Is the Phillips curve alive and well after all? Inflation expectations and the missing disinflation

Olivier Coibion, Yuriy Gorodnichenko 15 November 2013

a

A

“Prior to the recent deep worldwide recession, macroeconomists of all schools took a negative relation between slack and declining inflation as an axiom. Few seem to have awakened to the recent experience as a contradiction to the axiom.” (Bob Hall, 2013.)

“The surprise [about inflation] is that it’s fallen so little, given the depth and duration of the recent downturn. Based on the experience of past severe recessions, I would have expected inflation to fall by twice as much as it has.” (John Williams, 2010.)

According to the Phillips Curve, which links inflation and unemployment, advanced economies should have experienced severe disinflation – perhaps even deflation – in recent years. Such an outcome would not have been unprecedented.

  • During the Great Depression, the US experienced devastating levels of deflation – more than 10% per year in 1932.
  • Japan was mired in borderline deflation territory from the mid-1990s to the mid-2000s after its real-estate and stock-market bubbles popped in the early 1990s.

However, advanced economies have experienced little decline in inflation since the financial crisis of 2008–2009. Is the Phillips curve (Phillips 1958) broken?

Panel A of Figure 1 documents this ‘missing disinflation’ using an expectations-augmented Phillips curve, by plotting quarterly unemployment rates against the deviation of inflation from expected inflation, where the latter are modelled as ‘backward-looking’ – an average over the last four quarters’ inflation rates. The period from 2009 to 2011 stands out relative to the 1960–2007 pattern, with deviations of inflation from expected inflation being systematically higher over the recent period than one would have expected given the high levels of unemployment.

Figure 1. The missing disinflation in the US

Notes: Figures plot the deviation of inflation from expected inflation (y-axis) versus unemployment (x-axis). In Panel A, expectations are the average over inflation rates in the previous four quarters. In Panel B, expectations are from the Survey of Professional Forecasters for CPI inflation.

Explanations

Economists have proposed a number of explanations, but we argue that none are sufficient to explain the full extent of the inflation experience (Coibion and Gorodnichenko 2013). For example, the ‘anchored expectations’ hypothesis of Bernanke (2010) – that is, the credibility of modern central banks has convinced people that neither high inflation nor deflation are likely outcomes, thereby stabilising actual inflation outcomes through expectational effects – can only go some way in accounting for the absence of more significant disinflation between 2009 and 2011. Panel B of Figure 1 above, for example, illustrates that the missing disinflation is still present even when we condition on the ‘anchored’ expectations of professional forecasters.

Explanations based on the long-term unemployed having smaller effects on wages (Llaudes 2005) or downward wage rigidity preventing wages from falling as much as in prior downturns (Daly et al. 2012) imply that the missing disinflation in prices should have been accompanied by a missing disinflation in wages – a feature which we show is noticeably absent in the data.

Others have pointed to a flattening Phillips curve (IMF 2013), but no structural changes in the economy can account for large changes in the slope of the Phillips curve, and the quantitative effects of the estimated changes in the slope are themselves insufficient to account for much of the missing disinflation. This inability to explain the missing disinflation within the context of the Phillips curve has led some to conclude that this framework may have outlived its usefulness.

New research

We propose a novel explanation for the missing disinflation that remains fully within the context of traditional Phillips curve analysis (Coibion and Gorodnichenko 2013). We show that an expectations-augmented Phillips curve – using household inflation expectations as measured by the Michigan Survey of Consumers – can account for the absence of strong disinflationary pressures since 2009.

The primary reason for the success of a household inflation expectation-augmented Phillips curve is simple:

  • Household inflation expectations experienced a sharp increase starting in 2009, rising from a low of 2.5% to around 4% in 2013;
  • Other measures of inflation expectations, such as those from financial markets or professional forecasters, have hovered in the neighbourhood of 2% over the same period.

We illustrate these facts in Figure 2 below. Panel A plots the evolution of household inflation expectations, which experienced unusual increases relative to professional forecasts of inflation and inflation forecasts from asset markets during each run-up in oil prices – from 2003 to 2008, and again from 2009 to 2012. Panel B plots the expectations-augmented Phillips curve using household inflation expectations – the slope of the Phillips curve is stable over time, and the rise in household expectations since 2009 can fully account for the missing disinflation.

Figure 2. Household inflation expectations and the missing disinflation

Notes: Panel A plots inflation forecasts from the Michigan Survey of Consumers and inflation forecasts from asset prices and the Survey of Professional Forecasters. Panel B plots the expectations-augmented Phillips curve with household inflation expectations – deviation of CPI inflation from expected inflation (y-axis) vs. deviation of unemployment from CBO estimate of natural rate (x-axis).

Why focus on the expectations of households in the context of the Phillips curve, since the latter is meant to capture the pricing decisions – and therefore expectations – of firms?

