Fenix couldn't repossess collateral in rural Uganda, so it turned an already-installed solar panel into a remote-controlled lock on an unrelated school-fee loan

the problem

Lending in rural Uganda is hobbled by courts and physical repossession being too slow, costly, or unavailable to make most household assets usable as loan collateral, forcing lenders to either not lend or price in default risk with punishing rates

background

what everyone would do

The standard response to weak collateral enforcement in a market like rural Uganda is either to avoid lending for products without easily seizable, high-value assets, or to charge steep interest rates that price in the risk the lender knows it can't actually collect on through courts or physical repossession. Both responses treat the underlying enforcement problem as fixed and adjust price or availability around it, rather than solving the enforcement gap itself.

what they saw

Fenix and the researchers saw that the lender already possessed exactly the kind of asset good collateral requires — something the borrower genuinely valued and didn't want to lose — sitting inside the customer's home from an entirely separate, already-completed transaction: the solar home system. The insight wasn't building a new remote-lockout technology or inventing a new collateralizable good, it was recognizing that an asset already financed and installed for one purpose (power) could be repurposed as security for a completely different loan (school fees), purely because the lender already had the technical ability to disable it remotely — collapsing the entire physical-repossession problem into a software switch.

the move

Rather than try to invent a new collateralizable asset or extend physical repossession into rural areas where it wasn't practical, the lender repurposed an asset already inside the customer's home, financed for an entirely different purpose, as general-purpose loan security for a new product — because the asset was remotely controllable, a missed school-fee-loan payment could trigger the solar system's power output being throttled or switched off, with no lawsuit or repossession agent required.

why it works

Because the solar system could be remotely throttled or switched off without anyone traveling to the household, digital collateral eliminated the exact bottleneck that made conventional collateral unusable in rural Uganda — no court order, no repossession agent, no delay between default and consequence. This made the threat of enforcement genuinely credible in a way unsecured lending never could be, which is precisely why the study found roughly two-thirds of the effect came from reduced moral hazard: borrowers who knew a real, valued asset was genuinely at stake changed their repayment behavior, rather than the improvement coming mainly from the lender simply excluding worse borrowers upfront. Because the collateral was an asset the household already depended on for real daily value (lighting, phone charging, entertainment), the threat of losing it carried genuine weight without the lender needing to acquire, price, or manage a separate physical asset for each loan — the existing solar unit did double duty as the enforcement mechanism for an unrelated product, at essentially zero marginal cost to set up.

the payoff

Securing the school-fee loan with digital collateral cut default rates by 19 percentage points and raised the lender's rate of return by 38 percentage points compared with unsecured lending. Using a variant of the Karlan and Zinman (2009) decomposition methodology, the researchers found roughly two-thirds of the total effect came from reduced moral hazard — borrowers now had a real incentive to prioritize repayment because a real asset was genuinely at stake — with the remaining third from improved borrower selection. Access to the digitally secured school-fee loans also significantly increased school enrollment and school-related expenditures, without detrimental effects on households' overall balance sheets, per the study's own reported findings.

where it breaks

This mechanism depends on the borrower already possessing a remotely-controllable asset they genuinely value enough that losing access functions as real leverage — a household with no networked asset already in place has nothing this specific version of the mechanism can collateralize, and building one solely to enable future lending would reintroduce the cost problem digital collateral is meant to avoid. It also depends on the lender's control over the asset being technically reliable and legally acceptable — a system that can be tampered with, bypassed, or that raises consumer-protection concerns about disproportionate power over a household's basic utilities risks both practical failure and regulatory backlash. And because the study's own decomposition shows real behavioral change (moral hazard reduction) doing most of the work, this mechanism specifically requires the asset being pledged to matter enough to the borrower's daily life that the threat of losing it changes behavior — collateralizing something the borrower doesn't actually value or depend on would produce little of the same effect, since there would be nothing real at stake for the borrower to protect.

what came after

Published as Gertler, Green & Wolfram, 'Digital Collateral' (NBER Working Paper 28724, 2021; Quarterly Journal of Economics 139(3), 2024), the study is cited in development-finance and fintech literature as founding evidence for an entirely new class of lending — 'digital collateral,' relying on lockout technology rather than physical seizure — now used by pay-as-you-go solar and asset-financing lenders across multiple low- and middle-income countries to extend credit into markets where conventional collateral enforcement was previously impractical.

references

1. [1]Digital CollateralNational Bureau of Economic Research (Working Paper 28724; published in Quarterly Journal of Economics 139(3), 2024), 2021nber.org
2. [2]School Fee Loans to Increase Students Educational Outcomes in UgandaAbdul Latif Jameel Poverty Action Lab (J-PAL), 2021povertyactionlab.org

keep it

same kind of clever

• 2011 · Airbnb (Host Guarantee)Airbnb underwrote the nightmare so strangers would swap keys
• 350 BCE · Ancient Greek sea lendersGreek sea lenders wrote off the loan whenever the ship carrying it sank
• 483 BCE · Classical Athens (Laurion silver mine leases)Athens owned a silver mountain and never once tried to mine it directly