In traditional Phillips curve approaches, global slack only indirectly affects domestic inflation. The Phillips curve is broadly understood as the relationship between inflation and economic slack, where economic slack is traditionally defined only in domestic terms. In such a framework, global slack nonetheless has considerable influence on commodity prices, which then affect domestic inflation via import prices for commodities. Furthermore, foreign output gaps matter for short-run inflation dynamics by affecting import prices for these goods. Finally, global cyclical conditions affect the domestic output gap indirectly, since stronger global demand for goods and services supports domestic income via the net exports channel.
These traditional approaches have been challenged by arguments that globalisation has made national inflation responsive to global slack – in addition to domestic slack. In this literature (e.g. Borio and Filardo 2007), one argument is that any sudden expansion in domestic demand for goods and services could translate into higher imports rather than higher domestic prices. The degree to which this dampens domestic prices then depends on the global tightness or slack. Another argument is that globalisation constrains domestic wage or price increases in industries open to global competition. This lowers the sensitivity of wages to domestic demand pressures, giving foreign slack a direct influence on domestic inflation. In this context, the role of global value chains in increasing the global contestability of factor and labour markets (Auer et al. 2017) has received a lot of attention (for a more in-depth discussion of global versus domestic drivers of inflation, see ECB 2017 or Tagliabracci et al. forthcoming).
Some of the channels through which global slack could affect domestic inflation may already be captured (at least implicitly) in traditional Phillips curves. In particular, global slack affects import price inflation, which is usually included in Phillips curve models. Measures of domestic slack also incorporate indirectly some information about global conditions, since global demand for goods and services is reflected in net trade. Expectations of foreign demand affect investment decisions quite strongly. At the same time, the above mentioned channels of globalisation are not explicitly captured in the standard Phillips curve framework (ECB 2017).
Figure 1 Foreign and Eurozone output gap (percentages)
Sources: ECB calculations.
Notes: The EA output gap is based on data from the European Commission. Foreign output gap (data/projection) is trade-weighted and based on the latest IMF WEO data. The “EA output gap not explained by foreign output gap” is derived by an auxiliary regression (regressing the domestic on the global output gap).
One way of assessing the role of such global influences on domestic inflation is to augment the traditional Phillips curve in a thick-modelling approach with measures of foreign slack. A thick-modelling approach addresses the uncertainty about the most appropriate specification of the Phillips curve by estimating a large set of specifications including several different measures of (domestic and foreign) economic slack and inflation expectations (Ciccarelli and Osbat 2017). Also including foreign slack is complicated by the fact that domestic and foreign slack are highly correlated. To tackle the problem of multicollinearity, an auxiliary regression is run to obtain the part of domestic slack which is not explained by foreign slack. These residuals are then used as a measure of domestic slack. Figure 1 shows the development of foreign and domestic slack (in terms of the respective output gaps) as well as the series with the residual of domestic slack resulting from the auxiliary regression.
Overall evidence supporting the importance of the role of global slack for determining domestic inflation based on Phillips curve analyses is mixed. The literature provides only limited support for including a measure of foreign slack in traditional Phillips curve analyses. On the one hand, Borio and Filardo (2007) found that proxies for global economic slack added considerable explanatory power to traditional benchmark Phillips curve approaches in advanced economies and that the role of global factors had grown over time. The relevance of the global output gap was also supported by Milani (2009) in the case of the US after 1985. On the other hand, studies like Calza (2008), Gerlach et al. (2008), Ihrig et al. (2010), Martínez-García and Wynne (2010), and Eickmeier and Pijnenburg (2013) find conflicting evidence and suggest that the results of Borio and Filardo are likely to be specific to the estimation sample or measurement of the global output. More recently, Mikolajun and Lodge (2016) detect no appreciable direct effects of global economic slack on domestic inflation for the majority of advanced economies.
Figure 2 Significance of foreign slack and GVC integration measures in Eurozone Phillips curve specifications for HICPX
(share of specifications in which the respective variables are significant as a percentage of total specifications analysed; sample period: 2000-2016)
Sources: ECB calculations.
Notes: ECB staff calculations. For technical details, see Economic Bulletin issue 4/2017 – domestic and global drivers of inflation in the euro area – box 2. Analyses include up to 108 different Phillips curve specifications. ‘PC’ stands for Phillips curve. The results are based on standard significance tests (applying a 10% level of significance), which do not include a variance correction to account for the fact that auxiliary regressions were employed. The results should hence be interpreted as representing an upper bound with respect to the share of specifications in which foreign slack and GVC variables are significant.
