VoxEU Column Labour Markets Monetary Policy

Price and wage-setting in advanced economies: Selected takeaways from the ECB’s Sintra Forum

The origins and implications of the low inflation dynamics that characterised the post-crisis recoveries in many advanced economies were at the heart of the ECB’s 2018 Sintra Forum on Central Banking. In this column, two of the organisers highlight some of the main points from the discussions, including why measured economic slack did not translate into more vivid price and wage growth, which role inflation expectations play in the conduct of monetary policy, as well as where the challenges lie in reconciling changes in micro price-setting with aggregate inflation dynamics.

Editors' note: This column was first published in August 2018.

This year’s ECB Sintra Forum on Central Banking focused on the core issue for monetary policy of why inflation accelerated so moderately in the post-crisis recoveries of advanced economies. Policy makers, academics and market economists debated price and wage-setting both from macroeconomic and from microeconomic perspectives. In this post we summarise four of the main themes that were keenly debated: explanations for and implications of the flattening of the Phillips curve; sources and implications of low real wage growth; central bank communication with respect to inflation expectations; and key challenges in reconciling changes in micro price-setting behaviour with macroeconomic inflation developments. All papers, discussions and speeches can be found in the Conference e-book (ECB 2018) and video recordings of all sessions on the on the ECB’s YouTube channel

Slack and the flat Phillips curve: International factors, regime switches, slow adjustment or mismeasurement?

Against the background of a flattening Phillips curve among advanced economies over the last decades (see Constâncio et al. 2015, summarising the 2015 Sintra Forum), the discussion first revolved around the macroeconomic factors that can explain the subdued inflation dynamics during the post-crisis recovery. Several Sintra speakers agreed that international developments play an important role. For example, Kristin Forbes applied a trend-cycle model to inflation in 43 countries over the last 30 years and found that including variables like the real exchange rate, the world output gap, or commodity prices materially improves estimations (Forbes 2018a,b). Their relevance, however, varies across countries and over time. For example, the world output gap and non-oil commodities moved from insignificant to significant inflation determinants following the break-out of the crisis (see Table 1).

Table 1 Estimation of time-varying domestic and international inflation determinants

Note: The table shows the parameter estimates of an enhanced Phillips curve regression for quarterly inflation (as measured by consumer price indexes (CPIs)) for 43 countries between 1990 and 2017 according to equation (1) in Forbes (2018a). “ER” stands for exchange rate and “PPI” for producer price inflation. “Post” is a dummy variable equal to 1 for the years 2007 to 2017. The estimation uses random effects with robust standard errors clustered by country. ***, **, * indicate statistical significance at 1, 5 and 10 percent levels.
Source: Reproduced from Forbes (2018a), Table 1.

Aviv Nevo mentioned that since firms’ pass-through from marginal costs to prices for most sectors is smaller than one, the more global and complex supply chains that emerged should further reduce the overall pass-through rate (Nevo and Wong 2018). Philip Lowe reported that in relation to subdued price and wage-setting, Australian business leaders frequently refer to the much-increased global competition, in particular from Asia.

Both Lucrezia Reichlin (2018) and Kristin Forbes (2018b) agreed that trends in inflation are important. Based on a different trend-cycle model (Hasenzagl et al. 2018), Reichlin showed evidence that the domestic cyclical part of inflation is relatively small compared with the trend plus effects of global commodity prices. John Muellbauer pointed out that the equilibrium adjustment mechanism for inflation could involve very long lags (Aron and Muellbauer 2018), that the crisis of 2008/2009 in major economies may have acted like a “heart attack”, after which inflation would just take a long time to recover.

Many of these speakers confirmed that the Phillips curve could be re-established, after accounting for such factors. Jim Stock recovered the curve in his paper by excluding components from price indexes that do not respond to the domestic business cycle (Stock and Watson 2018, ECB 2014). The remaining ‘cyclically sensitive inflation’ indicator for the euro area has increased only very gradually in the last few years, much in line with the core inflation measure excluding energy and food. Based on the Lucas critique, Jim Bullard (2018) suggested that the empirically measured Phillips curve could have fallen victim to the central banks’ success in targeting inflation. Frank Smets and Michael Burda, however, referred to the possibility of non-linearities and regime dependence in the Phillips curve (see also Table 1). It could well be that at a more advanced stage of the cycle, the curve would steepen again.

