Understanding and overcoming low inflation
Remarks by Vítor Constâncio, Vice-President of the ECB, at the Conference on “Understanding inflation: lessons from the past, lessons for the future?”, Frankfurt am Main, 21 and 22 September 2017
Ladies and gentleman,
As this successful conference draws to an end, I wish to extend a warm thank you to all participants, for collaborating with this ECB research initiative, which has been very fruitful.
In my concluding remarks, I will discuss a few issues raised during the conference. Understanding inflation dynamics is especially important today due to the surprisingly low inflation levels prevailing in many countries. Moreover, there is evidence suggesting that in some countries, the relationship between slack in the economy and inflation seems to have weakened recently. In this context, a striking development is the emergence, of a “twin puzzle” after the Great Recession: first, “missing disinflation” from 2009 to 2011 (Coibion and Gorodnichenko, 2015), and second, “missing inflation or reflation” after 2012, particularly in Europe (Constâncio, 2015). During the Great Recession, inflation in advanced countries did not fall as much as the usual relation between inflation and economic slack, otherwise known as the Phillips curve, would have predicted, given the severity and length of the recession. Just as puzzling, more recent global developments point in the opposite direction. Between 2012 and 2017, both headline and core inflation in the euro area were lower than expected (i.e. lower than forecasted by the Eurosystem and other institutions). Despite the ongoing recovery, headline inflation rates in several advanced economies remain well below target.
In light of these developments, the ESCB set up a network of experts to study the causes of this low inflation and to research the implications for policy (Ciccarelli and Osbat, 2017). That network has helped us to better understand inflation dynamics in the euro area. In particular, we now know that low inflation in the euro area had different origins: initially, it was mainly due to external factors (the fall in foreign demand, energy and food prices) while, after 2012, inflation was predominantly driven by domestic sources in an environment of weak demand.
Still, a great deal remains to be understood. In this respect, the research presented at this conference has shed light on many angles. Several conference papers have added to our understanding of the drivers of inflation, the nature of inflation dynamics and inflation expectations, also highlighting how new data can help us improve our knowledge. The dynamics of inflation after the financial crisis and the apparent disconnect with real activity led to a belief that this relationship might have broken down or at least became unstable, for headline inflation as well as for core inflation and wages. From a policymaker’s perspective, such an apparent breakdown is serious. The presence of a structural Phillips curve relationship is part of the traditional transmission mechanism which central banks rely on to control inflation over the medium-term. The apparent disconnect between inflation and economic slack seems to have made interpreting and controlling inflation dynamics more difficult, with significant consequences for the conduct of monetary policy. Namely, the inflation trade-offs would worsen: on the one hand, if inflation rises as a result of cost-push shocks unrelated to domestic slack, then it would be more costly, in terms of output loss, to bring inflation down. On the other hand, the flip side is that a sharp dip in inflation would require a more substantial monetary policy stimulus, which in turn could be problematic given the lower bound on nominal interest rates. In sum, as I recently underlined, “a flatter slope of the Phillips curve would make controlling inflation either more costly or more difficult” (Constâncio, 2015). The reduced effect of economic slack is even more significant with regard to wages, a crucial factor of domestic underlying inflation. The fact that wages are not increasing more is an important puzzle in advanced economies. Higher wage increases were to be expected due to the strengthening of the economic recovery and are necessary in order to normalise inflation.
In any case, changes in the Phillips curve over time make the work of the central bank more difficult, especially as they have to be understood in real time. Fortunately, in recent years, , the slope of the Phillips curve has been increasing somewhat in what regards headline inflation in the euro area, especially if we use a broad concept of unemployment (Unemployment 6, which is now at 18%) to measure economic slack. This points to one of the difficulties in using Phillips curves: the choice of an appropriate measure of slack which is a multidimensional reality.
Many explanations have been put forward in the literature to rationalise the decrease of the Phillips curve coefficient of economic slack in all advanced economies. We can group them around three categories of factors: the increased importance of external shocks, including those related to globalisation; the enhanced role of expectations in influencing wage and price decisions, anchored by central bank targets, and finally, the existence of non-linearities and time-varying behaviour of different Phillips curve coefficients.
I will touch upon these three aspects of the Phillips curve relationships that can be at the root of the recent inflation puzzles, trying to link them with the papers and discussions at this conference. .
