Transcript Discussion of Atif Mian and Amir Sufi

Goldman Sachs, New York, NY
September 27, 2013
Discussion of
“The Effect of Interest Rates and Collateral Value
Shocks on Household Spending:
Evidence from Mortgage Refinancing”
by Atif Mian and Amir Sufi
Jonathan A. Parker
Amount of
Housing
Consumed
More income or
wealth shifts budget
constraint out
Slope is
relative
price
Budget constraint
Indifference Curve
Amount of other goods
and services consumed
Jonathan Parker
September 2013
Goldman Sachs Fellows Conference
New York, NY
Page 2
Amount of
Housing
Consumed
What happens when the price of housing rises?
Key: homeowners can always keep
spending the same because house
price rise increases wealth
But: cost of housing rises also, so
tend to reduce housing demand and
increase spending on other goods
Budget constraint
Indifference Curve
Amount of other goods
and services consumed
Jonathan Parker
September 2013
Goldman Sachs Fellows Conference
New York, NY
Page 3
• An increase in (only) house prices increases
spending
• A decrease in (only) interest rates increases
spending (intertemporal substitution and
relative price)
But this is microeconomics . . .
• In the US economy or in a local economy,
house prices are not causes, they propagate
deeper causes like increases in wealth/future
income/ability to borrow
What happens when the average consumer in a
region demands more?
Jonathan Parker
September 2013
Goldman Sachs Fellows Conference
New York, NY
Page 4
What happens when the average consumer in a
Amount of
region demands more if the housing supply is
Housing
Consumed perfectly elastic?
Price of a new house is its
construction cost, so budget
constraint makes a parallel shift out
Indifference Curve
Increase in income causes
consumption of housing and other
goods rise without house price rise.
Identifies and Engel curve
Amount of other goods
and services consumed
Jonathan Parker
September 2013
Goldman Sachs Fellows Conference
New York, NY
Page 5
What happens when the average consumer in a
Amount of
region demands more if the housing supply is
Housing
Consumed perfectly inelastic?
Quantity of housing fixed,
so price of housing rises so
demand for housing =
supply of housing
Price change causes increase in house
prices and a large increase in
consumption of other goods
Amount of other goods
and services consumed
Jonathan Parker
September 2013
Goldman Sachs Fellows Conference
New York, NY
Page 6
Mian, Rao, Sufi (2013) shows that in the crash, consumption fell
more where housing was less elastic and house prices fell more
•
•
Not an MPC out of housing wealth
Instead, confirms that changes in relative prices lead to substitution away from
more expensive goods
This (better) paper adds focus on low credit score households
and direct evidence on mechanism
•
•
Suppose low credit score households are impatient and credit constrained and
have access to mortgage lending up to some loan-to-value
Then house prices not only contribute to substitution to other consumption but
also change the demand of low-credit households for all goods
Prediction: in goods times, consumption largest where housing
elasticity is lowest and low credit scores prevalent
Jonathan Parker
September 2013
Goldman Sachs Fellows Conference
New York, NY
Page 7
In an inelastic area, when high credit score
Amount of
households demand more, house prices increase
Housing
Consumed
Increase in demand causes even greater
increase in spending on other goods
where low credit scores are prevalent
And low credit score
households can refinance and
spend more on other goods
=> Additional kick to
spending (only) in places
with low elasticity of housing
Jonathan Parker
September 2013
Goldman Sachs Fellows Conference
New York, NY
Page 8
Identification assumption: driving force as I have described
Identify: equilibrium co-movement or prices and quantities in
response to income/future income/wealth changes
Jonathan Parker
September 2013
Goldman Sachs Fellows Conference
New York, NY
Page 9
Questions:
• What if the driving force is differences in credit easing for low and high
credit score households?
• Why is the effect of low credit score so negative in a zip code with an
elastic housing supply?
• Paper captures spending increases inclusive of local multipliers. Would
the answers be quantitatively different at different levels of aggregation of
shocks?
– Would an idiosyncratic, exogenous house appreciation lead to an individual
response different for a household than a zip code than a county than a
country?
• Surprising finding for automobiles spending
– Spending is interest-rate sensitive and
– automobiles are easy to borrow against because they are good collateral
• Paper not clear on cross-section vs. time-series
– Suggest splitting into two cross-sections: boom and bust
– Bust may partly be about default!
Jonathan Parker
September 2013
Goldman Sachs Fellows Conference
New York, NY
Page 10
Conclusion
• Main predictions of theory confirmed:
– when a boom increases the relative price of housing, the increase in
consumption demand is larger in places where housing is more
constrained as relative prices lead to substitution
– and in these places, where there are more liquidity constrained
households, there is an even larger increase as borrowing constraints
are relaxed by increasing house prices
• This in no way invalidates representative agent short-cuts and
is consistent with New Keynesian models in which current
booms feed back more strongly into current demand
– designed to capture liquidity constraints and impatience
– more complex NK models with collateral looking even better
Jonathan Parker
September 2013
Goldman Sachs Fellows Conference
New York, NY
Page 11