On 21 September 2026 Bangladesh raised diesel, petrol, octane and kerosene by Tk 20 a litre — the third move in five months. What that does to prices is now reasonably well documented, including here. What nobody has captured is what it does to people: who cuts food first, whose daughter stops taking the bus, who gets blamed, and who quietly absorbs the shock on everyone else's behalf.
r1_Fuel is an open, independent research programme built to measure that — and the window to collect a pre-shock baseline closes in weeks, not months. Everything here is public: the analysis, the derivations, the code, the limitations.
The automatic pricing formula, introduced in March 2024, is supposed to adjust monthly. It was not applied between March and September 2026 while global prices doubled. Six months of suppressed adjustment were then released in a single Tk 20 move. The volatility people are experiencing is manufactured by the decision to freeze — not by the formula.
Furnace oil is the line that gets missed. It went Tk 70.10 → 94.69 → 113.54 between March and May — +62% in two months — and it is the fuel now carrying the power grid. That is the cost base underneath every loadshedding story, and underneath the next electricity tariff.
Each of these is worked through, with sources and derivations, in the full analysis.
The “% of income per 50-litre tank” index is useful for ranking countries and useless for predicting household distress here, because private car ownership in Bangladesh is a top-decile phenomenon. Almost nobody in the affected population buys 50 litres of octane. The transmission runs through diesel (bus fares, freight, irrigation, generators, trawlers), kerosene — the most regressive line in the gazette and the least commented on, burned by the poorest households for light during 7–10 hour rural outages — and furnace oil, which sets the electricity tariff.
BPC asked for Tk 187 diesel and got Tk 135. The flat Tk 20 design over-recovers on petrol, octane and kerosene — taxing the motorcyclist and the kerosene user to partly cross-subsidise diesel — and still leaves diesel Tk 52 short. BPC's working capital was Tk 12,368 crore on 6 September against a stated need of Tk 15,000–20,000 crore, and it has already drained Tk 19,500 crore out of development project accounts. A second hike, or an equivalent duty cut, is near-certain before end-2026.
Anyone planning on Tk 135 diesel being the new normal through the Boro season is planning on the wrong number. The Tk 20 buys roughly Tk 700–750 crore a month of cash flow est. against a loss run-rate near Tk 3,800 crore a month.
Duties and taxes on diesel were Tk 32.44 a litre in August — about 24% of the new pump price. The state collects Tk 32 at the revenue window while losing roughly Tk 69 at the BPC window: a net public-sector loss of about Tk 37 a litre est. with a Tk 32 wedge inside it. Zeroing the diesel duty would close 62% of the remaining gap est. with no increase in the pump price at all. That it has not been done is a revenue-protection choice, and it is the most defensible thing to criticise about the design.
About 70% of irrigation pumps run on diesel; Boro is around 55% of national rice output; peak irrigation is December to March, and planting is decided in November. The cost is already incurred; the price effect arrives in 2027. This hike adds roughly Tk 420 per bigha, the full move since March about Tk 840 — some Tk 2,140 crore across the coming crop est. Meanwhile the government held the Boro procurement price at the old level. Rising input cost against a fixed output price compresses margin, and margin decides acreage.
The National Pay Scale was gazetted on 19 September; the hike took effect on the 21st. Roughly 33 lakh public employees and pensioners were indexed to inflation — the first award in eleven years. Roughly 7.5 crore private and informal workers were not. National wage growth is 8.05% against 8.26% inflation, so real wages were already negative before this hike. That is the honest form of the grievance, and it is also, for a researcher, an unusually clean natural split.
The pay award did not cause the fuel hike — Hormuz did. It removed the fiscal cushion that would have let the state phase the shock more gently, and it did so in favour of about 2% of the population who are also the group most insulated from the consequences.
The fuel hike tightens a loop that was already running. Reading it as two coincidental nuisances is what makes the standard policy response — crack down on the rickshaws — actively harmful.