External shocks and exchange rate pass-through
Let me start with external shocks or global drivers of inflation. There are several dimensions of the growing importance of international developments in influencing domestic inflation rates notably those related with economic and financial globalisation. Some economists have gone as far as to maintain that the global economic slack became more important than the domestic one to explain inflation to the point of significantly eroding the role of monetary policy in controlling inflation. However, it is very difficult to capture, in quantitative analyses, all the effects of globalisation on inflation. The channels are various and as they overlap, not all of them can be included at the same time.
First, one can list the increased import content of economic activity as trade grows, the extension of global value chains and the augmented input-output linkages across countries. These factors are most visible through the impact of oil prices on energy inflation. Energy accounts for around 10% of the consumer goods basket used to measure the HICP. Energy inflation is very closely linked to oil prices (See ECB Economic Bulletin article 4/2017: “Domestic and global drivers of inflation in the euro area”). And then there are the indirect effects of oil prices on domestic product costs, which can then work their way through the supply chain to influence headline inflation. The global influence is, of course, not limited to oil and energy. Food prices in the euro area are affected by global food prices and prices of imported consumer goods are important for domestic inflation as well. The prices of imported goods and services directly affect headline inflation via their shares in final consumption and indirectly via their use as intermediate products.
Beyond commodity prices, the role of import prices has been enhanced by the extension of global value chains and the linkages that this creates. The paper by Auer and co-authors (2017) presented at this conference illustrates this point well. Starting from the worldwide synchronisation of producer price inflation (PPI), they conclude that input-output linkages explain 50% of the global component of PPI. In another recent BIS paper, Auer, Borio and Filardo (2017) show how the relevance of global value chains explains why the contribution of the global economic slack has gained importance. Nevertheless, the analysis of Bianchi and Civelli (2015) for 50 countries from 1971 to 2006, casts some doubt about the global slack coefficient having a consistent increase accompanying the progress of globalisation. Only when the two hypotheses of a decreasing significance of the domestic slack and the increased importance of the global slack are jointly tested, do they find “that in fact trade openness enhances the relative importance of the foreign output gap on domestic inflation”
However, Ciccarelli and Mojon (2010) and Ferroni and Mojon (2014) had previously shown that the commonality of inflation globally goes beyond what can be captured by commodity prices. They also report that the global inflation factor would improve forecasts of domestic inflation in different specifications, from augmenting an AR (1) model to using it in Phillips curve or BVARs. Furthermore, they demonstrate that the importance of the common factor does not depend on spillovers among countries but is rather the result of common shocks and convergence of monetary policy frameworks around the world. One could perhaps add that the subdued behaviour of wages can be partially attributable to a common factor related to the loss of labour pricing power at the national level, stemming from the relocation threat.
Moreover, the way in which exchange rate movements pass through into import prices at the border and at the final consumer level is also critically important to understand the influence of external shocks on inflation. Consequently, understanding exchange rate pass-through into aggregate prices is vital for forecasting inflation and setting monetary policy. In this conference, the paper by Forbes et al. (2017) shows that the Exchange Rate Pass-Through (ERPT) is variable, both across countries and across time. For policymakers this is a challenge, as it makes it significantly more difficult to predict the effect of fluctuations in currency markets on domestic inflation. However, an important insight from this paper and also previous work by Kristin Forbes and co-authors is that the ERPT to inflation may differ depending on the nature of the shocks that triggered the exchange rate movement, as it may reflect different transmission channels. The empirical evidence confirms that different shocks exert different effects on inflation, possibly even leading to inflation responses in the opposite direction of the exchange rate movement. For instance, results in the ESCB LIFT task force (Ciccarelli and Osbat 2017) point to a significantly higher ERPT than on average in case of a monetary policy induced currency appreciation, and even to upward pressure on prices following an appreciation driven by domestic demand. In the latter case, the direct impact of domestic demand on inflation dominates the one stemming from the exchange rate. This insight is especially relevant for the euro area, as it suggests that the recent euro appreciation may have a more limited dampening effect on inflation than what would be implied by historical averages.
Finally, I would like to mention the paper presented by Laseen and Sanjani (2016) that analyses whether the financial crisis, together with globalisation has broken the Phillips curve in the U.S. the economy in which traditional reduced form single equations have shown the highest decline in the economic slack parameter. Using multivariate, time-varying methods (a large BVAR, a DFM and a MS-BVAR) they show that changes in shocks of other variables are “a more salient feature of the data than changes in coefficients”. This is encouraging as it implies that, although financial and external variables have the highest forecasting power for inflation after the crisis, their effect could be more relevant in the short-term, leaving the coefficient of the economic slack in the Phillips curve still significant. In regime-switching models they show that “what changes across regimes is most importantly the variances of structural disturbances”. Together with the slight steepening of the Phillips curve in the euro area, this gives some hope that the future closing of the output gap will allow us to gradually reach our inflation target. In any case, as stated by the ECB President in the latest ECB press conference: “a very substantial degree of monetary accommodation is still needed for underlying inflation pressures to gradually build up and support headline inflation developments in the medium-term”.
