#527 2003 · Esther Duflo, Michael Kremer & Jonathan Robinson (SAFI field experiment) · Agriculture / development economics
Kenyan farmers agreed fertilizer paid for itself and still skipped it every season, until a voucher reached them the one day they had cash
the problem
Farmers agree fertilizer pays off but still don't buy it in time
background
Years of agricultural-extension campaigns in Western Kenya had told smallholder maize farmers, correctly, that fertilizer raised yields well above its cost. Surveyed farmers largely agreed. Adoption stayed low anyway — only about a quarter of farmers used any fertilizer in a given season — and the standard remedy was more of the same: more demonstration plots, more persuasion, subsidies offered at planting time, on the assumption that holdouts either didn't believe the numbers or couldn't afford the input.
Economists Esther Duflo, Michael Kremer and Jonathan Robinson suspected the real barrier was neither belief nor price but timing: by planting season, when fertilizer purchases and their price pressure actually arrived, harvest cash had usually already gone to school fees and debts, leaving farmers who genuinely intended to buy fertilizer without the money to do it. A subsidy experiment alone would only show whether price mattered, so they designed a field trial across two planting seasons in Western Province built specifically to separate the timing question from the price question.
what everyone would do
The standard levers against low fertilizer adoption were more persuasion (better information about the returns) or a bigger subsidy at planting time, both treating the holdouts as people who didn't know or couldn't afford it — but farmers already agreed fertilizer paid off, and a straight subsidy offered later still only lifted adoption by 13 points, less than the unsubsidized early voucher, showing price wasn't the binding constraint either.
what they saw
The researchers saw that the barrier wasn't belief or affordability but timing: a farmer's intention to buy fertilizer and their possession of the cash to do it occupied two different, non-overlapping moments, and by the time planting season created the price pressure, the harvest cash was already gone to other needs. The fix wasn't to change the offer, it was to change when the offer was made.
the move
The Savings and Fertilizer Initiative (SAFI) sent field officers to visit farmers right after harvest, while they still had cash from selling their crop, and sold them a voucher — at the regular, non-discounted price — for fertilizer to be delivered free at the start of the next planting season. This early, full-price, locked-in voucher was tested against a control group and against a 50%-subsidized voucher offered later, closer to planting time when cash was tight.
why it works
Selling the voucher immediately after harvest catches farmers at the one point their liquidity is highest, letting them lock in a purchase decision before competing near-term needs can consume the cash — the voucher itself functions as a commitment device against their own foreseeable impatience later. Because the fertilizer is delivered later at planting time, the farmer never has to hold onto both the cash and the resolve simultaneously across the gap where both would otherwise leak away, which is why the effect showed up even at full price and even grew in the second season as farmers who'd used it once sought it out again.
the payoff
The early full-price voucher raised fertilizer adoption 14 percentage points in the first season (from a base of 24%) and 18 points in the second (from 26%). A 50% subsidy offered later in the season raised adoption by only 13 points, and a free-delivery offer made later had no significant effect at all — moving when the offer arrived beat cutting what it cost.
where it breaks
The mechanism only helps people who already want to act but whose resources and their moment of resolve don't coincide — it does nothing for someone who doesn't believe the investment is worthwhile in the first place, and a late subsidy still beat nothing, so removing the early-timing option entirely isn't costless. It also requires a genuine, predictable gap between when cash arrives and when the decision is normally forced, plus a deliverer willing to visit at the inconvenient early moment rather than wait for the customer to come at the standard one — infrastructure many programs don't have.
what came after
SAFI became a foundational field demonstration of present-bias-driven procrastination in real economic decisions, cited across the behavioral-economics and development-policy literature that followed; Duflo and Kremer went on to share the 2019 Nobel Memorial Prize in Economic Sciences for the experimental approach to poverty this study exemplified. The core logic — offer a commitment device at the moment resources are highest, not the moment need is highest — has since been echoed in savings, health and education programs well outside agriculture.
references
- [1]Nudging Farmers to Use Fertilizer: Theory and Experimental Evidence from KenyaAmerican Economic Review, 2011aeaweb.org
- [2]Nudging Farmers to Use Fertilizer: Theory and Experimental Evidence from Kenya (NBER Working Paper 15131)National Bureau of Economic Research, 2009nber.org
- [3]Nudging Farmers to Use Fertilizer: Experimental Evidence from KenyaAbdul Latif Jameel Poverty Action Lab (J-PAL), MIT, 2011povertyactionlab.org