VoxEU Column Frontiers of economic research Macroeconomic policy

Hysteresis and fiscal policy during the Global Crisis

Conventional wisdom on supply and demand suggests that demand shocks are cyclical or transitory, and that only technology shocks are responsible for trend changes. This column argues that cyclical events can have permanent effects on demand, and therefore GDP. It is time for policymakers to start considering the possibility of hysteresis seriously.

More than eight years have passed after the Global Crisis started, and GDP is still very far from its pre-crisis trend in all advanced economies. No current forecast suggests that this gap will ever be closed. In fact, in some cases, there is the fear that even pre-crisis growth trends will not return.

The persistence of the Global Crisis

One way to visualise this growth pessimism is to look at revisions to IMF estimates of economic activity. Below we show for the Eurozone the change in both GDP as well as estimates of potential output as produced by the IMF World Economic Outlook (WEO) in three different vintages: April 2007 (before the crisis), April 2011 (after the first wave of the crisis), and April 2016 (today).

Figure 1. Eurozone GDP and potential GDP from three different IMF WEO vintages

Notes: The April 2007 WEO does not contain forecast beyond 2008 for GDP or Potential. In that case, we are extrapolating GDP and potential using the average growth rate since 1999. The April 2011 WEO contains forecasts up to 2016. We are extrapolating potential for the next three years using the average growth rate since 1999. GDP data prior to 2007 is not identical in all three vintages because of data revisions. Potential was also revised backwards for several of these years.
Source WEO IMF.

Relative to the 1999-2007 trend, Eurozone GDP today is about 15% below that level. In addition, potential has been revised downwards by a similar magnitude with no hope of any significant recovery. The IMF expects today that by 2021, the Eurozone will also be about 15% below the potential GDP level implied by its pre-crisis trend.

Ball (2014) documents that this is a consistent feature across all advanced economies, as do we in a recent paper (Fatás and Summers 2016). We have witnessed revisions to GDP projections over the last years, and these revisions have taken place in a series of steps where previous forecasts always underestimated the severity and persistence of the crisis. Either additional shocks, or simply the realisation that the crisis was worse than expected, have led to further downward revisions of potential output estimates.

Changes to GDP trends could be caused by many factors such as demographics and  changes in productivity growth, but what is interesting about this episode is that it coincides with the worst economic crisis since the Great Depression. Is there a connection between the two, or are they independent phenomena?

Is persistence a surprise? The academic literature

It has been well established since the work of Nelson and Plosser (1982) that fluctuations are persistent and they leave possibly permanent effects on the level of GDP. From a statistical point of view, this is best described by the presence of a unit root in the GDP series. From an economic point of view, this result was originally used to support the idea that technology shocks were a fundamental driving force of business cycles. This conclusion was reached using a class of models that treats growth as exogenous and independent of fluctuations, and where only technology shocks are permanent. Using this logic, Blanchard and Quah (1989) separated between technology (permanent) and demand (transitory) shocks using methodology that became the conventional wisdom on supply and demand shocks.

But is it possible that cyclical (demand) shocks can have permanent effects through hysteresis effects? The presence of hysteresis was originally discussed in the context of labour markets. Blanchard and Summers (1986) argued that cyclical unemployment could turn into long-term unemployment or that unemployed workers could lose some of their skills, making a cyclical shock persistent or even permanent. But we can also think about a broader concept of hysteresis, one that includes the effects on productivity and capital accumulation dynamics and establishes a much stronger connection between economic crises and long-term growth trends.

In fact, the moment we think about growth as an endogenous phenomenon, it seems very natural to think about hysteresis as the default outcome of any crisis. A reasonable hypothesis is that the forces that drive long-term growth slow down during recessions. Such a hypothesis is supported by evidence that investment in physical capital, R&D expenditures, and adoption of new technologies all tend to be cyclical (Fatás 2000, Anzoategui et al. 2016). A temporary slowdown of the forces that drive innovation or technology adoption must result in a permanent impact on GDP levels, leading to hysteresis.

If this is obvious, why is it that the academic literature has not incorporated hysteresis more often into its models? As we mentioned early, the mainstream economic model of fluctuations treats long-term growth as exogenous, and only technology shocks are responsible for trend changes. So, by definition, it rules out the possibility of hysteresis. Introducing endogenous growth and fluctuations together is challenging analytically and, in addition, it is not easy to take such a model to the data. The reason is that, unfortunately, the endogenous growth literature does not provide us with a clear set of stylised facts on what are the factors that ultimately drive growth.

But there is a recent empirical literature that shows that hysteresis effects are common and large. For example, Haltmaier (2013) and Martin et al. (2015) have shown that all US business cycles are very persistent and have long-term effects on GDP. Blanchard et al.  (2015) go even further and identify, in a sample of advanced economies, the source of different shocks to show that recessions more likely caused by demand shocks such as ‘intentional disinflations’ are also persistent and permanently lower GDP levels – strong evidence in favour of hysteresis.

