A Model of Secular Stagnation: Theory and Quantitative ... of \A Model of Secular Stagnation: Theory...
Transcript of A Model of Secular Stagnation: Theory and Quantitative ... of \A Model of Secular Stagnation: Theory...
Discussion of “A Model of Secular Stagnation:Theory and Quantitative Evaluation”
by Gauti Eggertsson, Neil Mehrotra and Jacob Robbins
Andrea FerreroUniversity of Oxford
Federal Reserve Bank of San Francisco Conference on
“Do Changes in the Economic LandscapeRequire a New Policy Framework?”
San Francisco, 21 April 2017
Introduction Summary Level & Decomposition Policy Conclusions
Introduction
Central tenet of Secular Stagnation hypothesis (Summers, 2014):
Low (possibly negative) equilibrium real interest rate
Data: Negative measured real interest rates
I Here to stay
Need a model to construct equilibrium real interest rate
I And think about “real interest rate gap”
EMR provide a model of Secular Stagnation
A new framework for policy analysis
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 2 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Introduction
Central tenet of Secular Stagnation hypothesis (Summers, 2014):
Low (possibly negative) equilibrium real interest rate
Data: Negative measured real interest rates
1990 1995 2000 2005 2010
perc
enta
ge p
oint
s
-4
-2
0
2
4
6
8
10Real Short-Term Yields
Median Mean
Source: Carvalho, Ferrero, Nechio (2016)
I Here to stay
Need a model to construct equilibrium real interest rate
I And think about “real interest rate gap”
EMR provide a model of Secular Stagnation
A new framework for policy analysis
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 2 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Introduction
Central tenet of Secular Stagnation hypothesis (Summers, 2014):
Low (possibly negative) equilibrium real interest rate
Data: Negative measured real interest rates
I Here to stay
Source: Summers (2015)
Need a model to construct equilibrium real interest rate
I And think about “real interest rate gap”
EMR provide a model of Secular Stagnation
A new framework for policy analysis
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 2 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Introduction
Central tenet of Secular Stagnation hypothesis (Summers, 2014):
Low (possibly negative) equilibrium real interest rate
Data: Negative measured real interest rates
I Here to stay
Need a model to construct equilibrium real interest rate
I And think about “real interest rate gap”
EMR provide a model of Secular Stagnation
A new framework for policy analysis
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 2 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Introduction
Central tenet of Secular Stagnation hypothesis (Summers, 2014):
Low (possibly negative) equilibrium real interest rate
Data: Negative measured real interest rates
I Here to stay
Need a model to construct equilibrium real interest rate
I And think about “real interest rate gap”
EMR provide a model of Secular Stagnation
A new framework for policy analysis
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 2 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Outline of Discussion
1 Brief summary and key findings
2 Decline of natural rate
3 Policy implications
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 3 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Summary
First part: Three-period OLG model with borrowing constraint
maxC yt ,C
mt+1,C
ot+2
Et(lnCyt + β lnCm
t+1 + β2 lnCot+2)
subject to
C yt = By
t = Dt/(1 + rt)
Cmt+1 = Ym
t+1 − (1 + rt)Byt + Bm
t+1
Cot+2 = Y o
t+2 − (1 + rt+1)Bmt+1
Get expression for equilibrium real interest rate
rt =(1 + β)(1 + gt)(1 + xt)Dt + (1 + xt+1)Y
ot+1
β(Ymt − Dt−1)
− 1
where xt ≡ At/At−1 − 1
Three factors that can push down real interest rate
1 gt : Demographics (Carvalho, Ferrero and Nechio, 2016)
2 xt : Productivity (Gordon, 2015)
3 Dt : Deleveraging (Eggertsson and Krugman, 2012)
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 4 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Summary
First part: Three-period OLG model with borrowing constraint
maxC yt ,C
mt+1,C
ot+2
Et(lnCyt + β lnCm
t+1 + β2 lnCot+2)
subject to
C yt = By
t = Dt/(1 + rt)
Cmt+1 = Ym
t+1 − (1 + rt)Byt + Bm
t+1
Cot+2 = Y o
t+2 − (1 + rt+1)Bmt+1