All in all, the fact that the simple variant of the Phillips curve is estimated to be flat does not mean that slack in the goods or labour markets is not at work as an inflation determinant. But there are many international and domestic factors operating and changing in importance over time, so it would be wrong to pinpoint only one or two to explain why euro area inflation recovered slowly. 

Low real wage growth, employment, and inequality

Many observers seem to agree that another key explanatory variable for low inflation has been wages until recently. Uta Schönberg emphasised that the decentralisation of the wage-bargaining process from the industry to the firm, or even to the individual, played an important role in this (Kuegler et al. 2018). She particularly contrasted the cases of Germany and France. At the time of re-unification, union wages were high in Germany, so with the threat of outsourcing to Eastern Europe firms could opt out of union agreements or use the increasingly included ‘opening clauses’ (which allow individual firms to pay salaries below those collectively agreed). Whereas the resulting German wage moderation led to a decoupling with productivity, the two remained aligned in France (see Figure 1, Panel a). Later, unemployment decreased substantially in Germany and increased in France (Figure 1, Panel b), but wage inequality increased in Germany and decreased in France. Schönberg attributed these different experiences to the extension of union agreements to virtually all firms in a sector, a relatively high inflation-adjusted minimum wage, and a more confrontational focus on wages in French industrial relations.

Figure 1 Labour productivity, wages and unemployment in France and Germany

a) Productivity and compensation


 b) Unemployment

Note: Data are in green for France and in blue for Germany. Labour productivity is measured as GDP at fixed prices (using the GDP deflator) divided by total hours worked by all employees (solid lines in panel a)). Compensation (abbreviated “Comp.”) per hour worked is defined as total labour costs (gross wages and salaries, plus employers’ social security contributions) divided by total hours worked by all employees (dashed lines in panel a)). Data in panel a) are normalised to 100 in 1995. Unemployment is defined as the share of unemployed people in the total labour force and measured in fractions (panel b)). Last observation is 2016 (GDP per hour 2015).
Source: Kügler et al. (2018), Figure 2.

Philippe Marcadent (2018) noted the broader trend of de-unionisation. In many advanced economies both union density and collective bargaining cover have declined significantly over the last two decades, including for many EU countries (but not in France or Italy, for example; see Figure 2).

Figure 2 De-unionisation in EU countries

Note: Union density is the share of workers that are union members. Collective bargaining cover is the share of wage agreements that are the result of collective bargaining.
Source: Reproduced from Marcadent (2018), Chart 4.

Michael Burda (2018), supported by Klaus Zimmermann (2018), stressed that the German wage ‘give-backs’, inequality, and resulting successful internal devaluation cannot be understood without the ‘Hartz’ labour market reforms of the early/mid-2000s. Two margins of flexibility resulted from the liberalisation of part-time work in 2003 and the reduction of unemployment benefits in 2005.

Several members of the audience wondered whether greater wage inequality was a necessary price to pay for bringing unemployment down. Uta Schönberg observed that the Nordic countries have both low unemployment and low wage inequality. But if unemployment is high to start with, then it is hard to reduce it without some rise in inequality. Michael Burda explained that countries like Denmark or Sweden are better prepared for avoiding the trade-off because they perform well in training low-skilled workers. Finally, both Michael Burda and Federico Fubini clarified that the re-distributional effects of the German social welfare system prevent the wage inequality observed by Schönberg from translating into an income inequality that is unusually high for OECD countries.

Klaus Zimmermann (2018) mentioned two further factors that weigh on wage growth. The first applies to countries with significant immigration. Second, the work by Bell and Blanchflower (2018) suggests that today employed people want to work more. Philippe Marcadent (2018) illustrated the downward wage flexibility that the structural increase of temporary contracts adds. Temporary workers earn 10% to 20% less than people on permanent contracts. Zimmermann suggested that the recent introduction of a minimum wage in Germany has the potential to lift up the whole wage distribution over time. But Marcadent recalled that in the case of the UK, the increase in the lowest wages was offset by a reduction of the next highest wages. Philip Lane, Philip Lowe and Klaus Zimmermann warned of the risk that, if wage negotiators’ inflation expectations were too low, this could hold back wage growth. 

Inflation expectations, central bank communication, and monetary policy

Central banks place great importance on the anchoring of inflation expectations, i.e. that they do not deviate much from their inflation objective. Yuriy Gorodnichenko noted that central banks tend to focus on the expectations of professional forecasters and of market participants (as embodied in asset prices; Coibion et al. 2018b). The inflation expectations of households or non-financial corporations (NFCs), however, can show large deviations from the inflation objectives pursued by the ECB or the US Fed (not uncommonly 1-2 percentage points higher; Coibion et al. 2018b).

