An Early Assessment of Residential Mortgage Performance in China

1
An Early Assessment of Residential Mortgage
Performance in China

• Yongheng Deng
• University of Southern California
• Della Zheng
• University of Southern California
• Changfeng Ling
• Tsinghua University

2
Consensus on Economic Analysis of Mortgage
Borrower Behavior- The Option Theory

• A rational borrower maximizes her welfare by
  refinancing when the call option is
  in-the-money, and default if the put option is
  in-the-money.
• Hendershott and Van Order (1987), Kau and Keenan
  (1995)
• The jointness of the prepayment and default
  options is important in explaining behavior
• Schwartz and Torous (1993), Deng (1997), Stanton
  and Wallace (1999)

3
Consensus on Economic Analysis of Mortgage
Borrower Behavior- The Proportional Hazard Model

• Duration or competing risks models provide a
  convenient analytical tool for analyzing borrower
  behavior
• Green and Shoven (1986)
• Schwartz and Torous (1989)
• Vandell (1993)
• Deng, Quigley and Van Order (2000)
• Deng and Gabriel (2002)
• Deng and Quigley (2002)

4
Discrete Cox Proportional Hazard Model

• Cox Proportional Hazard Function
• Cox Partial Likelihood Function

5
Proportional Hazard Function of Mortgage Default
or Prepayment
6
Our Strategy

• Apply a competing risk framework to model current
  Chinese mortgage performance.
• Test to what extend the option theory can explain
  the mortgage borrowers behavior in China.
• Identify major determinants of the mortgage
  prepayment and default risks in Chinas
  residential mortgage market.

7
The Data

• Single family mortgage loans issued between March
  1998 and October 2002, collected by a major
  residential mortgage lender in Beijing, China.
• Monthly loan history data that includes a large
  number of loans, borrowers, and property-related
  characteristics and also indicates termination
  date and reason for termination.
• Data includes borrowers characteristics,
  including household monthly income, borrowers
  age, marital status, education, occupation and
  job position, etc..
• Local house price performance, unemployment rate,
  interest rate, etc., measured by local government
  statistics branch.

8
Covariates

• Time-varying covariates
• Current equity ratio
• Slope of yield curve
• Unemployment rate
• Shanghai Stock Exchange Index
• Time-invariant covariates
• LTV ratio at origination
• Log value of original loan amount
• Origination year
• Household income
• Age
• Marital status
• Education
• Occupation
• Job position

9
Empirical Joint Survival Function of Prepayment
and Default
10
Empirical Joint Survival Function of Prepayment
and Default
11
Empirical Joint Survival Function of Prepayment
and Default
12
Empirical Joint Survival Function of Prepayment
and Default
13
Empirical Joint Survival Function of Prepayment
and Default
14
Empirical Joint Survival Function of Prepayment
and Default
15
Empirical Joint Survival Function of Prepayment
and Default
16
Empirical Joint Survival Function of Prepayment
and Default
17
Mean and Standard Deviations at Origination and
Termination
18
Mean and Standard Deviations at Origination and
Termination (Cont.)
19
Frequency of Loans by Major Categorical Covariates
20
Frequency of Loans by Major Categorical
Covariates (Cont.)
21
Frequency of Loans by Major Categorical
Covariates (Cont.)
22
Frequency of Loans by Major Categorical
Covariates (Cont.)
23
Frequency of Loans by Major Categorical
Covariates (Cont.)
24
Frequency of Loans by Major Categorical
Covariates (Cont.)
25
Proportional Hazard Estimates for Mortgage
Prepayment and Default
26
Proportional Hazard Estimates for Mortgage
Prepayment and Default (Cont.)
27
Proportional Hazard Estimates for Mortgage
Prepayment and Default -Extended Models
28
Proportional Hazard Estimates (Cont.)
29
Proportional Hazard Estimates (Cont.)
30
Proportional Hazard Estimates (Cont.)
31
Proportional Hazard Estimates (Cont.)
32
Conclusion

• Option theory fails to explain the prepayment and
  default behaviors in the residential mortgage
  market in China.
• Call option / refinance is infeasible due to
  imperfect market circumstances
• Put option is in general out-of-money since
  steady increases of property values.
• Other non-option theory related financial
  economic factors play major roles in explaining
  mortgage behaviors in China.
• Borrowers choose to pay off in the bear market
  and when the yield curve is flat
• Uncertainty averse higher unemployment rate
  raises payoff rate
• Policy changes affect mortgage behaviors
  significantly
• Borrowers characteristics are important
  determinants.

33
Conclusion (Cont.)

• Borrowers characteristics may be used as an
  effective tool for screening loan applicants.
• Jumbo loan borrowers are associated with high
  default risks
• median-high to high income borrowers as well as
  white-collar workers are more likely to prepay
  their mortgage debts
• younger households, blue-collar workers are less
  likely to prepay.

34
Policy Implication

• A risk-based pricing technique on residential
  mortgage lending can improve the efficiency of
  the mortgage market and enhance the mortgage
  credit available to the much needed population,
  such as young and lower-income households and
  blue-collar workers in China.