Population growth: Nt = (1 + gt)Nt−1 ⇒ (1 + gt)Byt = −Bm
t
Get expression for equilibrium real interest rate
rt =(1 + β)(1 + gt)(1 + xt)Dt + (1 + xt+1)Y
ot+1
β(Ymt − Dt−1)
− 1
where xt ≡ At/At−1 − 1
Three factors that can push down real interest rate
1 gt : Demographics (Carvalho, Ferrero and Nechio, 2016)
2 xt : Productivity (Gordon, 2015)
3 Dt : Deleveraging (Eggertsson and Krugman, 2012)
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 4 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Summary
First part: Three-period OLG model with borrowing constraint
maxC yt ,C
mt+1,C
ot+2
Et(lnCyt + β lnCm
t+1 + β2 lnCot+2)
subject to
C yt = By
t = Dt/(1 + rt)
Cmt+1 = Ym
t+1 − (1 + rt)Byt + Bm
t+1
Cot+2 = Y o
t+2 − (1 + rt+1)Bmt+1
Population growth: Nt = (1 + gt)Nt−1 ⇒ (1 + gt)Byt = −Bm
t
Productivity growth: Yt = At Y ⇒ Dt = At+1D
Get expression for equilibrium real interest rate
rt =(1 + β)(1 + gt)(1 + xt)Dt + (1 + xt+1)Y
ot+1
β(Ymt − Dt−1)
− 1
where xt ≡ At/At−1 − 1
Three factors that can push down real interest rate
1 gt : Demographics (Carvalho, Ferrero and Nechio, 2016)
2 xt : Productivity (Gordon, 2015)
3 Dt : Deleveraging (Eggertsson and Krugman, 2012)
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 4 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Summary
First part: Three-period OLG model with borrowing constraint
Get expression for equilibrium real interest rate
rt =(1 + β)(1 + gt)(1 + xt)Dt + (1 + xt+1)Y
ot+1
β(Ymt − Dt−1)
− 1
where xt ≡ At/At−1 − 1
Three factors that can push down real interest rate
1 gt : Demographics (Carvalho, Ferrero and Nechio, 2016)
2 xt : Productivity (Gordon, 2015)
3 Dt : Deleveraging (Eggertsson and Krugman, 2012)
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 4 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Summary
First part: Three-period OLG model with borrowing constraint
Temporary deleveraging shock ⇒ Permanently low real rate
Nice narrative:
I Real rate already on decline due to trends in demographics and productivity
I Becomes permanently negative because of crisis (deleveraging)
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 5 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Summary
First part: Three-period OLG model with borrowing constraint
Nice narrative:
I Real rate already on decline due to trends in demographics and productivity
I Becomes permanently negative because of crisis (deleveraging)
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 5 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Summary
Second part: Quantitative life-cycle model with
I Age-specific income profile
I Mortality risk
I Bequest motive
I Capital and CES production
I Exogenous process for relative price of capital
I Distortionary labor taxes
Calibrated to US data in 2015: Two options
I No output gap (Stock and Watson, 2012)
I Large output gap (Hall, 2016)
Legitimate to consider 2015 observed real rate as natural real rate but
I No output gap 6=⇒ Observed real interest rate = Natural rate
I Need additional assumption economy is in steady state
Results robust to alternative measures of output gap ∈ (−15%, 0)?
I Interesting that no deflation arises with large output gap
Paradox of wage flexibility (Galı and Monacelli, 2016)
I Need more flexibility to generate more deflation and larger output gap
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 6 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Summary
Second part: Quantitative life-cycle model
Legitimate to consider 2015 observed real rate as natural real rate but
I No output gap 6=⇒ Observed real interest rate = Natural rate
I Need additional assumption economy is in steady state
Results robust to alternative measures of output gap ∈ (−15%, 0)?
I Interesting that no deflation arises with large output gap
Paradox of wage flexibility (Galı and Monacelli, 2016)
I Need more flexibility to generate more deflation and larger output gap
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 6 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Summary
Second part: Quantitative life-cycle model
Legitimate to consider 2015 observed real rate as natural real rate but
I No output gap 6=⇒ Observed real interest rate = Natural rate
I Need additional assumption economy is in steady state
Results robust to alternative measures of output gap ∈ (−15%, 0)?
