Could AI help Bulog get cheap corn to Java’s egg farmers on time?

Cheap corn for egg farmers, often in the wrong place

On Java, Indonesia’s most crowded island, thousands of small and medium farms keep hens for eggs. Blitar, a district in East Java, is one of the biggest clusters. Many of these farmers mix their own chicken feed, and about half of that feed is corn. [6]

Corn is cheapest just after the harvest and dearer in the lean months, from about September to December. In Blitar it cost Rp6,400 to Rp6,500 a kilo in June 2026 [10] and Rp7,800 to Rp7,900 by November [5], a rise of more than a fifth. At the same time, egg farmers told Indonesia’s parliament that eggs were selling below what it cost to produce them. [6]

The government tries to soften the lean months with cheap reserve corn. Bulog, the state company that holds Indonesia’s food reserves, buys corn at harvest and sells it to small egg farmers at a fixed low price: Rp5,000 a kilo at its warehouse and at most Rp5,500 a kilo delivered to the farm (about 33 US cents). [2] Two government bodies set the rules: the National Food Agency (Bapanas) and the Ministry of Agriculture (Kementan). Farmers cannot buy directly. A registered cooperative, a member-owned group of farmers, orders for them through an online system called Klik SPHP Jagung. [2] [9]

The cheap corn is rationed. The government sized each farm’s allowance at three months of need for the smallest farms, two months for small farms and one month for medium farms. [2] This article follows Bulog’s side of the job: getting that rationed corn to the farms when they need it.

That is harder than it sounds. Much of Bulog’s corn comes from other islands, such as Sulawesi and Nusa Tenggara, while the hens are on Java. Bulog ships the corn across the sea to Java. In October, the agriculture ministry said Bulog’s stock in Lampung, on Sumatra, covered local farmers’ needs. Farmers in East and Central Java, meanwhile, had been short of corn. [3] The chair of an egg farmers’ cooperative in Blitar asked for smoother, steadier deliveries. [5]

By 5 November, cooperatives had bought about 235,000 tonnes, 60.75% of the year’s 387,000-tonne limit, for 11,658 farmers through 151 organisations. [5] Nobody has published why particular orders arrived late or were not filled.

Who does what for one order

The research describes one order of cheap corn like this. Step 4 is our reading of the published rules, because Bulog’s internal procedure is not public.

  1. Bulog’s buyers buy harvest corn from growers and traders at the government price of Rp6,400 a kilo, dried to 14% moisture. Bulog checks each lot’s moisture and stores it. [7]
  2. Bulog’s planners decide how much corn each region should get and ship it from the growing islands to Java. In one shipment, 15,000 tonnes landed at Tanjung Perak, Surabaya’s port, and Bulog split it into 5,000 tonnes each for East, Central and West Java. [3]
  3. A cooperative collects its members’ requests and places one purchase order in Klik SPHP Jagung. For the first release in Magetan, in East Java, two cooperatives ordered 250 tonnes. [9]
  4. A Bulog officer checks the order against the list of registered farmers and their remaining allowance. The officer then gives it stock from the local warehouse and sets a release date. If the warehouse is short, the officer makes the order wait for a shipment, fills it from another warehouse, or turns it down.
  5. The cooperative pays Rp5,000 a kilo, collects the corn at the warehouse and delivers it to members for no more than Rp5,500. [2]
  6. The farmer mixes the corn into feed. When the allowance runs out, the farmer buys corn at the market price, sometimes on credit at a higher price. [5]

Two features of this chain matter for Bulog. A ship from the growing islands takes time to load, sail and unload. And the first firm sign that a warehouse needs corn is a cooperative’s order. The published reports suggest that Bulog ships in large batches after shortages appear, rather than ahead of them. That is our inference from the reports, not a documented practice. [3]

Bulog’s problem in one sentence: Bulog has to get its cheap, rationed corn to eligible egg farmers on time through the lean months, but orders arrive one cooperative at a time while its stock is spread unevenly across the islands. The farms waiting for that corn are already squeezed by feed costs.

