FBR: Reforms at what cost?

Pakistan’s tax administration is again being offered a foreign-financed technological cure. The Central Development Working Party (CDWP) has recommended the Rs. 57.109 billion ‘Transforming and Digitalising Revenue Administration’ (TADRA) project to the Executive Committee of the National Economic Council (ECNEC). Its proposed cost is about $200 million, to be financed through an Asian Development Bank (ADB) soft loan.

This is a conditional recommendation, not final approval. ECNEC consideration, ADB Board approval, loan negotiations and signing still remain. That distinction matters. A more troubling matter, however, arose during scrutiny.

According to official records reported by a senior journalist, the Planning Commission observed that approximately $4.7 billion had already been obtained from development partners for reforming Pakistan’s tax system. It sought an impact assessment of the Tax Administration Reform Project, Pakistan Single Window, Integrated Transit Trade Management System and Pakistan Raises Revenue (PRR). With the proposed loan, the reported aggregate would reach $4.9 billion.

The government has not published the project-by-project reconciliation behind that figure. The chart below shows the position, obtained from the websites of lenders, etc. A publication by an independent think tank, ‘World Bank RRR Loan’ of November 2024 deserves attention. It asks whether a $400 million World Bank operation produced a sustainable increase in revenue, broadened the tax base, facilitated compliance and improved institutional efficiency.

Interviews with serving and former Federal Board of Revenue (FBR) officials describe slow information-technology spending, weak genuine broadening of the base and persistent institutional resistance. These concerns have become more relevant with the proposed TADRA loan.

The study should nevertheless be used with care. It is a prospective policy analysis, not an audit or causal evaluation. Its historical financing table does not establish the $4.7 billion figure. The listed amounts total about $3.535 billion and mix loans, grants, technical assistance, provincial public-finance programmes and general economic stabilisation.

One entry records the Second Balochistan Resource Management Programme at $900 million, whereas the ADB’s official 2008 announcement gives $100 million, plus a $0.8 million technical-assistance grant. The Planning Commission may possess a broader and corrected schedule that should be made public.

The official PRR record still reinforces the study’s central warning. The World Bank approved PRR in 2019 with $320 million tied to specified results and $80 million for conventional investment in FBR’s information and communication technology. In 2025, the Bank approved another $70 million, taking total financing to $470 million while the original project remained open.

The latest ‘World Bank implementation report’, dated June 4, 2026, presents mixed results. Progress towards the development objective was downgraded from satisfactory to moderately satisfactory. Overall implementation remained moderately satisfactory and overall risk substantial. FBR collections rose from 9 percent of gross domestic product in financial year 2024 to 10 percent in 2025.

Income-tax return filers increased from 4.738 million to 6.981 million. Data sharing expanded, some business processes were automated, and primary data-centre equipment was installed. It would thus be inaccurate and unfair to say that nothing was delivered.

The same report reveals why the headline numbers are insufficient. For financial year 2023, 4.435 million persons were classified as active taxpayers, but only 2.325 million paid any tax with their returns.

A larger filer count does not by itself prove a larger tax-paying base or additional revenue attributable to PRR. Results were still unavailable for two core outcomes: the time required to prepare, file, pay taxes, and efficiency in customs clearance. Taxpayer survey had not been conducted. Data gathering for customs had only begun. The single sales-tax portal remained confined to selected sectors, while 33 withholding lines survived against a target of zero.

Financial terminology has added to the confusion. By June 2026, $336.14 million, or 85.28 percent of the revised original credit, had been disbursed. The additional $70 million credit had disbursed nothing. Disbursement is not synonymous with technology installed or additional tax collected.

The $320 million results-based component reimburses eligible FBR expenditure after verification of agreed conditions. The Auditor-General’s ‘financial statements’ show employee costs, operating expenses and repairs among those expenditures. It is misleading to treat the entire disbursed amount as spending on digital systems.

The Auditor-General concluded that the financial statements fairly presented cash receipts and payments and that expenditure complied with the legal agreements. This is important: the available audit does not establish missing or misappropriated money. It also records cumulative expenditure of only Rs. 2.28 billion through June 2025 against the $80 million information-technology component, confirming serious implementation delay. Its clean opinion concerns financial compliance. It does not answer whether the project produced value for money or whether higher collections resulted from its interventions.

TADRA must be tested against this unfinished record. Reportedly, $81 million is earmarked for consultancy services and another $10 million for project management. Together, these items represent 45.5 percent of the proposed loan.

FBR says PRR created foundational systems; however, its data centres cannot support graphics-processing-unit-intensive machine learning. That may be technically correct, but an assertion is not a feasibility study. Reviewers identified the absence of a proper gap analysis, a comprehensive data-security framework, clear data governance and an assessment of open-source alternatives. The project must also be reconciled with FBR’s separately approved Rs350 billion Transformation Plan.

Digitalisation cannot repair defective tax policy. Pakistan’s system still relies excessively on withholding, minimum and presumptive taxation, indirect levies and coercive collection from already documented persons. Artificial intelligence can process data. It cannot supply constitutional tax policy, institutional integrity or political will.

As our book ‘Tax Reforms in Pakistan: Historic &Critical Review’ (PIDE 2020) demonstrates, imported reform programmes repeatedly fund systems without resolving fragmented administration, irrational laws, exemptions, blocked refunds and political protection for powerful untaxed sectors.

ECNEC should defer approval until four records are public. The first is a reconciled ledger of every tax-reform loan, grant and technical-assistance project, showing approvals, disbursements, outstanding liabilities, assets and systems delivered. The second is an independent value-for-money and impact audit of earlier programmes.

Notes: 1. The exact 4,699 match is an independent reconstruction, not proof of the Planning Commission’s method: no itemised Commission schedule was located. 2. PIDE’s 2024 paper itself implies about 1,100 and does not establish 4,700; its first 400 attributed to TARP conflicts with the World Bank’s 102.9 loan commitments. 3. PRIME’s table mixes macro, provincial and tax operations; the author’s correction says its heading should have referred to policy loans and rejects the private-email source. 4. Approved cost, commitment, guarantee, counterpart spending and disbursement are not interchangeable.

The third is a system’s inventory and reuse plan, explaining every proposed consultancy, licence, data centre and procurement. The fourth is an outcome framework measuring additional tax actually recovered, positive-liability taxpayers, compliance hours, refund time, audit yield, customs-clearance time, system availability and cybersecurity performance.

The controversy should not be reduced to an unproved assertion that $5 billion has disappeared. The documented failure is institutional: the state has sought another loan without first publishing a consolidated audit trail from past borrowing to assets, outcomes and additional revenue.

These liabilities belong to the people of Pakistan, who are entitled under Article 19A of the Constitution to know what was borrowed, what was delivered and who was accountable. ECNEC and ADB should require that record before approving another dollar.

Copyright Business Recorder, 2026