Reviving the Oswald Hypothesis?
by Ken Gibb
In a new working paper for the Peterson Institute for International Economics written with David Blanchflower, Andrew Oswald has returned to the forefront of the debate around the correlation between high home ownership rates and higher unemployment (‘Does high home-ownership impair the labour market?’).
The new paper can be found at http://www.piie.com/publications/wp/wp13-3.pdf.
Oswald set off quite a controversy after earlier papers in the late 1990s focused on evidence that there was such a link and also how it might be explained (i.e. that home ownership somehow led to labour frictions and other imperfections that pushed up joblessness). The response to the ‘Oswald hypothesis’ was sometimes sceptical and often openly critical.
In a paper we wrote about the role of home ownership and the economy in Housing Studies (2012, p.), Tony O’Sullivan and I concluded that:
“Oswald (1997, 1999) hypothesised that owner-occupation leads to higher levels of unemployment than otherwise would apply because of high house purchase transactions costs, high housing access costs for younger households and because housing-induced constraints on migration lead to poorer job choices. Oswald also suggested that areas with high homeownership levels may deter entrepreneurs from setting up new operations, and that homeowners commute more than renters, generating transport congestion”.
We agreed with Dietz and Haurin (2003) who “concluded that studies using micro-level data tend to reject the Oswald hypothesis while aggregate studies tend to be more supportive, and that while homeownership reduces mobility this is insufficient on its own to increase unemployment.”
The new paper is a careful empirical analysis of US States. It then goes on to speculate on the mechanics or explanatory forces that might explain the observed strong correlation. It also rules out a series of confounding arguments based on empirical counter-positions and, to be fair, identifies a number of possible flaws in the argument or evidence. While the authors continue to use some of the earlier explanations for the association between unemployment and home ownership, they do add some new arguments or refinements. The paper is essentially an empirical problem that is observed, its causes hypothesised and a general call made for further debate and detailed investigation by the economics community.
I want to make four main points in response. First, to what extent does the new empirical approach overcome earlier questions? Second, do earlier key problems remain? Third, what more might we say about the underlying causes of this relationship? Fourth, and to the extent there is some kind of validity in the core finding – what are the policy implications?
The paper looks at an annual panel of US states from 1985 to 2011 (including shorter sub-samples). They conclude that rises in home ownership are followed by a (long) lag to substantial increases (i.e. a doubling) in unemployment. This is caused by mobility problems, longer commutes and lower new firm formation (as well as some other factors – see below). The empirical section involves a series of tests to make the case that this correlation is not a spurious artefact. Nonetheless, the tone is one of provisional and tentative findings. They accept that, while they have eliminated some other explanations in the data (e.g. the 2006-07 crash in house prices), there may yet be omitted variables at play and that there may also be exogenous forces to do with changes in the structure of housing markets that are not measured in the data. Most importantly, they acknowledge that they have no fully-worked theory for how the housing market disturbs labour market equilibrium. But the point is that this statistical exercise is prima facie evidence for more research to get behind these findings.
I still think there are conceptual issues with this work. The main one for me is the lack of specification of the process not in terms of the labour market but rather the housing sector’s role. By this I mean that, while they say they do not expect the unemployed to be among the home-owners, they say little about the precise ways in which the housing system as a whole is causing labour market problems. What is the owner occupation rate a proxy for? Similarly, the authors say that high rates of home ownership re associated in the US state data with lower levels of new firm formation. Again what is actually being measured – what is the home ownership rate a proxy for? Third, they make the odd throwaway point about high renting low unemployment in the Swiss case and the converse in Spain. How can one make such a leap from state-level data? Particularly, as there is actually quite a lot of implicit regional and urban economics explanation going on in the US analysis and then make aggregate nation-level points where one would have thought institutional and other omitted variables are more likely to be working.
I did think however that in addition to returning to their earlier arguments (e.g. mobility and transactions costs) they raise some interesting spatial drivers of the correlation: spatial-mismatch, sprawl and congestion and other urban externalities relating to NIMBYISM (i.e. the argument that existing home owners do not want businesses in their locale that might adversely affect property values). Will urban economics provide better micro explanations that the previous work criticised for instance by Dietz and Haurin?
Finally, they argue that if this correlation is bone out by a plausible explantory model then it raises difficult questions for policymakers promoting home ownership. They even nod to taxing investment returns (e.g. imputed rents). This was taken up by the Guardian this morning as more evidence that the fixation on home ownership is misplaced. In our 2012 paper we concluded that there was precious little evidence that home ownership growth benefited the economy but perhaps had selective micro-social benefits e.g. for parenting and child outcomes. On this basis we could not see a justification for a general tenure-focused subsidy favouring home ownership per se. This does not rest on a specific argument that the housing system’s inflexibility leads to future higher unemployment. But were that to turn out to be the case – it would be another reason to wish to re-balance the unevenness of the distribution of housing policy and subsidy.
I am sure Blanchflower and Oswald’s paper will get its way and be the subject of much more debate and testing. If this should lead to closer conceptual work on the relationship between the housing system and the labour market that can be empirically modelled – it would be a very worthwhile result.
Dietz, R and Haurin, D (2003) ‘The Social and Private Micro-level Consequences of Home Ownership’, Journal of Urban Economics, Vol. 54 (3), pp. 401-50.
Oswald, A (1997) The Missing piece of the unemployment puzzle. An inaugural lecture. Available at: http://www2.warwick.ac.uk/fac/soc/economics/staff/faculty/oswald/inaugura.pdf.
Oswald, A (1999) The housing market and Europe’s unemployment: a non-technical paper. Available at:
O’Sullivan, A and Gibb, K (2012) ‘Housing Taxation and the Economic Benefits of Home Ownership’, Housing Studies, Vol.27 (2), pp. 267-79.