Improve: a warning before a warehouse runs out

This change targets step 2, when Bulog’s planners decide to ship corn to a Java warehouse, and the waits that decision can cause at step 4. Here is that part of today’s process again.

Today:

  1. Cooperatives order, and a Bulog officer fills each order from the local warehouse.
  2. A Bulog planner watches the regional stock figures and plans shipments. How the planner forecasts demand is not published.
  3. When the planner orders a shipment, the corn arrives days or weeks later. If the warehouse runs out first, orders wait.

With the change:

  1. Cooperatives order, and the officer fills orders as before.
  2. Every day, software reads each warehouse’s stock, the orders already placed, the ships due and how long past shipments took. It estimates a range of dates on which that warehouse could run out.
  3. When that date comes closer than the shipping time plus a safety margin, the software alerts the planner. The planner sets the margin. The planner then checks where spare corn and ships are available and decides whether to send corn. The software books nothing.

The planner’s timing matters because a ship cannot arrive faster than it sails. A warning is useful only if it comes before that time runs out.

What the AI would do. Three pieces of software would work together. A forecasting model would estimate how much corn the cooperatives around each warehouse will order over the coming weeks, with a range rather than one number. [A] A second model would flag a sudden jump in orders, so the forecast is redone at once. A third would estimate how soon the warehouse could run out, given the ships already on their way. [B] Ordinary software, not AI, would do the stock arithmetic and compare the run-out date with the shipping time.

Why AI, and not a spreadsheet? A planner could keep a simple sheet: stock left, divided by recent daily orders, gives days of cover. That sheet is the cheapest alternative, and the AI has to beat it. The AI’s possible advantage is that ordering changes with the season and with the programme’s rules, which changed several times in 2026. [2] [4] A model trained on past orders may catch those patterns. But past orders are capped by each farm’s allowance, so they understate what farmers wanted. That is why the forecast is the uncertain part. [A]

Related Business Opportunity:  Could local soybean deliveries cut costs for Central Java’s tofu makers?
Step Today With Improve
What starts a shipment A planner’s judgment Possibly after shortages appear A run-out warning Sent before shipping time runs out
Forecasting orders Method not published Regional plans and batches Forecast per warehouse A range for the coming weeks
Spotting a surge Reports and complaints From cooperatives Software flags it The forecast is redone
Deciding and booking Unchanged The planner decides; Bulog managers agree port priority; cooperatives order, pay and collect

Who would use it: Bulog’s own regional planners, with one analyst checking the data. Bulog could build it or buy it from a software supplier. The idea has a clear limit. A warning moves corn that already exists; it cannot create corn. In 2025 Bulog bought only 101,968 tonnes against a target of one million, while feed mills competed with it for the harvest. [7] If the corn was never bought, if a farm’s allowance is used up or if a cooperative lacks cash, an earlier warning does not help. Moving corn early can also leave the sending region short.

Reinvent: promise each delivery date before the order

The larger change does not speed up the old route. It changes what starts the work. Here is today’s process again, from the cooperative’s side.

Today:

  1. A farmer tells the cooperative how much corn they need.
  2. The cooperative places an order when it has enough requests and cash.
  3. A Bulog officer fills the order from whatever corn is in the warehouse that day. A cooperative that orders later can find the corn already gone.
  4. Bulog’s reports count tonnes that cooperatives bought, not corn that reached farms on time.

