Discussion of Campbell and Hercowitz

1
Discussion of Campbell and HercowitzsHome
Equity and Wealth During the Transition to a High
Debt Economy

• Erik Hurst
• November 2006

2
Overview

• Stylized Fact Use of debt has increased
  dramatically in the U.S. during the last two
  decades.

3
Overview

• Stylized Fact Use of debt has increased
  dramatically in the U.S. during the last two
  decades.
• My discussion
• - Why should this be of interest to
  macroeconomists?

4
Overview

• Stylized Fact Use of debt has increased
  dramatically in the U.S. during the last two
  decades.
• My discussion
• - Why should this be of interest to
  macroeconomists?
• - What are the potential causes for the increase
  in debt?

5
Overview

• Stylized Fact Use of debt has increased
  dramatically in the U.S. during the last two
  decades.
• My discussion
• - Why should this be of interest to
  macroeconomists?
• - What are the potential causes for the increase
  in debt?
• - Can we learn anything about the causes of the
  increase of debt from time series trends?

6
Overview

• Stylized Fact Use of debt has increased
  dramatically in the U.S. during the last two
  decades.
• My discussion
• - Why should this be of interest to
  macroeconomists?
• - What are the potential causes for the increase
  in debt?
• - Can we learn anything about the causes of the
  increase of debt from time series trends?
• - Throughout, I will discuss the contributions of
  the Campbell and Hercowitz paper to this
  literature?

7
Part 1 Why Should (Macro) Economists Care?

• An Explanation for The Great Moderation
• U.S. volatility was reduced dramatically starting
  around 1983 (see, for example, Stock and Watson
  2002).
• Common explanations focus on either 1) better
  monetary policy, 2) more favorable aggregate
  shocks, or 3) improvements in firm management of
  inventories.
• Given that consumption is the largest component
  of GDP, innovations in the ability of consumers
  to weather aggregate shocks will mitigate
  aggregate volatility.

8
Part 1 Why Should (Macro) Economists Care?

• An Explanation for The Great Moderation
• U.S. volatility was reduced dramatically starting
  around 1983 (see, for example, Stock and Watson
  2002).
• Common explanations focus on either 1) better
  monetary policy, 2) more favorable aggregate
  shocks, or 3) improvements in firm management of
  inventories.
• Given that consumption is the largest component
  of GDP, innovations in the ability of consumers
  to weather aggregate shocks will mitigate
  aggregate volatility. (Behavior of consumption
  during last recession)
• See recent work by Dynan et al (2006) and
  Campbell and Hercowitz (2006) for novel
  discussions.

9
Why Should (Macro) Economists Care?

• 2. Welfare Implications for Consumers (Including
  Sub Populations)
• - Not only smooth aggregate shocks, but better
  able to smooth idiosyncratic shocks or
  predictable lifecycle variation.
• - The increase in debt likely helped some sub
  groups much more than others. The median
  household likely always had some access to debt
  (store cards, etc.).
• - Historically, low income households were
  essentially excluded from credit market. May
  predict that the welfare gains would be largest
  for low income individuals.

10
Why Should (Macro) Economists Care?

• 3. Lower U.S. Savings Rates
• - Lower U.S. Investment?
• - Higher U.S. Interest Rates?
• - Increased Foreign Capital Inflows?
• Note I will return to the declining U.S.
  savings rate in a few minutes.
• Increased Bankruptcy (Default) Probabilities
• Changing Wealth Inequality within U.S.

11
Part 2 Natural Question

• What caused the sharp increase in debt (both
  collateralized and non-collateralized) among all
  groups of U.S. households during last
• forty years?
• - Supply side factors
• - Demand side factors

12
Supply Side Factors Legislation

• Monetary Control Act 1980
• Garn-St. Germain Act 1982
• - Both of above increased the competitiveness
  in consumer lending
• - Focus of the shock to credit market in this
  paper.