  • First, there is no quantitative measure of firm inflation expectations available in the US.

This means the question of how firms form their inflation expectations – and what may be the best proxy for them – is ex-ante ambiguous.

  • Second, we present new empirical evidence from estimated Phillips curves in the pre-Great Recession period that household forecasts are likely to be a better proxy for firm forecasts than either professional or backward-looking forecasts.

Regressions which include both household and professional forecasts systematically point to a larger role for household forecasts than other measures of inflation expectations.

  • Third, we present preliminary results from an ongoing survey of firms’ inflation expectations in New Zealand, and show that their properties resemble those of households more than professional forecasts – with relatively high levels of forecasted inflation and very high dispersion of forecasts across firms.

Thus, the available evidence is consistent with the use of household inflation forecasts as a proxy for firm forecasts of inflation in the Phillips curve.

The salience of oil prices

We then consider the source of the rise in household inflation expectations relative to the forecasts of professional forecasters since 2009, which is the main feature of the data which accounts for the missing disinflation. More than half of the historical differences in inflation forecasts between households and professionals can be accounted for by the level of oil prices, and the rise in oil prices since 2009 can fully account for the increase in household inflation expectations since then.

Why would households adjust their inflation forecasts more in response to oil price changes than professional forecasters? With gasoline prices being among the most visible prices to consumers, a natural explanation is that households pay particular attention to them when formulating their expectations of other prices. Consistent with this notion, we document using micro-data from the Michigan Survey of Consumers that individuals who spend more money on gasoline (in dollar terms) – and therefore frequent gas stations more often – adjust their inflation forecasts more in response to oil-price changes than do individuals who spend less money on gasoline.

Our suggested explanation for the missing disinflation has several appealing properties.

  • First, it fits naturally within the Phillips curve framework.
  • Second, it is quantitatively successful in explaining the missing disinflation.
  • Third, we present new econometric and survey evidence consistent with firms’ inflation expectations being similar to those of households. 
  • Fourth, the difference in household inflation expectations and those of professional forecasters since 2009 can readily be accounted for by the evolution of oil prices during this period.
  • Fifth, our explanation is consistent with the absence of strong deflationary pressures across a wide range of advanced economies since the recent financial crisis.

Implications of our explanation

One unusual implication of our explanation is that the absence of more pronounced disinflation – or even deflation – in advanced economies following the Great Recession likely reflected a unique set of factors (e.g. rapid recoveries in developing economies like China spurring global demand for commodities) which policymakers should not necessarily expect to be repeated in future crises.

To the extent that this rise in inflationary expectations may have prevented the onset of pernicious deflationary dynamics, the rise in oil prices could be interpreted as a lucky break – generating the very rise in inflationary expectations which policymakers have only recently begun to push aggressively toward in the form of forward guidance.

A second unusual feature of this interpretation is that – contrary to Bernanke’s ‘anchored expectations’ hypothesis – we show that household expectations have not been fully anchored, and continue to respond strongly to commodity price changes.

If our explanation is correct, anchored expectations on the part of households and firms would likely have delivered much worse economic outcomes through more pronounced disinflationary dynamics. So while anchored expectations likely remain a desirable outcome in most circumstances, the experience since 2009 presents a cautionary example of the potential downside of fully anchored expectations.

References

Bernanke, Ben (2010), “The Economic Outlook and Monetary Policy”, speech delivered at the Federal Reserve Bank of Kansas City Economic Symposium, Jackson Hole, Wyoming, 27 August.

Coibion, Olivier, and Yuriy Gorodnichenko (2013), “Is the Phillips Curve Alive and Well After All? Inflation Expectations and the Missing Disinflation”, manuscript.

Daly, Mary, Bart Hobijn, and Brian Lucking (2012), “Why Has Wage Growth Stayed Strong?”, Federal Reserve Bank of San Fransisco Economic Letter 2012-10, 2 April.

Dotsey, Michael, Robert G King, Alexander L Wolman (1999), “State-Dependent Pricing and the General Equilibrium Dynamics of Money and Output”, Quarterly Journal of Economics 114(2): 655–690.

International Monetary Fund (2013), “The Dog That Didn’t Bark: Has Inflation Been Muzzled or Was It Just Sleeping?”, Chapter 3 in World Economic Outlook, April.

Llaudes, Ricardo (2005), “The Phillips Curve and Long Term Unemployment”, ECB Working Paper 441.

Phillips, A W (1958), “The Relationship between Unemployment and the Rate of Change of Money Wages in the United Kingdom 1861–1957”, Economica 25(100): 283–299.

a

A

Topics:  Global crisis Monetary policy

Tags:  inflation, Phillips curve, expectations, oil, global crisis, disinflation, Great Recession

Assistant Professor, UT Austin

Associate Professor in the Dept. of Economics, University of California, Berkeley