With the change:

  1. Before the lean months, each participating cooperative keeps a calendar of its members’ needs: who needs how much, by when. Cooperative staff type it in or send it as ordinary messages. Software turns the messages into calendar entries, and the cooperative staff check each one. [C] Only registered farmers’ needs within their allowance count. Need beyond the allowance is noted separately and never promised.
  2. A forecasting model estimates a likely range of need for each cooperative and each delivery window. The model flags calendars that look incomplete, and cooperatives that sent none, so Bulog’s planners can follow up with those cooperatives. [A]
  3. Bulog’s planners turn the confirmed entries into dated delivery promises. Planning software proposes which lots of corn, already bought and checked, to set aside for each promise. It also proposes how much to keep free for surprises. The planners approve. [D]
  4. Scheduling software plans each promise’s whole route: warehouse, ship or truck, and a quality check. Shipping and trucking firms confirm space before the cooperative’s ordering deadline. [D]
  5. The cooperative still places its formal order in Klik SPHP Jagung. A Bulog officer re-checks eligibility, allowance, payment and quality as today. A promise never skips those checks.
  6. The cooperative delivers the corn. The farmer confirms the amount, the date and the quality. Only then does Bulog close the promise. A late, short or spoiled delivery stays open, with its reason, until a Bulog planner resolves it.
Step Today With Reinvent
What starts the work A cooperative’s order Placed when requests and cash allow A needs calendar Kept weeks before ordering
Who gets which corn First come, first served From that day’s warehouse stock Lots set aside For each dated promise; some kept free
Transport Booked after shortages Our inference from batch shipments Booked per promise Carriers confirm space in advance
When the job is done Corn collected Reports count tonnes bought Farm confirms receipt Failures stay open with a reason
Rules, prices, checks Unchanged The government sets allowances and prices; Bulog checks every order; cooperatives pay and deliver

Where AI does the work. Only three parts use AI. Software reads cooperatives’ messages into calendar entries. [C] A forecast estimates need where calendars are missing or optimistic. [A] Later, a risk model could warn when a promised date is likely to slip, but only after Bulog has a season of delivery records to learn from. [E] The core needs no AI: the allowance ledger, the planning and scheduling software, the record of each promise, and every approval. Proven planning software does those jobs. [D]

Why this is more than a set of fixes. Improvements to today’s process, such as the warning above, could deliver most of the benefit of moving corn earlier.

What they cannot do is hold a particular lot and truck for a farm’s date before the order arrives. With a promise, a cooperative that orders later is not left with corn that others took first. It can plan its cash and trucks around a known date. And Bulog remains responsible for each delivery until the corn reaches the farm, so its records show whether farms actually got their corn. If a trial showed that warnings alone deliver as much corn on time, the redesign would not be needed.

Where it reaches beyond today. The change goes outside Bulog’s own work. The government would have to approve a rule letting Bulog set corn aside before an order arrives, and whether Bulog may do that is not known. Cooperatives would take on new work, keeping calendars and recording receipts, and Bulog would have to pay for that time. Nothing can be charged to farmers above the Rp5,500 limit. People still make every decision that matters: the government sets the rules, planners approve every promise, warehouse staff decide whether corn is fit to release, and cooperatives place every order.

The limits worth stating. Well-organised cooperatives would send calendars first and could take corn that the smallest farms need. Cooperatives might overstate needs to secure corn, leaving it uncollected. Corn released into growing areas at harvest time could push down growers’ prices, which the rules already forbid. [2] And if the real cause of late orders is that Bulog never bought enough corn, promises cannot fix it. What Could Work’s stored research on this case also sketched a different business: a private corn trader selling farmers the corn they need beyond their allowance. That idea serves a different market. It would sit alongside this one rather than compete with it.

What each change could offer

Neither idea has been tried, so the table judges potential: how well each could do if it were built and run sensibly. The ratings draw on the research and on our AI Capability Atlas, What Could Work’s own reference on what today’s AI can and cannot do.

Related Business Opportunity:  Could a cashew-buying company pay for orchard upkeep in Dompu, Sumbawa?