Insights from household inflation expectations
Modern macroeconomic models attribute a central role to inflation expectations in price and wage setting. This literature has stressed the management of inflation expectations as a crucial monetary policy channel, and the importance of central bank credibility. When forward-looking agents expect the central bank to stabilise future inflation, they also react less to current disturbances in adjusting prices and wages, leading to more stable inflation outcomes. Therefore, an important element determining the effectiveness of monetary policy is the trust of the public in the commitment and the ability of the central bank to reach its target in the medium-term. In turn, measures of long-term inflation expectations are invaluable in helping to understand if expectations are well anchored.
The novelty of the work by Christelis, Georgarakos, Japelli and van Rooij presented in this conference, is that they measure both trust in the ECB and inflation expectations of individual consumers through a specifically designed survey. Indeed, their findings are reassuring because they establish a clear link between high levels of trust and inflation expectations that are anchored around the medium-term inflation target. Specifically, they find that trusting the ECB increases inflation expectations when these are below the ECB’s inflation objective, and lowers them when they are above it. Moreover, higher trust in the ECB significantly reduces uncertainty about future inflation. Their findings suggest that a central bank can influence the economy through people’s expectations, even in times when conventional monetary policy tools likely have weak effects. This is certainly very encouraging for policymakers.
The literature has also made some progress in incorporating data on actual inflation expectations into macro models, usually relying on survey data. In earlier work, Coibion and Gorodnichenko (2015) suggested a solution to the “missing disinflation” in the U.S., showing that household inflation expectations from 2009 to 2011 remained elevated, reflecting the increase in oil prices at the time. This conference has featured the paper by Slobodoyan and Wouters which embeds inflation expectations from the Survey of Professional Forecasters in a New Keynesian model with learning. This paper makes a strong case for entering survey expectations data into the standard dataset on which these macro-models should be estimated. What we learn is that survey information improves the model forecast for both inflation and the other macroeconomic variables, especially when expectations are modelled based on adaptive learning.
The work by Coibion and Gorodnichenko I mentioned before points to a source of stickiness in the formation of consumer inflation expectations that can lead to more stable inflation outcomes, independently of monetary policy actions, sometimes helping the monetary authority in achieving its inflation objective, sometimes making it more difficult. In this vein, the work by Weber and co-authors shows us that not all price changes are treated equally by households in forming their inflation expectations. Price changes that are more readily observed such as those that are larger or that are from goods that are purchased more frequently are more important in shaping inflation expectations. To some extent, the large cross-sectional variation of inflation expectations among households is somewhat worrisome for the policymaker, especially if (as the Weber et al. and the paper by Vellekoop and Wiederholt suggest) people change their consumption behaviour and make financial decisions based on these expectations. For policymakers, this seems to suggest that there is an important role of the central bank in shaping the expectations of the general public, not only that of financial markets. It also suggests that more research is needed to understand the different factors that shape the inflation expectations of individual households, in particular the role of volatile sub-series of the HICP in shaping the inflation expectations of the general public and on the decisions that households make, based on their inflation expectations. Only more detailed data that matches inflation expectations of households with their decisions will be able to shed more light on this issue.
Building better models of nominal price rigidities
This brings me to the last point of improving structural models of the Phillips curve, and the importance of non-linearities, particularly those that result from considering state-dependency of the Phillips curve and the possible different behaviour of price stickiness in different phases of the business cycle. Practically all models that are considered to be the core of modern monetary macroeconomics rely on the assumption of nominal rigidities, i.e. the fact that prices do not immediately adjust to changes in costs and demand. In such models, price stickiness is the ultimate cause of why monetary impulses have real effects and are transmitted slowly through the economy. Understanding the degree to which prices are sticky or inflation is persistent is important for central banks as it determines how interest rates need to be set to achieve the desired level of inflation.
Still, the exact nature and mechanisms that are at the core of price stickiness continue to be imperfectly understood. The debate among macro-economists on the foundations of nominal price rigidities, and therefore how to model them, is ongoing and far from settled. This is important for our understanding of the Phillips curve from a structural point of view, going beyond a simple correlation between inflation and economic activity. The structural Phillips curve relationship that comes out of the different models of nominal rigidities can be very different. These differences have implications for determining the degree of price stickiness and thus the slope of the Phillips curve.