I Interesting that no deflation arises with large output gap
Paradox of wage flexibility (Galı and Monacelli, 2016)
I Need more flexibility to generate more deflation and larger output gap
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 6 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Summary
Second part: Quantitative life-cycle model
Legitimate to consider 2015 observed real rate as natural real rate but
I No output gap 6=⇒ Observed real interest rate = Natural rate
I Need additional assumption economy is in steady state
Results robust to alternative measures of output gap ∈ (−15%, 0)?
I Interesting that no deflation arises with large output gap
Paradox of wage flexibility (Galı and Monacelli, 2016)
I Need more flexibility to generate more deflation and larger output gap
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 6 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Outline of Discussion
1 Brief summary and key findings
2 Decline of natural rate
3 Policy implications
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 7 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Level
Very low level of r∗ throughout sample (1970-2016)
I Compare with estimates from Holston, Laubach and Williams (2016)
I Large real interest rate gap since early 1980s?
Source: Justiniano and Primiceri (2010)
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 8 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Level
Very low level of r∗ throughout sample (1970-2016)
I Large real interest rate gap since early 1980s?
Source: Justiniano and Primiceri (2010)
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 8 / 13
Introduction Summary Level & Decomposition Policy Conclusions
What Explains a Falling r ∗?
Major role of demographics and productivity growth
Government debt only factor that avoided much lower level
Some factors “disappear” from discussion
I How would have r∗ looked like without crisis?
I Role of increased inequality?
I Would be interesting to see counterfactuals with major driving forces
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 9 / 13
Introduction Summary Level & Decomposition Policy Conclusions
What Explains a Falling r ∗?
Some factors “disappear” from discussion
I How would have r∗ looked like without crisis?
I Role of increased inequality?
I Would be interesting to see counterfactuals with major driving forces
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 9 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Demographics and the Natural Real Rate
Carvalho, Ferrero and Nechio (2016) find similar role for demographics
I Larger relative contribution of increase in life expectancy
1990 1993 1996 1999 2002 2005 2008 2011 2014
perc
enta
ge p
oint
s
2.4
2.6
2.8
3
3.2
3.4
3.6
3.8
4
4.2Real Interest Rate
Baseline Population Growth Life Expectancy
Main difference: In EMR, fixed lifetime horizon but decrease in mortality risk
I Cannot live more than 81 years
Life expectancy currently at about 80 in most OECD countries
I Increased over time and projected to keep increasing
I Makes natural rate likely to keep falling
Also, empirical consumption profile much less hump-shaped than in EFR
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 10 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Demographics and the Natural Real Rate
Carvalho, Ferrero and Nechio (2016) find similar role for demographics
I Larger relative contribution of increase in life expectancy
Main difference: In EMR, fixed lifetime horizon but decrease in mortality risk
I Cannot live more than 81 years
Life expectancy currently at about 80 in most OECD countries
I Increased over time and projected to keep increasing
I Makes natural rate likely to keep falling
Also, empirical consumption profile much less hump-shaped than in EFR
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 10 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Demographics and the Natural Real Rate
Carvalho, Ferrero and Nechio (2016) find similar role for demographics
I Larger relative contribution of increase in life expectancy
Main difference: In EMR, fixed lifetime horizon but decrease in mortality risk
I Cannot live more than 81 years
Life expectancy currently at about 80 in most OECD countries
I Increased over time and projected to keep increasing
I Makes natural rate likely to keep falling
Also, empirical consumption profile much less hump-shaped than in EFR
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 10 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Outline of Discussion
1 Brief summary and key findings
2 Decline of natural rate
3 Policy implications
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 11 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Policy Implications
Why is negative r∗ a problem?
I Zero lower bound
: Level of r∗ puts lower bound on π∗
I Bubbles and financial stability
: Better raise r∗ than increase π∗
Also, in this model, limited effects of forward guidance
I What about quantitative easing?
Implication: Expansionary fiscal policy
I Debt/GDP from 118% to 215% raises r∗ from -1.47% to +1%
I But risk premia likely to rise
Similar conclusions in Carvalho, Ferrero and Nechio (2016)
I Additional option: Raise retirement age
I But need increase well beyond currently contemplated reforms (OECD, 2010)
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 12 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Policy Implications
Why is negative r∗ a problem?