Favourable for the business Mixed Unfavourable

Potential Improve: run-out warning Reinvent: dated delivery promises
Technical feasibility ModerateForecasting and surge detection are proven for regular patterns, but the programme changed its rules in 2026 and past orders were capped by allowances; estimating when a warehouse runs out is less reliable because past run-outs are few. [A] [B] ModerateThe core is proven planning software that needs no AI. The forecast and message reading are conditional, and the delay-risk model needs delivery records that do not exist yet. [C] [D] [E]
Economics Unpriced, modest costCosts are a data review, software, a planner's and analyst's time and the extra shipping and storage of moving corn early. No cost or benefit figure exists; it pays only if avoided emergency moves and spoilage, or a budget decision that extra on-time corn is worth paying for, cover it. [F] Unpriced, heavier costAdds a regional system, paid cooperative time, corn held longer and advance transport bookings. Bulog pays everything, farmers cannot be charged above Rp5,500/kg, and no figure shows avoided costs. [F]
Customer value Moderate, if ships can actFarmers get their allowance sooner where spare corn and ships exist elsewhere. It adds no corn when none was bought or an allowance is used up. [7] High, if timing is the causeA cooperative knows its delivery date and can plan cash and trucks; later orders are not left with corn others took; Bulog is answerable until the farm confirms receipt.
Competitive positioning Easy for Bulog to adoptBulog holds the stock, records and planners and needs no new rules. A software firm could supply it, but there is little room for a new company. Bulog's alone to buildThe government runs the cheap-corn channel through Bulog, so no entrant can build an advantage; a software supplier could run the system. For Bulog the gain is doing the job the government gave it, not winning customers.
Risks ModerateFalse alarms waste planners' time; early shipments can leave the sending region short or hold corn longer. The planner decides every move and the warning can be switched off. HighOrganised cooperatives could take corn the smallest farms need; needs may be overstated; releases at harvest could depress growers' prices; Bulog's authority to set corn aside is unknown. [2]
How it would be tested Bulog's planners replay past planning dates at two Java warehouses and one shipping route against planners' real decisions and a days-of-cover spreadsheet, then run it beside a planner for a season with nothing booked automatically. Bulog first sorts 2026's late orders by cause in one region, then runs one lean season on paper with a group of cooperatives, comparing on-time usable tonnes at farms, by farm size, with today and with run-out warnings also given the cooperatives' calendars.
Our assessment of potential: how well each could do if it were built and run sensibly. Ratings are editorial judgments built from the research and from what our AI Capability Atlas says about each kind of AI; none is a measured result. The colour shows whether a rating helps or hurts the business case, so high risks show red. The last row says how each idea would be tested; it is not rated.

Which is worth pursuing?

We ask a conditional question of each: if the arrangement can be made to work and the AI part is practical, would it create enough value to justify the change?

Improve: potentially yes

The warning is cheap to try, changes no rules and can be switched off. But it only moves corn that already exists.

It is worth it if three things hold. Bulog must have spare corn somewhere and ships that can arrive within the warning time. The warning must beat a planner’s simple spreadsheet. And the extra cost of shipping and storing corn earlier must be covered, either by avoided emergency shipments and spoilage or by a government budget decision that extra corn delivered on time is worth paying for. It is best placed with Bulog itself, because Bulog holds the stock, the records and the planners. A software firm could supply it, but there is little room for a new company here.

Reinvent: potentially yes, but only after a check

Dated promises could get more corn to farms on time with the same stock and allowances. They would also give Bulog, for the first time, a record of whether corn reached each farm. But they cost more: new planning software, paid cooperative time, corn held for longer and transport booked in advance. Bulog would pay all of it, while farmers and cooperatives gain most.

It is worth it only if late orders are mainly caused by corn being in the wrong place at the wrong time. Bulog would also need permission to set corn aside, cooperatives would need to plan needs accurately, and the smallest farms must not lose out. Finally, the promises would have to beat simple warnings in a trial season. Only Bulog can run dated promises, because the government runs the cheap-corn channel through Bulog. A software supplier could run the system for it.

