Post Covid-19 Potential Output in the Eurozone

Question 1: How much lower will the potential level of GDP in the Eurozone in 2025 be due to Covid-19 relative to pre-Covid forecasts?

 

Question 2: How much lower will the potential growth rate of GDP in the Eurozone in 2025 be due to Covid-19 relative to pre-Covid forecasts?

 

Summary

A majority of the CfM-CEPR panel of macroeconomic experts on the European economy predict a 2-5% decline in the level of potential Eurozone GDP, but no impact on potential the long-run growth rate.

Background

The December 2020 CfM-CEPR survey asked members of its European panel to estimate the loss in the level and growth rate of potential GDP in the Eurozone due to the Covid-19 pandemic.  

Covid’s Long-run Impact on the Eurozone Economy

Global economic activity took a large hit during the Covid-19 pandemic. The Eurozone is no exception with the European Central Bank (ECB) currently projecting an 8% decline in GDP this year and that GDP will only recover to its pre-pandemic level at the end of 2021. But looking five years ahead, the ECB has made no change to its long run growth forecast (of 1.4% in five years’ time) throughout the pandemic. This particular forecast suggests a permanent loss in the level of GDP, but no effect on its growth in the long run. Yet, the long run effect of Covid-19 on the Eurozone’s economic potential remains very uncertain.

Bodnár et al. (2020) give an overview on the theory and evidence on the effects of Covid-19 on the Eurozone’s potential output. Potential output is defined as the maximal level of economic activity that an economy can sustain at current technology, labor supply, and capital stock, without leading to inflationary pressures. There are several reasons to think that Covid-19 may have persistent effects.

First, supply chain disruptions can cause a decline in the economy’s productive capacity, but how persistent these are is still uncertain. Vinci&Licandro (2020) point out the role of monetary policy interventions in preventing the destruction of productive capacities following large negative shocks. Second, it may take time for new entrants to replace firms that failed due to Covid-19. Third, unemployment tends to be persistent as workers’ skills deteriorate and their attachment to the labor force may weaken. Fatás and Summers (2017) give evidence of hysteresis effects of this sort.

Fourth, corporate debt overhang may create “zombie firms”, who have lesser incentives to invest in productive capital. However, Jordá et al (2020) find no historical support for post-crisis growth depending on corporate debt levels. Fifth, there is some theoretical support to the notion that low demand may have scarring effects on the economy due to underinvestment in capital or innovation, whether the recession originated from the demand or supply side (Benigno and Fornaro, 2018; Fornaro and Wolf, 2020; Benedetti-Fasil et al, 2020). Finally, pessimistic forecasts of long run growth can be self-fulfilling because policymakers are more likely to enact fiscal consolidation if they foresee a permanent loss in output. This, in turn, could secure the drop in GDP due to lower demand (Heimberger, 2020).

There are also reasons why Covid-19 may increase the economy’s long-run growth potential. It may have accelerated the implementation of new technologies. Further, it has and could lead to improvements in health investments.

It is difficult to come by forecasts of the long-run economic damage caused by Covid-19.  Pujol (2020) evaluates private sector forecasts and summarizes that they point to expectations of a permanent loss of 3-4% in the level of GDP, with large variation across countries. The European Commission estimates that German and French potential GDP have both declined by half a percent in the years 2020-21. The commission doesn’t attempt to forecast long-run potential economic growth and instead assumes no effect on long run growth. The World Bank’s Global Economic Prospects (June 2020) forecasts that the Eurozone will not regain pre-Covid levels of GDP until the second half of 2023. The UK Office of Budget Responsibility forecasts a permanent 2 percent loss in UK productivity and a 1 percent decline in the labor supply, amounting to a 3 percent drop in the level of potential GDP. But this too is forecasted to be a drop in the level of potential output rather than its trend and the forecast has large range of scenarios ranging from 0% to 6%.  NIESR forecasts a broad range of outcomes across countries, with fiscal support during the crisis the main factor determining how well a given economy recovers.

Question 1: How much lower will the potential level of GDP in the Eurozone in 2025 be due to Covid-19 relative to pre-Covid forecasts?