In general, time-dependent and state-dependent models have radically different implications in terms of the determinants of the slope of the Phillips curve, and its sensitivity to the business cycle. In line with the predictions of state-dependent models of price setting, recent research has shown that several features of the distribution of price changes vary over the business cycle (Berger and Vavra 2015). This would imply that the degree of price stickiness and thus the degree of non-neutrality of monetary policy is also different in expansions and recessions. This is a great example of how micro data can provide crucial evidence on nonlinearities that would be more difficult to detect decisively with less granular and more aggregate data. There is some evidence at the aggregate level that the slope of the Phillips curve might have changed during the period when we were over-predicting inflation. For the euro area, the evidence from several new papers points to a relative steepening recently, following the previous flattening, (see e.g. Oinonen and Paloviita (2014), Riggi and Venditti (2015) and Foroni and Porqueddu (2015)). This development is especially marked in those countries which experienced deeper and longer recessions and made greater efforts to reform their product and labour markets. Estimating the same specifications of Phillips curves over two samples, one stopping at 2012 Q1 (when we started systematically over-predicting inflation) and the other covering a longer sample, one finds evidence of an increase in the slope estimate. Indeed, regime-switching estimates, accounting for parameter change depending on the state of the business cycle, can help to explain the “excessive” disinflation since 2012 and justify the expectations about future normalisation.
Let me conclude. The findings of many papers in this conference would not have been possible without the use of detailed micro data. Micro data help us identify the channels of transmission better, such as when household expectations determine household choices or when micro price allow us to build better micro-founded models. The Eurosystem has already previously organised a number of large-scale research efforts with the aim of better understanding prices and price-setting at the micro-level: the Eurosystem Inflation Persistence Network (Angeloni et al. (2006); Álvarez et al. (2006); Dhyne et al (2006); Fabiani et al (2007); Vermeulen et al (2012)), and an expert group studying Nielsen scanner data in 2012 (ECB, Economic Bulletin, Issue I, 2015). There is a lot of potential for research on price-setting behaviour, especially given the recent opportunities to enhance access to relevant data. This is why we have made micro-price research a strategic priority in our research work plan for the next three years. A primary task will be a renewed effort in collecting a wealth of new price micro data for the euro area countries. The contribution of academic research to this ambitious agenda will be vital, and this conference is a great example of how much we can benefit from it.
Álvarez, L., E. Dhyne, M. Hoeberichts, C. Kwapil, H. Le Bihan, P. Lünnemann, F. Martins, R. Sabbatini, H. Stahl, P. Vermeulen and J. Vilmunen (2006), “Sticky prices in the euro area:a summary of new micro-evidence”, Journal of the European Economic Association, Vol. 4(2-3), pp. 575-584, 04-05.
Angeloni, I., L. Aucremanne, M. Ehrmann, J. Gali, A. Levin and F. Smets (2006), “New Evidence on Inflation Persistence and Price Stickiness in the Euro Area: Implications for Macro Modelling", Journal of the European Economic Association, Vol. 4(2-3), pp. 562-574, 04-05.
Auer, R., A. Levchenko and P. Saure (2017a), “International inflation spillovers through input linkages”, mimeo.
Auer, R., C. Borio and A. Filardo (2017b), “The globalisation of inflation: the growing importance of global value chains”, BIS Working Papers No 602.
Berger, D, and J.Vavra (2015), “Consumption Dynamics During Recessions," Econometrica, Vol. 83, pp. 101-154.
Bianchi, F. and A, Civelli (2015), "Globalization and inflation: evidence from a time varying VAR", Review of Economic Dynamics, Vol. 18, no 2, pp 406–433.
Carvalho, C. and O. Kryvtsov (2017), “Price selection”, presented this conference.
Ciccarelli, M. and B. Mojon (2010), “Global Inflation”, The Review of Economics and Statistics, 92(3): 524–535.
Ciccarelli, M. and C. Osbat (eds.) (2017), “Low inflation in the euro area: Causes and Consequences”, ECB Occasional Paper Series, No 181.
Christelis, D., D. Georgarakos, T. Jappelli and M. van Rooij (2017), “Trust in the Central Bank and Inflation Expectations”, presented at this conference.
Coibion, O. and Y. Gorodnichenko (2015), "Is the Phillips curve Alive and Well after All? Inflation Expectations and the Missing Disinflation", American Economic Journal: Macroeconomics, Vol. 7(1), pp. 197-232.
