Tax Incentives and Demand for Mortgage Debt: Evidence from the Finnish 1993 Tax Reform

Abstract

The 1993 Finnish tax reform reduced the incentives to use mortgage financing in home acquisition for high-income households. Before the reform, mortgage interest was deductible according to a progressive schedule which meant that the benefit from the deduction was the greater the higher was the taxpayer 's marginal income tax rate. After the reform, the deduction is made according to a flat schedule and the benefit no longer depends on taxpayer 's marginal income tax rate. This setting can be seen as a natural experiment, where one can distinguish multiple treatment groups and a control group. This paper uses household level repeated cross-section data from before and after the reform to study whether Finnish households responded to these changes in incentives to borrow. The results, based on difference-in-differences estimates, show that high income households with high marginal tax rates responded to the reform by clearly reducing their mortgage borrowing compared to the control group which was unaffected by the reform.

Key Words:

• Mortgage interest deduction
• tax incentives
• difference-in-differences

Acknowledgements

This study is part of the Academy of Finland research project ‘Economics of Housing’ led by Professor Heikki A. Loikkanen (University of Helsinki). I would like to thank Peter England, Aki Kangasharju, Heikki A. Loikkanen and two anonymous referees for helpful comments. Financial support from the Academy of Finland, Alli Paasikivi Foundation, OKO Bank Group Research Foundation and Yrjo Jansson Foundation is gratefully acknowledged.

Notes

^a Average deduction rate.

^b For first-time buyers the deduction rate is 30% from 1993 onwards.

^c After 1992 the limits apply only to the tax credit from labour income tax liability. Interest deduction from capital income is unlimited.

^d The limits for singles and married couples are inceared by the amounts in the last two colums.

** and * indicate that the effects are statistically significant at 1 and 5 percent level, respectively.

** and

* indicate that the effects are statistically significant at 1 and 5 percent level, respectively.

** and * indicate that the effects are statistically significant at 1 and 5 percent level, respectively.

1. See also Brueckner (1994).

2. Manchester and Poterba (1989) show that US households which obtain second mortgages are on average less wealthy than other households with similar characteristics. Although their result is only suggestive, they argue that second mortgages are used primarily for consumption not portfolio diversification purposes. See also Moriizumi (2000) for more discussion and Japanese evidence.

3. In the US, a taxpayer 's marginal tax rate affects the price of mortgage debt only if the taxpayer itemises its deductions. A taxpayer is allowed to itemise only if its overall amount of deductions exceed a certain limit. Otherwise it has to take the standard deduction which is equal to all taxpayers.

4. Stephens (2002) offers a more detailed analysis of Finnish housing markets developments in the past 20 years or so. See also Bengs and Loikkanen (1991) for an overview of earlier development.

5. The Finnish GDP collapsed about 12 per cent in 1991–1993. At the same time unemployment rate rose from 3.5 per cent in 1990 to 18.4 per cent in 1994. See Kalela et al. (2001) for more details on the Finnish recession and recovery.

6. In Finland there are only few financial firms concentrated on private housing finance. The biggest mortgage suppliers are general commercial, savings and cooperative banks.

7. According to the ECB (2003) mortgage stock to GDP ratio in 2001 was on average 33 per cent in the Euro area and 39 per cent in the EU. Highest ratios can be found in the Netherlands (74 per cent), Denmark (67 per cent) and the U.K. (60 per cent).

8. The stamp tax on interest income was first introduced in 1991 when the rate was 10 per cent. The rate was increased to 15 in 1992, to 20 in 1993 and finally to 25 per cent in 1994.

9. To be precise, Finnish house loans are not assumable mortgages but personal loans that are not tied to a particular dwelling. However, most Finnish house loans are secured by a home.

10. See Wooldridge (2002) for details of the method.

11. See Wooldridge (2002) for an overview of econometric methods for limited dependent variables.

12. See Wooldridge (2002), p. 130.

13. In this paper, this is done using the WALD command in LIMDEP 8.0. See Greene (2002) for details.

14. An overwhelming majority of Finnish households' mortgages are adjustable rate mortgages. For example, in 2001 94 per cent of new mortgages in Finland had adjustable rates (Stephens, 2002). In most contracts the interest rate is adjusted every 12 months. For households who have moved within the survey years, interest payments will not give an accurate picture of the amount of mortgage debt they hold.

15. See Wooldridge (2002), p. 522.

16. See also Fjærli (2004).

17. In fact, this assumption may not be valid. It is not clear whether income should have an isolated effect on mortgage demand. This assumption, however, is not necessary for our approach.

18. This interaction effect does not coincide with the coefficient of the interaction term in a probit or any nonlinear model. See Ai and Norton (2003).