Diagnostic is a series of essays and hosted conversations exploring the challenges of building more inclusive digital economies. For hosted (virtual) conversations, like the one described below, Caribou Digital convenes a diverse set of experts and thought leaders with unique insights on an issue and gives them the opportunity to explore a topic together. The conversation here occurred in June 2021 and included investment officers from large foundations as well as thematic experts and employees from multilateral and bi-lateral donor organizations. The conversation was private to facilitate open exchange so the following is an edited summary of key themes and not attributed to any participant.
The conversation we discuss here was kicked off by reviewing a post called “Fortune at the Middle of the Pyramid” that uses high-resolution income and consumption data to understand opportunities and limitations of various for-profit investment strategies in sub-Saharan Africa. While that post focused more on private sector concerns, in the Diagnostic discussion we tried to understand the implications for aid, government, and donor policy (i.e. the “development sector”).
A large set of impact investor, international donor, and government anti-poverty policy is based on the notion that for-profit companies can be induced to serve the poor with life changing services like banking or schooling but the limits of the for profit model are not always taken into account
According to the Global Impact Investor Network (GINN), there was $715B USD under management by global impact investors in 2019. The majority of the investors surveyed are targeting market-rate returns and attempting to improve the lives of low-income people as their primary impact. This is to say that most of them believe they can help low-income populations by investing in for-profit companies who serve them or their communities with some kind of good or service.
Beyond the impact investing sector, there is a much broader set of development actors (donors, governments, and other aid organizations) whose strategies implicitly or explicitly embody a similar belief — that with proper support, the private sector will serve low income people with products and services that improve their lives in meaningful ways and thus help them out of poverty.
Identifying the limits to this kind of for-profit model is thus quite important for a large number of impact investing and development strategies. In this Diagnostic, we summarize our research and the discussion of our panel of experts around the question of whether the poor can be a profitable market segment. This is a necessary condition for for-profit companies to want to serve the poor at all.
African data showing the shape of consumer purchasing power at different levels of income shows that the largest segment of consumers below $2 per day control only a small fraction of total purchasing power
The main data source we are going to look at is PovCalNet, a decades old project by the World Bank to compile comparable measures of income and consumption across countries. Using detailed household surveys, PovCalnet attempts to estimate daily consumption per adult in each household.
The example depicted above shows how we used this data to calculate the height of the below curve at each point along the x-axis. The height is the amount of discretionary spending power wielded by the entire population at each level of consumption (total income * y % which is discretionary * total number of people at this income). Thus, we believe this amount is a good proxy for spending power (and thus for the size of the market opportunity) for non-food, non-essential goods. (We address food and essentials below.)
Adding in realistic cost to acquire and serve low income customers shows most of the poor to be unviable as customers
Importantly, we can see that aggregate discretionary spending is relatively small — below $3/day — despite the fact that the majority of the population live at this level. Their incomes are low and spent almost entirely on necessities. The graph peaks around $4–$6, which is where the smaller number of people is best offset by higher income. Beyond that, everyone gets richer rapidly, but the population numbers fall even faster than income rises — thus absolute spending power falls. Interestingly, the spending power (calculated as the area under the curve) of the consumers in the $5–$10 band of income (which comprise only ~10% of the population) is about double that of the $0–$4/day group (which comprise ~85%).
This is the key observation of the analysis — this dynamic will create strong incentives to push for-profit companies to serve the much smaller group in the $5–$10+ band. This is especially true to the extent that product developers must make a trade-off and tailor their products to be more attractive to one group and less attractive to the other. (There are very few universal products;most must be tailored and thus a trade-off must be made)
Curves in this figure represent the total disposable income net of cost to acquire and serve where hypothetical costs increase from curve 1 (lowest) to 5 (highest).
Lowering the cost of acquisition and distribution is key!
