
Nine judges vote on every application for Tunisia’s startup label. Five votes either way settle a file at once, and the undecided go to an online pitch. That small rule let three World Bank economists compare firms that won the label with near-identical firms that narrowly lost it, and their answer is a rare causal estimate of what a startup law does in Africa. A national startup policy can change what the state itself controls: the cost of a hire, a border formality, the right to hold dollars. It can invite investors and companies who already exist and are waiting for terms. The job totals in a preamble depend on customers, founders and capital, and those sit outside any clause.
Key Takeaways
- In a 2025 World Bank evaluation, Tunisia’s startup label raised firms’ survival by 18 percentage points, from a 53 percent baseline, and added about two workers per firm.
- The gains came through levers the state holds: covered social security, foreign currency accounts and simpler customs. Profits showed no detectable change.
- Italy’s 2012 act helped startups raise equity and debt because Italian banks and investors were there to answer its guarantees and tax credits.
- Senegal adopted its startup law in December 2019 and opened the label desk in November 2025.
- In Uganda two of Tunisia’s three main levers lose force: the capital account has been open since 1997, and with 90 percent or more of employment informal, payroll relief reaches only firms already on the books.
- The Reach Sort files every promise in a draft policy as something the state controls, invites or hopes for, and moves the hopes into the evaluation plan.
What Tunisia’s label changed
Tunisia passed its Startup Act in April 2018, and Smart Capital, a public-private management company, began awarding labels in 2019. The label brings state cover of the full social security contribution (16.57 percent of gross pay from the employer, 9.18 percent from the worker), exemption from corporate tax, simpler customs status, and the right to hold a foreign currency account in a country where most firms need central bank approval to buy foreign exchange. Founders of very young firms can draw a stipend.
Nadia Ali, Massimiliano Cali and Bob Rijkers followed 466 firms that applied between March 2019 and December 2021, through 2022. Their preferred comparison uses the 118 firms on which the judges split in the first round. Among these, labeled firms were 18 percentage points more likely to survive, against a baseline of 53 percent. They added 2.0 workers on average, against a mean of 1.8, and about TND 69,000 (roughly $23,000) a year to their wage bill. Sales rose by an uncertain amount. Profits showed no detectable change.
Every labeled firm received the whole bundle, so the authors read its parts from who gained most: import-dependent, manufacturing and onshore firms, the ones that most needed customs relief and foreign currency. Founders named payroll cover, currency access and customs treatment as the useful parts. The label may have signaled quality to investors, but on the interview evidence “its magnitude is likely to be marginal.”
The cost was mostly payroll. Per firm per year, the state gave up about TND 23,415 in social security, paid TND 10,181 in stipends and TND 1,523 in administration, and lost only TND 659 in corporate tax, because most labeled firms made losses. The authors put the short-run cost at about TND 14,606 ($4,869) per full-time job, and they treat their estimate of 2.68 dinars of benefit per dinar spent as indicative.
Set that against the Act’s own goals: 10,000 jobs and 1,000 labels within five years. The ministry reported 1,018 labels by 31 March 2024, roughly on schedule. If every labeled firm had added the two workers the evaluation measured, the label would account for about 2,000 jobs. That is arithmetic on an average from early cohorts, and it lands at a fifth of the jobs target.
Personal savings and an old exchange code in Tunis
Teheni El Ghak and Najeh Souissi, studying Tunisian founders for the Friedrich Naumann Foundation, found that most startups still struggle with administrative complexity, slow funding, poor coordination between ministries and regulatory uncertainty. Many founders rely solely on personal funds, public and crowdfunding channels are seen as opaque or concentrated in the capital, and outdated foreign exchange laws make it hard to sell abroad.
On their account the Act built a protected class of firms inside a general climate it left largely as it was. The label opens a currency account for its holders, while the exchange code every other Tunisian firm faces stays in force, and the venture money came separately, through a World Bank and German-backed fund of funds starting at about €40 million. One figure in their summary, a fall in new labels from 224 in 2020 to 3 in 2023, conflicts with the ministry’s count of 30 labels in the first quarter of 2024 alone, so I set it aside.
