2017年-BIS国际清算银行_The_impact_of_macroprudential_housing_finance_tools_in_Canada_39页_612kb

Summary of "The Impact of Macroprudential Housing Finance Tools in Canada"

Core Content

This paper investigates the impact of macroprudential housing finance tools in Canada using a microsimulation model of mortgage demand. It combines loan-level administrative data with household-level survey data to assess how changes in mortgage regulations, specifically the loan-to-value (LTV) ratio and amortization length, affect first-time homebuyers (FTHBs).

Main Policy Tools and Changes

The study focuses on two key macroprudential tools:

• Loan-to-Value (LTV) ratio: Defines the proportion of a house's value that can be financed by a loan.
• Amortization period: The length of time over which a mortgage is repaid.

Key Rule Changes (2005–2010)

• 2006: CMHC increased the maximum amortization from 25 to 30 years, and Genworth increased it from 25 to 35 years. CMHC also raised the LTV ratio from 95% to 100%.
• 2008: CMHC and Genworth both tightened the LTV ratio back to 95%, and reduced the maximum amortization period from 40 to 35 years. A new debt-service ratio (TDS) constraint of 45% was introduced.

These changes mark a period of loosening (2006–2008) and tightening (2008–2010) in macroprudential policy.

Key Findings

1. Impact of Policy Changes on Borrower Behavior

• LTV policies have a larger impact on mortgage demand than debt-service ratio (PTI) policies.
• LTV changes significantly affect the number of FTHBs who can qualify for loans, with a 51% decrease in loan qualifications and an 8.1% decrease in mortgage debt when the LTV constraint was tightened from 100% to 95%.
• Amortization changes also affect demand, but to a lesser extent. Tightening amortization from 35 to 30 years led to a 4.8% decrease in loan qualifications and a 3.6% decrease in FTHBs.
• Average mortgage size fell by a similar amount following changes in both LTV and amortization.

2. Household Constraints

• Households are more constrained by wealth (equity) than by income.
• A small fraction of households take advantage of longer amortization periods to reduce monthly payments.
• The majority of FTHBs are not constrained by income, as they are able to afford larger payments.
• The LTV constraint plays a more significant role in limiting access to the housing market than the income constraint.

3. Microsimulation Model (HRAM)

• The Household Risk Assessment Model (HRAM) is used to quantify the impacts of macroprudential changes on mortgage demand.
• The model incorporates borrower and property characteristics, as well as macroeconomic variables, to simulate how households respond to policy changes.
• It allows for the assessment of the effectiveness of LTV and PTI policies in mitigating household risk.

4. Policy Implications

• Tightening LTV has a more substantial effect on reducing mortgage qualifications and debt than tightening PTI.
• Loosening LTV leads to an increase in mortgage demand, as more households can qualify with lower equity.
• Interest rate changes significantly influence mortgage payments, even with longer amortization and higher incomes, due to the increased cost of borrowing.
• Household arrears are lower in environments with tight LTV but loose PTI policies, suggesting that LTV policies are more effective in reducing risk for Canadian households.

Data and Methodology

Data Sources

• Loan-level administrative data from the Canadian Mortgage and Housing Corporation (CMHC) for FTHB residential purchases.
• Household-level survey data used to calibrate the microsimulation model.
• The data cover the period from 2005 to 2010, including:
  • A pre-period (2005–2006)
  • A loosening period (2006–2008)
  • A tightening period (2008–2010)

Variables Analyzed

• House price (nominal and real)
• Mortgage size
• Household income
• Interest rates
• Amortization period
• LTV ratio
• TDS ratio
• PTI ratio
• Conventional vs. unconventional down payments

Empirical Model

• A regression model is used to analyze the impact of macroprudential changes on mortgage characteristics and borrower behavior.
• The model includes:
  • Month and location fixed effects
  • Bank fixed effects
  • Borrower and property characteristics

Conclusion

The study concludes that:

• LTV policies are more effective than PTI policies in influencing mortgage demand and reducing default risk.
• Wealth constraints are more binding than income constraints for FTHBs.
• The Canadian mortgage insurance system, backed by the federal government, allows for a clearer assessment of the impact of macroprudential tools on household behavior without modeling endogenous credit supply.
• The HRAM model provides a useful framework for understanding how households respond to macroprudential policy changes and for evaluating the effectiveness of different policy instruments.

This paper contributes to the growing literature on macroprudential policy by providing empirical evidence on its impact on household finance, particularly in the context of Canada's unique mortgage insurance system.