Here are some highlights from last Friday’s joint What Works Centre / SERC seminar, which was given by Prof Rob Fairlie from UC Santa Cruz. The talk was part of the Spatial Economics Research Centre’s regular series of seminars, and the first in an occasional series of collaborations where we get the chance to dig into an innovative piece of local growth policy.
Rob’s talk looked at policies to promote entrepreneurship, and focused on whether training programmes for entrepreneurs can help these new businesses grow their workforce. This taps into some major policy concerns in the UK and elsewhere. Since the Great Recession, US and UK policymakers have been very keen to push entrepreneurial activity as a way to generate economic growth. Existing US evidence suggests that small and young firms account for the majority of new jobs generated; and most of that job creation comes from a very small number of those firms (although we should bear in mind that at least some of these jobs created come at the expense of jobs lost in incumbent businesses).
So are there policy interventions that can a) help move people into entrepreneurship and b) help those new businesses grow?
One cluster of interventions is accelerator and incubator programmes, in the tech sector and more broadly, which select a small number of participants for intensive training. In the States, entrepreneurship training has also been applied as part of active labour market policy, with programmes that aim to help long term unemployed people set up their own businesses. We covered some of these programmes in our evidence review on business advice. Rob’s talk focused on one of the largest of these programmes, Project GATE, which was set up as an RCT covering over 4,000 people, of whom over 50% were unemployed at the time. Project participants received a package of training, classes and seminars and 1:1 counselling / advice on setting up and growing a business.
A key question for policymakers is whether new entrepreneur-owned businesses generate additional employment, and where this comes from: an unemployed person who becomes a business owner has generated a job (for themselves), which is a success from a ‘welfare to work’ point of view. However, from a local growth perspective, what we really want to know is whether these firms also hire additional employees.
The conclusions from the evaluation are not hugely encouraging. There are positive but not statistically significant effects of the programme on hiring a new employee, and the training did not have an effect on the number of employees hired.
What might explain this?
The RCT results are average effects, so it is possible that the training might be more effective for firms in specific industries, or for specific types of people (the paper also uses descriptive evidence from the Kauffman Firm Survey which suggests quite a lot of heterogeneity in hiring patterns across different types of entrepreneurs).
Another issue is that this particular programme focused on moving participants into business ownership, not on growing workforces; many participants reported that the hiring advice they received wasn’t terribly helpful, and the number of hiring firms was low, making inference difficult.
A third issue is that over 50% of the control group also ended up getting some kind of entrepreneurship training. As one of the seminar participants pointed out, although they spent less time on training and reported lower satisfaction than the treatment group, it’s possible that a sub-section of the control group ended up more motivated – and likely to succeed – than the treatment group.
So what can policymakers take from all of this? First, be wary about claims that entrepreneurship training is a silver bullet for local growth. Second, try to set up some experiments so we can properly test such programmes in a UK context. Third, be very clear about what your entrepreneurship training programme is designed to achieve, and make sure that the policy design and evaluation strategy are able to pick this up. Finally, make sure that participant and performance data is available to researchers. Initial evaluations of Project GATE suggested it had positive impacts, leading the US Department of Labor to spend large sums on further rollout. After several years chasing down the data, Rob and colleagues found mistakes in the original studies and concluded the actual effects were substantially smaller than claimed.