Case #1: Building ApiCare — The Supply Chain Problem and Price of a Good Year

The good year

It is November in Jitlang, a village in Pakyong district in East Sikkim, and Tek Bahadur Rai has forty kilograms of honey.

This is the festival season. There are things he wants to buy for his family, which is the ordinary reason a farmer in the hills wants cash in November, and he has the produce to pay for them. Forty kilograms is a good harvest; better than most years, better than most of his neighbours. It is also considerably more honey than the people he knows can absorb.

A beekeeper with fifteen kilograms does not have a marketing problem. He sells to relatives, to neighbours, to the family two ridges over who remembers his father, and it goes slowly over several months in whatever bottles are to hand. Reused rum and beer bottles, mostly. It also goes at something close to its real worth, perhaps fifteen hundred rupees a kilogram, because everyone buying it knows exactly where it came from and from whom.

Forty kilograms cannot move that way, and it cannot wait. So the surplus has to travel, and the moment it has to travel, Tek Bahadur meets a buyer: a farmer producer organisation, a cooperative, an aggregator who takes volume. The buyer names a price. The reasoning offered, when any is offered, is that this is a bulk purchase, and bulk purchases attract wholesale rates. The offer is between five hundred and seven hundred rupees a kilogram.

Payment is cash. Sometimes, for smaller lots, UPI. There is no receipt; no grade; no record of moisture content or floral source; no note of who kept the bees or where. When the transaction is finished, nothing remains to show that it happened. Not for Tek Bahadur’s next negotiation, not for a bank, not for the buyer who will sell it on, and not for whoever eventually eats it.

This is the shape of the problem in the Eastern Himalayas, and its first cruelty is the direction it runs in. Producing more does not earn more per kilogram; it earns less. Success is the event that strips the value out.

A constraint system, not a landscape

Sikkim is a small state on India’s northeastern edge, bordered by Nepal, Bhutan and the Tibet Autonomous Region, and it is almost entirely mountain. This is usually described as scenery. It is more useful to read it as a set of physical constraints that determine how anything gets bought and sold.

Land comes in small, scattered, steeply pitched parcels. A household farms a fraction of a hectare, often split across terraces at different elevations, and grows what the season allows: cardamom, seasonal fruit and vegetables, a little livestock. Income arrives in pulses as each crop matures and is carried to market. Beekeeping sits inside this pattern as a supplement, never the main income, always the third or fourth line, and it is attractive precisely because it is cheap to run. Bees are not fed. For a household already working the land, hives are close to free upside.

Distance does the rest of the work. The nearest town is typically half an hour away, which sounds close and is, for buying salt. The nearest market of any commercial consequence is Gangtok, between two and seven hours by road depending on the village and the season. Those hours are why surplus cannot simply be carried to market by the person who produced it: forty kilograms does not justify a day of lost labour and a vehicle, and no individual smallholder in these hills produces enough to make the journey pay. Aggregation is not a choice; it is arithmetic.

Whoever does the aggregating therefore occupies a specific and powerful position. They are the only party who can see both ends of the chain. The farmer knows what he was offered. The buyer knows what the honey is worth in Gangtok, in Siliguri, in Kolkata. Nothing obliges that information to travel back down the hill, and nothing records the transaction in a form the farmer could use later.

There is a further turn of the screw in the timing. Harvest comes twice, in June and July and again in November and December, and the second of those lands squarely in festival season. So the farmer’s need for cash peaks in the same weeks that his supply does, and this is true of every beekeeper in the district at once. Maximum regional supply meets maximum household urgency on the same calendar. A farmer holding forty kilograms and facing Diwali is not in a negotiation; he is in a queue.

Two further facts complete the picture. Older beekeepers frequently do not own a phone, and any conversation about money happens in Nepali, or in some areas only in a tribal language. And Sikkim has been recognised by the UN FAO as the world’s first fully organic state, a distinction that, as of this case, nobody in the honey trade has managed to convert into a price.

The loop this produces runs as follows. Steep terrain gives small scattered plots; small plots give volumes too low to travel; low volumes force aggregation through an intermediary; the intermediary holds the only complete view of the chain; the farmer emerges with neither price information nor a transaction record; without a record he has no negotiating position and no credit history; without credit he cannot invest in more hives; without more hives his volume stays too low to travel.

Everything that follows is an attempt to cut that loop at one point.

What ApiCare set out to build

ApiCare Organic Farms and Honey Production Pvt. Ltd. was founded in Sikkim around 2021 with a straightforward plan: build one excellent farm and let it be copied.