13
Supply Side Factors Legislation

• Monetary Control Act 1980
• Garn-St. Germain Act 1982
• - Both of above increased the competitiveness
  in consumer lending
• - Focus of the shock to credit market in this
  paper.
• Federal Housing Enterprises Financial Safety Act
  (Mandate Fannie and Freddie better serve low
  income households) 1992
• - Change the composition of borrowers in the
  average mortgage pool
• Riegle-Neal Act (Interstate Banking) 1994
• - Further increase competitiveness among
  banks

14
Supply Side Factors Technology

• Technological advances reduced the cost of
  providing financial services.
• Computers Allowed lenders to store large
  amounts of data about perspective borrowers and,
  in doing so, allowed them to price borrower risk
  more effectively.
• - Invention and use of FICO scores (1990-ish)
• - Reduced credit rationing
• - Bennett, Peach, and Peristiani Structural
  Change in the Mortgage Market and the Propensity
  to Refinance (JMCB 01)

15
Supply Side Factors Technology

• Technological advances reduced the cost of
  providing financial services.
• Computers Allowed lenders to store large
  amounts of data about perspective borrowers and,
  in doing so, allowed them to price borrower risk
  more effectively.
• - Invention and use of FICO scores (1990-ish)
• - Reduced credit rationing
• - Bennett, Peach, and Peristiani Structural
  Change in the Mortgage Market and the Propensity
  to Refinance (JMCB 01)
• Securitization Innovations and managing risk
  through pooling loan portfolios (CMOs, etc.).
  Increasingly important for non-collateralized
  loans as well as high risk collateralized loans
  (early 1990s).
• Endogenous to regulations? Maybe/Maybe Not

16
Demand Side Factors

• Aggregate Volatility Declining (Starting in 1983)
  
• Declining volatility should result in declining
  precautionary savings.
• Although, evidence suggests that for some groups
  individual income volatility increased despite
  declining aggregate volatility.
• Bankruptcy Option
• Decline in Bankruptcy Costs (stigma, information,
  out of pocket expenses)
• Increases in Bankruptcy Exemptions (1978
  Bankruptcy Reform)
• Equilibrium would have higher debt and higher
  defaults (coupled with higher interest rates).

17
Summary

• Lots of reasons why debt could have increased
  during the last 20 years.
• - The role of technology (including the ability
  to credit score) and securitization are likely
  an important component of the story.
• - Along with changes in GSEs policies, changed
  the mix of borrowers.
• My read of the literature is that the innovations
  in lending occurred continuously throughout this
  time period.
• Cause of the increase in debt is important for
  interpreting the trends in the aggregate data
  (and for interpreting the results from calibrated
  models) .

18
Part 3 What is Campbell and Hercowitz About?

• Sets out to ask what is the response to
  consumption, work hours, debt, the wealth
  distribution, etc. from an exogenous increase in
  households ability to accumulate collateralized
  debt.
• Key All borrowing in the economy is
  collateralized.
• Extent of collateralization has two components
• p is the required equity needed to purchase a
  durable (i.e., the down payment).
• is the parameter that governs the speed of
  subsequent equity accumulation (think of this as
  the ability to refinance).
• Focuses on a shock in the early 1980s (financial
  deregulation) that causes both p and to
  change immediately.

19
Part 3 What is Campbell and Hercowitz About?

• Two types of households borrowers and
  savers.
• - Utility f(durables, non-durables, and
  leisure)
• The only reason borrowers accumulate debt
  within the model is impatience (savers are
  relatively more patient)
• Borrowers are always bound by the liquidity
  constraint in steady state.
• Model is general equilibrium (wages and interest
  rates adjust only borrowers work)
• Borrowers do no saving and savers do no
  borrowing. (All action in the model is between
  the trade of resources between borrowers and
  savers).

