Recovery from the global crisis is going well in some countries, but not all. Most advanced economies are projected to recover only slowly, with output remaining below potential for several years to come. According to traditional Phillips curve arguments (with support from New Keynesian models) this outlook raises concerns about inflation prospects – specifically that inflation might undershoot official targets for an extended period.
But not everyone agrees. Indeed, many official forecasts show inflation converging back to target rates well before output returns to potential, while some observers view risks tilted to the upside, predicting an inflationary sequel to the financial crisis (see examples from Buiter 2008 and Eijffinger 2009).
Importantly, the disagreement here relates not only, or even primarily, to the current cyclical outlook – most observers agree that there is sizeable slack. The divergence concerns the moderating impact such slack will have on inflation, relative to other factors.
New evidence on inflation in prolonged slowdowns
The standard way of analysing this issue is to estimate the slope of the Phillips curve, often in a single-equation approach using instrumental-variable methods. Yet these efforts are fraught with many difficulties, notably the scarcity of time series data for inflation expectations, weak instrument problems, and the complications arising from nonlinearities, nontrivial lag structures, drift in mean inflation, and nuisance factors such as exchange rate or oil price shocks.
In a recent IMF Working Paper, I sidestep such estimation issues by focusing on a simpler event study approach (Meier 2010). I trace inflation dynamics during historical episodes in advanced economies where output remained significantly (at least 1.5%) below potential for more than eight consecutive quarters. This limits attention to a segment of the Phillips curve that seems particularly relevant today, i.e., situations of protracted economic slack.
It is clear that stylised facts documented for the past can only be a rough guide to the present, but as my sample includes a total of 25 episodes in 14 countries from the last 40 years, it provides a broad perspective on inflation outcomes during episodes of persistent large output gaps.
A few key findings stand out:
- There is a clear and pervasive pattern of disinflation during historical episodes of persistent and large output gaps. Moreover, in the two atypical cases where inflation failed to decline, the observed increase was negligible and occurred from exceptionally low initial rates of inflation.
Figure 1. CPI inflation at the outset and end of persistent large output gaps (%)
- The disinflationary dynamics appear to be supported by weak labour markets, with high and/or rising unemployment, and falling nominal wage growth and real unit labour cost. This pattern points to the expected link between spare capacity and diminished cost pressures facing firms. In several cases, falling oil prices further helped the decline in inflation. While nominal exchange rates, in turn, show no uniform trend, appreciating currencies are systematically associated with faster disinflation in the cross section.
- Overall, the relationship between initial and final inflation rates seems roughly proportional, suggesting that countries with high initial rates of inflation experience greater disinflation in absolute terms, but not relative terms. This result remains intact if we control for the different length of individual episodes, by considering annualised changes in inflation rates.
- That said, the dynamics differ somewhat across time periods. In pre-1990 episodes (marked by relatively high initial inflation), disinflation tended to proceed rather steadily throughout the episode. By contrast, the more recent (low-inflation) episodes featured most of the disinflation early on, followed by a timelier bottoming-out. Indeed, inflation generally stopped falling, and instead stabilised or even increased, once it had reached a low positive rate. Although some of this pattern could be a statistical artefact related to temporary inflation shocks, the strong clustering of such observations in post-1990 episodes points to more fundamental reasons.
Figure 2. Quarterly CPI inflation during episodes of persistent large output gaps by time period (%)
Why might disinflation peter out at low positive rates of inflation?
Two explanations, in particular, come to mind.
- First, the literature has emphasised the enhanced credibility of central banks in preserving price stability in recent years.
Such credibility would be apparent not only in low average rates of inflation, but also in a strong anchoring of inflation expectations. If price-setters trust the central bank’s commitment, they have less reason to respond to short-term variation in marginal cost, and a weaker relationship between output gaps and inflation may ensue. This argument accords with the notion of a “flattening Phillips curve”, which some authors have documented (see for instance, Kleibergen and Mavroeidis 2008)1.
- Second, already-low inflation might well inhibit further disinflation because of downward nominal rigidities, which appear to be common in wage-setting (see Akerlof et al. 1996 and Benigno and Ricci 2010).
This explanation would also account for the scarcity of outright deflation in the sample. Given the resistance to nominal cuts, it may take truly exceptional circumstances (perhaps epitomised by Japan’s experience during the last two decades) to create negative wage and price dynamics.
