Across conversations in Kenya’s pubs and WhatsApp groups, debt is on everyone’s mind. The speed and ease of access to credit through new mobile apps delivers cash to millions of Kenyans in need, but many struggle to repay. Despite their small size, the loans come with a big cost—sometimes as much as 100 percent annualized. As one Nairobian told us, these apps “give you money gently, and then they come for your neck.”

He is not alone in his assessment of “fintech,” the ballooning financial technology industry that provides loans through mobile apps. During our research, we heard these emergent regimes of indebtedness called “catastrophic,” a “crisis,” and a major “social problem.” Newspapers report that mobile lending underlays a wave of domestic disarray, violence, and even suicide. One young man in Meru described it as a “can of worries.” His monthly salary was not enough to cover ordinary expenses such as rent and necessary contributions to extended kin networks—let alone leisure or investments in his own future. So, like millions of others, he turned to phone-based loans, at one point toggling between five different apps. Reeling as the costs added up, he struggled to repay, deleting the apps so he would not be tempted by repeated offers of dangerous debt.

One Kenyan argued the apps are ‘enslaving’ people—from the working poor to the salaried classes—by making claims on their future labor.

Relations of credit and debt are nothing new to Kenya. For ages, friends, family, and colleagues have lent and borrowed from each other, but what differs today is a lack of reciprocity. In peer-to-peer credit, everyone is eventually likely to be a debtor and a creditor; terms can be reworked according to timelines and margins that are subject to negotiation. In contrast, the fintech industry envisions ordinary Kenyans as first and foremost borrowers, leading many Kenyans to describe their predicament as a form of servitude. One Kenyan argued the apps are “enslaving” people—from the working poor to the salaried classes—by making claims on their future labor.

Indeed Kenya’s new experience of debt is worrying. It reveals a novel, digitized form of slow violence that operates not so much through negotiated social relations, nor the threat of state enforcement, as through the accumulation of data, the commodification of reputation, and the instrumentalization of sociality. Kenyans are being driven into circuits of financial capital that are premised not—as the marketing would have it—on empowerment, but on the profitability of perpetual debt. The eruption of over-indebtedness in Kenya marks the intersection of a faith in finance to ameliorate the lives of the poor and a recognition by techno-capitalists that those same populations are the source of runaway profits.

•••

It is perhaps no surprise that this confluence of technology and unregulated lending have emerged with such ferocity in Kenya. Since the early 2000s, Kenya has been touted as a hub of technological innovation from which novel financial infrastructures have emerged. Financialization through digitization is at the center of narratives of “Africa rising,” which have positioned Kenya as Africa’s “Silicon Savannah.” Both the origins and durability of this story can be largely attributed to what is now East Africa’s largest corporation, the telecommunications and financial services provider Safaricom. This corporation first drew international attention with the growth of the wildly successful and widely emulated service M-PESA, a mobile-to-mobile money transfer platform, but it has since grown far beyond this offering. Safaricom’s growth has been enabled by the Kenyan state, which proudly provides a permissive regulatory environment in the service of innovation. M-PESA, for instance, received only a “letter of no objection” from the Central Bank of Kenya that permitted—but did not regulate—the telecommunications firm’s entrance into the financial sector.

Since this entrée into payments, Safaricom has expanded to become a foundational infrastructure for lending. Growth was slow at first, but since 2016 its offerings have grown enormously. Its loan service, M-Shwari, launched with the Commercial Bank of Africa, has been replicated in partnerships with other banks. By mid-2018, M-Shwari alone had dispersed 230 billion Kenyan shillings (KSh) in loans. Its profitability depends, in part, on a discursive gymnastics that defines the 7.5 percent premium on borrowed funds as a “facilitation fee” rather than an interest rate. Were it judged to be the latter, it would be subject to a legal limit well below its current annualized cost of around 100 percent.

The loan service uses discursive gymnastics to dodge the legal interest rate well below its current annualized cost of around 100 percent.

Safaricom offers other loans too. At the start of 2019, Safaricom inaugurated a new overdraft facility, Fuliza, which lends to M-Pesa subscribers who have run out of digital value. Users are alerted of their inability to pay, and Fuliza is proffered as a real time solution, offering subscribers access to small loans to bridge the gap at a premium. In its first month alone, it lent more than KSh 6 billion, with each customer charged an initial one percent fee plus a daily fee of up to KSh 30. Fuliza expands the logic of Okoa Jahazi, an airtime credit service through which around one-third of Safaricom’s airtime is sold at a 10 percent mark-up to Kenyans short of cash. Okoa Jahazi is so popular, one investment banker told us, that were it regulated, it alone would make Safaricom one of the middle-tier banks in the country.