The model was a single demonstration apiary of two hundred to three hundred hives. At an expected six kilograms per hive, that implied twelve hundred to eighteen hundred kilograms of honey a year; enough volume to matter commercially, and enough of a working example that neighbouring farmers could see the thing operating and replicate it. A visible, profitable, replicable farm would do the persuading that no amount of advocacy could.

The design came from the literature, and the literature was European and American. It described Apis mellifera, the western honey bee: a species that tolerates dense apiary layouts, forages efficiently across open agricultural land, and has been bred for a century toward commercial beekeeping at scale.

The bees available in Sikkim were Apis cerana, the eastern honey bee, in its regional Himalayan variant. The colonies absconded, which is to say they abandoned their hives and left, for three reasons that only became legible afterwards.

The first was forage. The research assumed flat land and large monocultures within flight range. The Himalayan terrain is neither. It is vertical, forage is sparse and scattered across elevations, and there are no mega-farms to work. Bees navigating a mountainside do not find what bees navigating a plain find.

The second was density. Apis cerana colonies placed close together fight each other. The dense layout that makes a mellifera operation efficient makes a cerana apiary hostile to its own occupants.

The third was climate. The last several years in the region have not resembled the preceding twenty. Temperatures have risen, rainfall patterns have shifted, and flooding has become a recurring event. A model calibrated to historical conditions was being run in conditions that no longer held.

The redesign inverted the original logic. Instead of one large farm, many small ones: a maximum of twenty hives to begin with and up to fifty at maturity, spread across roughly 0.7 hectares, with hives spaced three feet apart and a minimum of five hundred metres between one cluster and the next. The layout is built around what cerana will actually tolerate rather than what mellifera permits.

But a cluster model has a requirement that a demonstration farm does not. A single farm can be run by the company that owns it. A network of clusters can only be run by the people who live where the clusters are, which meant that everything ApiCare had learned the expensive way had to be transferred to villagers who had never been asked to keep bees this way before.

That requirement is what turned a honey company into a training organisation, and the training is what brought the state government in.

What training could not buy

The trainings worked.

Between May 2025 and January 2026, ApiCare delivered four honey processing programmes under the MSME RAMP scheme across four districts, training 104 participants at Jitlang, Mangbrue, Kewzing and Beng. Log hives were replaced with movable-frame Newton hives. Hand-touch extraction, which participants had understood to be the organic method, was replaced with manual centrifugal extractors and moisture testing. Reused liquor bottles were replaced with food-grade glass. At Beng, twenty participants completed Udyam registration, and therefore formal enterprise status, inside seven days.

Then, in most villages, the thing stopped.

The obstacle was not knowledge and it was not willingness. It was that a trained farmer who wants to scale from ten hives to twenty needs capital he does not have and cannot borrow. The households ApiCare trains live week to week or day to day. Income arrives when produce matures and is spent by the time the next crop is ready. There is no pool of savings to draw on and, just as binding, very little tolerance for risk, because the downside of a failed expansion is not a bad quarter but a bad year in a household with no buffer.

The formal alternative closes the same way. Banks do lend to farmers, but they lend against collateral, and the only collateral a smallholder has is the land he lives off. Borrowing against it to buy hives means putting the household’s subsistence behind a bet on bees. Some farmers take that bet. Most, reasonably, do not.

This is where the case turns, and the turn is easy to miss because it looks like a success. ApiCare had diagnosed a production problem, solved it properly, and proved the solution across four districts with government financing and a documented audit trail. The farmers now knew how to produce clean, market-grade honey at several times their previous volume.

They still had no way to finance the hives, and no record with which to ask.

The constraint had never been in the bee yard. It was in the ledger, or rather in the absence of one. And the same absence that stops a bank lending is the absence that lets a bulk buyer name five hundred rupees for honey worth three times that, because in both cases the farmer arrives at the negotiation carrying nothing but his word.

Four problems wearing one coat

Once the constraint is located in the record rather than the hive, the problem decomposes into four parts that look like one problem and are not. Each fails independently, and no single intervention reaches all four.

Margin capture. The gap between what honey is worth and what its producer receives. ApiCare buys from its farmers at ₹1,000 to ₹1,100 per kilogram, cash on collection, and carries spoilage, adulteration risk, transport and payment delay itself. Bulk offers in the absence of a buyer like this run at half that or less. On the farmer’s side, the margin is not abstract: it pays school fees, it is the difference between eating well and eating adequately, and it is what gets spent at Diwali. On ApiCare’s side, it currently goes back into building the supply chain, into compliance and operations and staying solvent.

