.

We observe that the shadow rates in the control group countries moved below zero much earlier than in the Euro Area. On the other hand, the shadow rate for the Euro Area becomes consistently negative only since mid 2013 and lines up with the other rates only toward the end of period 9 (early 2015; of course since that period, the shadow rate of the US - but not of the UK - began to rise again). This stylized fact is broadly consistent with the interpretation we gave above for our empirical evidence.

Finally, we would like to point to potential avenues for future research. We use the baseline three-equation NK model as an illustrative motivation. Then we estimate the inflation output variability tradeoff based on an aggregate demand and supply shock, motivated by this baseline theoretical framework. However, the applied business cycle analysis routinely employs medium- to large-scale DSGE models with seven (see, e.g., Smets and Wouters 2007) or more shocks. Thus, one topic for future research could be to consider a larger number of shocks derived from a medium-scale SVAR (see, e.g., Canova and Paustian 2011). Next, a more general concern applies to both small- and medium-scale SVARs. These models include a relatively small number of variables, which may render the estimated shocks nonfundamental. However, increasing the number of variables without limit is not feasible because of the curse of dimensionality. Thus, another topic for future research would be to resort to large-scale models such as SFAVARs (see, e.g., Bernanke et al. 2005) or large-scale BVAR models. Nevertheless, we are confident that our parsimonious SVAR specification is not crucial for our results. Our findings are robust across different econometric specifications, shock identification schemes, and, detrending methods. Moreover, the Ramsey (1969) Reset test is almost invariably insignificant in our regressions, providing further evidence that variable omission is not likely to be a serious problem.

This paper conducts a counterfactual analysis providing evidence that the inflation output variability tradeoff of Euro Area countries deteriorated during the periods of the Great Recession and the European Sovereign Debt Crisis. This deterioration was more severe and long-lasting for the peripheral countries of the Euro Area, while basically ceasing with the Draghi and OMTs announcements as well as with the EAPP announcement and implementation.

Our findings are based on a novel empirical strategy that is consistent with a theory-based inflation output variability tradeoff whose position is influenced by structural supply and demand shocks. We develop a panel data set for twenty OECD countries which consists of variances for output gap and inflation deviations from target as well as variances for the structural supply and demand shocks. The shock variances are estimated via a structural model that uses sign restrictions to identify shocks. In the estimation of the tradeoff, we model the joint determination of the variability of inflation and output by structural shocks through a transformation function taken from the quantitative analysis of production processes. The counterfactual evidence relating to ATT’s is robust throughout various empirical specifications.

We interpret the higher inflation and output variability in the Euro Area during the periods of crisis as evidence that the ECB measures during these periods have not been effective to reduce inflation and output variability to levels comparable with other economies. The detrimental effect attenuates and finally vanishes after the Draghi announcement. This suggests that the policy moves subsequent to Draghi’s ‘whatever it takes’ announcement have been effective in reducing inflation and output variability in the Euro Area on average. We argue that this is the case, because these moves credibly signaled that the ECB was going to act as ‘buyer of last resort’.

Most importantly, our more detailed analysis shows that the detrimental effect of the Euro was more severe for peripheral countries of the Euro Area. Moreover, while the Draghi and OMTs announcements as well as the EAPP announcement and implementation immediately had a beneficial effect on the macroeconomic performance of the core, this effect occurred much later for the periphery. Hence, structural differences between core and periphery countries are likely to be the driving force behind the impact of the Euro. Investigation about the nature and role of these structural differences must however be left to future research.