20
Part 3 What is Campbell and Hercowitz About?

• Results from an exogenous decline in equity
  needed to purchase durables
• 1) Savers better off and borrowers worse off
  at the new steady state.
• 2) Borrowers are worse off because interest
  rates on debt increases, labor supply increases,
  and wages fall.
• Why do borrowers increase debt? Welfare gains
  during transition!
• Constraint is relaxed along the early part of the
  transition path borrowers can use current
  durables to expand current consumption.
• 3) Steady state wealth distribution becomes more
  unequal (borrowers go more in debt and savers
  increase wealth via loans)

21
Part 4 A Look at the Data

• Trends in collateralized debt
• - LTVs for mortgages (initial equity
  requirement, p)
• - Refinancing behavior (speed of subsequent
  equity accumulation, )
• Trends in non-collateralized debt
• - Levels
• - Access
• Trends in wealth distribution

22
Loan-To-Value (LTV) Ratios At Time of Purchase
Increase

• Notice that initial LTV is constant up through
  1989
• Model predicts debt (LTV) should start to
  increase immediately

23
Historical LTVs for New Mortgages (Including
Refis)
Increase starts 1992
1983
Source Federal Housing Finance Board
24
Some Facts Homeownership Rates
1994
1983
Source Census Bureau
25
Refinancings Over Time

• Define ? as the elasticity of refinancing
  propensity with respect to interest rate
  declines.
• Research shows that ?(2002) gt ?(1998) gt ?
  (1993) gt ? (1986)
• In other words, refinancing has continuously
  become more common over time.
• Bennett et al (2001) attribute this to the
  continuous decline in the cost of originating a
  mortgage.

26
Understanding The Role of Debt in the Macroeconomy
Initial Fees and Charges on Conventional
Single-Family Mortgages
1983
Notice Stead Decline
Note the Steady Decline
Source Federal Housing Finance Board
27
Non-Collateralized Debt Per Capita/Per Income
28
Access to Non Collateralized Credit

• Credit Card Access by Income Quintile by Year
  (fraction with card)
• Quintile 1970 1983 1989 1995
• Q1 2 11 17 28
• Q2 9 27 36 54
• Q3 14 41 62 71
• Q4 22 57 76 83
• Q5 33 79 89 95
• Source Durkin (2000)
• Note Trend continues since 1970.

29
Access to Non Collateralized Credit

• Credit Card Access by Income Quintile by Year
  (fraction with card)
• Quintile 1970 1983 1989 1995
• Q1 2 11 17 28
• Q2 9 27 36 54
• Q3 14 41 62 71
• Q4 22 57 76 83
• Q5 33 79 89 95
• Source Durkin (2000)
• Note Look at the Trend starting in 1970.

30
Share of Wealth Held By Top 10
1990
Figure 2 from Campbell and Hercowitz
31
Share of Housing Held By Top 10
1990
1990
Figure 2 from Campbell and Hercowitz
32
Mortgage Debt/Owner Occupied Real Estate
1983
1990
Steady increase starting around 1985
33
Comment 1 Measuring the Shock

• Are the legislative changes in the early 1980s
  the appropriate shock to calibrate the model?
• - For some analyzes, the distinction is not
  important.
• However, this papers focus (and interpretation
  of results) hinge on the transition dynamics.

34
Comment 1 Measuring the Shock

• Are the legislative changes in the early 1980s
  the appropriate shock to calibrate the model?
• - For some analyzes, the distinction is not
  important.
• However, this papers focus (and interpretation
  of results) hinge on the transition dynamics.
• - How can transition dynamics be isolated from
  subsequent shocks to lending technology or
  lending competition?

35
Comment 1 Measuring the Shock

• Are the legislative changes in the early 1980s
  the appropriate shock to calibrate the model?
• - For some analyzes, the distinction is not
  important.
• However, this papers focus (and interpretation
  of results) hinge on the transition dynamics.
• - How can transition dynamics be isolated from
  subsequent shocks to lending technology or
  lending competition?
• Data shows that most of the lending measures
  (LTV, Debt to Income, credit card debt) did not
  substantially change until the early 1990s or
  continuously evolved over the period?
• Timing of shock is important for interpreting
  magnitudes.
• Timing of shock is important for testing model
  predictions.