Put the two accounts side by side. The label changed outcomes for the firms inside it, and it changed them through exceptions to national rules the state chose to keep for everyone else. That is the shape of a startup policy’s reach.
Capital answers an invitation where capital already exists
Italy’s Startup Act took effect in October 2012 with tax incentives for equity investors, public guarantees on bank loans and flexible labor rules for registered “innovative startups”. Francesco Biancalani, Dirk Czarnitzki and Massimo Riccaboni found that registered startups raised more equity and more debt and hired more people because of the program. The difference from Tunisia sits on the other side of the table: Italian guarantees and tax credits had banks and investors ready to answer them.
The book under test makes a bolder version of the invitation argument. In Entrepreneurial Ecosystems: Resource Acquisition, Flow and Management (Springer, 2025), Gustavo Barboza and Alessandro Braga of Loyola University New Orleans describe Costa Rica’s free trade zone regime and its investment promotion agency, CINDE. From CINDE’s data they count US-owned firms rising from 6 in 1990 to 297 by 2022, and Costa Rican-owned exporting firms in the regime rising from 4 to 208. They read the second line as knowledge spilling from multinationals into local entrepreneurship.
The policy plainly invited the multinationals. The spillover is a heavier claim, and the authors’ own review records that firm-level studies have cast doubt on FDI spillovers in developing countries; two counts rising together leave the direction of cause open. The chapter also names a dual economy, with labor markets and productivity split between the free zone firms and those serving the home market.
Kenya’s export processing zones give an East African reading. In 2023 they held 170 operating enterprises employing 75,598 Kenyans, and 40.6 percent of the enterprises were wholly Kenyan owned. The zones imported KSh 58.3 billion of goods that year and bought KSh 16.3 billion locally, about three and a half shillings of imports for every shilling of local purchases. The invitation half of the Costa Rican story transfers: zone policy brought in firms, jobs and exports. The spillover half, which the authors tie partly to Costa Rica’s educated bilingual workforce, shows up thinly in Kenya’s purchase figures.
Senegal’s law, six years from vote to label desk
Senegal’s National Assembly adopted its startup law on 27 December 2019, and the implementing decree was signed in 2021. In December 2022 a commentary on OSIRIS, a Senegalese observatory of the information society, described the law as still on standby, with its guarantee fund out of reach, under a headline asking the country’s one labeled startup to raise its hand. The same piece credited Senegalese startups with raising more than $222 million in 2021. The money moved while the law waited. The operational launch came on 20 November 2025: published label criteria, an online one-stop desk, and targets of 500 labeled startups and 50,000 direct jobs.
Nigeria’s Startup Act was signed on 19 October 2022. Its labeling portal opened in November 2023, and by April 2024 the National Information Technology Development Agency said 12,948 startups had registered for labeling. Registrations measure appetite for the promised incentives. An evaluation would measure what they did, and for Nigeria I found none published.
A law starts causing things on the day an office can approve a named firm and a benefit reaches its payroll. The label desk, the portal and the guarantee fund are the instruments. The gazette notice is a promise about them.
Uganda opened its capital account in 1997
Run Tunisia’s three main levers through Ugandan conditions and two of them weaken.
Foreign currency first. In Tunisia a labeled firm’s currency account is a real privilege. The Governor of the Bank of Uganda recounted in May 2025 that Uganda opened its foreign exchange market and had a fully liberalized capital account by July 1997. A Ugandan label would have little to exempt firms from.
Payroll second. Uganda’s National Social Security Fund takes 10 percent of wages from the employer and 5 percent from the worker, and the amended Act requires every employer to register, whatever its headcount. Covering those contributions would lower the cost of a formal hire, as in Tunisia, and would reach only firms already paying. Across Uganda’s 2018/19 and 2021 labor force surveys, 90 percent or more of employment was informal in every geographic area.
