Record rise in tax returns, yet collection falls 10.3 percent

An unusual picture has emerged in Pakistan's tax system. According to the latest data released by the Federal Board of Revenue (FBR), a total of 7.56 million returns were filed in tax year 2026, about 27.2 percent more than the 5.94 million of the previous year. Yet admitted and collected tax stood at 100.6 billion rupees, 10.3 percent below the previous year's 112.18 billion rupees. In other words, the number of filers rose while the tax declared per filer fell. This contradiction—rising returns against falling revenue—is the real story, buried beneath the small headline figures. First, the arithmetic must be reconciled. Although the headline total is given as 7.56 million returns, summing the category figures gives about 7.5442 million—a shortfall of roughly 15,805 returns. The same one-directional gap (about 10,183) appears in last year's 5.94 million. The tax figures, by contrast, reconcile almost exactly. This asymmetry suggests the headline total is an independently sourced control figure rather than a sum, and that the category table is not a complete partition; presenting it as one would be wrong. The element given the most prominence—222 new filers and 9.25 million rupees under the Tax Assan scheme—is negligible against the aggregate. Of roughly 1.62 million additional returns, these 222 are just 0.014 percent; of 100.6 billion rupees of admitted tax, 9.25 million is only 0.0092 percent. Even all 1,945 scheme filers at 179 million rupees are only 0.178 percent of total tax. The scheme is arithmetically incapable of explaining either the 27.2 percent return growth or the 10.3 percent revenue decline. The category breakdown makes the picture clear. The non-salaried category filed 5.06 million returns, up 37.9 percent, but its tax was 46.07 billion rupees, down 16.2 percent. The salaried category had 2.37 million returns (up 10.2 percent) and tax of 9.51 billion rupees (up 21.6 percent). Company returns fell 15.6 percent to 15,243, with tax of 38.09 billion rupees (down 8.4 percent). The AOP category had 98,952 returns (up 7.8 percent) and tax of 6.91 billion rupees (down 10.3 percent). Of the total 11.59 billion rupee decline, the non-salaried category contributes about 9.00 billion (78 percent), companies 3.49 billion (30 percent), AOP 0.79 billion (7 percent), while salaried added 1.69 billion as an offset. The aggregate decline is therefore a non-salaried and corporate story; salaried is the only category improving on both margins. Most concerning is the fall in average tax per filer. Overall, average tax per filer fell from 18,886 to 13,307 rupees—a 29.5 percent drop, roughly three times the headline 10.3 percent. The headline understates the problem's intensity because the denominator is growing. In the salaried category, average tax is only 4,013 rupees—implausibly low for a withholding-taxed category, implying most salaried filers are near or below the exemption threshold and filing nil or nominal returns. Return count is a poor proxy for the tax base, yet the article treats 7.56 million returns as a strength. Decomposing into volume and rate shows that the filing-base expansion added about 15.8 billion rupees of arithmetic tax, while the fall in per-filer tax removed 27.4 billion—about 1.7 times more. Base-broadening is being outrun by base-dilution. The strongest, assumption-light conclusion is a bound. Assume the most pro-FBR case: all 1.39 million additional non-salaried filers paid exactly zero tax. Even then, the original 3.67 million filers' average is 12,553 rupees against last year's 15,005—still 16.3 percent lower. The decline cannot be explained away by new small filers; among pre-existing filers, per-filer declared tax fell at least 16 percent. The dilution argument is at best partial and arithmetically capped. The most policy-relevant figure in the scheme section was never computed: of 3,295 registered, 1,945 filed—a 59 percent conversion rate, with 1,350 (41 percent) non-filers, a clean, quantifiable compliance gap. New and incumbent scheme filers also differ sharply—222 new at 41,667 rupees versus 1,723 incumbents at 98,519, a 2.4-fold gap—consistent with partial-year liability or different slabs; flag it without speculating. Data-quality caveats are essential. 'Corresponding period' is asserted, not demonstrated—whether the TY2025 figure is a same-date snapshot or a full-year final is unclear; a full-year versus partial comparison would make return growth spuriously high. Admitted tax at filing is provisional—late filings, revised returns and audits move it; it usually rises at the deadline, making a 10.3 percent decline more notable. The nil/zero-return share is not given, so '27.2 percent more returns' cannot be read as broadening. Overall, this is a data-release rewrite rather than reporting—no FBR spokesperson, no methodology note, no target or total-collection context. The reader cannot distinguish 'broadening is working, revenue lags' from 'declarations are being compressed,' whose policy implications are opposite. Follow-up questions: Is the TY2025 comparator same-date or full-year? What share of 7.56 million returns are nil? Why did corporate returns fall 15.6 percent while per-company tax rose 8.5 percent—deregistration, non-filing, or threshold effects? Does admitted tax include minimum/fixed-tax regime payments? What happened to the 1,350 registered non-filers? Was the October 31 date statutory or an extension? Why is the new-filer average 2.4 times below the incumbent average? Without these answers, the true picture of base-broadening cannot be known.