Every second week for about two and a half months, groups of about sixty entrepreneurs from around Kampala met for roughly three hours at a partner institution’s premises and talked through one another’s business ideas. Researchers had split 568 Ugandan applicants into three groups at random: one met in person, one met online, and one carried on with the course alone. When the course asked for a business proposal, 34 percent of the control group submitted one, and the face-to-face group submitted at a rate 12.6 percentage points higher. The online group’s gain was too small to separate from chance. Community, in that study, behaved like an input: specified, scheduled, costed, and measurable against a control.
Key Takeaways
- In a randomized trial with 568 entrepreneurs near Kampala, fortnightly face-to-face peer meetings raised business-plan submission by 12.6 percentage points over a control group at 34 percent; the gain from online peer groups was too small to separate from chance.
- Savings groups are community built into a financial institution. A trial across 561 village clusters in Ghana, Malawi and Uganda found modest but real gains in household business activity, with no change in average consumption.
- Community helps most where demand gives people a reason to share. Ghanaian garment makers trained in a new technique taught few peers until orders for it rose.
- Clevenger and Fortunato’s Entrepreneurial Communities and Ecosystems treats community as the cultural ground an ecosystem grows from. In Uganda the bonding ties are already thick; the scarce input is the bridge between strangers in the same market and the link from the room to customers.
- An ecosystem budget can specify community the way it specifies a trainer: who meets, how often, what demand it serves, who pays, and which outcome is checked on which date.
Shopkeepers who buy from each other
Morgan Clevenger and Michael Fortunato, editors of Entrepreneurial Communities and Ecosystems: Theories in Culture, Empowerment, and Leadership (Routledge, 2022), draw a distinction that most ecosystem reports blur. An ecosystem, in their reading, is the working network of founders, funders and support organizations aimed at productive output. A community is the wider social body underneath it, the culture that decides whether an ecosystem can emerge at all.
Their evidence comes mostly from American towns. Fortunato’s earlier research compared rural places with high and low rates of entrepreneurship. In the low-rate places, founders were individualistic, treated each other as competitors and resisted collaboration; some residents called entrepreneurs “crazy people” for leaving a stable job. In the high-rate places, founders met in ordinary gathering spots and bought from one another’s businesses. The editors add a city case: Atlanta had the universities, capital and support organizations, yet a 2014 study by Dan Breznitz and Mollie Taylor found the cluster underperformed because local culture lacked routine interaction and “an ethic of mutual support.”
Their Table 2.1 sorts small-town practices into beneficial, neutral and deleterious columns. Regular founder meetups sit in the beneficial column. Founders who treat one another as rivals for a fixed market share, and who will not share information, sit in the deleterious one.
The editors are candid about their own tradition. The community literature, they write, “paints a very rosy picture of limitless collaboration with very little friction.” That admission is the right place to start testing the book in East Africa, because the friction is where the evidence lives.
Where a Kampala market and an American small town part ways
The book’s deficit is cultural esteem: towns where starting a business looks reckless and people rarely gather. A Kampala trader would recognize few of those symptoms in the form the book describes, and one kind of community institution is plainly widespread. According to the FinScope Uganda 2023 findings, as reported by Business Times Uganda, three in ten Ugandans belong to a village savings and loan association, and community-based groups hold an estimated Ushs 3.4 trillion in gross savings a year, about 9 percent of all savings across every channel.
So the American prescription, convene people and change the story about entrepreneurship, half transfers. Bonding ties among kin, neighbors and congregations are thick. The thin ties run between strangers in the same trade, between a stall owner and a buyer three districts away, and between the group and anyone holding growth capital. The same FinScope coverage reports that members borrow mainly for medical bills and school fees, which is community working as a household safety net more than as a business engine.
The Table 2.1 judgment about rivals also reads differently on a Kampala street where many shops sell the same goods. Refusing to share information with the shop next door may be a sensible reading of a crowded market rather than a cultural flaw. The evidence below suggests both readings hold, depending on who sits in the room.
Sixty strangers near Kampala
The trial in the opening is the Kampala arm of a pan-African experiment by Fernando Vega-Redondo, Paolo Pin, Diego Ubfal and five co-authors, covering almost 5,000 entrepreneurs from 49 countries enrolled in an online business course. Only the Ugandan sample could compare meeting in person with meeting online, because only there could the researchers convene groups physically.