But this lending is not banking as usual. While M-Shwari is offered in conjunction with a regulated bank, huge amounts of debt are now offered in Kenya outside the purview of state regulation. Services such as Fuliza and Okoa Jahazi do not accept customer deposits and are therefore not subject to the same oversight as banks.

Those lending apps in which Safaricom is a partner are the tip of an iceberg. Dozens of other mobile lending companies now operate in Kenya, though loose regulation makes the full extent difficult to know with certainty. Speculation on their provenance and intention abounds among ordinary Kenyans; on a recent visit, we heard rumors that politicians were trying to launder money by launching lending apps, and Russians were seeking willing Kenyan partners to advise them on the contours of this new frontier market in debt. These tales point not merely to the industry’s lack of transparency but also to a sense among Kenyans that the real beneficiaries are distant and unaccountable.

Two of the most prominent fintech apps are Tala and Branch. From their California headquarters, these firms export Silicon Valley’s curious nexus of technology, finance, and developmentalism. Small shops across the country are painted in Branch’s brand of blue, with slogans offering “loans for the way you live.” Quickly downloaded onto Kenya’s proliferating smartphones and utilizing the country’s ubiquitous mobile money transfer system, these apps mine people’s devices and social media accounts for signs of their creditworthiness. While their lending algorithms are closely guarded secrets, industry insiders suggest an ambitious effort to track everyday behavior and social relations. In line with the belief that “all data is credit data,” these firms seek to analyze everything from whether you call your family regularly, go to the same workplace every day, and have an extensive network of contacts. Tala’s CEO reported that “repayment of a loan is more likely by someone whose contacts are listed with both first and second names.” Branch, for its part, relies on a user’s smartphone data, though what among the likes, links, locations, and browsing is noteworthy—let alone whether financiers should have such access—is less discussed. While these firms offer little transparency to the public, they tell investors that money is pouring in: Tala has raised more than 109 million U.S. dollars while Branch has received nearly 260 million U.S. dollars from investors keen to capitalize on poor people’s debt and data.

Crucial to the fintech business model is an endless stream of nudges, exhortations, and incitements to borrow. Unsolicited text messages interrupt people throughout the day, enticing those in need to borrow at extraordinary rates. Many pointed to the high rates of borrowing on weekend nights as evidence that loans are marketed and taken in moments of inebriated revelry.

Those at risk of default receive just as much hectoring (one study found 50 percent of Kenyans repaid a loan late). A University of Nairobi graduate told us how embarrassing it is to be in a meeting—or worse, a job interview—and have repayment reminders pop up on your screen. “It’s so embarrassing! They text all the time. You get stressed.” Another of these apps, Okash, took this logic of stigmatization even further, harvesting users’ contacts and calling bosses, parents, and friends to shame defaulters into repaying.

Crucial to the fintech business model are the endless nudges and incitements to borrow. Unsolicited text messages arrive throughout the day, enticing people to borrow at extraordinary rates.

Beyond the excessive fees and stress, Tala, Branch, and the others are responsible for a crisis of credentialing. In 2010, the Central Bank of Kenya began licensing credit reference bureaus (CRBs). Premised on the idea that credit was poorly priced due to information asymmetries, these databases collect loan histories and share them among financial institutions. The clearest result, however, has been the widespread blacklisting of borrowers. One report found that already 2.7 million Kenyans had been negatively listed by CRBs by 2017 (a year before fintech really took off). One loan officer at an upcountry bank branch was only somewhat exaggerating when she told us 80 percent of would-be customers were listed with CRBs due to fintech loans. Whatever the figures may be, those we spoke to have a widespread sense that easy lending is foreclosing borrowers’ futures. CRB reports are not being used only by loan appraisers; rather, they are being enrolled by a range of institutions, including would-be employers who demand applicants’ CRB report in order to be considered. At more than 20 U.S. dollars for a clearance certificate, such requirements create a financial burden and a create an unfair roadblock for a generation seeking meaningful employment.

All of this bodes poorly in light of Facebook’s newly announced financial service, Libra. Like fintech apps, Facebook draped Libra in socially conscious rhetoric, promising to incorporate “the unbanked” through its blockchain offering. Indeed, Tala’s boss was quick to extol the possibilities of Libra, telling Fortune that “[t]he promise of it is exciting,” allowing her company to “accelerate” their expansion into new populations of borrowers. The global ambitions of Libra promise fintech firms regulatory arbitrage and massive amounts of data, circumventing the paltry protections of national legislators.