Why is that? Research suggests that households’ inflation expectations are particularly influenced by price changes of frequently-purchased goods (such as groceries or gasoline), creating a ‘veil of inattention’ with respect to aggregate inflation and monetary policy announcements. For example, Michael Weber (2018) presented an analysis suggesting that the high expectations are driven by the members of households who do the main grocery shopping (D’Acunto et al. 2018).

Therefore, Gorodnichenko and co-authors argue that influencing these expectations actively would change real interest rates, consumption, and investment as well as firm pricing, implying a great potential as a new central bank ‘policy tool’. Recent research suggests that one can break through the ‘veil of inattention’. For example, Coibion et al. (2018a) find that US consumers who are informed of past inflation rates, the Federal Reserve’s inflation target, or the inflation forecast of the Fed’s Federal Open Market Committee (treatment groups) revise their inflation expectations towards those levels, but consumers who have not been given this additional information (control group) do not (see Table 2). To be successful in this, however, central bankers would need to use simple messages that are regularly repeated and directly target to the relevant firms or population subgroups (e.g. using social media). Today, however, the necessary data and know-how are not yet ready for such a policy to be implemented.

Table 2 Estimation of the scope for influencing household inflation expectations

Note: The table reports the estimated effects of providing information (indicated in the left column) to households participating in the Nielsen Homescan panel for the US. For treatment “Irrelevant 2% figure”, households are informed that population in the US grew by 2% over the last three years. The dependent variable is equal to (post-treatment one-year-ahead inflation expectations) minus (pre-treatment one-year-ahead inflation expectations). Pre-treatment expectations are computed as the implied mean of the expected inflation distribution over the next year. Post-treatment expectations are represented as point forecasts. Treatment effects tend to be negative, because households typically have too high inflation expectations (before receiving additional information). Robust standard errors are reported in parentheses. ***, **, * indicate statistical significance at 1, 5 and 10 percent levels.
Source: Reproduced from Coibion et al. (2018b), Table 4.

The idea of targeted influencing of households and NFCs via social media was met with scepticism by a number of central bankers, such as Jim Bullard, Benoît Cœuré, Mario Draghi, and Otmar Issing. Concerns ranged from potential risks for credibility if central banks used means of populist politicians to reductions of transparency when influencing different groups differently. Moreover, the observed inattention of NFCs and households could be a sign of central banks’ success in stabilising inflation in advanced economies.

The discussant, Ricardo Reis (2018), questioned whether the experimental event studies in Gorodnichenko et al.’s survey paper could really ascertain a strong inattention of households and NFCs. For example, in the ‘Volker disinflation’ of the early 1980s it took more than three years for US consumer inflation expectations to stabilise around a lower mean (Mankiw et al. 2004). Such slow-moving effects could be better captured by including monetary policy communication events in vector-autoregression models. In terms of central bank communication, Reis supported the influencing of inflation expectations via the revelation of fundamental information about the economy and, when needed, via the revelation of future monetary policy (forward guidance). But he warned of trying to stimulate ‘animal spirits’, behaviours unrelated to the first two.

Do we understand the microeconomic factors influencing inflation well enough?

The last issue addressed in Sintra was whether changes at the micro level could increase our understanding of the ‘missing disinflation’ during the Great Recession or the ‘missing inflation’ during the subsequent recovery. What about substitution biases in price indexes? Aviv Nevo showed with grocery data scanned by individual consumers that during the Great Recession, US households increasingly used sales, coupons and generic products (Nevo and Wong 2018). This reversed in the recovery period – partly, fully or even more, depending on the shopping indicator considered. With the available data, however, it cannot be established that the same fluctuations in ‘lower-level’ substitution biases for groceries would also hold in a full price index or when accounting for potential utility differences.

Erica Groshen (2018) asked what the evidence on ‘upper-level’ substitution bias was. Her comparison of the US consumer price index (CPI), which adjusts consumption baskets only bi-annually, and the chained consumer price index (C-CPI), which adjusts them monthly, suggests that it is on average moderate and, if anything, has mildly decreased since the crisis. In sum, the available evidence does not seem to support the idea that part of the ‘missing disinflation’ or part of the ‘missing inflation’ were statistical artefacts.

