Accelerators are an important component of startup ecosystems. Since the establishment of California-based Y Combinator in 2005 – widely considered the original archetype – the accelerator model has spread across the globe, bringing to life hundreds of thousands of startups which might otherwise never have launched.

Many of the core features of accelerators – a time-limited curriculum, usually involving intensive mentoring and peer-learning, together with a selective admissions process – have remained more or less the same since the birth of Y Combinator. But business models have evolved: whereas Y Combinator makes small equity investments into startups, research in the U.K. and globally finds that governments and corporate sponsors often limit their funding to the operational cost. However, Startup Genome research suggests that equity-based accelerators, including those backed or created by seed VCs, generally achieve a greater impact on startups and are more sustainable. This has led us to develop new, more effective models for funding accelerators.  

Objectively Assessing Accelerator Performance

Carefully-designed public support of accelerators can be justified by an increase in the skills of entrepreneurs and an increase in the creation and the performance of startups, in addition to the existence of spillovers: research shows that accelerators within an ecosystem can help connect and coordinate actors (increase Local Connectedness) even beyond the accelerator itself, leading to benefits such as increased VC funding and even increase Global Connectedness. Such public goods provide a classic economic rationale for public subsidies, giving accelerators a place in the toolbox of governments of all political colors.

That said, policymakers looking to subsidize accelerators face a number of questions: how can one tell whether a specific accelerator is adding value? What types of programs should they support? How many are too many?

On the level of individual accelerators, public funders have a responsibility to ensure value for money, but it is often difficult to tell whether a program is creating the desired impact. Impact itself usually has many dimensions, typically encompassing economic value creation, job creation, founder education, and firm longevity – some of which may not be apparent until several years downstream. On occasion, desired impact may also conflict, for example, if a public funder wants to encourage local economic regeneration, while a founder or private investor wants relocation for faster firm growth. Having clarity about their intended purpose is therefore important.

If the program is selective (as most accelerators are), then it is inadequate to compare the performance of a program’s graduates with the performance of startups that have not worked with an accelerator. A robust analysis must control for this selection effect, for example, by using techniques such as Propensity Score Matching or Regression Discontinuity Design. If such analysis is not employed, a likely overestimation of the program’s impact will occur.

Unfortunately, program managers rarely have the time, skills, or incentive to do this for themselves, although the process is much simplified if, at their outset, accelerators consider how they will measure their performance and keep data to use for controls, such as lists of rejected startups.  

An Appropriate Mix of Startup Support

On an ecosystem level, policymakers should design an appropriate mix of startup support, taking into account the development stage of the ecosystem and the number of startups it produces. For example, in the early phase of ecosystem maturity – what Startup Genome terms the Activation phase – there is relatively limited startup experience and low startup output. One of the main efforts should be to grow the number of locally-produced startups. In such an ecosystem, there will be a greater need for attention on early-stage accelerators and pre-accelerators to persuade more prospective founders to take initial steps.

In contrast, later-phase ecosystems have different priorities. In an Attraction-phase ecosystem, the main objective is to drive global resource attraction to increase the size and performance of the ecosystem. Here, the mix of startup support is likely to include more specialist, industry-vertical accelerators, formed in collaboration with corporate partners. Such specialist accelerators typically can serve a larger geographic area than generic programs, can attract startups from all over the world, and are only feasible once an ecosystem has reached a certain stage of development and/or is highly-connected to others.

While there is evidence that accelerators can add value at both the firm and ecosystem level, accelerators cannot operate effectively in a vacuum. They need to interface with local sources of talent, draw on local mentors, connect with local VCs, and be run by experienced program managers. Policymakers should beware the temptation to see accelerators as a ‘silver bullet,’ and instead consider a wider portfolio of interventions tailored to the specific needs of their own ecosystem.

To learn more about startup funding sources, read Friends vs. Family: Who to Raise Funds From.