Constâncio, V. 2015, “Understanding inflation dynamics and monetary policy”, Annual Economic Policy Symposium, Federal Reserve Bank of Kansas City, available at https://www.kansascityfed.org/~/media/files/publicat/sympos/2015/2015constancio.pdf?la=en
Corsetti, G., L. Dedola and S. Leduc (2008), "International Risk Sharing and the Transmission of Productivity Shocks", Review of Economic Studies, Vol. 75(2), pp. 443-473.
D’Acunto, F., U. Malmendier, J. Ospina and M. Weber (2017), “Salient Price Changes, Inflation Expectations, and Household Behavior”, presented at this conference.
Dhyne, E., L. J. Álvarez, H. Le Bihan, G. Veronese, D. Dias,J. Hoffmann, N. Jonker, P. Lunnemann, F. Rumler and J. Vilmunen (2007), “Price setting in the euro area: some stylised facts from consumer price data”, Journal of Economic Perspectives, Vol. 20(2), pp. 171-192.
ECB Economic Bulletin, Issue I (2015), “Grocery prices in the euro area: findings from the analysis of a disaggregated price dataset”.
ECB Economic Bulletin, Issue 4 (2017), “Domestic and global drivers of inflation in the euro area”.
Fabiani, S., C. Loupias, F. Martins and R. Sabbatini (2007), “Pricing decisions in the euro area. How firms set prices and why”, Oxford University Press.
Ferroni , F. and B. Mojon 2014, “Domestic and Global Inflation”, mimeo.
Forbes, K., I. Hjortsoe and T. Nenova (2017), “Shocks versus Structure: explaining Differences in Exchange Rate Pass-Through across Countries and Time”, presented at this conference.
Forbes, K., I. Hjortsoe and T. Nenova (2015), “The Shocks Matter: Improving our Estimates of Exchange rate pass through”, Bank of England External MOC Unit Discussion Paper, No. 43.
Foroni, C. and M. Porqueddu (2015), “Inflation Dynamics in the Euro Area: the Role of Inflation Expectations and Nonlinearities in the Phillips curve”, mimeo.
Kamber, G. and B. Wong (2017), “An Open Economy Model of Trend Inflation”, presented at this conference.
Laseen, S. and M. T. Sanjani 2016, “Did the global financial crisis break the US Phillips curve?“, IMF Working Paper No. 16/126.
Oinonen, S. and M. Paloviita (2014), “Updating the Euro Area Phillips curve: The Slope Has Increased”, Bank of Finland Research Discussion Paper No. 31/2014.
Riggi, M. and F. Venditti (2015), “Failing to Forecast Low Inflation and Phillips curve Instability: A Euro-Area Perspective”, International Finance, Vol. 18, Issue 1, pp. 47–68.
Slobodyan S. and R. Wouters (2017), “Adaptive Learning and Survey Expectation Inflation”, presented at this conference.
Vellekoop, N. and M. Wiederholt (2017), Inflation Expectations and Choices of Households: Evidence from Matched Survey and Administrative Data, presented at this conference.
Vermeulen, P., D. Dias, M. Dossche, I. Hernando, R. Sabbatini, P. Sevestre and H. Stahl (2012), “Price setting in the euro area: some stylised facts from individual producer price data“, Journal of Money, Credit and Banking, Vol. 44(8), pp. 1631–1650.
 See Borio, C. (2017), “How much do we really know about inflation?”, Presentation on the BIS Annual Report available at http://www.bis.org/speeches/sp170625a_slides.pdf.
 Shocks versus Structure: explaining Differences in Exchange Rate Pass-Through across Countries and Time, presented by Kristin Forbes, Bank of England, MIT-Sloan School of Management and NBER, with Ida Hjortsoe, Bank of England and CEPR, Tsvetelina Nenova, Bank of England.
 This concept was put forward by Corsetti, Dedola and Leduc (2008) and taken up empirically by Forbes et al. (2015).
 Adaptive Learning and Survey Expectation Inflation, presented by Raf Wouters, National Bank of Belgium, with Sergey Slobodyan, CERGE-EI.
 Inflation Expectations and Choices of Households: Evidence from Matched Survey and Administrative Data, presented by Nate Vellekoop, Goethe University Frankfurt, with Mirko Wiederholt, Goethe University Frankfurt and CEPR.
Europese Centrale Bank
- Sonnemannstrasse 20
- 60314 Frankfurt am Main, Duitsland
- +49 69 1344 7455
Reproductie is alleen toegestaan met bronvermelding.Contactpersonen voor de media