The cash cycle. Honey pays twice a year. Household need does not, and in November the two peaks coincide across the whole district at once. This is the mechanism that makes a low offer rational to accept rather than merely unfair.

Provenance. A buyer can today verify the village, the area and the cluster a batch came from. What nobody could verify was the thing Sikkim’s organic status is supposed to guarantee: that this specific honey was handled the way the claim implies. The state has a designation; individual batches had no evidence. A claim without batch-level evidence commands no premium, which is why a decade of organic certification has moved no money toward the beekeeper.

Creditworthiness. A productive farmer is invisible to a lender, not because lenders are unreasonable but because the risk is real and unpriced. Four good years do not protect against one bad season, and with no record of the four good years there is nothing to price against except the bad one a loan officer can imagine. The farmer’s history exists; it has simply never been written down.

Exhibit 1 · What a kilogram of honey fetches, by channel

The row that matters is the second one, and what makes it interesting is that it is triggered by producing more. The farmer moves from the first row to the second by succeeding.

The response

ApiCare’s answer began with its own buying rather than with technology, and the sequence is the point.

The first move was to pay well and pay immediately. Farmers receive ₹1,000 to ₹1,100 per kilogram, in cash, at the moment of collection, and ApiCare absorbs everything downstream: spoilage, adulteration risk, transport, the delay before anybody pays ApiCare. For a household living week to week, the speed matters as much as the rate. A good price in ninety days is not a good price.

The second move was the brand. Honey is sold by village of origin, so that a jar carries the name of the place and the cluster that produced it. This is not sentiment; it is the commercial mechanism that makes provenance worth paying for. If a buyer in Delhi or Dubai will pay more for honey from a named village in an organic state, then the record that proves the village is worth keeping, and the farmer who keeps it captures part of that premium. Without a buyer who pays for provenance, a traceability system is a cost with no counterparty.

Only then does the record layer make sense, and this is where FarmLedger begins. Its design principle is worth stating precisely, because it is the part that is usually got wrong: verification comes from the buyer’s side, not the farmer’s. A farmer can record his produce and send it into the system, but the confirming event is the purchase, entered by the party who paid for it. This does two things at once. It removes any incentive to inflate, because a farmer cannot verify his own transaction, and it anchors the record in the one event that both parties have reason to get right. Over time, a farmer accumulates a documented history of what he produced and what he was paid, which is precisely the artefact a lender needs and has never had.

The traceability has a second use that runs in the opposite direction, up the chain rather than down. Sikkim’s organic status is currently a claim about a jurisdiction. Batch-level records make it a claim about a jar. Giving the state’s designation something verifiable underneath it is how a regional reputation becomes a price, and that premium is the thing that has to exist before any of the rest of this pays for itself.

What this does not solve should be stated plainly. It does not create capital; a documented farmer still needs a lender willing to act on the document, and none has yet. It does not shorten the state’s disbursement cycle. It does not reach a beekeeper who has no phone, and many of the oldest and most skilled do not. And it asks a household living week to week to maintain a record whose benefit arrives months later, which is a harder request than it sounds and is the subject of the last section of this case.

The public and the private

In 2025 the Government of Sikkim selected ApiCare as a private-sector partner under the MSME RAMP scheme, a partnership with the state financed by the World Bank. The selection assessed what a delivery partner would need: equipment, prior training history, adequate facilities.

The structure is worth setting out because the division of labour is the substance of the arrangement.

Exhibit 2 · Who does what under RAMP

What the state supplied that ApiCare could not buy was access and trust. A private company arriving in a village to talk about bees is a stranger with a proposition. The same company arriving with the district administration, sitting first with the Zilla Panchayat and the panchayat, informing the local agriculture officers and then meeting the farmers, is something else entirely. At Mangbrue, the agriculture inspector went on to mobilise more than fifty further farmers for subsequent training, which no amount of marketing spend would have achieved.

What ApiCare supplied that the state could not was delivery: the equipment moved to each venue, trainers hired from within Sikkim who taught in the right language, a functioning apiary to demonstrate on, and the willingness to carry the cost until somebody reimbursed it.

That last clause is where the arrangement strains. The state disburses slowly. Receivables run three to four months behind delivery, and a small business that has already paid for halls, food, transport, equipment and trainers is financing the state’s programme out of its own working capital in the interim. For a company whose farmers need cash on collection, that lag propagates into everything: procurement, growth, the pace at which clusters can be supported. The concept works when both parties operate at the same speed. They do not.