Tax third. Tunisia lost about TND 659 per labeled firm a year in corporate tax, because young firms mostly make losses. A profit-tax holiday is cheap and changes little, in Tunis or in Kampala.
What transfers is the method: aim each instrument at a national barrier that binds for young firms, and measure it against a comparison group. A Ugandan act would have to find its own binding barriers; what its text should carry is argued in Why Uganda Has No Startup Act, and What One Should Say.
Joseph’s fifth and the seven years
Pharaoh asked Joseph for an interpretation and received a budget. Joseph’s first words drew the boundary: “I cannot do it,” he told Pharaoh, “but God will give Pharaoh the answer he desires” (Genesis 41:16). The seven good years and the seven lean ones were “firmly decided by God” (41:32). What Joseph then proposed sat wholly inside the administration’s reach: appoint commissioners, take a fifth of the harvest in the good years, store the grain in the cities (41:34-35). The rain belonged to God and the fifth to Pharaoh. A secular reading of that division rests on credibility: promise what you control and you will be believed. Genesis adds a second boundary that credibility has no reason to draw. Years later the same granary bought every field in Egypt for Pharaoh, and Joseph “reduced the people to servitude” (47:20-21). A lever that works keeps pulling. Humility about outcomes keeps a preamble honest, and restraint about levers keeps the founder’s field her own. The fifth fed Egypt. The fields went to Pharaoh.
The Reach Sort
The Reach Sort is a way to read any draft startup policy, preamble first. Take each promise in it and file it in one of three places.
Controls. Things the state does with a signature and a budget line: a payroll contribution it covers, a fee it waives, a customs status, the date a label desk opens. These can carry dates and be measured directly. Tunisia’s working levers lived here, and so did Senegal’s six-year gap.
Invites. Outcomes that need another party to say yes: an investor answering a tax credit, a bank answering a guarantee, a multinational answering a free zone. Write the terms, name the party expected to respond, and check that it exists at the scale the clause assumes. Italy’s equity effect lived here.
Hopes. Totals that depend on customers, founders and luck: 10,000 jobs, 50,000 jobs, regional hub status. Move them from the preamble into the evaluation plan, with a comparison group like Tunisia’s pitch-stage firms, so a later reader can separate what the policy caused from what the economy did anyway.
Cornwall carries a warning for anyone tempted to count a building as an outcome. Daniel Prokop’s chapter in the same book describes a technology park opened there in 2018 with European regional development money, and adds: “The success of it is yet to be established.” The park belongs in the first bin. Its success belongs in the third.
By this sort, AVODA’s 600+ trained and 100+ startups supported are controls: they record what we delivered. A claim about what that work caused would need a comparison group, and we would have to build one before making it.
Frequently Asked Questions
Does a national startup act create jobs?
It can, at a modest scale. In the World Bank’s evaluation of Tunisia’s Startup Act, labeled firms added about two workers each and were 18 percentage points more likely to survive than comparable rejected applicants. The gains came mainly through covered payroll contributions and easier currency and customs rules.
Why did Senegal’s Startup Act take so long to work?
The law was adopted in December 2019, and its instruments came years later. In December 2022 an OSIRIS commentary described it as still on standby, with its guarantee fund out of reach. The operational launch, with published label criteria and an online one-stop desk, took place on 20 November 2025.
Would Tunisia’s startup label work in Uganda?
Parts of it would. Covering social security contributions would lower the cost of formal hires, yet 90 percent or more of Ugandan employment is informal, so the benefit would reach few firms. The foreign currency privilege carries little value in Uganda, which has had a fully liberalized capital account since 1997.
What should a national startup policy promise?
Promise what the state controls, with dates: fees, payroll relief, permits and a working label desk. Offer terms to investors and banks where they exist at the scale the clause assumes. Put job and turnover totals in the evaluation plan, measured against a comparison group, so the preamble holds only commitments the state can meet.