•••

This capacity to escape jurisdiction marks a rupture in the business of debt. While traditional banks secure their lending with a combination of customer deposits and pledged collateral, digital lenders such as Safaricom forego these. Instead, through the assembly of a vast archive of user data, the firm claims it can analyze behavior in a predictive fashion, premising loans not on property—whether deposits or collateral—but on quantitative assessments of credit worthiness.

This “reputational collateral,” to use Keith Breckenridge’s term, is strengthened each time a customer engages Safaricom. Since the 2007 launch of M-PESA, Safaricom has broadened the realms from which it accumulates its cache of user data. Information on users’ rates of sending and receiving money, the quantities moving through their accounts, and their reliance on Okoa Jahazi (the short-term loan for emergency airtime) were all mobilized to assess the creditworthiness of potential borrowers when M-Shwari was first launched in 2012. The result is that Safaricom is coming to look more and more like a credit bureau—while avoiding being regulated as such.

Digital lenders forgo the traditional ways that banks secure their loans, instead assembling a vast archive of user data to predict consumers’ behavior.

Safaricom’s use of reputational collateral has proven to be remarkably reliable in the case of M-Shwari, with the company reporting rates of repayment above those of traditional banking institutions. People work hard to repay their loans, if not on the timelines dictated by lenders. Many report taking out loans with one mobile service to repay existing debts when the due dates come up with another lending service. This is a debt treadmill, and it is hard to escape.

It is also a cycle of indebtedness which the lenders have little incentive to break. While borrowers scramble to repay, fintech firms have structured the market to benefit from iterative borrowing. Each time a loan is taken out, more user data is harvested, allowing companies to develop better predictions on the rates of repayment of a given customer. Timely repayment grants a user access to loans of higher values, but the fixed percentage facilitation fee means the profit only proportionally increases. In other words, the greater the use of the service, the higher the returns to the corporations.

Proponents of mobile lending argue that it has solved the knowledge asymmetries in conventional lending that required deposits or collateral to secure consumer loans. New practices of data harvesting and analytics allow lenders to assess consumer credit-worthiness and people’s capacity to repay. People once excluded from credit, the narrative goes, can now access capital to improve their short and long term prospects. This narrative is silent on fintech’s exorbitant premiums. Moreover, reliance on multiple sources of credit at any given time seems to suggest not “financial inclusion,” but new forms of both dependence and exclusion: borrowing from Tala to pay Branch is indicative of a costly dependence; and those borrowers piling up in CRB lists are being excluded from other opportunities, with their poverty marshalled against them.

Expanding access to bank credit has long been a goal of international organizations, government, and industry, united by the goal of “financial inclusion,” or “banking the unbanked.” This agenda valorizes the role of markets to improve people’s lives—an ideology buoyed by the conservative turn in places such as the U.K.’s Department for International Development and the outsized influence of the Gates Foundation.

Our research shows, however, that people targeted by fintech are not simply unbanked: many are regularly broke. More specifically, the profitability of digital debt depends on the routine shortage of cash among Kenyans. While some are using credit to invest in businesses, many consumers of easy credit turn to lenders when unable to pay a bill, make rent, or even afford charcoal to cook an evening meal. Such a predicament is not merely the reserve of the especially poor; the need to buy time with expensive loans unites Kenyans up and down the class ladder. More than a third of digital debtors are using the loans to meet day-to-day household needs—the type of routine expenses that are unlikely to disappear with borrowing.

Across the country, millions of Kenyans work in a condition that Michael Denning has referred to as “wageless life.” Whether hawking mitumba (used clothing) along Kenya’s streets, working in the privatized transport sector, or operating as a mama mboga (vegetable seller) in Kenya’s markets, people in this labor market make money on the day. But their fiscal horizons are unpredictable and subject to volatility. Instead of selling their labor power in exchange for a wage, these men and women toil in what more closely approximates a piece-work regime, making a small margin every time the negotiation for a piece of clothing is finalized, a vehicle is boarded, or a bag of potatoes is sold.

Wageless workers are often unable to accumulate large sums of money because profits made one day are often spent by the next day. Consumer product firms—in part spurred by the promise of a “fortune at the bottom of the pyramid”—have capitalized on these monetary dynamics by resizing their offerings. They have inaugurated what in Kenya is sometimes called the “kadogo economy,” from the Swahili for “small.” Single-use packets of laundry detergent, beef stock, and cooking fat allow those who make money on the day to consume these products which would be out of reach in their more conventional sizes and quantities. Safaricom, too, was a “pioneer” in this regard, allowing customers to purchase low-value airtime scratch cards, costing as little as KSh 10. The irony, however, is that it is expensive to be poor: while accessible due to their small size, products for the kadogo economy cost proportionally more than the conventionally sized goods available to wealthier buyers.

