Does solo self-employment serve as a ‘stepping stone’ to employership?

Highlights

• Solo self-employment has a small impact on the probability of becoming an employer.

• The probability of the solo self-employed remaining an employer is small.

• Entry into employership is greater among those with a history of self-employment.

• The size of our key result is similar to that found in a previous German study.

Abstract

This paper examines the extent to which solo self-employment serves as a vehicle for job creation. Using panel data from the Household, Income and Labour Dynamics in Australia (HILDA) Survey, a dynamic multinomial logit model of transitions between labour market states is estimated. The empirical strategy closely follows that used in a previous study employing household data from Germany by Lechmann and Wunder (2017). Estimates of true cross-state dependence between solo self-employment and employership are obtained that are relatively small. Further, the results imply that the probability of a male worker being an employer just two years after transitioning out of solo self-employment is only about 4% (and among women, it is just 2%). The extent of both true cross-state dependence and true state dependence in employership is, however, much greater among individuals who have demonstrated a preference for self-employment in the past. This implies that pro-entrepreneurial policies that target more ‘entrepreneurial’ individuals will have more pronounced and long-term effects in stimulating job creation.

Introduction

Governments in many Western countries devote taxpayer funded resources to supporting self-employed entrepreneurs on the grounds that new start-up businesses are critical for the creation of new jobs. Yet most self-employed workers do not hire anyone. Data available from the International Labour Organization (ILO) website (ilostat.ilo.org/data), for example, indicate that, among developed nations, the percentage of self-employed workers who employ others ranged, in 2018, from just 14% in the UK to 44% in Germany, with the multi-country average being around 31% (and with Australia being above average, at around 35%).¹ Policies that incentivize transitions out of unemployment into self-employment, while obviously reducing unemployment in the short run, may thus have little additional job creation impacts. Indeed, and as noted by Congregado, Golpe and Carmona (2010), such measures, by attracting persons who are not well suited to self-employment, may have employment effects that are only temporary.

For entrepreneurship policies to be successful, in the sense of creating firms that generate new jobs, requires targeting those entrepreneurs who are most likely to transition from working alone (i.e., the solo self-employed, often also referred to as own-account workers) to employing others (i.e., employers). Somewhat surprisingly, however, especially given the many studies seeking to unpack the mechanisms that encourage individuals to enter self-employment (for reviews see Parker, 2009; Simoes et al., 2016; van Praag and Versloot, 2007), there has been relatively little research on the factors influencing transitions out of solo self-employment into employer status.²

Notable exceptions here are a series of studies involving analysis of data from the European Community Household Panel (ECHP) by Millán and colleagues (Congregado, Millán and Román, 2010; Millán et al., 2013, 2014, 2015) – though only one (Millán et al., 2014) provides a direct test of whether solo self-employment has any effect on the probability of becoming an employer in the future – and analyses of household panel data from both the German Socio-Economic Panel (SOEP) (Lechmann and Wunder, 2017) and the UK Household Longitudinal Study (Henley, 2019).

Millán et al. (2014) estimate discrete-time models of the hazard of exit out of both solo self-employment and employership within a competing risks framework. Prior experience is represented by a simple dummy variable indicating whether the respondent reported ever previously working as an own-account worker (i.e., solo self-employed worker), an employer or a paid employee. To the authors’ surprise, prior experience as a solo self-employed worker is not found to influence the probability of transitioning from solo self-employment to employership. At the same time, the estimated probability of switching from solo self-employed to employer is found to be markedly higher than the probability of exiting to other states, which the authors take as a “sign of success” and so conclude that the results are consistent with the notion that there is persistence in entrepreneurship and hence provide support for public investments in new start-ups. We suggest such a conclusion is not warranted. First, it is inconsistent with their finding that previous experience in self-employment is irrelevant, and second, their estimation method does not allow the “true” effect of solo self-employment on employership in the future (true state dependence) to be isolated from the effects of other factors that influence both solo self-employment in the past and employership in the future (spurious state dependence).

Henley (2019) too estimates a discrete-time model of the hazard of exiting solo self-employment into employership, but with the sample restricted to those observed in self-employment.³ Furthermore, he includes a person-specific random error component as a way of controlling for unobserved heterogeneity. In addition, prior self-employment experience is captured with a cumulative measure of years of experience rather than a dummy. In contrast to Millán et al. (2014), a positive relationship between transition into employer status and elapsed duration in self-employment is found, providing arguably stronger evidence for the hypothesis that prior experiences of self-employment have a causal effect on the probability of employing workers in the future. Nevertheless, the methods applied are still problematic. Like Millán et al (2014), the data are censored for cases where the self-employed are never observed becoming employers over the (6-year) data period, and the use of random effects imposes the unlikely assumption that the explanatory variables are uncorrelated with the unobserved error component.

Far more convincing is the estimation strategy employed by Lechmann and Wunder (2017). They employ a very broad sample, which is not just restricted to persons observed entering self-employment, and estimate a dynamic multinomial logit model that controls for both the previously observed employment state and initial conditions as well as unobserved heterogeneity, but this time using correlated random effects. They find that the “genuine effect” of experiencing solo self-employment (relative to wage employment or non-employment) on future employership is both small (about a 5 percentage point increase for men and a little over just 2 percentage points for women) and not long-lasting. As such, these results are discouraging for proponents of subsidies and policy measures designed to encourage individuals to enter self-employment. Of course, this is only one study utilising data from one country, and thus replication both in Germany and in other institutional settings is required before it can be confidently concluded that such measures will typically have little lasting effect on employment.

Providing such a replication is the aim of this study. More specifically, we utilise a similar household panel survey data set to that used by both Lechmann and Wunder (2017) and Henley (2019), but from Australia (a country where self-employment rates have tended to be higher than the average across other industrial nations, and certainly much higher than in Germany), and apply the same estimation strategy used by Lechmann and Wunder. We obtain estimates of true cross-state dependence that are strikingly similar in size to those found in the German study.