Implications for the current inflation outlook…
Taken together, the historical evidence points to a clear disinflationary effect from persistent large output gaps, at least until inflation has declined to very low positive rates. For countries currently facing protracted economic slack, this would suggest limited upside inflation risk. Yet, such a prediction must obviously be taken with a bit of caution, mainly for two reasons:
- Historical experience, especially from the 1970s, shows that real-time assessments of spare capacity may be subject to large ex-post revisions. Similarly, economists might be overestimating the extent of slack in advanced economies today. Yet countering this concern is the profession’s awareness of the lessons from the 1970s. Indeed, most economists already factor in that the global crisis has not only depressed demand, but also curtailed supply capacity.
- Time-invariant relationships are scarce in macroeconomics, and even patterns reliably documented for the past might not persist under today’s particular circumstances. One often-cited argument relates to the exceptionally aggressive policy response to the recent crisis. Indeed, with policy rates (essentially) at zero, several central banks have resorted to money-financed bond purchases. Although there is no obvious, let alone mechanical, link from these unconventional policies to high inflation, they might conceivably interact with fears about high public debt to undermine trust in the currency. Yet policymakers are well aware of this tail risk and, in many cases, have already laid out concrete consolidation plans to ensure fiscal sustainability. As a result, fiscal policy is likely to support, rather than counteract, disinflation over the coming period. Meanwhile, inflation expectations have shown no signs of being unhinged by quantitative easing.
…and a look at recent data
With these aspects in mind, it is worthwhile to consider actual inflation trends during recent quarters. Using the same definition for the historical sample, we identify 15 ongoing output-gap episodes in advanced economies. This time round the decline in output is unusually large, although labour markets have held up better in relative terms. In fact, widespread labour hoarding appears to have driven up average unit labour costs in many countries, even as nominal wage growth has eased. Another striking feature is the rollercoaster ride of oil (and other commodity) prices, which first fell precipitously but have since recovered some of the lost ground. While these swings had a considerable impact on headline inflation, a general downward trend is nonetheless apparent (see Figure 3).
Figure 3. CPI inflation during the global crisis relative to historical output gap episodes (%)
And once food and energy prices are stripped out from the CPI, the pace of disinflation actually looks very similar to the historical precedent. In other words, median inflation has eased by about 20% of the initial inflation rate p.a. so far – about the same rate as in earlier output-gap episodes.
Historical episodes of persistent large output gaps in advanced economies show a clear pattern of disinflation, supported by weak labour markets and, in many cases, falling oil prices. The most recent experience since the beginning of the financial crisis is consistent with this pattern. Indeed, inflation in many advanced countries has now declined to the very low rates at which disinflation typically petered out during past output-gap episodes, probably reflecting well-anchored inflation expectations and downward nominal rigidities. Thus, while upside inflation risks should be limited in countries facing continued economic slack, a slide into outright deflation should not be taken for granted either.
The views expressed herein are those of the author and should not be attributed to the IMF, its Executive Board, or its management.
Akerlof, G, W Dickens, and G Perry (1996), “The Macroeconomics of Low Inflation”, Brookings Papers on Economic Activity 1:1-59.
Benigno, P and L Ricci (2010), “The Inflation-Output Trade-Off with Downward Wage Rigidities”, NBER Working Paper 15672.
Buiter, Willem (2008), “Central banks, the financial crisis and the threat of inflation”, Vox Talks with Romesh Vaitilingam, 18 July.
Eijffinger, Sylvester (2009), “Deflation or stagflation in the Eurozone?”, VoxEU.org, 15 January.
Kleibergen, F and S Mavroeidis (2009), “Weak Instrument Robust Tests in GMM and the New Keynesian Phillips Curve”, Journal of Business and Economic Statistics, 27:293-311.
Meier, A (2010), “Still Minding the Gap—Inflation Dynamics during Episodes of Persistent Large Output Gaps”, IMF Working Paper 2010/189.
Nason, JM and GW Smith (2008), “The New Keynesian Phillips Curve: Lessons from Single-Equation Econometric Estimation”, Federal Reserve Bank of Richmond Economic Quarterly, 94(4):361-395.
1 By contrast, Nason and Smith (2008) find no evidence for a parameter change. This study also shows similar overall disinflation in relative terms during pre-1990 and post-1990 output gap episodes.