Forty-three panelists responded to this question. A majority (51%) of the panel predicts that the level of potential GDP will be 2-5% lower that in 2025 than it would have otherwise been absent Covid-19. In other words, the predict a permanent loss in income due to Covid. However, nearly 40% of the panel predicted a small or negligible cost in terms of the level of GDP, with 21% predicting that potential GDP will decline by 2% or less, and 19% that it will not decline at all.

The more pessimistic responses note that GDP losses in deep recessions tend to be permanent. Roger Farmer (University of Warwick) points out that “After a major shock, and COVID certainly qualifies as major, there is little or no tendency to return to a give path.” Several participants pointed to factors that were unique to Covid-19 and may have a longer-term effect. David Miles (Imperial College) referred to the “Education disruption and failure of firms allied with a sharp rise in unemployment” as factors affecting the economy in the long run. Francesco Lippi highlights other scarring factors including “higher public and private debts [and] more unemployment.” Other respondents voiced concern that political and social factors may further hamper the recovery. Etienne Wasmer (Sciences Po) argues that “This is a break in the trend, and Europe has traditionally be[en] slow to adjust to a new organization, it will indeed take 5 to 10 years to adjust.” Ramon Marimon points directly to EU politics: “The main problem is not the potential level in the Eurozone in 2025, but how the Eurozone divide, which already increased after the financial-euro crisis, will be in 2025?”

The optimists noted that Covid-19 may also be viewed as a creative destruction shock. Francesco Lippi, while noting the risks of scarring effects above, believes potential GDP could be higher because “The shock pushed many firms to adopt new technologies that may be cost effective in the medium run and increase potential output and welfare.” Robert Kollmann (Université Libre de Bruxelles), while predicting a minor loss to long run GDP, adds that “households and firms have been forced by the Covid crisis to upgrade their IT skills and equipment, and new ways of organizing office work, production and distribution have been invented. In addition, the crisis could accelerate the transition to greener technologies.” Others credit good macroeconomic policy to their rosier forecasts. John Hassler (IIES, Stockholm University) predicts that “Most likely, the fundamental cause of the crisis will vanish this year. Due to very forceful policies, the initial shock did not lead to self-reinforcing feedbacks -- a quite possible scenario with a financial crisis leading to perhaps even a depression was avoided.” However, Thorsten Beck (Cass Business School) warns that the outcome depends on future policies as well: “The impact of COVID-19 on potential GDP will depend a lot on policy responses. So far, these policy responses have been appropriate, trying to minimise negative effects. But political doubts remain whether they can be continued long enough to avoid damage.”

Finally, several panel members argued that the Eurozone should not be viewed as a single economy and that there will be large differences across countries in the pace of recovery. Jagjit Chadha (National Institute of Economic and Social Research), for example, points to “considerable heterogeneity across EA countries because of their respective reliance on socially intensive activities, the scope for available fiscal responses and the effectiveness of the health and social care infrastructure.”

Question 2: How much lower will the potential growth rate of GDP in the Eurozone in 2025 be due to Covid-19 relative to pre-Covid forecasts?

Forty-seven members of the panel responded to this question. The vast majority (81%) predicted that the Eurozone’s potential growth rate will be unaffected by Covid-19 or forecast small losses of half a percentage point or less.

Some panelists based their responses on past recessions. Jumana Saleheen (CRU Group) reminds us that “In previous crisis we have seen an impact on the level of potential GDP but not on its growth.”  Other panelists argued that none of the long-run drivers of long-run economic has been adversely affected. As Jordi Galí CREI, (Universitat Pompeu Fabra and Barcelona GSE): “I do not see any reason to believe the forces behind innovation and, hence, potential growth, will be much different after the pandemic, one way or another.”. John Hassler adds: “It is hard to see that the fundamental causes of growth, namely the creation of new ideas and the ability to adopt and exploit these for commercial and social use, are affected in the long-run by the coronacrisis.”

A couple of participants were even more optimistic. Fabrizio Coricelli (Paris School of Economics) argues that Covid-19 will end up being good for economic growth: “Covid-19 may induce much needed infrastructural public investments, speed up productivity enhancing technological innovation, also driven by environmental concerns.” Maria Demertzis (Bruegel) adds: “There are important reasons to fill that new technology adaption has been accelerated as a result of Covid19.”

However, several participants warned that these optimistic forecasts depend on the policy response. Etienne Wasmer conditions that “if we are in the middle of public finance crisis, it will be a -2 to -3% [hit to GDP growth] but hard to tell whether this is in 2025, 2024 or 2023.”

