Tackling Informal Economy With AI: Rs. 35 Trillion Test, Revenue Without Fiscal Coercion— Part V

Pakistan’s Rs. 35 trillion revenue ambition hinges on expanding the tax base through AI-driven transparency, lower predictable rates and provincial cooperation, rather than relying on fiscal coercion and repeatedly burdening existing taxpayers

Tackling Informal Economy With AI: Rs. 35 Trillion Test, Revenue Without Fiscal Coercion— Part V

The last part of this series highlights that promise of a low-rate tax system must survive arithmetic. Pakistan can expand revenue through wider bases, better information and provincial cooperation but Rs35 trillion is a demanding long-run scenario, not money waiting to be collected.

Pakistan has been promised fiscal transformation too many times. Each budget raises rates, multiplies withholding obligations, adds new penalties and forecasts a breakthrough in collections. Each subsequent shortfall produces another demand for coercive powers. A country cannot build a credible tax system by endlessly increasing the burden on those already visible to the State.

This concluding Part V of the series asks whether an alternative can pass the simplest test of all: arithmetic. Our five-part examination of artificial intelligence and the informal economy began with a National Economic Registry, distinguishing workers from taxpayers and lawful activity from crime, and proposed an accountable method of identifying taxable capacity. The final task is to test what a broad, low-rate tax architecture could yield.

The official starting point is sobering. The federal budget for fiscal year (FY) 2026–27 sets an FBR tax-revenue target of Rs. 15.264 trillion, against revised estimates of Rs. 12.983 trillion for FY 2025–26. These are budget figures, not proof of final collections. Provincial shares in federal taxes are transfers, not additional national tax revenue, and must never be added again when calculating Pakistan’s combined tax receipts.[1]

Our study, Towards Broad, Flat, Low-rate and Predictable Taxes, proposes a different structure: 10% personal income tax, 20% corporate income tax, a broad-based 10% value-added tax (VAT), low and relatively uniform customs duties, and carefully limited excises. Provinces would strengthen their own taxes within the constitutional division of powers.[2]

A lower rate does not, by itself, produce more revenue. It works only when the taxable base expands sufficiently, exemptions and privileges contract, and compliance becomes more attractive than concealment. No honest economist should promise that reducing an 18% sales-tax rate to 10% automatically increases the yield.

Part II set out an ambitious illustrative revenue envelope: Rs. 10 trillion from personal income tax, Rs. 5 trillion from corporate income tax, Rs. 10 trillion from VAT, roughly Rs. 1 trillion from customs and up to Rs. 2 trillion from selective federal excise duty. Together these amount to Rs. 28 trillion. The remaining Rs. 7 trillion would have to come from genuinely additional provincial and local tax bases. The figures require testing, not repetition as established capacity.

Consider personal income tax. At a uniform 10%, Rs. 10 trillion of revenue requires Rs. 100 trillion of taxable personal income. Divided among 30 million paying individuals, that implies average taxable income of approximately Rs. 3.33 million a year.

Pakistan cannot assume that distribution exists merely because tens of millions work informally. The registry must establish it from evidence, taking care not to count corporate profits again as personal income.

The corporate calculation is equally transparent. Rs. 5 trillion at 20% requires Rs. 25 trillion in taxable corporate profits. These profits must be distinguished from company turnover, asset values and distributions already reflected in personal income. The objective is a stable effective rate after removing distortions, not a concession to profitable corporations.

A 10% VAT producing Rs. 10 trillion requires an effective net taxable value-added base of Rs. 100 trillion after the operation of input credits, exemptions and refunds. It is not enough to point to gross retail sales or bank payments. The same rupee may pass through several firms before reaching the consumer. Proper VAT taxes net value added and allows timely credit and refunds; without that discipline, a supposedly low rate can conceal cascading taxation.

Customs illustrates the danger of attractive round numbers. Merchandise imports around US$70 billion, roughly Rs. 20 trillion at an illustrative exchange rate, would yield about Rs. 1 trillion at a uniform 5% duty before behavioural and administrative adjustments. A claim of Rs. 3 trillion from that same rate would require an import base near Rs. 60 trillion. Lower tariffs should be defended principally for reducing smuggling, under-invoicing and the cost of productive inputs—not through invented revenue.[3]

Selective excises can discourage harmful consumption and raise funds, but a Rs. 2 trillion yield needs product-by-product estimates and an allowance for behavioural responses. Tobacco, sugary drinks, environmental harms and narrowly defined luxury consumption require different tax instruments. The revenue from reducing harmful activity cannot be assumed to rise indefinitely.

The remaining Rs. 7 trillion is the hardest part of the stretch scenario. High agricultural incomes, urban land values, immovable property, motor vehicles and suitable environmental or local levies offer possibilities. But their current bases, exemptions, collection costs and legal incidence must be measured. Taxes already included in the harmonised VAT cannot be counted again as provincial sales taxes on services. Provincial shares of federal revenue are not provincial own-source taxes.

Artificial intelligence can improve that reconciliation. Properly governed analytical systems can match economic identities, distinguish profits from turnover, identify unexplained wealth, expose duplicate invoices and detect inconsistencies across tax declarations.

Martin Grote and Jean-François Wen’s 2024 International Monetary Fund note makes a practical case for broad property-tax bases at modest rates, linked to local services.[4] This approach deserves serious consideration in Pakistan. Taxing high-value land and economic rents need not mean imposing unaffordable bills on low-income homeowners. Valuation, hardship relief, local accountability and the constitutional allocation of powers are indispensable.

