Estimating Hysteresis Effects

Francesco Furlanetto, Ørjan Robstad, Pal Ulvedal

Norges Bank

November 2019

The views expressed in this presentation are those of the authors and do not

necessarily reflect the views of Norges Bank

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 1 / 27

The independence assumption

The natural rate of unemployment (and potential output) isindependent of monetary policy

’...the ”natural rate hypothesis” quickly became widely accepted andhas been the dominant paradigm in macroeconomics ever since. It isembodied in the thinking and models used by central banks, and it isthe basis of the inflation-targeting framework...(Blanchard, 2018)

The assumption has been challanged by theories of hysteresis

Hysteresis in European unemployment in the 80s (Blanchard andSummers, 1986)

Great Moderation

Great Recession

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 2 / 27

The independence assumption

The natural rate of unemployment (and potential output) isindependent of monetary policy

’...the ”natural rate hypothesis” quickly became widely accepted andhas been the dominant paradigm in macroeconomics ever since. It isembodied in the thinking and models used by central banks, and it isthe basis of the inflation-targeting framework...(Blanchard, 2018)

The assumption has been challanged by theories of hysteresis

Hysteresis in European unemployment in the 80s (Blanchard andSummers, 1986)

Great Moderation

Great Recession

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 2 / 27

Alternative interpretations: Summers vs Fernald

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 3 / 27

The independence assumption in empirical work

Blanchard and Quah (1989): There are two types of shocks

Shocks with temporary effects on output (demand shocks)

Shocks with permanent effects on output (supply shocks)

Coibion, Gorodnichenko and Ulate (2018): Counterfactual outputdriven only by permanent shocks is interpreted as potential output

Blanchard (2019):

There may be supply shocks with temporary effects

There may be demand shocks with permanent/long lasting effects

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 4 / 27

The independence assumption in empirical work

Blanchard and Quah (1989): There are two types of shocks

Shocks with temporary effects on output (demand shocks)

Shocks with permanent effects on output (supply shocks)

Coibion, Gorodnichenko and Ulate (2018): Counterfactual outputdriven only by permanent shocks is interpreted as potential output

Blanchard (2019):

There may be supply shocks with temporary effects

There may be demand shocks with permanent/long lasting effects

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 4 / 27

The independence assumption in empirical work

Blanchard and Quah (1989): There are two types of shocks

Shocks with temporary effects on output (demand shocks)

Shocks with permanent effects on output (supply shocks)

Coibion, Gorodnichenko and Ulate (2018): Counterfactual outputdriven only by permanent shocks is interpreted as potential output

Blanchard (2019):

There may be supply shocks with temporary effects

There may be demand shocks with permanent/long lasting effects

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 4 / 27

Our approach

Our goal: identify demand shocks with potentially permanent effectson output. This tells us something about the empirical relevance ofhysteresis effects.

Combine short-term (sign) and long-term (zero) restrictions in a VARmodel (Arias, Rubio-Ramirez and Waggoner, 2019)

Two types of demand shocks

A demand shock with temporary effect on outputA demand shock with potentially permanent effect on output

Two types of supply shocks

A supply shock with temporary effect on outputA supply shock with potentially permanent effect on output

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 5 / 27

Benchmark model

US data, 1983Q1-2019Q2

Variables:

Real GDP per capitaPCE deflatorEmployment-population ratioInvestment per capita

First differences

3 lags (BIC)

Non-informative priors (Jeffreys)

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 6 / 27

Identification

Demand - permanent Supply - permanent Demand - temporary Supply - temporaryShort-term