References

Benedetti-Fasil, Cristiana, Giammario Impulitti, Omar Licandro and Petr Sedlacek, “Heterogeneous Firms, R&D Policies and the Long Shadow of Business Cycles,” unpublished manuscript

Benigno, Gianluca and Luca Fornaro, “Stagnation traps", The Review of Economic Studies, 2018

Bodnár, Katalin, Julien Le Roux, Paloma Lopez-Garcia and Bela Szörfi, “The impact of COVID-19 on potential output in the euro area”, ECB Economic Bulletin, 2020

Fatás, Antonio and Lawrence H. Summers, “The Permanent Effects of Fiscal Consolidations,” Journal of International Economics, 2017.

Fornaro, Luca and Martin Wolf, “Covid-19 Coronavirus and Macroeconomic Policy: Some Analytical Notes”, manuscript, 2020

Heimberger, Philipp, “Potential Output, EU Fiscal Surveillance and the COVID-19 Shock”, 2020

Jordà Òscar, Martin Kornejew, Moritz Schularick and Alan M. Taylor, “Zombies at Large: Corporate Debt Overhang and the Macroeconomy,” CEPR discussion paper DP15518, 2020.

Pujol, Thierry, “The Long-Term Economic Cost of Covid-19 in the Consensus Forecasts,” Covid Economics 44, CEPR, 2020. 

Vinci, Francesca and Omar Licandro “Switching-track after the Great Recession,” CFCM Working Paper 02/2020.

