A European Cross-Country Comparison of the Impact of Homeownership and Transaction Costs on Job Tenure

Abstract

De Graaff T. and van Leuvensteijn M. A European cross-country comparison of the impact of homeownership and transaction costs on job tenure, Regional Studies. This paper looks at the impacts across European countries of homeownership and transaction costs on job tenure. The analysis is based on a representative household sample of fourteen European countries covering the period 1994–2001. These data are merged with country-specific transaction tax data and aggregate homeownership rates. Similar to previous microeconomic studies, the results indicate that homeownership reduces job mobility as well as the probability of becoming unemployed or economically inactive. It is found, however, that this effect is severely mitigated by two factors. First, it decreases when aggregate homeownership rates are higher. Thus, homeowners have larger job tenures, but more so in countries with low homeownership rates. Second, transaction costs on the housing market are positively associated with the probability of becoming unemployed.

De Graaff T. and van Leuvensteijn M. 关于所有权与交易成本对任期长短的影响的欧洲跨国对比研究，区域研究。本文着眼于欧洲国家房屋所有权和交易成本对就业期限的影响。分析是基于14个欧洲国家在1994-2001年间一个代表性的家庭样本。这些数据与国家的交易税数据合并汇总了住房拥有率。与以前的微观经济学研究相似，本研究结果表明，住房拥有率降低了就业的流动性以及失业或弱经济活动的概率。然而，研究发现，这种影响严重受两个因素的弱化。首先，当总的住房拥有率较高时这种影响会降低。因此，购房者工作任期较长，但此种情况更多地存在与住房拥有率较低的国家。其次，二，住房市场的交易成本与失业的概率呈正相关。

De Graaff T. et van Leuvensteijn M. Une comparaison transnationale européenne des effets sur l'ancienneté professionnelle des coûts des transactions et de l'accession à la propriété, Regional Studies. Cet article cherche à examiner les effets des coûts des transactions et de l'accession à la propriété sur l'ancienneté professionnelle. L'analyse est fondée sur un échantillon représentatif auprès des ménages en quatorze pays européens sur la période de 1994 à 2001. Ces données sont confondues avec des données sur les coûts des transactions et les taux de propriété spécifiques des pays. Les résultats laissent indiquer que l'accession à la propriété réduit la mobilité du travail ainsi que la probabilité d'être au chômage ou bien inactif du point de vue économique, ce qui correspond aux études microéconomiques antérieures. Cependant, il s'avère que cet effet est sérieusement atténué par deux facteurs. Primo, il s'affaiblit quand les taux de propriété globaux sont plus élevés. Ainsi, ceux qui sont propriétaires ont une ancienneté professionnelle plus longue, d'autant plus dans les pays où les taux de propriété sont moins élevés. Secundo, les coûts des transactions sur le marché du logement sont en corrélation étroite avec la probabilité d'être au chômage.

De Graaff T. und van Leuvensteijn M. Ein europaweiter, länderübergreifender Vergleich der Auswirkungen von Hauseigentum und Transaktionskosten auf das Beschäftigungsverhältnis, Regional Studies. In diesem Beitrag werden die Auswirkungen von Hauseigentum und Transaktionskosten auf das Beschäftigungsverhältnis in verschiedenen europäischen Ländern untersucht. Die Analyse beruht auf einer repräsentativen Stichprobe von Haushalten in 14 europäischen Ländern im Zeitraum von 1994 bis 2001. Diese Daten werden mit landesspezifischen Daten über Transaktionssteuern und den Gesamtanteil an Hauseigentümern kombiniert. Ähnlich wie bei früheren mikroökonomischen Studien geht aus den Ergebnissen hervor, dass sich durch Hauseigentum die berufliche Mobilität sowie die Wahrscheinlichkeit von Arbeitslosigkeit oder wirtschaftlicher Inaktivität verringern. Gleichzeitig stellen wir jedoch fest, dass dieser Effekt durch zwei Faktoren stark abgeschwächt wird. Erstens nimmt der Effekt ab, wenn der Gesamtanteil an Hauseigentümern höher ausfällt. Hauseigentümer verfügen also über längere Beschäftigungsverhältnisse, dies jedoch stärker in Ländern mit einem niedrigen Anteil an Hauseigentümern. Zweitens stehen die Transaktionskosten auf dem Wohnungsmarkt in einem positiven Zusammenhang mit der Wahrscheinlichkeit einer Arbeitslosigkeit.

