Inflation in a changing economic environment: Insights from a conference at the ECB
Michael Ehrmann, Marek Jarociński, Christiane Nickel, Chiara Osbat, Andrej Sokol05 February 2020
In the aftermath of the financial crisis, inflation developments worldwide repeatedly surprised policymakers, practitioners, and academics alike, leading to renewed intellectual scrutiny and heated policy debates. A famously puzzling episode was the ‘missing disinflation’ shortly after the onset of the Great Recession, when inflation in advanced economies fell by less than expected given the severity of the preceding downturn (Coibion and Gorodnichenko 2015, Bobeica and Jarocinski 2019). More recently, the greatest surprise has been the disconnect between measures of slack and underlying inflation in the euro area (Bobeica and Sokol 2019). While most measures of economic slack have moved into neutral territory or even shown mild signs of excess demand in recent quarters, underlying inflation in the euro area has remained well below its historical average, which a standard Phillips curve framework struggles to account for (Figure 1).
Figure 1 Phillips curve decomposition of HICP inflation excluding energy and food in the euro area
Note: annual percentage changes and percentage point contributions; all values in terms of deviations from their averages since 1999. Source: ECB calculations, as detailed in Bobeica and Sokol (2019).
Against this backdrop, the ECB recently hosted a high-level conference titled “Inflation in a changing economic environment”, aimed at enhancing our collective understanding of the drivers and dynamics of inflation. The Phillips curve model underlying Figure 1, while stylised, provides a useful organising framework to understand the contributions presented at the conference: the relationship between economic slack and inflation requires further qualifications and scrutiny; what measure of inflation to study is not a foregone conclusion; there’s more to inflation expectations than a simple reading of survey measures can reveal; other factors, both cyclical and secular, can cloud the relationship between slack and inflation; and a better understanding of individual price-setting behaviour might be required to understand aggregate inflation dynamics.
Identification and measurement
The Phillips curve was born as an empirical regularity observed between unemployment and wage developments in the UK in the 1950s (Phillips 1958). It has since evolved into a cornerstone of modern macroeconomic theory, and in the textbook New Keynesian model it denotes a structural relationship between real activity and prices. Silvana Tenreyro’s keynote address (based on McLeay and Tenreyro 2019) offered a compelling reminder of the importance of maintaining that distinction, and of the difficulties involved in inferring a structural relationship from the observed data – thus, for example, one should avoid attributing a causal role to the slack contribution shown in Figure 1. Recent work on the euro area Phillips curve has taken direct inspiration from their approaches to identification using both monetary policy surprises and regional data (Lane et al. forthcoming).
We often take for granted what we mean by inflation, sometimes forgetting the plethora of measures available to gauge price developments. Even within the more restricted universe of underlying inflation measures, designed to strip out the most volatile and transitory components, choices abound (Nickel and O’Brien 2018). Larry Ball’s keynote address (based on Ball and Mazumder 2019) focused on the merits of one such measure, weighted median inflation, and the ability of Phillips curve models based on that measure to account for both the missing inflation and missing disinflation episodes in the euro area.
Economic theory is adamant that economic decisions are forward-looking, and that expectations are a key element of those decisions. But how such expectations are formed and updated, measured, and ultimately impact decisions remains an exciting area of research, as highlighted in the discussion by Michael Weber (Weber 2019).
Using randomised experiments in household and firm surveys, Coibion et al. (2019) find that higher household inflation expectations are associated with more pessimistic views of households’ real incomes and lead to reduced spending on durables. In an earlier paper, Coibion et al. (2018) find that higher inflation expectations lead firms to increase prices, borrow more, and reduce employment.
Focusing instead on how firms’ expectations respond to news about monetary policy, Enders et al. (2019) show that expectations of German firms do respond to policy surprises, but that the response becomes weaker the larger the surprise – thus, for example, they find only small effects from most of the ECB’s announcements of unconventional policy measures. The expectations of Italian firms are also responsive to monetary policy surprises, with stronger effects since 2009 (Bottone and Rosolia 2019).
Kucinskas and Peters (2019) and Mertens and Nason (2019) study the expectations of professional forecasters. The former shows that forecasters in the US Survey of Professional Forecasters underreact to aggregate shocks but overreact to idiosyncratic shocks, while the latter demonstrates that longer-horizon forecasts inform estimates of trend inflation. Finally, Almosova and Andresen (2019) show that non-linear machine learning techniques can produce competitive forecasts of US inflation, contributing to a growing debate about the usefulness of such techniques with typical macroeconomic time series (e.g. Giannone et al. 2018).