Long-term

Short-term

Long-term

Short-term

Long-term

Short-term

Long-term

GDP - - - 0 - 0Inflation - + - +Employment 0 0Investment

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 7 / 27

A demand shock with permanent effect on output

0 10 20 30 40

Quarters

-1.4

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

Per

cent

GDP

0 10 20 30 40

Quarters

-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

0

Per

cent

Price Level

0 10 20 30 40

Quarters

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

Per

cent

Employment

0 10 20 30 40

Quarters

-5

-4.5

-4

-3.5

-3

-2.5

-2

-1.5

-1

-0.5

0

Per

cent

age

poin

ts

Investment

0 10 20 30 40

Quarters

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

Per

cent

age

poin

ts

Ouput per worker

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 8 / 27

A supply shock with permanent effect on output

0 10 20 30 40

Quarters

-1.4

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

Per

cent

GDP

0 10 20 30 40

Quarters

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

Per

cent

Price Level

0 10 20 30 40

Quarters

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

Per

cent

Employment

0 10 20 30 40

Quarters

-5

-4.5

-4

-3.5

-3

-2.5

-2

-1.5

-1

-0.5

0

Per

cent

age

poin

ts

Investment

0 10 20 30 40

Quarters

-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

0

Per

cent

age

poin

ts

Ouput per worker

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 9 / 27

Variance decomposition

GDP

0 10 20 30 400

0.2

0.4

0.6

0.8

1

Temporary DemandTemporary SupplyPermanent DemandPermanent Supply

Price Level

0 10 20 30 400

0.2

0.4

0.6

0.8

1Employment

0 10 20 30 400

0.2

0.4

0.6

0.8

1

Investment

0 10 20 30 400

0.2

0.4

0.6

0.8

1Output per worker

0 10 20 30 400

0.2

0.4

0.6

0.8

1

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 10 / 27

Channels for hysteresis

Through employment:

Insider/outsider effects (Blanchard and Summers, 1986, Gali, 2016)

Skill depreciation: unemployment duration reduces employability(Krueger at al., 2014, Abraham et al, 2016)

Labor force participation hysteresis (Yagan, 2019)

Disability insurance (applications and acceptances)

Through labor productivity:

Persistent drop in investment (Benigno and Fornaro, 2017)

Spending in research and development (Moran and Queralto, 2018)

Speed of technology adoption (Anzaotegui et al., 2018)

Composition effects: larger effects on less productive firms (Caballeroand Hammour, 1994)

Composition effects: larger effects on low skilled workers (Charles etal., 2016)

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 11 / 27

Investigating the channels

0 10 20 30 40

Quarters

-1.4

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

Per

cent

GDP

0 10 20 30 40

Quarters

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

Per

cent

Employment

0 10 20 30 40

Quarters

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

Per

cent

Output per worker

0 10 20 30 40

Quarters

-5

-4.5

-4

-3.5

-3

-2.5

-2

-1.5

-1

-0.5

0

Per

cent

Investment

0 10 20 30 40

Quarters

-1.6

-1.4

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

Per

cent

R&D Investment

0 10 20 30 40

Quarters

-0.7

-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

0

Per

cent

age

poin

ts

Routine employment share

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 12 / 27

Unemployment and participation

0 10 20 30 40

Quarters

-0.1

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Per

cent

Unemployment rate

Unemployment rate

0 10 20 30 400

0.2

0.4

0.6

0.8

1

0 10 20 30 40

Quarters

-0.05

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

Per

cent

Long-term unemployment rate

Long-term unemployment rate

0 10 20 30 400

0.2

0.4

0.6

0.8

1

0 10 20 30 40

Quarters

-0.25

-0.2

-0.15

-0.1

-0.05

0

0.05

Per

cent

Participation rate

Participation rate

0 10 20 30 400

0.2

0.4

0.6

0.8

1

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 13 / 27

Permanent

demand shock:

Variance

decomposition:

Robustness: Are the results driven by the great recession?