Next, Nevo addressed the effects of three different structural changes. The recent paper by Goolsbee and Klenow (2018) suggests that US online inflation between 2014 and 2017 could have been one percentage point lower than regular CPI inflation, but this is not in line with the conventional wisdom of previous literature suggesting that online and regular shop prices do not differ markedly (e.g. Cavallo 2017, Gorodnichenko et al. 2017). Moreover, ECB staff did not find e-commerce to lower euro area inflation (ECB 2015). Erica Groshen remarked that statistical offices catch up by including online prices in their sampling (amounting to 8% in the US in 2016).

Recent trends towards market concentration and power (e.g. De Loecker and Eeckhout 2017, 2018) seem less pronounced in the euro area (Valletti 2018). While many competition economists seem to believe that market concentration reduces pass-through from marginal costs to prices, this does not lead to unambiguous conclusions. For example, rising wages would put less upward pressure on prices and declining marginal costs (e.g. through digitisation) less downward pressure on prices. Third, while the availability of ‘big data’ allows NFCs to step up price discrimination – focusing on consumers’ willingness to pay rather than costs – it is not clear that this will necessarily lead to aggregate price increases.

The discussant, Michael Weber (2018), brought up another structural change – namely, ageing societies. Based on Schoefer et al. (2018), he showed that US producer price inflation is economically and statistically significantly lower in industries with higher ‘senior-to-all ratios’ than in other industries. Figure 3 visualises the relationship for three recent time periods. Wages seem to play an important role in it.

Figure 3 Workforce age structure and producer price inflation in the US

Note: The charts show the relationship between the share of total hours worked by senior employees (aged between 55 and 64 years) in the total hours worked by all employees (“Old-to-All ratio”) and producer price inflation for averaged US industry data over the time periods indicated above the three panels, as estimated by Schoefer et al. (2018). The total number of points is much lower than the total number of industries covered in the underlying estimations, because a “binning” technique is used to ensure clarity of exposition. The plots are conditional, i.e. effects of other variables included in the underlying regression are taken out (see column 6 of Table 1 in Weber 2018).
Source: Reproduced from Weber (2018), Chart 8.

Charles Wyplosz and some other macroeconomists wondered whether these effects of competition and micro price developments were too limited and temporary to be of relevance for monetary policy. Jan Eeckhout responded that the phenomenon of increasing mark-ups has been observed for several decades, with an average annual growth rate of about 1%. John Muellbauer added that accounting for industry concentration improves the modelling and forecasting of US inflation (Aron and Muellbauer 2018). Finally, Tommaso Valletti (2018) argued that lower cost pass-through and greater price discrimination should reduce the transmission of central bank interest rate policy via non-financial corporations.

Overall, much still needs to be understood about the links between micro price and aggregate inflation developments, in line with Aviv Nevo’s call for better data and collaborative research efforts between industrial organisation and macroeconomists. For example, the important question of whether increasing mark-ups or declining marginal costs dominate in the aggregate remains unanswered so far.

A more comprehensive summary of the 2018 Sintra Forum discussions can be found in the first chapter of the Conference e-book (Hartmann and McAdam 2018).

Authors’ note: All views expressed are summarised to the best of our understanding from the various Sintra participants’ Forum contributions and should not be interpreted as the views of the ECB or the Eurosystem.


Aron, J and J Muellbauer (2018), “The reincarnation of the Phillips curve, with new forecasting evidence for US core inflation”, in progress, Oxford University.

Bell, D and D Blanchflower (2018), “Underemployment in the US and Europe”, mimeo, Dartmouth College.

Bullard, J (2018), “The case of the disappearing Phillips curve”, forthcoming in Price and Wage-setting in Advanced Economies, ECB.

Burda, M (2018), Comment on “Productivity growth, wage growth and unions“ by A Kügler, U Schönberg and R Schreiner, forthcoming in Price and Wage-setting in Advanced Economies, ECB.

Cavallo, A (2017), “Are online and offline prices similar? Evidence from large multi-channel retailers”, American Economic Review 107(1): 283-303.

Coibion, O, Y Gorodnichenko and M Weber (2018a), “Monetary policy communications and their effects on household inflation expectations”, mimeo, University of California at Berkeley.

Coibion, O, Y Gorodnichenko, S Kumar and M Pedemonte (2018b), “Inflation expectations – a policy tool?”, forthcoming in Price and Wage-setting in Advanced Economies, ECB.