Contact us for more information

How the experts responded

Question 1

Participant Answer Confidence level Comment
Paul De Grauwe's picture Paul De Grauwe London School of Economics No different Confident
Jumana Saleheen's picture Jumana Saleheen CRU Group Between 2% and 5% Very confident
Roger Farmer's picture Roger Farmer University of Warwick Between 2% and 5% Confident
GDP is well described by non-stationary process. After a major shock, and COVID certainly qualifies as major, there is little or no tendency to return to a give path.
Dawn Holland's picture Dawn Holland NIESR 2% or less Confident
While there will be some permanent loss, this will be partly offset by productivity gains from the experience of remote working in many sectors.
Jagjit Chadha's picture Jagjit Chadha National Institute of Economic and Social Research Between 2% and 5% Not confident
There will be considerable heterogeneity across EA countries because of their respective reliance on socially intensive activities, the scope for available fiscal responses and the effectiveness of the health and social care infrastructure. So this range is a broad guide to what we might expect overall without necessarily being specific about any one country.
Philippe Bacchetta's picture Philippe Bacchetta Université de Lausanne Between 2% and 5% Not confident
GDP will probably not return to its 2019 level before early 2022, which means about two years of "lost growth". Full catching up in the 2022-2025 is not very likely.
Harris Dellas's picture Harris Dellas University of Bern No different Very confident
Covid-19 is the latest addition to a very long list of things (that started with the endogenous growth literature) that are supposed to affect the natural level of output/average rate of growth.In reality these two have proved very insensitive to many many plausible shifters. I am skeptical that any of the channels suggested in the Covid 19 literature has any long term quantitative bite.
Mario Forni's picture Mario Forni Università di Modena Between 2% and 5% Not confident
Cédric Tille's picture Cédric Tille The Graduate Institute, Geneva 2% or less Not confident
John Hassler's picture John Hassler Institute for International Economic Studies (IIES), Stockholm University No different Confident
The coronacrisis led to the largest and most globally synchronized fall in GDP ever recorded. In contrast to previous large economic downturns, we know the reason for this one. Social contacts need to be reduced to contain the pandemic. Most likely, the fundamental cause of the crisis will vanish this year. Due to very forceful policies, the initial shock did not lead to self-reinforcing feedbacks -- a quite possible scenario with a financial crisis leading to perhaps even a depression was avoided. Bankruptcies and unemployment have not been avoided completely. However, from the perspective of long-run consequences, it has had had a more favorable composition than many other crises. Young individuals and small scale service sectors are more hit than others, but also more flexible. The long-run consequences of the crisis are the likely to be fairly small.
Wouter Den Haan's picture Wouter Den Haan London School of Economics Between 2% and 5% Not confident
Sir Charles Bean's picture Sir Charles Bean London School of Economics Between 2% and 5% Not confident
Alexander Ludwig's picture Alexander Ludwig Goethe University More than 5% Confident
Federica Romei University of Oxford No different Confident
It is not the shock that permanently decreases GDP, but the policies can. If Europe will play the right policies, I doubt that there will be a big drop in potential output.
Ramon Marimon's picture Ramon Marimon European University institute and UPF-BarcelonaGSE Between 2% and 5% Confident
The main problem is not the potential level in the Eurozone in 2025, but how the Eurozone divide, which already increased after the financial-euro crisis, will be in 2025?
Moritz Schularick No different Very confident
Ethan Ilzetzki's picture Ethan Ilzetzki London School of Economics Between 2% and 5% Not confident at all
Jean Imbs's picture Jean Imbs Paris School of Economics Between 2% and 5% Not confident at all
Michael Wickens's picture Michael Wickens Cardiff Business School & University of York 2% or less Confident
Fabrizio Coricelli's picture Fabrizio Coricelli University of Siena and Paris School of Economics No different Not confident
David Cobham's picture David Cobham Heriot Watt University Between 2% and 5% Confident
Thorsten Beck's picture Thorsten Beck Cass Business School 2% or less Not confident