Persistence during the Global Crisis and the fiscal consolidation of 2009-2011

What about the persistence of the Global Crisis? Was it caused by cyclical or structural factors? As with any other empirical study that tries to separate the sources of macroeconomic shocks and their relative contribution, we face the challenge of properly identifying exogenous events. The work of Anzoategui et al. (2016) shows that the recent slowdown in productivity can be explained by a model where growth reacts to the contraction in demand. In our paper, we also present supporting evidence of this hypothesis by looking at the persistence of a particular demand shock (Fatás and Summers 2016). A candidate for such a shock is the large fiscal consolidation that took place during the 2010-11 period.

We know from the work of Blanchard and Leigh (2013) that changes in fiscal policy in those years led to significant changes in short-term GDP. But how persistent were those changes? We follow their methodology to study the effects on GDP over long horizons. We look at actual GDP in 2015, GDP forecasts for 2021, as well as estimates of potential GDP for the same horizon.

In all cases we confirm that the effects of the fiscal consolidation identified by Blanchard and Leigh (2013) extend over longer horizons, up to the latest data available (2015) or even to current projections for 2021. What is more interesting is that the same effects are present when we replace GDP with measures of potential output.

Our results show that in advanced economies, every 1% decline in GDP during the years 2010-11 caused by fiscal policy consolidation translated into a 1% decline in potential output (as estimated at the beginning of 2016). The results are stronger for the Eurozone. The large size of these long-term fiscal policy multipliers is likely to be a consequence of the specific circumstances during these years. We are looking at a very large crisis and one where monetary policy was constrained by the zero lower bound.  

The fact that the effects of fiscal consolidation on GDP during these years were permanent and large raises the question of how effective they were at reducing the debt-to-GDP ratio. Following the analysis of DeLong and Summers (2012), we show that given the size of our estimates, it is very likely that that the 2010-11 fiscal consolidation was self-defeating. Contrary to its goals, it led to an increase in the debt-to-GDP ratio via its negative long-term effects on GDP.


It is time to consider the possibility of hysteresis seriously. There is strong evidence that cyclical events have permanent effects on GDP. The results are very natural in any model where growth is endogenous and its driving forces are affected by cyclical shocks.

Incorporating the possibility of hysteresis into policymaking can substantially change the way we think about the role of monetary and fiscal policy. Our results on the potential negative effects of the most recent waves of fiscal consolidations have deep implications for the way we think about the damage caused by the wrong policies during large crises. More generally, attempts to smooth cyclical fluctuations can have much larger benefits than previously thought because of their effects on long-term outcomes. The benefits of expansionary policy, fast recoveries, and long-lasting economic booms on GDP have to be factored in when designing optimal stabilisation policies.

It is positive to see that policymakers are finally considering the possibility of hysteresis seriously and are raising concerns about how weak demand can led to sluggish growth and permanent scars. From Lagarde (2016):

 “The longer demand weakness lasts, the more it threatens to harm long-term growth as firms reduce production capacity and unemployed workers are leaving the labour force and critical skills are eroding. Weak demand also depresses trade, which adds to disappointing productivity growth.”

Understanding that stabilisation policy can have stronger and more long-lasting effects might just be the first step. Further academic research needs to provide policymakers with strong tools to translate this intuition into concrete and actionable policy frameworks.


Anzoategui, D, D Comin, M Gertler, and J Martinez (2016), "Endogenous Technology Adoption and R&D as Sources of Business Cycle Persistence", NBER Working Paper no. 2205.

Ball, L (2014), Long-Term Damage from the Great Recession in OECD Countries, National Bureau of Economic Research, Cambridge, MA.

Blanchard, O, E Cerutti, and L Summers (2015), "Inflation and Activity - Two Explorations and they Monetary Policy Implications", NBER Working Paper.

Blanchard, O J, and D Quah (1989), "The Dynamic Effects of Aggregate Demand and Supply Disturbances", American Economic Review 79(4): 655-673.

Blanchard, O J, and D Leigh (2013), "Growth forecast errors and fiscal multipliers", National Bureau of Economic Research.

Blanchard, O J, and L H Summers (1986), "Hysteresis and the European unemployment problem", NBER Macroeconomics Annual 1986, Volume 1, MIT Press.

DeLong, J B, and L H Summers (2012), "Fiscal Policy in a Depressed Economy [with Comments and Discussion]", Brookings Papers on Economic Activity, Spring: 233-297.

Fatás, A (2000), "Do business cycles cast long shadows? Short-run persistence and economic growth", Journal of Economic Growth 5(2): 147-162.

Fatás, A and L H Summers (2016), "The Permanent Effects of Fiscal Consolidations", NBER Working Paper 22734.

Haltmaier, J (2013), "Do recessions affect potential output?", FRB International Finance Discussion Paper 1066.

Lagarde, C (2016), "We Need Forceful Policies to Avoid the Low-Growth Trap", IMF Direct.

Martin, R, T Munyan, and B A Wilson (2015), "Potential Output and Recessions: Are We Fooling Ourselves?", International Finance Discussion Papers.

Nelson, C R, and C R Plosser (1982), "Trends and random walks in macroeconomic time series: some evidence and implications", Journal of Monetary Economics 10(2): 139-162.

14,173 Reads