Phase 1: baseline and outreach.

Hard deadlines

W0 by ~20 October. P3 in villages in November. Neither can slip without losing the measurement.

Binding constraint

An institutional host. It gates IRB, most funders, and the sampling frame.

4 / 8

Phase 1 milestones done

Everything that could be done unfunded and alone is done.

Households surveyed

The entire point of the programme, and entirely ahead of us.

Phase 1 · in progress

Baseline data collection and proposal outreach

Done

Impact analysis across inflation, fares, agriculture, industry, power and macro
Full audit trail: every search, figure, source and derivation
Illustrated PDF report
Six research proposals with designs, budgets and ethics
Collaborator and funder landscape mapped and flagged for verification
P4 corpus collected — 1,996 articles, three outlets, two windows
P4 pipeline: volume, frame shares with Wilson intervals, exploratory topics
Single-coder validation pass, showing exactly where the classifier fails
Public site with the analysis, designs, data and limitations

Not done

Outreach emails sent to the identified partners
Institutional host secured — the blocker
IRB or local ethics approval
P1 W0 questionnaire finalised, bilingual, piloted with 30 respondents
Survey partner contracted and W0 fielded
Two human coders through the P4 sample for a real kappa
BanglaBERT-class frame classifier and a validated sentiment layer
ACLED registration and a real protest-event series
P3 sites scoped ahead of November planting

Critical path

What has to happen, and by when

Dates are set by the shock, not by preference. Anything after them measures a different thing.

This week

Outreach to hosts and co-investigators

Ask each partner for a sampling frame, an affiliation or host, and quick feedback on the W0 design. Start P4’s human coding in parallel — it needs no permission from anyone.

By ~5 October

W0 instrument finalised and ethics cleared

Bilingual, piloted with 30 respondents, IRB through a host institution or a local ethics board. Fieldwork does not start before approval.

By ~20 October

W0 fieldwork complete

Ahead of the October BPC letter-of-credit cliff and whatever price or duty decision follows it. If W0 slips into November, P1 loses its baseline — that is the whole design.

November

P3 in the field before planting

Boro acreage is decided in November–December. A survey after planting measures regret, not decision-making. P2 scoping runs alongside, in two or three pilot sites.

November → June 2027

Panel waves W1 through W4

November, January, March, then June after the Boro harvest — capturing the October cliff, LDC graduation, peak irrigation and the harvest in one dataset.

Rolling

Public dashboard and policy briefs

Refreshed after each wave, and timed before the Boro and budget decisions rather than after them. Evidence delivered late is history.

Later phases

After the baseline

Phase 2

Field collection

Panel waves across sectors and regions, the Boro survey, and the e-rickshaw ethnography and diaries running in parallel.

Phase 3

Analysis and comparison

Panel fixed effects, sector × wave difference-in-differences, the framing experiment, and a genuine 2022-versus-2026 comparison using the PPRC–BIGD panels.

Phase 4

Publication and release

Papers, policy briefs timed to real decisions, and a de-identified public dataset so the next shock does not start from nothing again.

Changelog

What has changed, including corrections

Corrections are logged here rather than quietly patched. If a number on this site is wrong, this is where it gets recorded.

Date	Change
21 Sep 2026	Public site published — overview, analysis, research plan, collaboration, funding, data and method.
21 Sep 2026	P4 corpus rebuilt on the full source set; Samakal footer boilerplate stripped and outputs refreshed.
21 Sep 2026	Single-coder validation pass run against the v0 keyword classifier. Result: only the protest frame is usable.
21 Sep 2026	Six research proposals drafted with sampling, instruments, budgets and ethics. Collaborator and funder landscape mapped.
21 Sep 2026	Impact analysis and illustrated PDF report published, with the full audit log behind them.

One standing caveat Some outlets describe 21 September as the government’s second fuel hike of 2026. This analysis counts three since March, because the June change of Tk 5 applied only to petrol and octane — for diesel and kerosene it is the second. Both counts are defensible; they are counting different things.