Figures the daily Policy Advisor carried from the press, checked against the twin's own tables for the same quantity and period. Listed here: every figure that differs by more than 2%, largest gap first.
Every advisor fact
35 of 1,398 facts name a quantity and period that a twin table holds; the rest have no series and get no verdict. The checks cover NBR revenue collection (National Board of Revenue statements), the FY2026-27 budget aggregates (Finance Division, Budget in Brief), the FY2026-27 social safety net total (Social Security Budget Report) and ministry allocations (Annual Financial Statement). NBR figures are the grand total of customs, VAT and income tax, cumulative from July; the monthly statements are provisional and may be revised. A “revenue target” that says neither total nor NBR could mean either, so it gets a verdict only when its amount matches the figure the twin holds for that year. A sector named without its ministry gets none: the twin holds ministry allocations, not sector totals.
Press vs data
In FY2025–26, the National Board of Revenue (NBR) collected Tk 476,000 crore against an initial budget target of Tk 564,000 crore (revised to Tk 518,000 crore), generating an Tk 88,000 crore shortfall from the original target.
Daily Industry, 2 Aug 2026 · in the advisor of 15 Aug 2026
NBR revenue collection, cumulative, grand total, FY2025-26, July to June
Press Tk 476,000 crore · Data Tk 415,473 crore · Gap +Tk 60,527 crore (+14.6%)
Data: grand total, July to June = customs 112,119 + VAT 157,734 + income tax 145,620; provisional statement, subject to revision. NBR monthly revenue statement, listed 4 Aug 2026.
Claims as reported in the press and summarised in BDPolicyLab's daily Policy Advisor; they are not official statistics unless their source is an official one. Each figure is checked against the twin's data for the same quantity and period: it agrees within 2% of the data value and differs otherwise, with the gap read as press minus data. A fact with no series is one no reviewed rule matches to a data series, so it gets no verdict.