De Graaff T. y van Leuvensteijn M. Comparación entre países europeos del impacto de la propiedad de la vivienda y los costes de transacción en la actividad laboral, Regional Studies. En este artículo analizamos las repercusiones en los países europeos de la propiedad de la vivienda y los costes de transacción en la actividad laboral. El análisis se basa en una muestra representativa de hogares de catorce países europeos en el periodo de 1994 a 2001. Estos datos se combinan con los datos sobre los impuestos de transacción para cada país y las tasas agregadas de la propiedad de vivienda. Al igual que en estudios microeconómicos previos, los resultados indican que la propiedad de la vivienda reduce la movilidad laboral, así como la probabilidad de convertirse en desempleado o inactivo económicamente. Sin embargo, observamos que este efecto está seriamente mitigado por dos factores. Primero, disminuye cuando son más altas las tasas agregadas de la propiedad de la vivienda. Por tanto, los propietarios de viviendas tienen una actividad laboral más larga, pero más en países con bajas tasas de la propiedad de vivienda. Segundo, los costes de transacción en el mercado de la vivienda están positivamente asociados a la probabilidad de quedarse en paro.

• Housing tenure
• Job tenure
• Competing risk
• Transaction costs

• 住房使用权
• 工作任期
• 竞争风险
• 交易成本

• Logement
• Ancienneté professionnelle
• Risques
• Coûts des transactions

• Hausbesitz
• Beschäftigungsverhältnis
• Konkurrierendes Risiko
• Transaktionskosten

• Tenencia de viviendas
• Actividad laboral
• Riesgo competitivo
• Costes de transacción

JEL classifications:

• J6
• R2

Acknowledgements

The authors would like to thank Namkee Ahn, Harminder Battu, Casper van Ewijk, Henri de Groot, Pierre Koning, Jacob Munch, Anthony Murphy, Euan Phimister, Andrew Oswald, Ada Ma, Michael Rosholm and Michael Svarer for very useful comments. Financial support (Grant Number 461-03-571) was provided by the European Collaborative Research Projects in the Social Science. The usual disclaimer applies.

Notes

1. The only contribution known by the authors is Coulson and Fisher (2009), who looked at differences between metropolitan areas – metropolitan statistical areas (MSAs) and primary metropolitan statistical areas (PMSAs) – whereas the present paper looks at differences between countries. Arguably, the latter is better suited to look at differences between institutions.

2. A somewhat separate, but still related, literature is concerned with the possibility that falling home prices and rising interest rates, events associated with housing market declines, can ‘lock-in’ people to their homes, reducing housing and labour mobility (for an overview, see, among others, Quigley, 1987; Genesove and Mayer, 1997; and Ferreira et al., 2008).

3. For the results to hold, it is sufficient to assume that transaction costs are higher for homeowners than for renters.

4. Obviously, the impact of homeownership on the outflow out of current job spells and on the inflow into unemployment gives only half the story. For the other half: the impact on the outflow out of unemployment, see van Leuvensteijn and Koning (2004), Svarer et al. (2006), and Battu et al. (2008).

5. This leaves the number of probabilities to be estimated still up to 4^K.

6. Because, in the case of job and housing mobility, this assumption might be a bit strong, as many variables as possible are incorporated into models (10) and (11) that might reflect changes in preference structures, that is, because of life cycle effects.

7. To avoid confusion, censoring is not modelled as another competing risk. In other words, transitions to state b do not include censoring, while the destination vector d_b does include censoring.

8. Data for Sweden were derived from the Swedish Living Conditions Survey and transformed into ECHP format.

9. This holds except for the Netherlands and Sweden. For the Netherlands it was still possible to construct monthly activity states from other questions; for Sweden it was not. Therefore, Sweden was dropped from the final estimation. Because of this procedure, small job spells may have been unrecorded for the Netherlands, resulting in an underestimation of the number of job spells and an overestimation in the length of the average job spell. Further, information on out-of-labour force exits was not available for the Netherlands. Because it does not affect the other exit rates, the competing risk was left out for the Netherlands.

10. As one referee pointed out, transaction taxes might be endogenous to mobility. Theoretically, this is perfectly conceivable. However, the scant data available on the history of transaction taxes (for example, European Central Bank (ECB), 2003) suggest that these taxes are rather constant over time and that policy-makers are reluctant to use them as instruments. And if countries changed these taxes, then it was usually combined with other changes in taxation or subsidies for the housing market. Moreover, the direction of the change seems random at well. For example, Germany and France changed their transaction tax in an opposite direction in the 1990s, although their regional mobility levels at that time were more or less the same. Therefore, at least in the short-run, these taxes can be considered as exogenous.

11. Note the drop after eight years in all three Kaplan–Meyer survival functions. These are caused by the end of the ECHP survey in 2001. Some spells may, however, last longer than eight years because these started before the survey (and with a recorded starting date).

12. Risk into the out-of-labour force is somewhat inconsistent across studies. In this case, the coefficient is insignificantly negative. Van Leuvensteijn and Koning (2004), for example, find for the Netherlands that the coefficient is insignificantly positive. In any case, the coefficient is smaller than to enter unemployment or change jobs, pointing to the limited effect of homeownership in leaving the labour force.

13. Basically, these effects are derived from a similar estimation, as shown in Table 2, but then with country-specific individual homeownership effects instead of a generic individual homeownership effect.

14. The national homeownership rate is measured here as the mean national homeownership rate taken over the period 1994–2001.

15. In the case, there is a maximum of (K – 1) × 4 support points.