Many Kenyans toil in what closely approximates a piece-work regime of wageless work. Profits made one day are often spent by the next day.

While the development of these products matches the daily financial rhythm of Kenya’s poor, this population is constantly having to hedge their financial futures. A bad day at the market and unexpected expenses—such as an illness—upset this delicate balance. So, too, do more predictable expenses such as school fees and rent, which are premised on other temporal logics, that of the semester and the month. As a result, running out of cash—whether to pay for a bus home at the end of the day or to fuel a car mid-way through the month—is a frequent occurrence.

We think of this in terms of the zero-balance economy. Unlike the kadogo economy, which names the resizing of goods, the zero-balance economy is characterized by the temporal disconnect between available cash and necessary expenditure. Members of the zero-balance economy routinely find themselves lacking liquidity with which to meet costs. While some cash shortfalls are unexpected, many Kenyans think of financial volatility in more patterned ways: pecuniary irregularity is the norm. Within this context, Kenyans have developed repertoires to manage and make sense of routine volatility.

Kenya’s salaried workers too, are part of the zero-balance economy. While a salary surely reduces economic precarity, it is not enough for most employed people. Recent figures from the Kenya National Bureau of Statistics indicate that three-quarters of salaried workers (nearly 2 million people) make less than KSh 50,000 a month (roughly 485 U.S. dollars) . People tend to get paid towards the end of the month, and they use much of their salaries to pay their rent by the first of the following month. Furthermore, the whittling away of public services—what one economist we spoke to called “internally-imposed structural adjustment”—means increasing costs for individuals, who spend their hard-earned shillings to, for example, pay water and electricity bills, and school and medical fees. By the middle of the month, bank accounts are dwindling and wallets thin. There is a sort of respiration across the Kenyan economy that reflects the monthly flow of zero balance life: from the crowds at bars waning to the increased reliance on the familiar, inexpensive Kenyan dish sukuma wiki (literally, to push the week). We were routinely directed to look to the roads: by mid-month, Kenyans are less beleaguered by “the jam”—the national shorthand for traffic—because many can no longer afford gas for their cars and commute via bus instead.

If the kadogo economy was a commercial means of capitalizing on the paucity and daily uncertainty of wageless life, digital lenders, armed with user data, are now profiting from the temporalities of the zero-balance economy. Unlike those peddling single-serve products, merchants of debt have a battery of digital data and algorithms at their disposal, not to mention a largely unregulated lending frontier. They offer customers a means to buy time, but at a premium so costly it would be illegal for a Kenyan bank.

•••

So worrying are the trends in Kenya that even proponents of digital lending are calling for caution. The Central Bank of Kenya has demanded borrowers are made aware of the apps’ terms and conditions, but at present a legislative mandate does not exist that would require much more than disclosure by non-bank lenders. Donors, including the groups that have done the most to valorize digital finance, are beginning to advocate similar consumer protections, from transparency requirements to rules around reckless lending. A group of fintech lenders have tried to head off official regulations by signing a voluntary code of conduct and forming a lobbying group. Promising not to lend more than 40 U.S. dollars to first-time borrowers, or to clearly present the terms of a loan, may help around the margins, but commercial self-regulation will not transform the profiteering of globally ambitious venture capitalists.

Kenya’s indebted class is united by its inclusion in the zero-balance economy.

The obstacles to getting relief to Kenyan borrowers go beyond regulatory policy. Regulators, donors, and industry share an underlying assumption that individuals get into trouble with debt because they are confused or imprudent. Financial inclusion advocates and venture capitalists only see people as borrowers or entrepreneurs. Yet, the shared predicaments of so many Kenyans are better understood in terms of the emergence of a new class. Kenya’s indebted class is united by its inclusion in the zero-balance economy: whether working as a hawker who cannot accumulate sufficient stock or an employee whose salary is not enough to last the month, the irregularity and paucity of income in Kenya compels borrowing. With rent coming due, school fees on the horizon, and relatives asking for help with medical costs, digital loans are not taken for lack of information or awareness; rather, Kenyans must borrow to make ends meet. Their constraints on social reproduction cannot be solved by appeals to liberal aspirations for informed consent. Their indebted status is not a sign of empowerment; it is evidence of their subordination to economic arrangements not of their choosing.