Worth pursuing: Improve, first. Both ideas rest on the same unproven belief: that corn arrives late because it is in the wrong place, not because Bulog never bought enough. The warning is the cheaper way to test that belief and needs no new rules. Bulog should build dated promises only if a trial season shows that warnings still leave cooperatives that order later without corn.

What would have to be tested

Before either: Bulog’s regional team would rebuild its 2026 order records for one region, such as East Java, and sort the late or unfilled orders by cause. Possible causes are corn in the wrong place, no corn bought, an allowance used up and a cooperative without cash. If corn in the wrong place is rare, both ideas fail.

For Improve, Bulog’s planners would replay past planning dates at two Java warehouses and one shipping route. At each date, the warning sees only what was known then. They would compare its warnings with their real decisions and with a simple days-of-cover spreadsheet. They would measure how much usable warning time it gives beyond the shipping time, how many shortages it misses and how many false alarms it raises. Then Bulog would run it beside a planner for a season, with nothing booked automatically.

For Reinvent, Bulog would run one lean season on paper with a group of cooperatives. The system makes promises, Bulog releases nothing on them, and farms record what actually arrives. Bulog would then compare on-time, usable tonnes at farms, by farm size, with today’s process and with run-out warnings that are also given the cooperatives’ calendars. The idea fails if it does no better than warnings, or if the smallest farms lose out.

The missing evidence, and what each piece decides:

  • Why orders were late or unfilled in 2026. No order-level record is public. This decides both ideas.
  • What Klik SPHP Jagung and Bulog’s warehouse records hold, with real dates and times. This decides whether either system can be built.
  • Shipping times and spare stock by route and month. This decides whether a warning can be acted on.
  • Bulog’s cost of an emergency shipment and of corn lost in storage. These figures are not public. They decide whether either idea pays for itself in cash.
  • Whether Bulog may set corn aside before an order. This decides whether Reinvent is allowed.
  • How accurately cooperatives can state needs weeks ahead. This decides whether promises can be kept.
Related Business Opportunity:  East Java's feed mills already buy local fishmeal. Could a new supplier sell it cheaper?

Research Notes

What this investigation is based on

What Could Work’s stored desk research on Indonesia’s cheap-corn programme for small egg farmers (the news reports, government releases, one academic study and one US Department of Agriculture report listed below), and our own analysis of how the work is done and where AI could change it. No interviews, internal Bulog data, farm surveys or pilots were used. Our own archive’s search was unavailable while we wrote this, so no earlier What Could Work article is cited.

Whose problem this article takes, and why. We considered four parties: the egg farmers, the cooperatives that buy for them, a private corn distributor in Blitar, and Bulog. All four carry part of the problem and have a reason to act. Only Bulog runs the process, makes its operating decisions, has the money and staff, holds the records, and reaches every cooperative in the programme. The farmers, the cooperatives and the distributor each lack most of these. So the article follows Bulog, acting for the farmers and cooperatives it serves.