The pattern favored the room. Face-to-face groups raised submission by 12.6 points and raised a combined measure of submission and graded quality as well. In the five largest countries, Ghana, Kenya, Nigeria, South Africa and Tanzania, online groups of compatriots added 3.6 points. Online groups that mixed nationalities showed an effect indistinguishable from zero there, and in the 44 smaller countries they lowered submission by 5.7 points.
Two cautions belong beside those numbers. Submitting a plan is an early outcome, well short of revenue or jobs, and among plans that were submitted, the face-to-face groups did not score significantly higher. The peers moved people to finish; whether finishing leads to firms is a later question. The design still answers the budget question directly: the same course, with and without a room, produced different results.
A larger study outside Africa adds the revenue line. Jing Cai and Adam Szeidl randomized 2,800 young Chinese firms into small groups whose managers met monthly for a year. Revenue rose 8.1 percent, the gains persisted a year after the meetings ended, and managers shared business information most readily with peers who did not compete with them.
Trust that lends and insures
Community also works as a financial institution. Dean Karlan, Beniamino Savonitto, Bram Thuysbaert and Christopher Udry evaluated savings-group promotion across 561 village clusters in Ghana, Malawi and Uganda. Treated villages showed a 0.06 standard-deviation rise in a business outcomes index, 6.3 percent more businesses per household and 8.6 percent more months of operation. Average consumption did not move. In Uganda, 85 percent of members had taken a loan from their group by the end of the study.
Kenya supplies the insurance version. William Jack and Tavneet Suri found that when households were hit by a shock, consumption fell 7 percent for those without M-Pesa while users held steady, because money arrived from more people in their network. The network existed before the phone; cheap transfers turned it into working insurance.
Both studies report small effects, and both are useful for that reason. They put a size on community as a factor of production, and the size is modest and depends on a mechanism that lowers the cost of trusting someone.
Know-how moves when the orders arrive
Morgan Hardy and Jamie McCasland designed a new weaving technique, trained a random set of Ghanaian garment makers inside a real network of firm owners, and watched who they taught. The result cools any warm reading of community. Trainees adopted it a little and taught almost nobody. Then the researchers created demand for the technique. Trained owners who received orders taught roughly 400 percent more of their peers.
On that evidence, know-how between rivals moves when someone is paying for the result. The book’s founders who treat each other as rivals are often responding to the market they are in, and a meetup will change little about that. Cai and Szeidl’s finding points the same way from China, where information flowed most freely between non-competitors. A builder who funds a monthly mixer for twenty tailors selling to the same buyers should expect handshakes and very little transfer.
The concession has limits. Hardy and McCasland’s network was a set of competitors with one technique at stake; the Kampala groups mixed trades and stages and still moved people to finish. Community without demand moves little, and community joined to demand moves a lot.
The Input Specification
If community is an input, an ecosystem budget can describe it with the same precision it gives a trainer or a grant. The Input Specification asks five things of any line that claims to build community.
| Row | What the budget states | What the evidence suggests |
|---|---|---|
| Membership | Who is in the room, and whether they compete for the same customers | Mixed trades and non-rivals share more; Cai and Szeidl, Hardy and McCasland |
| Cadence and form | How often, how long, in person or online | Fortnightly, about three hours, face to face beat online in Kampala |
| Demand hook | The buyer, order, tender or finance product the group exists to reach | Diffusion followed demand in Ghana |
| Owner and cost | Which line pays, who convenes, what it costs per member per meeting | A specified line survives budget cuts better than “refreshments” |
| Outcome and date | One measurable result and the date it is checked against a comparison | Submission at ten weeks in Kampala; revenue at a year in China |
The five rows need no new data, only decisions the hub already makes. Suppose, purely as an illustration, a hub runs a monthly evening for food processors. The specification would name the membership rule, keep meetings in person, attach each session to one buyer or a supermarket’s supplier requirements, cost it from the program line, and check a stated outcome, such as the number of members who completed a supplier application, after six months against applicants who joined later.
Ecclesiastes states the economics before it states the comfort: “Two are better than one, because they have a good return for their labor.” The Teacher puts the return first and the warmth second, the same order in which the Kampala and Ghana results arrived. The weavers also mark where an account built on incentives runs out. Demand can make sharing profitable for a season, and it cannot turn the tailor next door into someone you would lift up when she falls. The next verse is written for that season: “If either of them falls down, one can help the other up.” Buyers cancel orders. The verse was written for the month after they do.
