In 2016, a Californian startup facing regulatory constraints on drone delivery in the United States signed a deal with the government of Rwanda. Within months, Zipline was delivering blood to rural hospitals [1], and Rwanda had a technology story that governments across Africa have wanted to repeat ever since.
It is a compelling story. Nevertheless, it also reveals a problem in how technology policy is often judged. The signing ceremony, the pilot project and the foreign company opening a local office are frequently announced as if they were the outcome. They are only the beginning.
Three things are routinely confused. A foreign company entering a market is one outcome. Its technology being successfully deployed is another. Local firms and workers gaining new knowledge and capabilities because it was there is a third. None follows automatically from the one before it. A country can become very good at hosting technology developed elsewhere without becoming much better at adapting, improving or eventually building that technology itself.
So the question worth asking is not simply whether foreign technology arrived. It is what capability it left behind.
Four things policy has to get right
My research in Rwanda drew on interviews with government officials, startup founders, universities, development agencies and a foreign technology firm, alongside policy documents and public data. It suggests that policy shapes partnerships between foreign technology companies and domestic startups through four closely related mechanisms.
The first is credibility. A company entering an unfamiliar market needs confidence that rules will be applied consistently, that institutions will function, and that agreements will outlast the announcement. Rwanda ranked 29th globally in the World Bank’s Doing Business 2019 report and 38th in 2020, second in Africa both years [2,3]. However, the people I interviewed pointed to something more practical than rankings: institutions able to turn policy promises into action. Agencies such as the Rwanda Development Board helped investors with registration, facilitation and introductions to other actors in the economy. Credibility is not about having attractive policies on paper. It is about whether institutions can actually implement them.
The second is opportunity signalling. When regulators adapt their rules so that an emerging technology can be tested, they tell other companies that experimentation is possible here. Zipline is the obvious example.
But this is also where the argument turns. A country can become an excellent testbed for foreign technology while capturing very little of the knowledge being generated. Flexible regulation can bring experimentation in. It does not automatically make local companies learn from it.
The third is operational feasibility. No regulatory sandbox can compensate for unreliable electricity, poor connectivity or inadequate roads. World Bank Enterprise Survey data indicate substantial improvements in some infrastructure conditions facing firms in Rwanda between 2006 and 2019, particularly in the reliability of electricity supply [4,5]. Innovation policy and infrastructure policy are usually discussed separately, but in practice they are deeply connected. An attractive framework for emerging technologies achieves little if companies cannot rely on power, internet access and transport.
The fourth, and the one that matters most for long-term learning, is absorptive capacity. This is the ability of firms to recognise useful outside knowledge, take it in and apply it [6]. Access to foreign knowledge is not the same as the ability to use it. If domestic firms lack engineers who understand the technology, or deal with foreign companies only as customers or suppliers, little knowledge will spread through the local economy.
This is where skills policy becomes technology policy. The number of ICT graduates from technical and vocational training in Rwanda rose markedly between 2010 and 2020 [7,8], which widened the pool of trained people that local firms can draw on when a foreign partner arrives. Absorptive capacity is usually treated as a trait of individual firms, but government can build part of it from outside the firm. There is a limit, though. Research on partnerships has long shown that firms also have to want to learn [9]. Government can create the conditions for learning, but it cannot force firms to learn.
Not more state, but better-performed functions
This is not a call for government to replace entrepreneurs or make technological decisions for them. Innovation researchers have long argued that innovation comes from interaction between firms, universities and public institutions, not from companies working in isolation [10,11]. More recent work asks how states can actively shape markets rather than only correct their failures [12]. The useful distinction is between policy that simply stimulates activity and policy that also creates the conditions for learning.
One function stood out in Rwanda: brokerage. Foreign technology companies often know little about what local startups can do. Local startups often have limited access to foreign companies, investors or major projects. Public institutions helped connect them. In mature ecosystems, those connections form through existing business networks. In younger ones, somebody has to create them deliberately.
Diagnose before you copy
Technology policy has a copying problem. Governments see another country introduce a startup law, a regulatory sandbox, a technology park or a tax incentive, and decide to reproduce it. But policies do not work independently of the institutions and conditions around them. A better approach is to diagnose the actual constraint:
If foreign technology companies are not entering, examine credibility.
If they enter but will not experiment, examine what the regulation signals.
If projects are announced but fail to scale, examine infrastructure.
If the technology is deployed successfully but little knowledge reaches local companies and workers, the problem is no longer attraction. It is absorptive capacity.
Different problems need different solutions. A startup law will not repair an unreliable electricity system. A sandbox will not make up for a shortage of engineers. Tax incentives will not create knowledge transfer between companies that barely interact. Policy instruments can travel from one country to another. The conditions that made them work usually cannot.
Rwanda is evidence, not a template
It would be a mistake to turn Rwanda into another model to copy. Its relatively centralised government and strong capacity for coordination make some interventions easier than they would be in larger, federal or more politically fragmented countries.
Tunisia offers a useful contrast. It built an extensive formal innovation framework, including a national innovation strategy and a dedicated Startup Act in 2018. Yet available figures for 2019 record around US$8 million in tech startup equity funding for Tunisia, against roughly US$126 million for Rwanda [13]. The two countries differ in market size, institutions, industrial structure and access to finance, and the data are incomplete, so a simple comparison cannot show that one policy caused one outcome. It does show that sophisticated laws on paper do not, by themselves, explain results. What set Rwanda apart was not any single instrument, but the way policy direction, institutions, regulation, infrastructure and skills reinforced one another.
The lesson that travels is therefore not a model but a set of questions. Do foreign companies trust the institutions enough to enter? Does regulation show that experimentation is possible? Can the technology actually operate under local conditions? And once it is operating, can domestic firms and workers learn from it?
Every government can ask these questions. The answers will rarely look exactly like Rwanda’s.
The measure of success, then, should not be how many technology companies a country has attracted. It should be what local firms and workers can do today that they could not do before those companies arrived.