A more credible discussion should present scenarios rather than one triumphal forecast. At a 10% personal rate, a Rs. 25 trillion taxable base yields Rs. 2.5 trillion; a Rs. 45 trillion base yields Rs. 4.5 trillion; Rs. 100 trillion yields Rs. 10 trillion. At a 20% corporate rate, taxable profits of Rs. 12 trillion, Rs. 18 trillion and Rs. 25 trillion yield Rs. 2.4 trillion, Rs. 3.6 trillion and Rs. 5 trillion respectively. These are assumptions for sensitivity analysis, not estimates of actual taxable incomes. Likewise, net VAT bases of Rs. 55 trillion, Rs. 75 trillion and Rs. 100 trillion yield Rs. 5.5 trillion, Rs. 7.5 trillion and Rs. 10 trillion at 10%. Add illustrative customs of Rs. 1 trillion in each case, selective excises of Rs. 1 trillion, Rs1.5 trillion and Rs2 trillion, and separately measured provincial and local taxes of Rs. 2 trillion, Rs. 4 trillion and Rs. 7 trillion.

The resulting national totals are Rs. 14.4 trillion, Rs. 22.1 trillion and Rs. 35 trillion. The first two scenarios are not forecasts either. Their purpose is to expose precisely which assumptions must be demonstrated before a target becomes credible.

These scenarios cannot be compared mechanically with existing receipts as if all categories were incremental. They represent alternative full-system tax structures, not extra collections on top of FBR’s present yield. Some taxes would be abolished, others merged, and revenues redistributed between levels of government. Transitional losses, refunds, behavioural change and enforcement costs would all have to be modelled.

There is a further constraint: taxable personal incomes, corporate profits and consumption value added arise from a connected national economy. Their assumed magnitudes must be reconciled with national accounts and the size of gross domestic product. If a model cannot reconcile its bases with production, income and expenditure data, its impressive total is not a revenue forecast. It is an accounting warning.

Artificial intelligence can improve that reconciliation. Properly governed analytical systems can match economic identities, distinguish profits from turnover, identify unexplained wealth, expose duplicate invoices and detect inconsistencies across tax declarations. They can also identify erroneous demands, delayed refunds and taxpayers entitled to relief. The technology must be evaluated on net revenue after refunds, compliance costs and successful appeals—not on gross notices issued.

Pakistan also needs a new institutional settlement. A harmonised VAT may offer major compliance gains, but provincial rate-setting and legislative competence over services cannot be sacrificed to administrative convenience. Integration of data and common standards should be negotiated, with enforceable safeguards over access, appointments, refunds and dispute resolution. Provincial consent is a condition of legitimacy, not an obstacle to efficiency.

The first reform is not another tax. It is public accounting. FBR and provincial authorities should publish timely collections by tax type, refunds claimed and paid, advances, arrears, exemptions and comparable national-accounts data. The absence or delay of basic reports undermines both research and trust. Article 19A cannot coexist comfortably with a fiscal system that asks citizens to believe figures it will not adequately disclose.[5]

A National Tax Court with independent adjudication, simplified laws, transparent audits and an enforceable taxpayer charter would do more for voluntary compliance than another round of discretionary enforcement. Repeated amnesties must end. So must the assumption that every banking transaction or electricity payment proves taxable income.

The real prize is not the number Rs. 35 trillion. It is a tax system in which the State knows who has capacity to pay, citizens know what they owe, provincial autonomy is respected, honest businesses can compete, and collection no longer depends on punishing visibility.

Rs. 35 trillion should remain a demanding long-run proposition subject to independent validation. A smaller amount raised fairly, predictably and without destroying productive incentives may be worth more than an extravagant target achieved through coercion, unpaid refunds and advances borrowed from future years. Artificial intelligence can help Pakistan see its economy. Only constitutional governance, low predictable rates and institutional credibility can persuade that economy to come into the light. That is where this series ends—and where genuine fiscal reform must begin.

[Concluded]

Notes

1. Government of Pakistan, Annual Budget Statement 2026–27, Summary of Receipts: FBR budget Rs15,264bn; revised 2025–26 Rs12,983bn. 

2. Huzaima Bukhari and Ikramul Haq, Towards Broad, Flat, Low-rate and Predictable Taxes, third edition (PRIME, 2024), especially tax-gap and rate proposals.

3. Government of Pakistan, Pakistan Economic Survey 2025–26, external-sector and fiscal chapters; import-base example uses rounded values for illustrative arithmetic.

4. Martin Grote and Jean-François Wen, How to Design and Implement Property Tax Reforms, IMF How-To Note 2024/006 (19 September 2024). 

5. Constitution of Pakistan, Articles 19A, 160 and relevant legislative entries; Ministry of Finance, Fiscal Operations annual series.

6. Government of Pakistan, Fiscal Policy Statement 2026, showing 2024–25 FBR tax receipts and provincial own-source tax receipts as distinct categories. 

7. Pakistan Bureau of Statistics, Labour Force Survey 2024–25, for the distinction between economic participation and taxable capacity. 

8. Ikramul Haq, Tackling Informal Economy With AI, Parts I–IV, The Friday Times series (2026).

Dr. Ikramul Haq, Advocate Supreme Court, Adjunct Faculty at Lahore University of Management Sciences (LUMS), member Advisory Board and Visiting Senior Fellow of Pakistan Institute of Development Economics (PIDE), holds LLD in tax laws. He was full-time journalist from 1979 to 1984 with Viewpoint and Dawn. He also served Civil Services of Pakistan from 1984 to 1996.