A practical observation for anyone designing this kind of partnership: the institution sanctioning the funds should have someone physically present on the ground, and that person needs high integrity and no sense of entitlement about the position. In the Indian context there are also third-party intermediaries around such programmes who require care. Development financing that is administered entirely at a distance from the villages it is meant to reach will be administered by whoever is standing closest to them.

What rural digital transformation actually costs

The obvious question about a system like FarmLedger is why a farmer would maintain a record whose benefit arrives months later. The honest answer is that he would not, and that anyone who begins with the record has already failed.

The sequence that works runs in the other direction. First, give the farmer the value he is owed, quickly, in cash, for work he has already done. A household living week to week needs disposable liquid money, and if you provide it against their produce they will respect the arrangement and do more of it. Second, teach: sanitation, processing, scale. Third, and only then, ask for the record, at a point where the system is already visibly working in the farmer’s favour and the request is a small addition to something that has been paying him for a year.

Trust here is not a soft variable and it is not built by communication. It is built by execution and by the recognition of value created. Every element of it has a cost that shows up in the field: the trainers hired locally so the teaching happens in Nepali; the entry through the Zilla Panchayat and the agriculture officers before a single farmer is approached; the visits that continue after the certificates are handed out; the cash carried up the hill on collection day.

The hardest thing to change turned out not to be a technique. It was dependency: the expectation, built by years of schemes that arrive, distribute and leave, that assistance is something that happens to a village rather than something it does. A hive distributed is a gift. A hive bought with money earned from honey sold at a fair price is a different proposition, and it produces a different farmer.

Exhibit 3 · Four programmes, four outcomes

Read across the rows and the variable that moves the outcome is not the length of the training. Jitlang ran fifteen days and Beng ran seven, and Beng registered every participant. What predicts adoption is what already existed in the village before ApiCare arrived, and who remained after it left: a working farm next door, dormant skill waiting for a reason, or an official with a stake in the follow-through. Mangbrue received a hive each and a promise, and produced the weakest confirmed result of the four.

Kewzing is the most instructive of them. Beekeeping there had not failed. Honey had been produced traditionally for generations and could not be sold, because of handling and processing rather than any shortage of bees or skill, and so farmers became discouraged and stopped keeping bees altogether. The binding constraint had been downstream of production the entire time. One participant put the whole case in a sentence: “Unhygienic is not an organic; cleanliness plays vital role to product quality.”

What is still open

Three tensions remain unresolved, and stating them is more useful than claiming they are not there.

The first is a question about the model’s dependence on its own buyer. The sequence that makes the record worth keeping begins with someone paying well and paying immediately. ApiCare can do this for its own clusters. It is not obvious what happens in a village where no such buyer exists, and that describes most villages. If the answer is that the record has no value until a fair buyer arrives, then the record is not the intervention; the buyer is, and the record is what the buyer needs in order to scale.

The second is a governance question that the current design does not answer. In the present arrangement, ApiCare is both the principal buyer and the keeper of the record that prices the transaction. The buyer-side verification rule is sound, and it removes the obvious problem of farmers inflating their own numbers. It does not remove the structural one. A ledger that a farmer could take to a lender, or to a competing buyer, has to be credible to people who have no reason to trust the company that wrote it.

The third is about reach. If adoption tracks what a village already had, then the villages with the least are the ones where training will do least, which is the opposite of how a development programme is meant to work. The four sites in Exhibit 3 suggest that what determines the result is presence after departure: an officer, a neighbouring farm, a reason. That is expensive, it does not scale the way a training calendar does, and nobody has costed it.

Unsolved by intention rather than tension: foreign markets and distribution. The honey is good, the story is real, and the route to a buyer outside India remains the gap.

What most people working on rural supply chains do not know, in the view of this operator, is what the problems look like at the grassroots. Programmes are designed for a farmer who can wait for payment, read the form and answer the phone. That farmer exists; he is not the median beekeeper in these hills.

RAMP’s current round of work ends in March 2027. The plan if nothing replaces it is to continue, because the position in 2027 is not the position in 2021. There are now four clusters, 104 trained participants and a substantial body of data about what happened in each village and why.

Which returns the case to where it began. Kewzing’s post-training income was reported at fifty-eight thousand rupees; that figure was an estimate, and the actuals turned out to be different. The estimate was not dishonest, it was simply what was available, because there was no instrument that could have produced a real number. That is the thing the record layer is for, and the measure of whether it works will not be a training statistic. It will be that the next time somebody asks what a village earned, the answer is not an estimate.