The impact of COVID-19 on potential GDP will depend a lot on policy responses. So far, these policy responses have been appropriate, trying to minimise negative effects. But political doubts remain whether they can be continued long enough to avoid damage.
Mirko Wiederholt's picture Mirko Wiederholt Science Po 2% or less Not confident
Ricardo Reis's picture Ricardo Reis London School of Economics 2% or less Not confident
Forecasting levels of output four years out is hard. So many bad and good policies, lucky and unlucky shocks could happen until then. See https://personal.lse.ac.uk/reisr/papers/18-wrong.pdf
Nicholas Oulton's picture Nicholas Oulton London School of Economics Between 2% and 5% Not confident
David Miles's picture David Miles Imperial College Between 2% and 5% Not confident
Education disruption and failure of firms allied with a sharp rise in unemployment are unlikely to leave no medium term (and indeed long term) costs to capacity.
Antonio Fatás's picture Antonio Fatás INSEAD, Singapore Between 2% and 5% Confident
Still uncertain given the status of health crisis in many parts of the world. This crisis deeper but possible shorter than GFC. That's why my answer is more optimistic than what we saw after GFC.
Omar Licandro's picture Omar Licandro University of Nottingham Between 2% and 5% Very confident
Francesco Lippi's picture Francesco Lippi LUISS Potential GDP will be higher in 2025 forecast pre-Covid Not confident
The long run effects could be positive. The shock pushed many firms to adopt new technologies that may be cost effective in the medium run and increase potential output and welfare (less time commuting, more efficient meetings, etc). Such beneficial effects have to be weighted vs the medium run consequences of the scars that will be left by the shock (higher public and private debts, more unemployment). My main point is that there are forces pushing both ways, even though in the long run I do not see why the negative ones should we. It seems feasible to go back to doing whatever was done before the shock. Of course there is a lot of uncertainty on how things will work out as it depends on key policy choices.
Lucio Sarno's picture Lucio Sarno Cambridge University More than 5% Extremely confident
Isabelle Méjean's picture Isabelle Méjean Ecole Polytechnique Between 2% and 5% Not confident
Uncertainty on how the pandemic itself will slow down (ie whether there will be a third wave) makes it difficult to answer on the level effect.
Robert Kollmann's picture Robert Kollmann Université Libre de Bruxelles 2% or less Not confident
Obviously, there is a lot of uncertainty. But households and firms have been forced by the Covid crisis to upgrade their IT skills and equipment, and new ways of organizing office work, production and distribution have been invented. In addition, the crisis could accelerate the transition to greener technologies. This could partly offset the negative effects of higher corporate and government debt. The medium- to long-term damage could therefore be fairly moderate.
Gernot Müller's picture Gernot Müller Eberhard-Karls-Universität Tübingen No different Not confident
Marcel Fratzscher's picture Marcel Fratzscher DIW Berlin and Humboldt University Berlin Between 2% and 5% Not confident
Chryssi Giannitsarou's picture Chryssi Giannitsarou University of Cambridge, Faculty of Economics Between 2% and 5% Not confident
Natalie Chen's picture Natalie Chen University of Warwick 2% or less Not confident
Gino A. Gancia's picture Gino A. Gancia Queen Mary University of London Between 2% and 5% Not confident
Etienne Wasmer's picture Etienne Wasmer Sciences Po, Paris Between 2% and 5% Confident
This is a break in the trend, and Europe has traditionally be slow to adjust to a new organization, it will indeed take 5 to 10 years to adjust, not to talk about a risk of public finance crisis over the 5 year window of the forecast exercise.
Costas Milas's picture Costas Milas University of Liverpool 2% or less Not confident
Ugo Panizza's picture Ugo Panizza The Graduate Institute, Geneva (HEID) Between 2% and 5% Not confident
My guess is that medium-long term growth will go back to pre-COVID. It will be substantially higher in 2021-2022, but not high enough to compensate for output lost in 2020. There is massive uncertainty and probably substantial cross-country variation
Philip Jung's picture Philip Jung University of Dortmund No different Not confident
Linda Yueh's picture Linda Yueh London Business School No opinion Confident
Jorge Braga de Macedo's picture Jorge Braga de ... Nova School of Business and Economics, Lisbon Between 2% and 5% Confident
what is due to covid-19 is difficult to ascertain given the current uncertainty about public health responses across the eurozone let alone the outsider world

