Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=

# Startups in Africa

Authors:   
Discussant: None  
Video: https://www.youtube.com/watch?v=PHYYOyrSItw&t=19287s


## Talk (05:21:27 – 05:32:33)

[05:21:27] during the break. Thanks. >> Last presenter, Emmanuel Coronelli.  
[05:21:44] Thank you. Hello everyone. Thanks to the organizer for putting our paper on the program. This is joint work with Tomaso Columbia GSB and Maro and Mariana at AFC was a partner in a sense on the project.  
[05:21:56] You will see I'll talk about startups in Africa. Uh apologies for going fast. Uh it's been a problem regardless of the nine minutes. Uh so high growth tech entrepreneurship or startups have been very much the central attention when it comes to the recent discussion of what  
[05:22:11] matters for the future of growth in Africa. uh policy interest is correspondingly rising very rapidly. So from government uh to developmental institutions to also private investors we've been thinking quite a bit recently as to how to foster or what works and  
[05:22:25] what doesn't for developing a startup ecosystem in this context. Uh problem is that as development economists we have very little to say for these type of firms. Most of the work as we know that we've done has been mostly on micro firms and when we looked at bigger firms we typically look at more traditional  
[05:22:39] sectors. And so what we do in this project is we try try to take sort of like a first step in trying to shed some light with new facts on this type of like new firms that we know surprisingly little about. Um and we're going to focus on finance uh for this project.  
[05:22:53] This is the start of an agenda I think but hopefully it's a start that matters because it's very much central to the policy debate. Um what do we do?  
[05:23:01] Essentially we build new data and today I'm going to give you the the very descriptive facts about it. uh we work with two main broad sets of data that try to capture u very much early stage startups as well as like more advanced  
[05:23:13] startups. So the first uh set of uh data is what we call data on early stage startups. This comes from from a survey an experimental survey that we conducted with the IFC. The international finance corporation is the private sector arm of the World Bank. As many of you uh know  
[05:23:27] uh we did two rounds in 2023 2025. We focus on uh young companies that are looking to grow fast uh that have at least two full-time employees. We reach more than 4,000 of these entrepreneurs across 51 countries with a pretty good  
[05:23:41] response rate for the type of companies we try to to survey. Um then we look at the more what we call established startups. These are startups who received venture financing of of different sorts. But we basically built the near universe of VC deals, venture  
[05:23:55] capital deals that cover the typical global data we've used as well as new data that is from the counterparts that have been sort of developing over the last few years in Africa. We do a lot of manual work uh and then we match this to uh LinkedIn as well as to various  
[05:24:09] website and through manual data collection we try to capture the CV the history of all the founders as well as most of the employees of these firms. So this is the data infrastructure. Um this data infrastructure allow us to study both the demand and the supply as we're going to call it for today. The demand  
[05:24:23] is what do startups want? A lot of the policy makers are struggling over how to provide let's say financing to these type of firms. Do they want equity? Do they want debt? Do we develop local investor ecosystem? Do we try to increase the participation of foreign  
[05:24:36] investors and the like? We take a first step in the absence of like causal evidences to see what the real prevalence of what entrepreneurs want tells us. uh on the the supply side, we're going to study where venture capital is going, who's providing it, who's getting it. Okay. So today, I'm  
[05:24:51] going to try to show you two facts and then I'm going to say something with a simple framework that we use to uh give us a sense of whether these frictions we identify might potentially matter. Uh for the bigger picture, I'm not going to talk about summary stats today. This is just to give you a sense of like we have  
[05:25:05] a lot of descriptives in the paper. This is what the concentration of activity looks like across the continent. Uh it's very concentrated in a few urban apps.  
[05:25:13] uh in particular think of it as four cities Cairo, Lagos, Nairobi and Cape Town accounting for a disproportionate amount of activity much more so than what GDP would predict.  
[05:25:23] So first uh fact I want you to I want to leave you with uh what do startups in Africa want. We're not going to just ask them. We're going to try to develop an experimental setting that's very simple to identify what startups care about and to sort of like how much they trade off  
[05:25:36] different characteristic of a potential investment opportunity or investor. Uh we do this in collaboration with AFC. We create this connective connecting Africa's entrepreneurs experiment which is essentially a matchmaking experiment between firms and investors. And so the  
[05:25:50] idea is quite simple. Uh we basically apply a version as you are very familiar with of the resumeit study without deception. Uh where instead of having let's say firms we evaluate jobseker CVs whose characteristics are randomized. We  
[05:26:04] basically have entrepreneurs evaluating potential investment opportunities as well as investors under the promise that it would be matched to characterist investors or investment opportunities that match the preferences they they they display. Okay. So we're going to do this this type of like the science only  
[05:26:19] work when you have a strong incentive. We think the partner the IFC makes it such the incentive is quite strong in this context. The IFC is essentially the most highly recognized investor in this setting uh for startups as well as like bigger funds and the incentive of  
[05:26:34] actually facilitating the matching between these companies as well as investors is something that we validated in various ways matter quite a bit for these of companies for a number of reasons. Uh then once what we do is that we create a bunch of these random profiles. These are just two examples.  
