Syndicate content

Should Development Organizations be Hunting Gazelles?

David McKenzie's picture

Last week I presented some early findings from ongoing work at the IADB, at Innovations for Poverty Action’s SME initiative inaugural conference. There was a lot of interesting discussion about early results from efforts to improve management and skills in small and medium firms, discussion of the most appropriate ways of financing these firms and the extent to which a personal vs automated approach to determining creditworthiness can be used, and an interesting panel on policies towards the missing middle. However, the one theme that has got me thinking the most is something that seems to come up a lot in discussions of microenterprise development and SME programs recently, namely should development institutions and policymakers be directing fewer resources at microfirms and more at high-growth-potential enterprises or gazelles?

Gazelles are defined by the OECD to be all enterprises up to 5 years old with average annualised growth greater than 20% per annum, over a three year period, and which have 10 or more workers. Recent work in the US, and looking at firms around the world have emphasized the role of a subset of dynamic, fast-growing young firms in net job creation, leading to policymakers and practioners focused on job creation to think we should be devoting more effort to identifying and supporting these gazelles, and decrying the lack of venture capital markets in developing countries. For example, see this scoping note by Tom Gibson and Hugh Stevenson at the IFC.

I have had mixed feelings about such efforts for a while, mostly because we don’t really know whether we have policies which can succeed. In order to succeed in helping these high-growth enterprises we need to be able to do two things:

(a)    Identify who these gazelles are in advance: some progress is being made in this front – Chris Woodruff presented work from a business plan competition in Ghana showing that both survey questions and a panel of experts were able to judge which firms would on average grow faster over the subsequent year or so. In this NBER chapter, Chris, Suresh de Mel and I use data from Sri Lanka and species classification techniques from biology to separate micro-firm owners who look more like SME owners from those who are more like wage earners. Building on these ideas, the entrepreneurial finance lab at Harvard is testing the role of psychometric testing in identifying better growth prospects for lenders. There is still much to learn, and even venture capitalists in Silicon Valley identify more duds than they do successes, but there is reason to be hopeful we can have some success in targeting here.

(b)    But then we also need to have policies which help these firms: this second step is one that I think is overlooked in the rush to jump onboard success stories. My recent insight on this is that this is a step that matters to development institutions much more than to venture capitalists. If I find the next Google (step A) and make some investments in it, it is second-order to me whether my investments or advice really change the trajectory of this firm from great to greater – I earn a return even if my advice and money don’t help the firm at all. That is, my claim is counterfactuals don’t matter much to venture capitalists, but they should to development institutions and Governments. With resources scarce, the justification for investing IFI or Government resources in policies targeted to gazelles has to be that it helps them overcome some market failure – such as lack of insurance against the high risks they take on perhaps, or informational and credit constraints that prevent them from hiring the consultants they need to get their business growing at full potential.

What are the implications for this?

The consequence is that I think development institutions and Governments ought to be concerned about marginal returns, not absolute returns. This has implications for targeting of these programs and for measuring their impacts:

•    We therefore want to know how the impact of various programs varies over the ability/gazelleness-type distribution.

•    Once we know this, it may be that we want to target our programs at firms which don’t have the highest likelihoods of being gazelles, but rather at those who will only succeed with our help. So if the Government is having a business plan competition to select winners to receive grants and consulting advice, a somewhat radical proposition might be to score all the firms, and then NOT give the prizes to those with the highest scores, but perhaps those somewhere in the middle of the distribution who are good enough that they have a base that help can build on, but not so good that they will succeed anyway.

Currently we are far from being able to target such programs well and make such recommendations, as the evidence base on whether any of these government programs design to catalyze gazelles in developing countries work is very limited. This makes it crucial to build rigorous impact evaluations around ongoing efforts to conduct such programs – something we are doing through the DIME-FPD cluster at the World Bank and the IPA SME initiative is also working on. So stay tuned…

Comments

Submitted by Arvind Gupta on
If you want to hunt then be ready to bear the credit risks as well as the costs and consequences of missing the Gazelle. Are we willing to give up our preferred creditor status and sovereign guarantees?

I am so happy to see these questions of job creation hit the research agenda. As I have noted on my site, it's highly likely that the effect of SME financing and BDS on job creation is greatly affected by firm and market characteristics, e.g., agriculture v. tech, local cost of capital, etc. I am skeptical that an RCT would be able to show whether or not SME finance "works" for job creation unless it takes these traits into account, as the average treatment effect simply would not be very meaningful. I do think a smartly designed RCT could test a number of different contexts and client types to understand where (if anywhere) and what context it works. Like in medicine, the challenge is to know how the treatment effect varies by patient, so you understand how to best leverage scarce resources. Unfortunately, many are put off by the idea that research will show their treatment "doesn't work" rather than excited by the idea of better targeting their resources for maximum impact. This post really hit home with my discussion note (see the link for my "homepage") on "Cost-Effective Impact Assessments for the Impact Investor," which may be brazenly titled as what I propose certainly would not substitute for actual impact evaluations, but could be a useful complement by integrating a counterfactual into organizational performance monitoring. I think this is key for the birth of any "social impact" market, and speaks directly to your point on "counterfactuals don’t matter much to venture capitalists, but they should to development institutions and Governments. " While impact evaluations can validate the assumptions that underpin the theory of change of SME financiers, for example, I would like to see organizational performance monitoring plans validate that the organizations are indeed creating incremental value for their clients.

Submitted by T.Althefery on
It is worth noting that this aspect of developmental institutions is very similar to that of conventional financial institutions. However, FDI's need to balance between the return on their investment whilst making the most developmental impact in the sector. This can be better performed by engaging the entrepreneur through a partnership agreement versus a financing agreement. This is what we try to do in our funds.

Add new comment