Lessons From The Great Recession:

Macro In The Shadow Of Wealth Hoarding

Atif Mian

University of California, Berkeley

Causes: Why did household leverage go up?

Three broad hypotheses The “fundamentals” (credit demand) hypothesis

Leverage (and house prices) was driven by fundamentals, the economy got a bad unlucky shock leading to the recession

“House price increases largely reflect strong economic fundamentals …” - Oct. 20, 2005

The “excess credit” (credit supply) hypothesis

Supply curve for credit shifted out – credit supply became cheaper and more abundant. Why? Could be due to one or more of: securitization, agency problems, lose monetary policy, global imbalances etc.

The irrational exuberance in housing market hypothesis

People (irrationally) expected house prices to go up forever, and credit just followed these hyper-expectations.

Evidence Against Fundamentals View

Mian and Sufi QJE 2009

Areas with largest growth in credit have declining relative (and

absolute) incomes.

i.e. correlation of income and credit growth turned negative in 02-05.

Only period in last 20 years.

Inconsistent with standard demand side explanations

-.5

0.5

1

Mo

rtg

ag

e C

red

it G

row

th C

orr

ela

tio

n

with

Ho

use

ho

ld I

nco

me

Gro

wth

1991 1993 1995 1997 1999 2001 2003 2005

Year

-.2

5-.

2-.

15

-.1

Incom

e G

row

th:

Sub

pri

me m

inu

s P

rim

e z

ip c

od

es

1991 1993 1995 1997 1999 2001 2003 2005

Year(growth is cumulative since 1991)

-.0

5

0

.05

.1

Deb

t to

Incom

e R

atio:

Sub

pri

me m

inu

s P

rim

e z

ip c

od

es

1991 1993 1995 1997 1999 2001 2003 2005

Year

Subprime and Prime zip codes are defined to be the highest and lowest quartile zip codes within each County based on the fraction of residentswith a credit score below 660 in 1991.

Credit Growth And Income Growth Over Time

Figure 3

Evidence in favor of supply expansion

Cross-sectional growth strongly correlated with growth

in non-GSE securitization and initial mortgage denial

rates.

Why supply expanstion?

Securitization? (financial innovation more broadly)

Yes

Loose monetary policy? (e.g. Taylor rule not followed)

However, low and declining interest rates not a sufficient explanation

Global imbalances?

Likely

Is Securitization To Blame?

The market imposed a strong credit supply constraint on

prospective borrowers in 1996, esp. subprime.

The constraint kept tightening ….

Until 2002

That is when (private) securitization kicks in

as well …

Growth in private securitization strongest in

sub-prime neighborhoods

Result driven by mortgages sold to “unrelated” parties!

Loose Monetary Policy? Sub-prime expansion did not happen in previous monetary expansion

Source: Mian and Sufi (2009)

12

34

5

3 m

on

th T

-bill

yie

ld

01jan2001 01jan2002 01jan2003 01jan2004 01jan2005 01jan2006

34

56

78

3 m

on

th T

-bill

yie

ld

01jan1990 01jan1991 01jan1992 01jan1993 01jan1994 01jan1995

2001-2005

1990-1994

-.2

0.2

.4

Mo

rtg

ag

e O

rig

ina

tio

n G

row

th:

Su

bp

rim

e m

inu

s P

rim

e z

ip

co

de

s

Year0 Year1 Year2 Year3 Year4

(growth cumulative since Year0)

Subprime and Prime zip codes are defined to be the highest and lowest quartile zip codes within each County based on the fraction of residentswith a credit score below 660 in the base year.

Relative Mortgage Origination Growth For Subprime Zip CodesIn Falling Interest Rate Periods

Figure 4

Global Imbalances?

Securitization may be a “proximate cause” of sub-prime

credit expansion and house price appreciation

But securitization has been around for years

Why did it jump up all of a sudden around 2001-02?

A “deeper cause” of the rise in securitization and hence the housing

bubble is likely to be related to “global imbalances”

Global imbalances

Fast-growing and oil-rich Asian economies start saving large amounts

of capital (primarily through their central banks)

This capital is pushed into western countries … primarily the U.S.

Why did Asia do that?

We need to get understand a bit of history to appreciate the full

backdrop. [See Appendix Slides: Will cover time-permitting]

U.S. became an Emerging Market!

House price appreciation expectations?

There are two Americas

Flat America (Atlanta)

Rocky / coastal America (San Francisco)

Flat America has high housing supply elasticity

Any house price appreciation pressure leads to more housing

that can be easily built

House prices are not expected to go up (Caveat: no one understands Nevada!!)

Hence, is house price appreciation created the sub-prime

boom, we should not observe it in Flat America (e.g.

Texas, Atlanta)

But we do! (Source: Mian and Sufi 2009)

Flat America saw no house price bubble

Flat America saw no relative boom in sub-

prime house prices either

Yet Flat America received its fair share of

sub-prime credit boom

Consequences:

The Rich Save “Too Much”

Source: Chris Carroll

Marginal Propensity to Consume (MPC) declines with Permanent Income

Financial shocks in the presence of leverage redistribute resources from net borrowers (low wealth) to net lenders (high wealth)

Heterogeneity in MPC implies financial shocks translate into aggregate real shocks.

We do not need a banking sector to amplify shocks here.

These effects are purely driven by “aggregate demand” fluctuations

Also gives us “liquidity trap”

Implications …

U.S. inequality over time (1913-2008)

Source: Emmanuel Saez and Thomas Piketty

U.S. HH Debt to GDP: 1916-today

Source: Credit Suisse

U.S. HH Debt to GDP: 1916-today

Is it just a coincidence?

Consistent with the speculative theory I presented

more evidence ….

OECD HH debt and recessions

Source: Glick and Lansing (2010)

High and Low HH debt within the U.S.

Durable consumption

Source: Mian and Sufi (2011)

High and Low HH debt within the U.S.

Local employment

Source: Mian and Sufi (2011)

What can be done about HH debt?

Monitor / Regulate HH leverage?

But who decides what level is appropriate?

Who announces “the party is over”?

Past experience not very promising (U.S., Europe, Japan)

Better to see the problem as one primarily of efficient risk sharing.

We may not know when households have over-borrowed

But we can design mortgage contracts such that if a systemic shock were to occur, risk (losses) will be distributed between borrowers and lenders more evenly.

e.g. linking mortgage principal write downs (ex-ante) to aggregate (or regional) house price indices.