![Sample our Built Environment journals, sign in here to start your access, Latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5926/TOC-Subject-Sample-2021-Built-Environment.jpg"})
ABSTRACT
Homeownership has long been associated with a myriad of economic, social and civic benefits, prompting countries such as Norway to expand access to homeownership for socio-economically disadvantaged households. In this paper, we explore the impact of homeownership on residential stability using a longitudinal data-set of renters who applied for a state mortgage programme in Norway between 2004 and 2010. These data allow us to specifically address the issue of selection bias in our analysis. We find that even after controlling for a wide range of demographic, socio-economic and housing market characteristics, homeownership has a substantial, positive impact on residential stability. This effect is stronger for groups that are more marginalised in Norwegian housing and labour markets, including East European and non-Western immigrants. The Norwegian case suggests homeownership policy can help to promote social goals, but also highlights the importance of providing welfare supports in tandem with access to mortgage credit in order to reduce the risks of homeownership for lower-income households.
Acknowledgments
Dr. Aarland's work on this project was funded by The Norwegian State Housing Bank, and was conducted while Dr. Aarland was a visiting scholar at the Institute of Urban and Regional Development (IURD) at University of California, Berkeley. The authors would like to thank IURD for providing the opportunity for us to collaborate on this work. We would also like to thank colleagues and reviewers for their helpful comments on our earlier drafts.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1. As a result, renters with similar characteristics may be approved for a mortgage in one municipality but rejected in another.
2. The homeownership rate has been fairly stable for many decades, measured at 80% in 1990 and 77% in both 2001 and 2011. While single-family homes are common in smaller towns and rural areas across the country, in the larger cities owner-occupation is often in apartments and through housing cooperatives.
3. This includes a high share of senior citizens.
4. The share of small-scale landlords is 65% according to the Census 2001, which is the last Census that classified landlords. Many rent out a separate self-contained unit or part of their own home, for example, a basement suite. This is due to a provision in the Norwegian tax code that renders rental income tax exempt if derived from renting out less than half of one's own dwelling.
5. In general, social housing is considered temporary shelter for tenants who are unable to secure adequate quarters in the private rental market. The average tenure in the Oslo social housing system is nine years (Personal Communication, 2016).
6. Elderly that are no longer fit to live independently is another group that tends to live in rented dwellings.
7. Rents are market-based, and most social rental contracts are time-limited, most commonly for a period of three or five years, as municipalities have a goal of high turnover in their social housing stock.
8. For example, the Community Advantage Program (the largest effort to expand access to low-income families in the United States) only originated approximately 45,000 mortgages (Quercia, Freeman, and Ratcliffe Citation2011).
9. In the case of repossession, any loss is shared between the local authority (25%) and the State Housing Bank (75%) (Barlindhaug and Astrup Citation2010)
10. See Aarland (Citation2012), Astrup and Aarland (Citation2013) and Astrup, Monkerud, Ruud and Aarland (Citation2015) for details.
11. Housing allowances in Norway are tenure-neutral, and repayments of principal count as eligible housing expenses (Barlindhaug and Astrup Citation2010).
12. Applicant categories reflect the starter mortgage guidelines and include first-time buyers, refugees, disabled household member, re-establishment after household dissolution, financially vulnerable household, homeless, substance abuse, and combination of substance abuse and mental health problems.
13. The starter mortgage is also used as a refinancing tool in special circumstances, most notably in the case of household dissolution, and more rarely as a way to restructure debt to financially sustainable terms for high-debt households.
14. We drop applicants who were living with their parents prior to the application.
15. In compliance with the regulation of The Norwegian Data Protection Authority, the official Norwegian personal number was replaced by a project-specific, unique person-identifying code by Statistics Norway before the data were handed over to the researcher.
16. Register data is not available for all mortgage applicants, and 51.7% of the sample is matched successfully.
17. Mean household income in Norway in 2015 was NOK794,400. (NOK589,900 for after tax income).
18. Stata estimates ln(α) instead of α.
19. We define high welfare users as those who have more than 40% of their income deriving from transfer income (75th percentile in our sample).
20. It is not uncommon to follow up mortgage applications with phone calls or personal interviews; applicants may also be referred to the mortgage application system from other municipal offices.
21. Single person, couple, single parent and couple with children.
22. As a robustness check, we re-ran the regression on the sample of applicants within the East Interior region of the country. This did not substantively change the results. For example, for the model presented in and , the estimated coefficient on D approved is −0.5033 and it is significant at <0.000 level. The corresponding predicted reduction in moves for accepted applicants is 39.6%, compared to 37% for the full sample.
23. A Chow-test reveals that the coefficient estimates differ substantially by immigrant group and estimating the model separately by subsample yields a better fit. The test statistic is LR = −2(L1−∑L0) ∼ χ2 with ∑d0 – d1 degrees of freedom, where the summation is over the log likelihoods and degrees of freedom, respectively, for the four subsamples. In our case LR = 234.2, corresponding to a P-value of 0.0000, since the test statistic exceeds the critical value 82.4.
24. Full model results are available from the authors upon request.