0 5 10 15 20 25 30 35 40

Quarters

-1.4

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

Per

cent

1983Q1-2019Q2

1983Q1-2019Q2

0 5 10 15 20 25 30 35 400

0.2

0.4

0.6

0.8

1

0 5 10 15 20 25 30 35 40

Quarters

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

Per

cent

1983Q1-2007Q4

1983Q1-2007Q4

0 5 10 15 20 25 30 35 400

0.2

0.4

0.6

0.8

1

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 14 / 27

Permanent

demand shock:

Variance

decomposition:

Taking stock

Main results so far:

Demand shocks with potentially permanent effects are important

Propagation through employment

Gali and Hammour (1993): Cholesky identification scheme

Demand shocks have long-run effects

Negative shocks increase productivity in the long run

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 15 / 27

What kind of demand shocks generate hysteresis effects?

Monetary policy shocks (Jorda, Singh and Taylor, 2019)

Fiscal shocks (Fatas and Summers, 2018)

Financial shocks (Guerron-Quintana and Jinnai, 2019)

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 16 / 27

Extending the model to 7 variables and 7 shocks

Extend the benchmark model to include:

Real wagesFed funds rate/shadow rate (Wu and Xia, 2016)Participation rate

5 shocks with potentially permanent effects on output3 demand shocks:

Monetary policy shockFinancial shockOther demand shocks

2 supply shocks:

Technology shockLabor supply shock

2 shocks with temporary effects on output

Demand shockSupply shock

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 17 / 27

Identification

Demand - permanent Supply - permanent Demand - temp. Supply - temp.Other

demandMonetary

policyFinancial Tech.

Laborsupply

Demand - temp. Supply - temp.

Short-term

Short-term

Short-term

Short-term

Short-term

Short-term

Long-term

Short-term

Long-term

GDP - - - - - - 0 - 0Inflation - - - + + - +Employment 0 0Investment/GDP + -Fed funds rate - + -Real wage - +

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 18 / 27

Variance decomposition

GDP

0 5 10 15 20 25 30 35 400

0.2

0.4

0.6

0.8

1

Temporary DemandTemporary SupplyPermanent Other DemandPermanent Monetary PolicyPermanent FinancialPermanent TechnologyPermanent Labor Supply

Employment

0 5 10 15 20 25 30 35 400

0.2

0.4

0.6

0.8

1

Output per worker

0 5 10 15 20 25 30 35 400

0.2

0.4

0.6

0.8

1

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 19 / 27

A financial shock with permanent effect on output

0 10 20 30 40Quarters

-1

-0.8

-0.6

-0.4

-0.2

0P

erce

ntGDP

0 10 20 30 40Quarters

-0.8

-0.6

-0.4

-0.2

0

Per

cent

Price Level

0 10 20 30 40Quarters

-0.8

-0.6

-0.4

-0.2

0

Per

cent

Employment

0 10 20 30 40Quarters

-4

-3

-2

-1

0

Per

cent

Investment

0 10 20 30 40Quarters

-0.6

-0.4

-0.2

0

0.2

Per

cent

Ouput per worker

0 10 20 30 40Quarters

-0.25

-0.2

-0.15

-0.1

-0.05

0

0.05

Per

cent

age

poin

ts

Participation rate

0 10 20 30 40Quarters

-0.2

0

0.2

0.4

0.6

Per

cent

Real wage

0 10 20 30 40Quarters

-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

0

Per

cent

age

poin

ts

Fed funds rate

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 20 / 27

A labor supply shock with permanent effect on output

0 10 20 30 40Quarters

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0P

erce

ntGDP

0 10 20 30 40Quarters

-0.6

-0.4

-0.2

0

0.2

Per

cent

Price Level

0 10 20 30 40Quarters

-0.8

-0.6

-0.4

-0.2

0

Per

cent

Employment

0 10 20 30 40Quarters

-4

-3

-2

-1

0

Per

cent

Investment

0 10 20 30 40Quarters

-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1

Per

cent

Ouput per worker

0 10 20 30 40Quarters

-0.3

-0.25

-0.2

-0.15

-0.1

-0.05

0

Per

cent

age

poin

ts

Participation rate

0 10 20 30 40Quarters

-0.2

-0.1

0

0.1

0.2

0.3

0.4

Per

cent

Real wage

0 10 20 30 40Quarters

-0.6

-0.4

-0.2

0

0.2

Per

cent

age

poin

ts

Fed funds rate

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 21 / 27

Variance decomposition: Participation rate

Participation rate

0 5 10 15 20 25 30 35 400

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 22 / 27

Conclusion

We identify demand shocks that have sizeable long lasting effects onoutput

These shocks affect output primarily through the employment channel

In a larger model, financial shocks seem to explain a significantamount of the variation in output, employment and labor productivityin the longer term