Sources

  1. Poultry Indonesia: Kementan Siapkan Skema Integrasi Jagung dan Peternak, Bulog sebagai Penyangga Pasokan (November 2026). The news report behind this investigation: a prepared farmer–cooperative–Bulog scheme, not yet operating.
  2. Investor Trust: Pemerintah Tambah 150.000 Ton Jagung SPHP, Peternak UMKM Jadi Prioritas (8 September 2026). Prices, distribution through registered cooperatives, allowances of three, two and one months, priority for the smallest farms, care for harvest-time grower prices.
  3. Ministry of Agriculture: Mentan Amran Percepat Penyaluran SPHP Jagung (12 October 2026). Stock in Lampung covering local need while East and Central Java had been short; shipments from the growing islands through Tanjung Perak; the 15,000-tonne split; port priority. Official account.
  4. Periskop: Bulog Gelontorkan 150 Ribu Ton Jagung Murah, Peternak Mikro Diprioritaskan (20 October 2026). Bulog’s stock and progress figures, which use different totals over time.
  5. Liputan6: Peternak Bisa Titip Jagung di Bulog, Diambil saat Paceklik (6 November 2026). Progress to 5 November; Blitar prices in November; a farmer on buying on credit; a cooperative chair on smoother deliveries.
  6. Fortune IDN: Peternak Mengadu ke DPR Soal Harga Telur Murah dan Pakan Mahal (7 October 2026). Farmers’ testimony to parliament: feed about half corn, eggs below cost. Figures are as reported, not audited.
  7. US Department of Agriculture, Indonesia Grain and Feed Annual 2026 (copy hosted by APEDA). The government purchase price; Bulog’s 2025 purchases against its target. Read in sections.
  8. InfoPublik: Pemerintah Perkuat Tata Kelola Sektor Peternakan dari Hulu ke Hilir (27 October 2026). Stated farmer need of about 1.8 million tonnes a year, without a method.
  9. ANTARA: Bapanas salurkan 250 ton jagung SPHP di Magetan (9 May 2026). The first release in Magetan; ordering through Klik SPHP Jagung.
  10. ANTARA Jatim: Peternak Rakyat Blitar minta perlindungan pemerintah (1 June 2026). Blitar corn prices in June, as stated by farmers.

Our explanations and assumptions

Where a note discusses a kind of AI, it draws on our AI Capability Atlas, What Could Work’s own reference on what today’s AI can do. For each kind of task it records what goes in and what comes out, how proven the approach is, its typical failures and how to check for them. Established means proven methods exist for a narrow, well-defined task. Conditional means it can work, but reliability depends heavily on the setting and has to be tested there. Exploratory means it is still unproven beyond experiments. The Atlas describes AI in general; nothing in this article has been tested in the business it describes.