I Zero lower bound: Level of r∗ puts lower bound on π∗
I Bubbles and financial stability
: Better raise r∗ than increase π∗
Also, in this model, limited effects of forward guidance
I What about quantitative easing?
Implication: Expansionary fiscal policy
I Debt/GDP from 118% to 215% raises r∗ from -1.47% to +1%
I But risk premia likely to rise
Similar conclusions in Carvalho, Ferrero and Nechio (2016)
I Additional option: Raise retirement age
I But need increase well beyond currently contemplated reforms (OECD, 2010)
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 12 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Policy Implications
Why is negative r∗ a problem?
I Zero lower bound: Level of r∗ puts lower bound on π∗
I Bubbles and financial stability: Better raise r∗ than increase π∗
Also, in this model, limited effects of forward guidance
I What about quantitative easing?
Implication: Expansionary fiscal policy
I Debt/GDP from 118% to 215% raises r∗ from -1.47% to +1%
I But risk premia likely to rise
Similar conclusions in Carvalho, Ferrero and Nechio (2016)
I Additional option: Raise retirement age
I But need increase well beyond currently contemplated reforms (OECD, 2010)
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 12 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Policy Implications
Why is negative r∗ a problem?
I Zero lower bound: Level of r∗ puts lower bound on π∗
I Bubbles and financial stability: Better raise r∗ than increase π∗
Also, in this model, limited effects of forward guidance
I What about quantitative easing?
Implication: Expansionary fiscal policy
I Debt/GDP from 118% to 215% raises r∗ from -1.47% to +1%
I But risk premia likely to rise
Similar conclusions in Carvalho, Ferrero and Nechio (2016)
I Additional option: Raise retirement age
I But need increase well beyond currently contemplated reforms (OECD, 2010)
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 12 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Policy Implications
Why is negative r∗ a problem?
I Zero lower bound: Level of r∗ puts lower bound on π∗
I Bubbles and financial stability: Better raise r∗ than increase π∗
Also, in this model, limited effects of forward guidance
I What about quantitative easing?
Implication: Expansionary fiscal policy
I Debt/GDP from 118% to 215% raises r∗ from -1.47% to +1%
I But risk premia likely to rise
Similar conclusions in Carvalho, Ferrero and Nechio (2016)
I Additional option: Raise retirement age
I But need increase well beyond currently contemplated reforms (OECD, 2010)
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 12 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Policy Implications
Why is negative r∗ a problem?
I Zero lower bound: Level of r∗ puts lower bound on π∗
I Bubbles and financial stability: Better raise r∗ than increase π∗
Also, in this model, limited effects of forward guidance
I What about quantitative easing?
Implication: Expansionary fiscal policy
I Debt/GDP from 118% to 215% raises r∗ from -1.47% to +1%
I But risk premia likely to rise
Other policies options are challenging
I Hard to increase fertility rates and productivity growth
I Probably don’t want to increase mortality...
Similar conclusions in Carvalho, Ferrero and Nechio (2016)
I Additional option: Raise retirement age
I But need increase well beyond currently contemplated reforms (OECD, 2010)
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 12 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Policy Implications
Why is negative r∗ a problem?
I Zero lower bound: Level of r∗ puts lower bound on π∗
I Bubbles and financial stability: Better raise r∗ than increase π∗
Also, in this model, limited effects of forward guidance
I What about quantitative easing?
Implication: Expansionary fiscal policy
I Debt/GDP from 118% to 215% raises r∗ from -1.47% to +1%
I But risk premia likely to rise
Similar conclusions in Carvalho, Ferrero and Nechio (2016)
I Additional option: Raise retirement age
I But need increase well beyond currently contemplated reforms (OECD, 2010)
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 12 / 13
Introduction Summary Level & Decomposition Policy Conclusions
Conclusions
Very nice paper, definitely useful to think about current policy challenges
Decline in natural real interest rate product of
I Financial crisis
I Interacting with long-term trends
May still require some fine tuning on quantitative part
If Secular Stagnation is relevant scenario
I Limited options for monetary policy?
I Shift to more activist fiscal policy?
Andrea Ferrero (Oxford) Discussion of Eggertsson, Mehrotra and Robbins 21 April 2017 13 / 13