So far we’ve only considered spending figures — for-profit actors care about profits, so we have to consider customer acquisition costs as well as the cost of goods sold. Factoring these costs into the equation changes the picture quite a bit, as seen in the curves above. As costs rise, some consumers at the very lowest end will become infeasible (i.e. even if we captured all their discretionary spending, which is nearly impossible, the cost to acquire and serve would still be higher.) Also, the peak of the curve will shift to higher income brackets for two reasons. Not only is there more money, but also it is distributed across fewer people, so the cost to acquire a dollar of discretionary spending is assumed lower. In the graph above, we can see that the peak goes from $5 to closer to $20 as costs rise and the area under the curve (the total potential market value) shrinks dramatically.
Another way to look at this graph is in reverse — any entrepreneur that can design distribution models that dramatically lower cost to acquire and serve will create huge market value and make many more people feasible as customers (perhaps there are government or donor investments that can also have this effect).
From the Diagnostic conversation:
There may be business models that can serve populations despite a lack of significant purchasing power and thus show promise as poverty alleviation strategies based in the private sector
The financial inclusion field is one area of development policy that seeks to identify for-profit delivery models for services that will benefit low-income consumers. One area of discussion centered on the kinds of financial services models that could be successful. One observation was that this logic essentially rules out savings-based models, as they depend almost entirely on discretionary income and tend to have high acquisition costs due to most people’s low urgency for savings. An exception might be the voluntary savings circles, but most participants actually see those as credit. Insurance seems to suffer a similar fate. The very basic package of P2P remittances, bill payments, and credit that has become the standard mobile money offering seems to be pretty close to what is sustainable for lower-income populations. This is partly because these products are high-margin due to the degree of need and can almost be seen as necessities (sending money or borrowing it in an emergency can be very urgent needs).
The discussion also touched on products (like airtime or fast moving consumer goods) that can be sold as individual package pricing where profit is guaranteed at the level of the individual sale. Because there is no risk of acquiring a customer that is unprofitable, producers of these services are likely to be more open to serving the poor and see them as a limited but potentially profitable subsegment.
Finally, the group discussed cross-subsidy models (like Facebook) where a richer segment of the user population can cross-subsidize unprofitable lower-income users. Caribou Digital’s analysis of Facebook users in the developing world shows that while Facebook makes lots of money on users in North America, Europe, and Latin America, they likely lose money on users in Africa and a few other parts of the world where user discretionary income is too low. Thus advertisers are not willing to pay the premiums they are willing to pay elsewhere (very much in-line with this analysis indicating they have low discretionary income), and lower-quality devices and constrained usage patterns make them harder to serve. We estimate that Facebook loses something like a dollar a year on hundreds of millions of users in the developing world. But because Facebook’s revenues are tens of billions of USD per year, these losses are likely considered a cost of doing business when running a global platform that benefits from the network effects of having everyone on board. But this strategy depends on the money-losing population of users being a relatively small fraction of the total user base, and one where the total losses are limited. Thus, outside of the handful of massive global platforms where this applies, cross-subsidy models are not a feasible way to serve the population below the poverty line with digital services.
Very low-cost acquisition models might also break this logic, at least to the extent they assure the population in the $4–$8/day range are feasible as customers. Finally, models focusing on selling food and daily necessities also have potential, as shown below. There was some excitement about models that focus on agriculture value chains as a perennial development sector that has recently seen more activity from new technology startups. It was noted that the development sector has not paid as much attention to models focusing on small business supplier marketplaces and last-mile logistics even though they often have necessities and food at the core — perhaps because they have mostly focused in urban areas to date.
Focusing on models that deliver food and basic necessities helps expand the opportunity for the private sector, but it is still an uphill battle.
The above analysis assumes that most companies are creating services that target discretionary expenditures. Above we show the same curve when we include all expenditures, including primary necessities like food. When considering basic necessities, we now see a much larger market and one where the peak of consumer spending power is at a lower level of income — now in the $2–$5 range. This comports with the “Bottom of the Pyramid” argument that there is significant spending power among the lower income populations when it comes to food, and other primary necessities. Nevertheless, we still show relatively limited aggregate spending power for the large group of people who live on less than $2/day, which biases the for-profit sector to higher incomes. Most digital and technology powered models do not focus on necessities, but we believe there is significant scope to do so.