  1. Forecasting orders, and spotting surges. A forecasting model learns from past orders, with their dates, to estimate future orders and a range around them. A second model watches the stream of orders and flags a sudden change. Our AI Capability Atlas rates forecasting as established for regular, repeating patterns and conditional when data are sparse or the rules keep changing. It warns that a single figure hides uncertainty and that a forecast should be checked against a simple baseline. It rates spotting a defined change in a stream of events as established for steady signals and conditional when conditions shift. A seasonal swing can look like a surge.
    How this shaped the article: the cheap-corn programme changed its rules several times in 2026, and past orders were capped by allowances, so we treat the forecast as conditional and rate both ideas’ feasibility Moderate. It is why the AI must beat a days-of-cover spreadsheet in Improve and a simple seasonal forecast in Reinvent, or be dropped.
  2. Estimating when a warehouse runs out. A time-to-event model estimates how long something will last, here a warehouse’s stock, as a range of dates. The Atlas rates this established only for equipment with long records of past failures, and conditional where such events are rare. It warns that acting on an estimate changes the future it predicts: a shipment sent early means the run-out never happens. The stock arithmetic and the simulation of arrivals are proven planning software, not AI; the Atlas marks that as conventional.
    How this shaped the article: it is why the warning shows a range, why the planner sets the safety margin and decides every shipment, and why we describe the stock sums as ordinary software. The warning is tested first on past dates, using only what was known at each date.
  3. Reading cooperatives’ messages into calendar entries. A language model reads a message and pulls out who needs how much, by when, and shows the words it used. The Atlas rates this established for well-defined kinds of entry and conditional for complicated conditions. It warns that reading a message does not make what it says true or complete. The analysis also notes that messages in regional languages need testing first.
    How this shaped the article: it is why cooperative staff check every entry, and why only registered farmers’ needs within their allowance enter the calendar.
  4. Setting corn aside and planning routes. Planning software decides which lots of corn to hold for which promises, and in what order to move them, within rules and limits. The Atlas rates allocation, scheduling and tracking each step of a workflow as established, but as proven planning software, not AI, when the rules are clearly written down. They become conditional when conditions keep changing. It warns that forecast errors can outweigh what the planning gains, and that a plan can break when a ship is late.
    How this shaped the article: it is why we say the core of Reinvent needs no AI, why planners approve every promise, and why some corn is kept free for surprises.
  5. Warning when a promised date may slip. A risk model estimates the chance that a promised delivery fails, from what is known when the promise is made. The Atlas rates this established for stable, well-recorded outcomes and conditional when past records are biased or the population changes. It warns that a risk score does not explain the cause. It needs a record of past deliveries and their outcomes, which Bulog does not publish today.
    How this shaped the article: it is why this model comes last in Reinvent, only after a season of delivery records exists.
  6. What each idea would cost. No cost or benefit figure is available, so neither idea’s payback can be calculated. The analysis listed what each would need. Improve: a data review, links to stock, order and shipment records at two warehouses and one route, the forecasting software and its testing, training for a planner and an analyst, then running costs for computing, maintenance, reviewing alerts, and the extra shipping and storage of moving corn earlier. Reinvent, for one region: a system linking promises to Klik SPHP Jagung and warehouse records; the forecast and risk models; about 8 planner-months a year (two planners from September to December); about 40 cooperative staff-months (ten cooperatives for four months); about 17 weekly planning runs; about 1,250 tonne-months of extra storage (ten cooperatives ordering about 125 tonnes each, held one month longer, scaled from the two Magetan cooperatives’ 250-tonne order [9]); and about 40 advance transport bookings. All quantities are the analysis’s assumptions, not quotes. Spreading one-off costs over three years, each idea pays in cash only if Bulog’s avoided emergency shipments, storage losses and failed collections cover its yearly cost. Otherwise it pays only if the programme decides the cost per extra on-time tonne is worth it.
  7. Other ideas we considered. For Improve: a record of why each order was late or turned down, with AI suggesting the reason from an officer’s notes. It could reveal the real cause, but labels alone do not deliver corn, and a simple drop-down menu may do as well. We judged the warning to have the clearer link to on-time corn. Other options, such as sensors in warehouses and checks of harvest images, lacked the records or equipment they need. For Reinvent: small local stores run by cooperatives were dropped because the shortage is between islands, not between a warehouse and a farm, and storing corn in more places raises the risk of mould. Swapping ownership of corn between regions instead of shipping it was dropped because farms and surplus corn are on different islands, so few swaps would match.
  8. Dollar amounts use about Rp16,500 to the US dollar, our rounding. Bapanas’s and Bulog’s roles, and the description of the egg clusters, are general background from the sources. The “Today / With the change” tables and the steps of each change are our summaries of the recorded process and the proposed designs.

Evidence status of the main claims

Claim Status
Cheap corn’s fixed prices and cooperative channel Observed [2] [9]
Allowances equal three, two and one months of need by farm size Observed [2]
East and Central Java ran short while stock sat elsewhere Reported by the ministry [3]
60.75% of the year’s limit bought by 5 November Reported by Bulog [5]
Bulog ships in batches after shortages appear Inferred [3]
Why individual orders were late or unfilled Unknown
Run-out warnings get more corn to farms on time Hypothesis
Dated promises beat warnings alone Hypothesis
Bulog may set corn aside before an order Unknown

Open questions

  • How much of the late or unfilled corn in 2026 was caused by location and timing, and how much by too little corn bought, used-up allowances or cooperatives without cash? Bulog’s regional operations could answer this from its records.
  • Can small farms afford to buy and store harvest corn themselves, and how much does buying on credit add? No farm survey was found.
  • Do private distributors already sell corn in advance at a fixed price, and would farmers pay a deposit for it? This bears on the private trader idea, not on Bulog’s channel.
  • How much corn does Bulog lose in storage, and what does an emergency shipment cost?
  • Is the 50 ppb limit for aflatoxin, a mould poison, that the research reports actually applied at intake and release?

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *