Transcript Discussion of household leveraging and deleveraging

Discussion of household
leveraging and deleveraging
Discussion by
Gauti B. Eggertsson
Brown University
This paper (from my perspective)
• Can a deleveraging cycle generate the great recession in a
standard DSGE mode? (paper addresses other issues too!)
• Ask this questions in a quantitative model.
• Early models (e.g. Eggertsson and Woodford (2003)) model
trigger of the crisis as preference shock.
• Free variable, little discipline
• Recent work aimed at modeling the origin (debtdeleveraging). Why?
– Puts more discipline on the shocks (can model it to match
features of the data rather than a residual).
– Important? Yes may have important interaction with policy
• Mortgage write-downs, Fisher Debt deflation, fiscal multipliers, etc
etc.
• This paper is one of the first to deliver on this promise.
• First order importance, exciting stuff.
This paper
• Can a household debt-deleveraging cycle
generate the great recession in a standard
DSGE model?
• Answer: No
• But not the last word …
• Focus here on plausible change so that the
answer is “yes”.
• Build on Eggertsson and Krugman (2012) (web
Appendix)
Story in Eggertsson-Krugman
• Part of the economy “need” to cut down
spending
• For output to be at potential somebody need
to make up for it
• Who?
• Those unconstrained.
• How? By a drop in the real interest rate
Thought experiment
• Imagine an endowment economy.
• One agent more patient than the other.
• Impatient (borrowers) subject to a debt limit.
At steady state at the limit.
• Now
D
high
®D
low
D
high
®D
low
C
C
b
t
s
t
rt
n
Basic mechanism EK
• Some part of the population stops spending
(“deleveraging shock”)
• Somebody else needs to make up for it.
• How do we make those other guys “make up for
it”.
• By a drop in the interest rate. Can trigger a zero
bound.
• With nominal frictions: Big problem!
• Liquidity trap.
• Show this also in a “standard looking” NK model
This paper
• Augmented by housing sector. Deleveraging
happens via deleveraging of the households.
• Model drop in D more seriously.
• Can it generate a meaningful response?
• Answer: No
• Why? Because even if households are cutting
down their spending on housing ….
 Somebody else is making up for it even
without a reduction in the real interest rate.
This paper
• Very little drop in real interest rate
• Unconstraint agents (savers) don’t need to see
much of a drop in interest rate to start spending.
• Why?
• Because they will start investing in productive
capital.
• Investment is tied to the real interest rate via
marginal productivity of capital.
• Question: What happens to investment in the
model?
Main comment
• A key feature of the recession is the drop in
investment and real interest rate.
• You want to make sure that your model
delivers this.
• How to do this?
• One approach: Eggertsson and Krugman - web
appendix
Eggertsson and Krugman
– web appendix
• Does incorporating productive capital change the
result?
• Yes, if savers can invest, drop in the real interest
rate small.
• Similar to Christiano (2004) result in discussing
Eggertsson and Woodford (2003)
– same preference shock will mean the zero bound no
longer binding.
• My response at the time
– Introduce shock to the “capital adjustment function”.
– Not really a convincing thing to do here.
Eggertsson and Krugman Web
Appendix
• We should think of “investors” as constrained?
• How to do it?
• Most obvious way: Eggertsson and Krugman
make the constrained agents the ones that
have access to capital investment.
• Deleveraging now applies not only to
“household” but also the “investors” – they
are the same person.
Result form EK
• Now deleveraging shock has an even bigger
effect.
• Investment responds by even more than
“regular” consumption”
• Fraction of savers is 0.7
with fixed capital stock
• It is 0.9 with flexible
• Bigger effect with
flexible investment
Suggestions for next paper
• Introduce entrepreneurs (use e.g. Iacoviello (AER,
2005)).
• The entrepreneurs are also constrained.
• Conjecture: Deleveraging will make zero bound
binding and effect can be quite big.
• Details will of course matter (does land show up
in collateral constraint – housing prices etc)
• Bottomline:
– Can deleveraging explain crisis,?
– Might make sense to have an alternative, e.g.
households vs. entrepreneurs?
Conclusion
• Exciting research agenda.
• Discipline models with data on deleveraging to
search for the origin of the crisis.
– Can we use the Mian-Sufi evidence to impose
some discipline?
• Result so far: Difficult to put the usual
leveraging-deleveraging stories into a
quantitative model and get much action.
• For story to fly need to add more features.