In line with the mixed evidence found in the literature, a thick-modelling approach for the Eurozone finds that including a measure of global slack in the Phillips curve for the harmonised index of consumer prices excluding food and energy (HICPX) inflation establishes a significant role for foreign slack only in around one third of the specifications (see Figure 2). If indicators for the integration in global value chains and foreign slack are included simultaneously, they are significant in around 50% to 60% of the specifications. The upward trending global value chain measure captures a downward sloping trend in inflation, which in some specifications is significant if included in addition to the weighted foreign slack measure. However, the global value chain measure is almost never significant when interacted with global slack, implying that the integration in global value chains does not seem to have an amplifying effect on the role of foreign slack. Against the background of these findings, we will in the following discuss whether an augmentation of a Phillips curve by foreign slack actually improves the performance of Phillips curve approaches in explaining the recent period of low inflation in the Eurozone.
Augmenting traditional Phillips curve approaches in a thick modelling approach by measures of foreign slack would have slightly improved the fit of Phillips curve based forecasts for HICPX over the last years. Figure 3 illustrates that after 2012, actual HICPX developments were at the lower bound of forecasts based on a broad range of fixed-coefficient Phillips curve specifications conditioned on the outturns for different measures of domestic slack. Also including foreign slack slightly shifts the range of forecasts down, and actual developments are then somewhat more in the middle of the range of estimates.
Figure 3 HICPX: Actual and conditional out-of-sample projection (thick modelling approach; annual percentage changes)
Sources: ECB calculations based on Eurostat, IMF, Consensus Economics and SPF data.
Notes: Results based on a thick-modelling approach including a broad range of fixed-coefficient specifications of the Phillips curve including either only domestic or domestic and foreign slack (see Figure 1). The parameters are estimated over the sample 1995Q1 to 2016Q4. The conditional out-of-sample forecast is done for 2012Q2 to 2016Q4. The range depicts forecasts for HICPX coming from differently specified Phillips curves. The specifications include permutations across the expectation formation (backward- or forward-looking) and the variables representing economic activity/slack. For more details on the thick-modelling approach see Ciccarelli and Osbat (2017) or the speech of B. Coeuré “Scars or Scratches - hysteresis in the euro area “(Geneva 19 May 2017).
Summing up, there is some tentative evidence that augmenting Phillips-curve approaches by measures of foreign slack could help to slightly better explain past developments in underlying inflation. However, these results have to be interpreted with some caution. First, it is driven only by a small share of specifications, at the upper and the lower bound of the range of model estimates of a thick-modelling approach, while a majority of specifications with and without foreign slack yields very similar results (as reflected in the overlapping range of model estimates). Furthermore, even for a period when developments of domestic slack differed substantially from developments of foreign slack, the effects seem to be rather small. Looking ahead, further analysis is needed for a solid assessment of the potential role of foreign slack for domestic inflation in the Eurozone, but also at a country level.
Author’s note: The views expressed here are the author’s own and do not necessarily reflect those of the ECB.
Auer, R, C Borio and A Filardo (2017), “The globalisation of inflation: The growing importance of global value chains”, BIS, Working Papers no 602.
Auer, R, A Levchenko and P Sauré (2017), “International inflation synchronisation through global value chains” VoxEU.org, 19 May.
Borio C and A Filardo (2007), “Globalisation and inflation: New cross-country evidence on the global determinants of domestic inflation”, BIS Working Papers no 227.
Calza, A (2008), “Globalisation, domestic inflation and global output gaps: Evidence from the Euro Area”, Federal Reserve Bank of Dallas, Working Paper no 13.
Ciccarelli, M and C Osbat (2017), “Low inflation in the euro area: Causes and consequences”, ECB, Occasional Papers no181.
Eickmeier, S and K Pijnenburg (2013), “The global dimension of inflation – Evidence from factor-augmented Phillips Curves”, Oxford Bulletin of Economics and Statistics 75(1): 103-122.
European Central Bank (2017), “Domestic and global drivers of inflation in the euro area”, Economic Bulletin, Issue 4.
Gerlach, S, A Giovannini, C Tille and J Viñals (2008), Low inflation: Testing times for central banks, 10th Geneva Report on the World Economy.
Ihrig, J, S Kamin, D Lindner and J Marquez (2010), “Some simple tests of the globalization and inflation hypothesis”, International Finance 13(3): 343-375.
Martínez-García, E and M Wynne (2010), “The global slack hypothesis”, Federal Reserve Bank of Dallas, Staff Papers no 10.
Mikolajun, I and D Lodge (2016), “Advanced economy inflation: The role of global factors”, ECB, Working Paper Series no 1948.
Milani, F (2009), “Global slack and domestic inflation rates: A structural investigation for G-7 countries”, Federal Reserve Bank of Dallas, Working Paper no 33.
Tagliabracci, A, C Osbat and G Koester (forthcoming), “The role of foreign slack and global integration for domestic inflation”.