From costs to prices
As pointed out above, multiple factors can confound the link from a change in costs to a change in prices (see the discussions by Katia Peneva (2019) and Klaus Adam (2019) ranging from adjustments along the pricing chain in a production network (Smets et al. 2018), to slow-moving secular factors not fully offset by monetary policy (Juselius and Takáts 2018).
Because they are a sizeable component of firms’ costs, the dynamics of labour costs are a key element to understanding inflation. The ECB Wage Expert Group’s recent report (Nickel et al. 2019) documented how standard wage determinants (such as labour market slack, productivity, and past inflation) can go a long way to explaining much of the weakness in euro area wage growth over the 2013-2017 period, but that other factors (such as compositional effects, the possible non-linear reaction of wage growth to cyclical improvements, and structural and institutional features) also played a role. Both Bobeica et al. (2019) and Conti and Nobili (2019) find that the pass-through of wages (or labour costs) to aggregate (or sectoral) inflation depends on the nature of the shocks hitting the economy, and that demand shocks tend to induce a stronger positive co-movement. Bulligan et al. (2019) demonstrate that the propensity of firms to adjust hours (i.e. the intensive margin) is more important to understanding wage dynamics in the euro area than the adjustment of employment (i.e. the extensive margin).
External prices are another important driver of inflation (discussion by Natalie Chen 2019). Food commodities are a case in point: Peersman (2018) finds that unanticipated shocks to food commodity prices account for 25-30% of euro area HICP inflation volatility, and that large autonomous swings in international food prices contributed significantly to the twin puzzle of missing disinflation and missing inflation in the euro area. On the other hand, both Kim et al. (2019) and De Walque et al. (2019) offer fresh perspectives on the classic issue of exchange rate pass-through: the former document an asymmetry in the pass-through of dollar appreciations and depreciations to product-level import prices in terms of both timing and completeness, while the latter show how introducing trade in intermediate inputs into a new open economy macroeconomics (NOEM) model can help generate both high pass-through to import prices and low pass-through at the consumer level.
What can be learned from individual prices?
With a wealth of information about individual prices becoming more broadly available and amenable to econometric and big data techniques, studying such data both in their own right and with a view to answering more traditional macroeconomic questions has turned into a new and exciting area of research (discussions by Fernando Alvarez 2019 and Volker Hahn 2019).
Dedola et al. (2019) exploit the data underlying construction of the Danish Producer Price Index to study the extent of price synchronisation in multiproduct firms and its underlying mechanisms, also providing novel evidence on the pass-through of both common (energy) and idiosyncratic shocks to firms’ prices. Further, Bonomo et al. (2019) embed multiproduct pricing in a general equilibrium model and find that a trade-off between the desire to change many prices at a time and to adjust misaligned prices can reduce the sensitivity of inflation to monetary surprises. Hong et al. (2019) also propose using information from both aggregate and micro price data when building aggregate models, and find that a Calvo-pricing mechanism is superior to a menu-cost setup in matching the data.
Using detailed micro data on French imports and exports, Carluccio et al. (2018) find that imports from low-wage countries resulted, on average, in about 0.17 percentage point lower consumer price index (CPI) inflation each year over 1994-2014, while Berardi and Sevestre (2019) study price dispersion across French supermarkets, finding extensive price dispersion. Knotek et al. (2019) analyse the data underlying the Israeli CPI and document how a preference for certain price endings (the final digit of a posted price) can be a source of both micro and macro price rigidity. Finally, Adam and Weber (2019) document how individual product prices tend to fall over their life-cycle relative to closely competing products, and use this observation to estimate an optimal inflation target for the UK in the range of 2.6% and 3.2% over the 1996-2016 period.
Conclusion and outlook
Inflation developments have been puzzling economists for nearly a decade. The ECB conference on “Inflation in a changing economic environment” has tried to shed light on this from many angles. It has helped to better understand some aspects, but also opened many avenues for research. Enhancing our understanding of inflation further will therefore remain a key research area at the ECB. Current large-scale projects in this direction are the establishment of a consumer expectations survey as well as the analysis of micro price data in a coordinated research network among Europe’s central banks.