Frequently Asked Questions
Does community actually improve business outcomes in Africa?
Experimental evidence says yes, in measured amounts. Fortnightly in-person peer groups near Kampala raised business-plan submission by 12.6 percentage points over a 34 percent baseline. Savings-group promotion in Ghana, Malawi and Uganda raised household business activity modestly. Effects depend on face-to-face contact, compatible members and a reason to share.
Are online founder communities as effective as in-person ones?
In the Kampala sample, online groups showed no significant gain while in-person groups did. In five larger African countries, online groups of compatriots added 3.6 points, and online groups mixing nationalities lowered submission by 5.7 points in smaller countries. Online formats can supplement a room, and they replace it poorly.
Why do competing entrepreneurs refuse to share knowledge?
Often because sharing costs them customers. In Ghana, garment makers trained in a new technique taught few peers until demand for it rose; then trained owners taught about 400 percent more. In China, managers shared most with non-competitors. Builders get more transfer by mixing trades and attaching groups to real buyers.
How should an ecosystem builder budget for community?
Write it as a specified input: who is in the room, how often it meets and in what form, which buyer or finance product it serves, which line pays for it, and which outcome is checked on which date. A line written that way is easier to defend to a funder than “events and refreshments.”
How is an entrepreneurial community different from an entrepreneurial ecosystem?
Clevenger and Fortunato treat the ecosystem as the working network of founders, funders and support organizations aimed at productive output, and the community as the wider social body whose culture decides whether that network can form. In Uganda, the wider community is dense; the missing pieces are links between strangers and to demand.
Next Step
Before next year’s budget goes to a funder, rewrite its community line as the five rows above, with one outcome and one date. Argue the outcome row out loud with Joshua on a strategy call first, since that is the row a program officer will press on, and a line that can be tested is harder to cut.
Related Reading
- Community Over Networking
- Chama Savings Groups: Covenant Economics Africa Built
- Co-opetition for Ecosystem Builders: Accelerators, Funds, and BDS Providers
- Cohort Peer Effects: The Evidence for the Room
Sources and Evidence
- Morgan R. Clevenger and Michael W-P Fortunato (eds.), Entrepreneurial Communities and Ecosystems: Theories in Culture, Empowerment, and Leadership, Routledge, 2022: community versus ecosystem, rural high- and low-entrepreneurship places, the Atlanta case, Table 2.1, the “crazy people”, “ethic of mutual support” and “rosy picture” quotations (DOI 10.4324/9781351045711).
- Vega-Redondo, Pin, Ubfal, Benedetti-Fasil, Brummitt, Rubera, Hovy and Fornaciari, “Peer Networks and Entrepreneurship: A Pan-African RCT”, IGIER Working Paper 648, 2019: Uganda sample of 568, fortnightly face-to-face groups, submission effects of +12.6, +3.6 and minus 5.7 points by arm.
- Karlan, Savonitto, Thuysbaert and Udry, “Impact of Savings Groups on the Lives of the Poor”, PNAS, 2017: 561 clusters in Ghana, Malawi and Uganda; business index, business count and months of operation; no average consumption effect; 85 percent of Ugandan members borrowed.
- Business Times Uganda, “How Village Savings Associations are Boosting Uganda’s Local Economies”, 16 September 2024: reports FinScope Uganda 2023 figures on VSLA membership, Ushs 3.4 trillion in annual community-group savings and main loan uses.
- Jack and Suri, “Risk Sharing and Transactions Costs: Evidence from Kenya’s Mobile Money Revolution”, American Economic Review, 2014: 7 percent consumption drop after shocks for non-users, none for users, via remittances from more senders.
- Cai and Szeidl, “Interfirm Relationships and Business Performance”, NBER Working Paper 22951: 2,800 Chinese firms, monthly meetings, 8.1 percent revenue gain, persistence and sharing among non-competitors.
- Hardy and McCasland, “It Takes Two: Experimental Evidence on the Determinants of Technology Diffusion”, Journal of Development Economics, 2021: Ghanaian garment makers; little diffusion after training, about 400 percent more peer teaching after a demand shock.
- Ecclesiastes 4:9-12, New International Version, Bible Gateway: “a good return for their labor” and the verse on lifting a fallen companion.