Do homeowners cope better with economic crises in terms of employment? An analysis of micro panel data from Spain, 2004–2013

Abstract

This paper investigates the effects of housing tenure choice on the employment status of individuals through the empirical estimation of different dynamic panel data models using Spanish data covering the period 2004–2013. I find that there is no sign of any significant spillover effect of the regional homeownership rate on the probability of employment, whereas the average effect on this probability for the lagged housing tenure choice is robust and significantly positive, ranging around two percentage points for the whole homeowners group and for those homeowners with a mortgage outstanding. The effect vanishes for the outright homeowners. Some implications for policies supporting home ownership are briefly discussed.

Keywords:

• Homeownership
• housing tenure choice
• unemployment
• dynamic probit model
• Spain

Notes

1. See Bentolila et al. (2012) for a brief review of the institutional settings of the Spanish labor market.

2. Studies include: Henley (1998), Coulson & Fisher (2002, 2009), Munch et al. (2006, 2008), Battu et al. (2008), Van Ewijk & Van Leuvenjstein (2009), Rouwendal & Nijkamp (2010), Farber (2012), Head & Lloyd-Ellis (2012), Coulson & Grieco (2013), Isebaert et al. (2015), among others.

3. See Rogerson et al. (2005) for a recent survey on search-theoretic models of the labor market.

4. The head of the household is the person owning or renting the accommodation. There is no way to deduce from the data if the other household members incur in any housing cost and, as a consequence, the moving cost and the mobility pattern could be very different for them. In addition, family ties could exacerbate this problem. Therefore, the inclusion of the other household members may lead to errors in the conclusions of the paper.

5. Renting with a reduced price rent includes renting social housing, renting at a reduced rate from an employer or other organizations, and renting in accommodation where the actual rent is practically frozen by law (old-rent contracts signed before 1985, when the so-called Boyer Decree liberalized the Spanish rental market, see Pareja & San Martín, 2002). These households are excluded because, due to their peculiar characteristics, especially the relative immobility stemming from rents below market rents, high waiting lists and the restricted transferability within public housing, and security of tenure, as shown in Battu et al. (2008), their consideration may distort the conclusions of the paper.

6. The reason I use this geographical/spatial scale is that the only geographical location information about the individual contained in EUSILC is the region of residence of the household aggregated at the NUTS 2 level.

7. Specifically, the housing price index for the autonomous region of residence, based on the average price index for new and existing housing (€/m²) published by the Ministry of Public Works Transport and Housing (Ministerio de Fomento).

8. It does not seem logical to assume that, for example, the unobserved innate ability, is uncorrelated with control variables as education or the number of years the individual spent in paid work before.

9. In fact, if we estimate specification 3 without the true state dependence, i.e. removing E_it − 1 and H_it − 1 as explanatory variables in the employment and housing tenure equations in (1), respectively, then, it results in a significant correlation between both equations, but the specification 3 is validated against this model by a likelihood ratio test.

10. This is not as surprising as it may seem at first glance. What is reflected in this result is that those who remain employed in one year are more likely than the rest to be employed the next year.