Homeownership and new entrepreneurs: first empirical evidence of a bidirectional interaction

ABSTRACT

Literature linking housing and professional choices focuses on the supply side of the labour market (the search for a job as employee), thus considering the unemployment rate as the only measure of labour market outcomes. Indeed, the usually neglected link between homeownership and entrepreneurship is equally important for assessing labour market performance. This empirical article represents one of the first attempts to test the interplay between homeownership and (new) entrepreneurs. Effectively, using a panel data analysis in Italy, we find a first empirical evidence of a positive and bidirectional interaction between homeownership and new entrepreneurs (precisely, small businesses and the self-employed).

KEYWORDS:

• Homeownership
• new entrepreneurs
• small businesses
• self-employment

JEL CLASSIFICATION:

• C33
• J64
• L26
• M13
• R21
• R31

Acknowledgments

The author wishes to thank both the anonymous referee and the Editor, Professor Mark P. Taylor.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

¹ The ‘Oswald puzzle’ refers to the discrepancy between the results obtained at the macroeconomic and microeconomic levels. Indeed, at the macroeconomic level, the empirical evidence points out a positive relation between homeownership and unemployment rates; while, at the microeconomic level, many empirical studies report that homeowners have better labour market outcomes than tenants (for a comprehensive survey of the related literature, see Havet and Penot 2010).

² Unemployment refers to searching for a salaried job and is an alternative to creating its own job or opening its own business, thus posting its own vacancy.

³ Bracke, Hilber, and Silva (2014) find a strong and statistically significant negative association between homeowners with a mortgage and business start-up (whereas the negative correlation between outright homeownership and new business is small and insignificant). Instead, Harding and Rosenthal (2013) find that housing capital gains encourage self-employment.

⁴ Both of them draw opposite conclusions to Oswald’s hypothesis; namely, they find a negative and significant correlation between unemployment and homeownership.

⁵ Garcia and Hernandez (2004) and Coulson and Fisher (2009) make use of the three-stage least squares procedure to test the bidirectional interaction between homeownership and unemployment. Both works, however, use cross-sectional data (Spanish data and U.S. data, respectively). In a previous and related work, Coulson and Fisher (2002) used microeconomic data and a single equation.

⁶ We use the so-called ‘fixed-effects model’. This empirical specification has a very useful property with respect to the random-effects model. If the unobservable specific effects represent omitted variables and the cross-section effects are correlated with the other explanatory variables of the model, the fixed-effect estimator remains consistent, while the random-effect estimator becomes inconsistent (Judson and Owen 1999). Obviously, in Equations (1) and (2), we cannot add other variables that do not vary over time, since they will be perfectly colinear with the fixed effects. Also, in panel applications characterized by a small number of temporal observations, it is often preferable to use the more simple one-way individual specification (Baltagi 2008).

⁷ An increase in ‘actual’ taxation also occurs when the reduction in the official GDP is greater than the reduction in tax revenue.

⁸ The Error Component Regression Models assume that the regression disturbances are homoscedastic, with the same variance in time and between individuals. This hypothesis, although restrictive in some cases, will determine (in the case of heteroscedasticity) estimators that are always consistent but inefficient (Wooldridge 2001; Baltagi 2008). The same can be said about serial correlation.

⁹ The shadow economy could concern tenancy but not homeownership.

¹⁰ The production scheme that focuses on undeclared work and the firm that operates simultaneously both in the regular and in the underground economy are termed as ‘moonlighting production scheme’ and ‘moonlighting firm’, respectively (Busato et al. 2005; Ciccarone, Giuli, and Marchetti 2012).

¹¹ At the moment, specific Stata routines that are able to carry out the 3SLS procedure with panel data do not exist. An alternative is the use of the Stata command ‘xtdata’ that produces a transformed data set of the variables specified in varlist or of all the variables in the data. Precisely, the option ‘fe’ converts data into a form suitable for fixed-effects (within) estimator. Once the data are transformed, the usual Stata command ‘reg3’ may be used. For more details, see: http://www.stata.com/manuals13/xtxtdata.pdf.

¹² Instead, the positive effect of the number of successfully attained mortgages on the number of homeowner households is very intuitive.

¹³ Instead, offensive entrepreneurship is an alternative to employment, since the so-called ‘offensive entrepreneurs’ start a business activity despite having better employment prospects or being already employed (see, e.g. Lisi (2017a) and references therein).