Next Step
Drafters of a startup policy, and the hub leads, investors and bank officers invited to comment on one, can test a draft on a strategy call with Joshua. We go through it promise by promise, file each one against the Tunisian, Italian and Senegalese record, and mark the job targets that belong in the evaluation plan. Doing it while the draft is still open to comment spares the final act a preamble its instruments cannot keep.
Related Reading
- Why Uganda Has No Startup Act, and What One Should Say
- Special Economic Zone Tax Expenditure Audits: What the Holidays Actually Bought
- How Governments Should Buy Acceleration
- Tax Amnesties and the Formalization Bargain: When Forgiveness Is Policy
Sources and Evidence
- Ali, Cali and Rijkers, “Promoting Innovative Startups: Quasi-Experimental Evidence from Tunisia,” World Bank Policy Research Working Paper 11117, May 2025: label benefits, selection design, survival, employment, wage bill, profit and cost-benefit estimates, the Act’s targets.
- Kapitalis, “Tunisie : 1018 entreprises bénéficient du label Start-Up (au 31 mars 2024)”: 1,018 labels by 31 March 2024, 30 in the first quarter, and the fund of funds.
- Friedrich Naumann Foundation, “Tunisian Startups and Investment Climate: An Exploratory Qualitative Study,” El Ghak and Souissi: founders’ funding, administrative and foreign exchange barriers under the Act.
- Biancalani, Czarnitzki and Riccaboni, “The Italian Start Up Act: a microeconometric program evaluation,” Small Business Economics, 2022: effects of Italy’s act on equity, debt and employment.
- Baig, Montoya-Martinez, Belitski, Theodoraki and Godley (eds.), Entrepreneurial Ecosystems: Resource Acquisition, Flow and Management, Springer, 2025: the book tested here.
- Barboza and Braga, “Foreign Direct Investment and Multinational Enterprises’ Impact on Local Entrepreneurship Development via Knowledge Spillover Effect: The Transformational Experience of Costa Rica,” chapter 8: free trade zone regime, CINDE firm counts, spillover caveat, dual economy.
- Prokop, “The Economic Development Challenges of Coastal Entrepreneurial Ecosystems: The Case of Cornwall and Isles of Scilly,” chapter 2: the 2018 technology park and its unestablished success.
- OSIRIS, “Sénégal : La loi sur les Startups adoptée,” January 2020: adoption of the law on 27 December 2019.
- OSIRIS, “Startup Act : Que la seule startup sénégalaise labellisée lève la main !”, December 2022: the law on standby, the guarantee fund, the 2021 decree and the 2021 fundraising figure.
- OSIRIS, “Startup Act : le Sénégal entre dans la phase opérationnelle le 20 novembre,” November 2025: the label desk launch and its targets.
- Nairametrics, “Nigeria Startup Act: NITDA says 12,948 startups now registered,” April 2024: registrations for labeling and of investors and hubs; portal opened November 2023.
- Nairametrics, Startup Support and Engagement Portal launch, November 2023: the Act signed on 19 October 2022.
- Bank for International Settlements, Michael Atingi-Ego, “From financial liberation to economic transformation: Uganda’s journey toward prosperity,” May 2025: Uganda’s capital account fully liberalized by July 1997.
- Kakuru Advocates, “NSSF Contributions Mandatory for all Formal Entities and Employers,” May 2023: contribution rates and registration of every employer under the amended Act.
- Koolwal and Vanek, “Workers in Informal Employment in Uganda: A Statistical Profile, 2018-2021,” WIEGO Statistical Brief No. 41: 90 percent or more of employment informal in all areas.
- Kenya Export Processing Zones Authority, Annual Performance Report 2023: enterprises, employment, imports, local purchases and ownership.
- Genesis 41:16, 32-36 and 47:20-21, NIV: Joseph’s answer to Pharaoh, the fifth, and the land bought for Pharaoh.