Question 2

Participant Answer Confidence level Comment
Paul De Grauwe's picture Paul De Grauwe London School of Economics No different Confident
Jumana Saleheen's picture Jumana Saleheen CRU Group ½ percentage point or less Very confident
In previous crisis we have seen an impact on the level of potential GDP but not on its growth. There is no reason to expect anything different this time.
Roger Farmer's picture Roger Farmer University of Warwick No different Not confident
We have little understanding of what determines growth and all projections of future GDP growth should be give very wide confidence bounds. There are too many unknowns to make a confident projection.
Dawn Holland's picture Dawn Holland NIESR No different Confident
I think it unlikely that the Covid experience will permanently impact potential growth.
Jagjit Chadha's picture Jagjit Chadha National Institute of Economic and Social Research No different Confident
Growth will continue to be pinned down by trends and prospects for productivity. It is not clear that Covid-19, which will be ultimately be a short run shock, will significantly affect the evolution of manufacturing, market and non-market service sector productivity. We might though want to improve the measurement in these sectors.
Philippe Bacchetta's picture Philippe Bacchetta Université de Lausanne No different Not confident
Harris Dellas's picture Harris Dellas University of Bern No different Very confident
Mario Forni's picture Mario Forni Università di Modena No different Confident
Cédric Tille's picture Cédric Tille The Graduate Institute, Geneva No different Confident
John Hassler's picture John Hassler Institute for International Economic Studies (IIES), Stockholm University No different Confident
It is hard to see that the fundamental causes of growth, namely the creation of new ideas and the ability to adopt and exploit these for commersial and social use, are affected in the long-run by the coronacrisis.
Wouter Den Haan's picture Wouter Den Haan London School of Economics No different Not confident
Sir Charles Bean's picture Sir Charles Bean London School of Economics No different Not confident
Alexander Ludwig's picture Alexander Ludwig Goethe University ½ percentage point or less Confident
Federica Romei University of Oxford No different Confident
Once again it depends on the policies played by the European policymakers. There is a lot of uncertainty about them, but if they will play the right policies growth rate can be back to the "normal" level.
Ramon Marimon's picture Ramon Marimon European University institute and UPF-BarcelonaGSE No different Confident
Historically has been like this and even if the COVID crisis is different in many respects, is not clear why should it be much different regarding growth trends.
Moritz Schularick No different Very confident
Ethan Ilzetzki's picture Ethan Ilzetzki London School of Economics No different Not confident at all
Looking at economic behavior in France and elsewhere when lockdown restrictions were relaxed suggests large pent up demand. I would predict that this will more than outweigh the growth loss due to debt overhang and other accumulated risks during Covid-19.
Michael Wickens's picture Michael Wickens Cardiff Business School & University of York No different Confident
Francesco Giavazzi's picture Francesco Giavazzi IGIER, Università Bocconi, Milano Potential GDP growth will be higher in 2025 than forecast pre-Covid Confident
Fabrizio Coricelli's picture Fabrizio Coricelli University of Siena and Paris School of Economics Potential GDP growth will be higher in 2025 than forecast pre-Covid Not confident
Covid-19 may induce much needed infrastructural public investments, speed up productivity enhancing technological innovation, also driven by environmental concerns.
David Cobham's picture David Cobham Heriot Watt University Between ½ and 2 percentage points Confident
Thorsten Beck's picture Thorsten Beck Cass Business School No different Not confident
As before, I am not confident as it is hard to predict policy actions over the next few years. My estimate is a best-case scenario where policies are adopted to minimise damage and resolve the crisis (corporate insolvencies, bank fragility) heads-on. Under such a scenario there should be no reduction in potential growth rate.
Mirko Wiederholt's picture Mirko Wiederholt Science Po No different Confident
Ricardo Reis's picture Ricardo Reis London School of Economics No different Not confident
Forecasting is hard. For growth rates, using a random walk, so expecting no difference seems a good way to reflect how not confident I am.
Nicholas Oulton's picture Nicholas Oulton London School of Economics Between ½ and 2 percentage points Not confident
David Miles's picture David Miles Imperial College No different Not confident
I suspect lasting impacts of COVID are largely on the level of capacity rather than its rate of growth.
Jordi Galí's picture Jordi Galí CREI, Universitat Pompeu Fabra and Barcelona GSE No different Not confident
I do not see any reason to believe the forces behind innovation and, hence, potential growth, will be much different after the pandemic, one way or another.
Antonio Fatás's picture Antonio Fatás INSEAD, Singapore No different Confident
No strong reason or evidence to believe that growth will be permanently affected.
Omar Licandro's picture Omar Licandro University of Nottingham No different Confident
Francesco Lippi's picture Francesco Lippi LUISS No different Confident
Hard to find a sound reason why the long run growth rate should change.
Lucio Sarno's picture Lucio Sarno Cambridge University ½ percentage point or less Very confident
Isabelle Méjean's picture Isabelle Méjean Ecole Polytechnique No different Not confident
Robert Kollmann's picture Robert Kollmann Université Libre de Bruxelles No different Confident
While there is much uncertainty about the speed of recovery over the next 1-4 years, there is no reason to expect a decline in the growth rate of potential output over a longer period.
Gernot Müller's picture Gernot Müller Eberhard-Karls-Universität Tübingen No different Not confident
Jose Luis Peydro Potential GDP growth will be higher in 2025 than forecast pre-Covid Confident
Marcel Fratzscher's picture Marcel Fratzscher DIW Berlin and Humboldt University Berlin More than 2 percentage points Not confident
Chryssi Giannitsarou's picture Chryssi Giannitsarou University of Cambridge, Faculty of Economics ½ percentage point or less Not confident
Natalie Chen's picture Natalie Chen University of Warwick ½ percentage point or less Not confident
Gino A. Gancia's picture Gino A. Gancia Queen Mary University of London No different Not confident
Etienne Wasmer's picture Etienne Wasmer Sciences Po, Paris ½ percentage point or less Confident
same as above. If we are in the middle of public finance crisis, it will be a -2 to -3% but hard to tell whether this is in 2025, 2024 or 2023.
Costas Milas's picture Costas Milas University of Liverpool ½ percentage point or less Not confident
Ugo Panizza's picture Ugo Panizza The Graduate Institute, Geneva (HEID) No different Not confident
I don't think that there will be long-term negative effect on growth, but not confident.
Philip Jung's picture Philip Jung University of Dortmund No different Not confident
Maria Demertzis's picture Maria Demertzis Bruegel Potential GDP growth will be higher in 2025 than forecast pre-Covid Very confident
There are important reasons to fill that new technology adaption has been accelerated as a result of Covid19. However, understanding the short to medium-term effects of the transition will also be important. On the whole GDP potential ought to increase by comparison to what we thought before Covid19.
John VanReenen's picture John VanReenen London School of Economics No different Confident
Linda Yueh's picture Linda Yueh London Business School No opinion Confident
Jorge Braga de Macedo's picture Jorge Braga de ... Nova School of Business and Economics, Lisbon Between ½ and 2 percentage points Confident
the growth rate of potential output is harder to predict that the level and the estimates of the output gap peter out pretty quickly after 2021 but the existing disruption across the world economy, incluiding Brexit, seem to rule out a quick recovery