Constâncio, V, P Hartmann and O Tristani (2015), “Selected takeaways from the ECB’s Sintra Forum on ‘Inflation and Unemployment in Europe’”,, 28 October.

D’Acunto, F, U Malmendier, J Ospina and M Weber (2018), “Salient price changes, inflation expectations and household behaviour”, paper presented at the Federal Reserve Bank of Cleveland conference “Inflation: Drivers and Dynamics”, 17 May.

De Loecker, J and J Eeckhout (2017), “The risk of market power and the macroeconomic implications”, NBER Working Paper 23687.

De Loecker, J and J Eeckhout (2018), “Global market power”, NBER Working Paper 24768.

Dustmann, C, J Ludsteck and U Schönberg (2009), “Revisiting the German wage structure”, Quarterly Journal of Economics 124(2): 843–881.

ECB (2014), “The responsiveness of HICP items to changes in economic slack”, Monthly Bulletin, September, Box 5: 65-67.

ECB (2015), “Effects of e-commerce on inflation”, Economic Bulletin 2, Box 6: 51-54.

ECB (2018), Price and Wage-setting in Advanced Economies, Frankfurt am Main.

Forbes, K (2018a), “Has globalization changed the inflation process?”, paper prepared for 17th BIS Annual Research Conference, draft 10 June.

Forbes, K (2018b), “Time for a new astrolabe?”, forthcoming in Price and Wage-setting in Advanced Economies, ECB.

Goolsbee, A and P Klenow (2018), “Internet rising, prices falling: measuring inflation in a world of e-commerce”, NBER Working Paper 24649,.

Gorodnichenko, Y, S Sheremirov and O Talavera (2017), “Price setting in online markets: does IT click?”, forthcoming in Journal of the European Economic Association.

Groshen, E (2018), “Views on advanced economy price- and wage-setting from a reformed central bank researcher and national statistician”, forthcoming in Price and Wage-setting in Advanced Economies, ECB.

Hasenzagl, T, F Pellegrino, L Reichlin and G Ricco (2018), “A model on the Fed’s view on inflation”, CEPR Discussion Paper 12564.

International Labour Organisation (2016), “Non-standard employment around the world: understanding challenges, shaping prospects”, Geneva.

Kügler, A, U Schönberg, and R Schreiner (2018), “Productivity growth, wage growth and unions”, forthcoming in Price and Wage-setting in Advanced Economies, ECB.

Lane, P (2018), “The macroeconomics of price and wage-setting”, forthcoming in Price and Wage-setting in Advanced Economies, ECB.

Mankiw, G, R Reis and J Wolfers (2004), “Disagreement about inflation expectations”, in M Gertler and K Rogoff (eds), NBER Macroeconomics Annual, MIT Press: 209-248.

Marcadent, P (2018), “Wage dynamics and labour market institutions”, forthcoming in Price and Wage-setting in Advanced Economies, ECB.

Nevo, A and A Wong (2018), “Measuring inflation in the modern economy – a micro price-setting view”, presentation at the ECB Forum on Central Banking on “Price and Wage-setting in Advanced Economies”, Sintra, 20 June.

Reichlin, L (2018), Comment on “Slack and cyclically sensitive inflation” by J Stock and M Watson, forthcoming in Price and Wage-setting in Advanced Economies, ECB.

Reis, R (2018), Comment on “Inflation expectations - a policy tool?” by O Coibion, Y Gorodnichenko, S Kumar and M Pedemonte, forthcoming in Price and Wage-setting in Advanced Economies, ECB.

Schoefer, B, M Weber and X Yin (2018), “Changes in the age composition of workers and industry inflation”, mimeo, University of Chicago.

Stock, J and M Watson (2018), “Slack and cyclically sensitive inflation”, forthcoming in Price and Wage-setting in Advanced Economies, ECB.

Valletti, T (2018), “Concentration in markets: trends and implications for price-setting”, forthcoming in Price and Wage-setting in Advanced Economies, ECB.

Weber, M. (2018), Comment on “Measuring inflation in the modern economy – a micro price-setting view”, forthcoming in Price and Wage-setting in Advanced Economies, ECB.

Wyplosz, C. (2018), “Learning from stubborn inflation”, forthcoming in Price and Wage-setting in Advanced Economies, ECB.


Zimmermann, K. (2018), “Reflections on wage-setting”, forthcoming in Price and Wage-setting in Advanced Economies, ECB.

3,360 Reads