The operating-cost wedge that drives it. At Tk 165 a litre and about 15 km/l, a petrol three-wheeler costs roughly Tk 11 per km in fuel est. A battery three-wheeler draws 4–9 kWh a day for 100+ km — about Tk 0.4–0.8 per km est., and for the 93% of Dhaka charging points that are unauthorised (48,136 against 3,300 legal ones) the marginal cost to the operator is effectively zero. The wedge was already 15 to 1. This hike widened it by another 14%.
Factories and shops fire up generators, diesel demand rises precisely when BPC cannot fund October letters of credit, physical shortage follows, and black-market premiums appear — diesel has been selling Tk 15–80 above the regulated price in rural markets. Tk 135 is a floor, not a ceiling, wherever supply is short. Which is exactly where the people with the least buying power are.
It is the mechanism by which a household on Tk 12,000 a month keeps moving through a 38% petrol shock. Around 21% of battery-rickshaw drivers are ex-farmers — and farm incomes are being squeezed by the same diesel shock. Suppressing the fleet without a substitute would transmit the oil shock to the informal poor far more completely than it currently does. The fix is metering and charging infrastructure, not prohibition: it converts 500 MW–1 GW of theft into billed revenue, makes the load visible to dispatch planning, and leaves the livelihood intact.
These are the dates a measurement programme has to be standing before, not after.
How households cope, what they give up first, who inside the house absorbs the cut, whom they blame, and whether they think it is fair — none of that is in a CPI print. That is the gap r1_Fuel is built to fill.
Two policies, 48 hours apart, split the population cleanly by occupation: ~33 lakh public employees indexed, ~7.5 crore private and informal workers not. Relative deprivation is usually hard to identify. Here it was handed to us, dated and gazetted.
BPC cannot fund its October letters of credit, so another adjustment is near-certain, and Boro will pass irrigation costs through after that. Anything measured in the next few weeks is a genuine pre-shock baseline. That window closes fast.
The 2022 hike (diesel +42.5%) produced protests, fare disputes and a lot of commentary — and very little systematic household-level sociology. 2026 can be done properly, and compared back to 2022 using panels that already exist.
P1 is the backbone the others can attach to. P4 needs no fieldwork and has already started. Full designs, sampling, instruments, analysis plans and ethics are in the research plan.
| # | Study | The question it answers | Indicative cost | State |
|---|---|---|---|---|
| P1 | The Fuel Shock Panel 1,500–2,500 households, 5 phone waves to mid-2027 |
How coping, well-being and trust in the state move over the shock cycle — and how that path differs between public, formal-private and informal households. Carries an embedded framing experiment on perceived fairness. | $35–60k | Needs a host + W0 by ~20 Oct |
| P2 | Charging the City Ethnography, 4 sites, 60–80 life histories |
How drivers, garage owners, charging-point operators and local power brokers govern an illegal energy economy — and how the hike redistributes risk along that chain. | $15–30k | Scoping |
| P3 | Boro Decisions 6 villages, ~600 plots, pre-plant and post-harvest |
How diesel cost, supply uncertainty and a frozen procurement price change how much Boro gets planted — and who in the agrarian hierarchy absorbs it. | $20–35k | Must field in November |
| P4 | Who Gets the Blame? Bangla + English news corpus, 2022 vs 2026 |
How responsibility and fairness are framed in public discourse, and which frames track offline protest. | $5–15k | Corpus built · 1,996 articles |
| P5 | Cash or Cheap Fuel? Conjoint experiment, n≈2,000 |
Whether people prefer universal cheap fuel or targeted cash, and how trust and perceived corruption shape that — directly informing the choice the government faces this quarter. | $8–15k | Can embed in P1 |
| P6 | Whose Mobility? Gender module, paired-respondent subsample |
How the fuel shock reshapes gender inequality inside the household — who cuts back first, whose mobility is curtailed, and the gender gap in awareness. | $20–35k | Pairs with P1 |
Costs are indicative and built from per-interview rates, not a target. The full breakdown and what each increment actually buys is on the funding page.