36
Comment 1 Measuring the Shock

• Even if we believe qualitative results and the
  timing of only one shock, do quantitative
  magnitudes make sense given the data?
• Model estimates that the liquidity constraint
  re-binds for borrowers after 30 quarters (7
  years). Yet most of the action does not take
  place on the borrowing side until after 1990?
• If successive shocks were hitting the economy,
  how do we interpret the model parameters? The
  imputed welfare gains?
• Understanding the origins of the shock is
  important for shaping future policy
  recommendations. Is it deregulation or
  computers?
• Question Can you estimate the model where the
  lending technology is evolving (perhaps at some
  constant rate) over the last twenty years?

37
Comment 2 Other Motives for Accumulating Debt

• Other potential reasons households accumulate
  debt
• 1) To smooth idiosyncratic labor risk
• 2) To smooth predictable income changes over
  the lifecycle.
• 3) To smooth aggregate shocks.

38
Comment 2 Other Motives for Accumulating Debt

• Other potential reasons households accumulate
  debt
• 1) To smooth idiosyncratic labor risk
• 2) To smooth predictable income changes over
  the lifecycle.
• 3) To smooth aggregate shocks.
• Welfare gains from smoothing income could be
  huge.
• - Borrowers could be better off even in the new
  steady state
• - Only reason borrowers accumulate debt in this
  model is impatience.

39
Comment 2 Other Motives for Accumulating Debt

• Other reasons households accumulate debt
• 1) To smooth idiosyncratic labor risk
• 2) To smooth predictable income changes over
  the lifecycle.
• 3) To smooth aggregate shocks.
• Welfare gains from smoothing income could be
  huge.
• - Borrowers could be better off even in the new
  steady state
• - Only reason borrowers accumulate debt in this
  model is impatience.
• To provide quantitative results for policy
  purposes, it would be important to model other
  reasons to accumulate debt besides impatience.
• Question Why not set up an OLG lifecycle
  analysis? What about the role for
  non-collateralized debt?

40
Comment 3 Testing the Mechanism

• Sharp decline in U.S. savings rate
• (From Figure 1 of Maki and Palumbo, 2001)

1983
1992
41
Comment 3 Testing the Mechanism

• Savings rates by income quintiles (Table 2 from
  Maki and Palumbo, 2001)

42
Comment 3 Testing the Mechanism

• Savings rates by income quintiles (Table 2 from
  Maki and Palumbo, 2001)

43
Comment 3 Testing the Mechanism

• Savings rates by income quintiles (Table 2 from
  Maki and Palumbo, 2001)

44
Comment 3 Testing the Mechanism

• Contribution to aggregate savings rates for each
  income quintiles (Table 4 from Maki and Palumbo,
  2001)

45
Comment 3 Testing the Mechanism

• What would your model predict about the savings
  rates for borrowers and savers during this time
  period?
• Would they match the aggregate data?
• My guess is no what does that imply? Should we
  think about the foreign sector being the
  savers? Does your model match foreign inflows
  into the U.S.?
• Why are rich U.S. households decreasing their
  savings rate so much?
• Question Could your model predict an increase in
  returns that would generate the rich decreasing
  their savings and increasing their wealth (at the
  same time that the poor increased their savings
  and increased their wealth?)

46
Conclusions

• The expansion of debt (both collateralized and
  non-collateralized) should be an important area
  of research for macroeconomists.
• I applaud the authors for working on this topic!
• For their research agenda (in both this paper and
  the previous paper) which employs calibrated GE
  models it is important to model the shock to
  lending correctly.
• To gauge welfare gains from expansion to credit,
  it would be good to include more realistic
  demands for borrowing (e.g., life cycle model
  with idiosyncratic labor income risk).
• I would put their mechanism to the test and see
  how it does at matching the trends in saving
  rates for borrowers and savers (as well as
  the aggregate).