Implications for policy: overshooting of inflation (Rudebusch andWilliams, 2016) , ”run the economy hot” to reverse the damagecaused by high unemployment (Yellen, 2016)

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 23 / 27

EXTRA

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 24 / 27

A monetary policy shock with permanent effect on output

0 10 20 30 40Quarters

-0.8

-0.6

-0.4

-0.2

0

0.2P

erce

ntGDP

0 10 20 30 40Quarters

-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1

Per

cent

Price Level

0 10 20 30 40Quarters

-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

0

Per

cent

Employment

0 10 20 30 40Quarters

-3

-2

-1

0

1

Per

cent

Investment

0 10 20 30 40Quarters

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

Per

cent

Ouput per worker

0 10 20 30 40Quarters

-0.2

-0.15

-0.1

-0.05

0

0.05

Per

cent

age

poin

ts

Participation rate

0 10 20 30 40Quarters

-0.4

-0.2

0

0.2

0.4

0.6

Per

cent

Real wage

0 10 20 30 40Quarters

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

Per

cent

age

poin

ts

Fed funds rate

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 25 / 27

A technology shock with permanent effect on output

0 10 20 30 40Quarters

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0P

erce

ntGDP

0 10 20 30 40Quarters

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

Per

cent

Price Level

0 10 20 30 40Quarters

-0.8

-0.6

-0.4

-0.2

0

0.2

Per

cent

Employment

0 10 20 30 40Quarters

-4

-3

-2

-1

0

Per

cent

Investment

0 10 20 30 40Quarters

-0.6

-0.4

-0.2

0

0.2

Per

cent

Ouput per worker

0 10 20 30 40Quarters

-0.2

-0.15

-0.1

-0.05

0

0.05

Per

cent

age

poin

ts

Participation rate

0 10 20 30 40Quarters

-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

0

Per

cent

Real wage

0 10 20 30 40Quarters

-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1

Per

cent

age

poin

ts

Fed funds rate

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 26 / 27

Data

GDP: Real Gross Domestic Product, Billions of Chained 2012 Dollars, Quarterly, Seasonally Adjusted Annual Rate

Population: Population, Thousands, Monthly, Not Seasonally Adjusted

Inflation: Personal consumption expenditures (implicit price deflator), Index 2012=100, Quarterly, Seasonally Adjusted

Employment: All Employees: Total Nonfarm Payrolls, Thousands of Persons, Monthly, Seasonally Adjusted

Investment: Real Gross Private Domestic Investment, Billions of Chained 2012 Dollars, Quarterly, Seasonally AdjustedAnnual Rate

R&D investment: Real Gross Domestic Product: Research and Development, Billions of Chained 2012 Dollars,Quarterly, Seasonally Adjusted Annual Rate

Participation rate: Civilian Labor Force Participation Rate, Percent, Monthly, Seasonally Adjusted

Unemployment rate: Civilian Unemployment Rate, Percent, Monthly, Seasonally Adjusted

Long-term unemployed: Civilians Unemployed for 27 Weeks and Over, Thousands of Persons, Monthly, SeasonallyAdjusted

Real wage: Employed full time: Median usual weekly real earnings: Wage and salary workers: 16 years and over,1982-84 CPI Adjusted Dollars, Quarterly, Seasonally Adjusted

Fed funds rate: Effective Federal Funds Rate, Percent, Monthly, Not Seasonally Adjusted

Furlanetto, Robstad, Ulvedal (Norges Bank) Estimating Hysteresis Effects November 2019 27 / 27