What are the key takeaways from this data?
- Purely B2C models that seek to sell discretionary goods or services directly to low-income African households are unlikely to be profitable due to low discretionary income and the relatively high costs of acquiring and serving customers.
- The largest block of discretionary income is in the hands of the population who live at $4-$8/day. These folks are not formally poor by global development standards.
- With realistic acquisition costs factored in, the incentive is more often to target $10+/day, which represents only about 5% of the population in most of Africa.
From the Diagnostic conversation:
Two conclusions for the wider implications of impact investing and the poor as a profitable market segment
- You can’t spend your way out of poverty — you have to increase income!
One area the group went into in some depth was whether models that increase income could break the logic above. Digital services that include remittances, gig work platforms, marketplaces where SMEs can be sellers, and productive lending all potentially increase income and thus don’t depend as much on the initial income of the user (in fact, they increase it). We believe these models should be pursued aggressively by the development sector to the extent it can be shown any of these models increase welfare along with increasing income (many rigorous studies within the microcredit sector have not shown increases in welfare from productive lending, for example, despite the narrative that they were a way for low income people to pull themselves out of poverty).
2. The development sector could use a strategic repositioning from financial inclusion towards digital livelihoods
Some of the participants noted a trend of development actors pivoting away from financial inclusion as an example of a strategy focusing on delivering a service via for-profit providers to low-income populations. As one participant noted: “We did focus groups at some point and financial services was never anything the poor asked for. They ask over and over for jobs and ways to increase income.” So that participant’s organization is now more focused on broader financial sector development that will lead to increased economic growth and therefore jobs.
The group noted a number of development actors who had undergone similar strategic repositionings towards digital livelihoods or jobs. Often the repositioning was towards digital livelihoods, including platform marketplaces where people can go to find gig work (e.g. Uber, Gojek, Upwork) or seller marketplaces (e.g. Alibaba, Jumia) or a wider range of informal digital livelihoods (MSMEs who transact over WhatsApp or Facebook). Many of these opportunities are new and seem to present significant opportunities, though the potential downsides of the digital platform economy are also important to understand. Caribou Digital maintains a site with a trove of research and references in this new area called Platform Livelihoods and DFS Lab has looked at the Indonesian landscape in this area here.
Finally, there was some discussion about what would be difficult about such a transition for many donors. Even if it does turn out livelihoods have a better business model and are what people actually want, it can be hard for donors to move in this direction. Since livelihoods cut across sectors, donors will need to understand multiple sectors at once, as well as the fast-moving dynamics of digital platform economies. It can also be more nuanced because dealing with digital platforms has potential positives and negatives where data privacy, working conditions, and the general balance of power is tilted in favor of platforms in setting the terms of engagement with workers or sellers. These nuances can be hard to navigate. For donors and governments it is relatively easy to defend giving a digital savings account, and harder to justify demanding gig-work with relatively low wages (even if this is clearly something that the workers benefit immensely from and want because it improves their lives).
What do we take away from this exercise?
This data helps quantify some of the key reasons most startups and technology companies as well as more traditional corporations target only a relatively small subset of the population at the higher end of the income spectrum in Africa.
In doing so, it also calls into question the notion that funding innovation or subsidizing the private sector in other ways will result in reaching the lower-income population. Realistic levels of customer acquisition costs may make much of this population unprofitable to serve regardless of the model or the technology. And, even if they are feasible, for-profit models will almost always be pushed by incentives to focus their marketing on the population above key global poverty line measures (e.g. $1.90 or $3.10) and potentially even try to discourage lower income people from using their services (especially if the total cost to acquire and serve make those people unprofitable).
In light of this logic, one of the areas our group identified as high potential for further research and donor intervention was digital livelihoods which give people a chance to improve their income directly. We hope to see more work in this area down the road.