The Impact of the Interest and Capital Mortgage Deduction on Belgium borrowing behavior Annelies Hoebeeck & Carine Smolders ERES Conference 26/06/2015
Context Belgium home ownership around 70-73% Becoming an owner gives access to important tax benefits
Context Before 2005: different system for interest cost, capital amortization and mortgage insurance premiums Policy reform in 2005: Mortgage Interest, Capital amortizations and insurance Premiums assembled in one single Deduction: the MICPD ≠ Previous system: Fixed deduction, independent of loan size Double deduction for all couples Only for mortgages of 10 years or more… To obtain the owner-occupied dwelling… … which has to be the only dwelling of the household
Context € 2080 (in 2010 prices) + € 690 first ten years + € 70 for 3 or more children MICPD extensive housing subsidy over the loan life
Context Hoebeeck & Smolders (2014) MICPD did not promote Belgian homeownership due to its capitalization into higher house prices Capitalization is mostly attributed to household behaviour (e.g. Durning & Quigley, 1985) Survey at a Belgian housing fair (October 2013): households do not add the MICPD to their housing bid Different transmission channel at work
Context Different transmission channel at work: more mortgages higher mortgages longer duration MICPD MORTGAGE MARKET HOUSE PRICE HOUSEHOLDS
Context MICPD induced HH to take out more mortgages Average loan length ↑ from 18 (2005) to 25 years (2010)
Research Question MICPD = ineffective & unfair subsidy Higher income households get a larger subsidy No effect on homeownership Household pay more for their house… … and for their loan Sellers and financial institutions are subsidized instead of homebuyers To which extent did the MICPD change HH borrowing behavior?
Dataset First wave of the HFCS(2010) Questions 2327 HH Extensive set of loan characteristics Research unit: mortgage for the household main residence Loan origination date: 5 multiple imputed datafiles to treat missing values N=534
Methodology step 2: cope with selection bias
Methodology step 2: cope with selection bias First solution: restrict sample L_year > 2000 Length ≥ 10 years
Tax bracketTax rate (t) % % % % > % Max yearly deduction Basic amount2080 € First 10 years+ 690 € First 10 years, ≥ 3 children+ 70 €
Methodology step 4: cope with simultaneity and endogeneity Cross-equation correlation: SUR Endogeneity of D,M, H, MICPD variable and R: 2SLS Instruments for D,M,H: all exogenous variables of the model Instruments for R: interest rate on long term/short government bonds Instrument for MICPD over loan life: maximum yearly deduction SUR + 2SLS= 3SLS
Results Permanent income Harel R²= 0.43
Results Selection regression % correct predicted= 0.834% Intercept model: 0.587²+( )^2= 0.515
Results Mc Elroy R²= N=534
Results
Conclusion MICPD ↑mortgage maturity and demand MICPD has no direct effect on the house price Capitalization happens through the mortgage market Financial institutions= real beneficiaries of the MICPD MICPD should be eliminated