[05:26:48] Each entrepreneur will see about 10 to 15 of them one at a time sequentially.  
[05:26:52] And basically we randomize characteristics of the investor of the investment opportunities. So you can see whether you're offering $100,000 in debt versus equity. You're seeing whether the money is coming from an investor from Italy, an investor from Nigeria and so on and so forth. All of this is randomized. We apply all the best  
[05:27:06] practices. But what we can see essentially is that these will then translate into a set of dams or characteristics that entrepreneurs might care about. And then we regress on the left hand side a variable for the rating. How much do I want to match with this type of investor investment  
[05:27:20] opportunity on a bunch of characteristics that might predict my preference for the given investor investment opportunity? I'm going to summarize the findings in this one slide. These are point estimates ordering from higher magnitude to lower magnitude and what we see essentially is  
[05:27:34] that there is a massive preference for equity financing. Uh this was not how we selected a sample. We can discuss more during the break but this is what comes up quite strongly. a strong preference for equity. Mostly people entrepreneurs care about deal terms. They don't really  
[05:27:49] care about the identity of the investor. A lot of the discussion is about this foreign versus local type of investors.  
[05:27:53] Entrepreneurs could not care less. We discuss we randomize issues such as mentorship and other type of noninancial support. They don't really care. And if anything, what we see is that there is a preference for local type of support. So what they want and we do various tests  
[05:28:07] to to get deeper into this is that this equity seems to represent something they want which is flexible capital from investors who have a skin in the game who are able to potentially support more actively. Uh so this is finding number one on the demand side. Now let's look at the supply meaning how is venture  
[05:28:21] capital actually allocated across the continent. In this case the data is global. So for for Africa we are even more careful to sort of like build a new data that is not just what off the shelf but for a lot of the other continents other contexts around the world this is  
[05:28:35] still also pretty good and can allow us to have some decent benchmark. So what we see uh and it's going to be essentially one fact I want you to to I want to leave you with is that u in Africa when it comes to funded entrepreneurs we see that Africa has very few funded startups meaning  
[05:28:50] relative to GDP this is an exponential scale we see the low uh amount of funding to startups in Africa relative to GDP much lower in terms of volume and much lower when it comes to how much of the funding comes from local investors.  
[05:29:03] So essentially another way of saying this is that most of the funding in Africa comes from foreign investors. You can cut this in many different ways.  
[05:29:10] When it here I say foreign, we're talking about outside of Africa, but most of the funding for this type of companies is coming from foreign investor much more so than what it seems to be in context that are other emerging markets as well as other developed markets. And you can do it by countries  
[05:29:23] as well. You can see a very very uh strong relationship in relationship with GDP. Uh in addition to the capital being mostly foreign, thanks to the LinkedIn data as well as like the human capital data we construct, we can see who the capital goes to. And what we see is that most of the capital is going to  
[05:29:38] essentially foreign entrepreneurs. We define foreign mostly with where they studied and where they worked. And so what we see is that like about twothirds of entrepreneurs who receive venture financing either studied or have worked abroad. We even spend a lot of time  
[05:29:51] looking at pictures on LinkedIn and so on and so forth about half of the capital goes to white entrepreneurs. We don't have of course nationality. We cannot observe that. This is bigger than what you see in other emerging markets as well as other developed markets. We do this for employees as well and you see a similar pictures. While for  
[05:30:06] example example in context let's say like India you have still quite a few foreign entrepreneurs but mostly local labor in Africa you have mostly foreign entrepreneurs as well as foreign labor.  
[05:30:16] Um so a very foreign ecosystem and so in the last part of the paper we try to think a bit more about what this means for both the size as well as the composition of the sector. Uh we do this with a very simple model uh where we think about what the potential drivers of foreigness might be. So we think  
[05:30:30] about the role of preferences the potential for differential returns of different type of entrepreneurs is a model in which entrepreneurs and investors are matching from different types. We have connections playing a strong role and then we have entry and as well as capital wedges capturing  
[05:30:43] various margins that are left out. Uh we discipline this both with the experiment that for example rules out a lot of preferences for differential type of capital as well as multiple other tests in the paper. For example, what we see is that a lot of the capital is  
[05:30:56] modulated by investor from a certain region investing specifically from to with entrepreneurs that studied or or worked in those specific regions. So connections matter a lot. Then we recover the wedges and we can do simple exercise of counterfactual where we try  
[05:31:10] to say what happens if let's say entrepreneurs from Africa have the same connections to investors from the US similar to what invest entrepreneurs who study in the US would have and so on and so forth. But mostly this tells us where to where to look. Um in this paper  
[05:31:24] essentially we're trying to build some of this new facts. U the story we try to paint with this first paper is to say that startups want equity capital. Most of the equities provided by foreign investors. Mostly this flows to foreign connected type of entrepreneurs. Capital  
[05:31:37] and access friction seem to generate this foreign dominance and this also matters for not just the composition but also the size of the startup sector.  
[05:31:44] That's pretty much it. Thanks again for your time.  
[05:31:56] Okay. >> 250.  
[05:32:03] >> Thank you. Thank you.