A Wrong Tag on Sindh's Property-Tax Programme: How a $150m Land Registry Landed in a Football File

The header on the file carried a single word: football. Thirteen years of habit made me read it frame by frame. My first guess was a Morocco 4-1-4-1, or perhaps a story about unpicking a mid-block. By the thirty-seventh information point the arithmetic refused to close. Not one pressing trigger, not one corner structure, not one transfer fee. Instead: property attribute enumeration, urban immovable property tax, land-record digitisation, a stakeholder engagement plan. The gap between label and content was so wide that the only honest verdict on that file was to write insufficient information into all eight analytical axes. I stopped at information point twelve, and the classification admitted its own error.

What surfaced was not football. It was the expansion of Sindh's property tax system — and a question that belongs to both domains: when does a digital register actually earn its value? Not in its wiring. In the match rate between what it is labelled and what it holds.

Context: the programme that lost its label

The initiative is the Sindh Property Revenues Enhancement Program, SPREP, backed by the World Bank with an envelope of roughly USD 150 million. Internally it splits into two instruments: USD 110 million under Program-for-Results, PforR, and USD 40 million under Investment Project Financing, IPF. The Local Government Department implements it; the Board of Revenue oversees the revenue side.

The PforR-versus-IPF distinction is quiet but decisive. IPF disburses against inputs — specific investments and technical assistance. PforR disburses against results, meaning the programme must prove the agreed indicators were genuinely achieved. For a tax survey, that means the count of registered properties is not the terminal argument.

One more figure from the context deserves attention. In the covered divisions, only about one-fifth of properties have been surveyed so far. Forty-five councils are participating: twenty-five inside Karachi, twenty outside it, spread across five Sindh divisions. The geographic reach is wide. The foundation is still shallow.

Core: where a survey manufactures tax reach

The field work looks harmless. A team goes out, records a property's address and attributes, takes photographs, matches the owner's identity. In tax administration this step is the most political of all, because this list decides which property enters the net and which slips out. A property absent from the roll is legally non-existent, even while it stands in brick and stone.

This is where the CLICK precedent becomes the most valuable fact in the file. Before the survey, registered properties numbered roughly 900,000; after it, roughly 4.2 million. Many read that jump as discovery. I read it as a question. Moving from 900,000 to 4.2 million does not mean property outgrew population. It means the earlier register was profoundly incomplete — and that incompleteness was the true measure of the old system's failure.

Two data-hygiene tasks follow. First, duplicate removal: the same property must not appear twice under two names, two addresses or two wards. Second, removal of properties sitting outside the councils' mandate. Both sound technical. Both are fundamental. A duplicate does not expand the tax base; it quietly devalues the whole register. Including out-of-mandate property inflates statistics while adding no revenue.

Under institutional strengthening, the programme builds capacity across the forty-five councils, including the rollout of IFMIS — an integrated budget, accounting and payroll platform. To collect tax you must first know what arrives and who sits where. A council that cannot keep its payroll books straight will not keep property ownership straight.

Citizen participation is specified in detail too. Town Citizen Committees comprise two male citizen members, two female citizen members and one council member, meeting monthly. Alongside sit a Stakeholder Engagement Plan, grievance channels, safeguards for vulnerable groups, verifiable enumerator identification and confidentiality requirements. Citizen concerns about survey accuracy are documented in the same plan.

Now the analytical layer. A land register and an analysis pipeline share one dependency: both rest on labels. A file arriving in a football pipeline under a football label gets processed through a football framework. If it holds thirty-seven property-tax points, the output is pure fiction. A cadastre behaves identically. A property filed under the wrong class either escapes taxation or is taxed without cause. In both systems the failure is not in storage. It is in classification.

Blockchain can be pulled in here, carefully. The programme documents mention no distributed ledger; that connection is my analytical extension, not a claim in the report. For land tenure, a distributed ledger offers three real benefits: a tamper-evident audit trail, detection of double-selling on one plot, and a single shared history across agencies. The limit is equally plain — a ledger cannot repair registration quality. Seal bad data into an immutable chain and the error inherits authority. Garbage in, gospel out.

I don't trust a proposition until I can rebuild it with clips and cold coffee. Evidence first, decision second.

Deciding against the evidence: two precedents, one lesson

In January 2026 I filed a report on a Brazilian midfielder after watching twenty-seven matches. The tape showed his pressing trigger firing eight-tenths of a second later than the league average. The club signed him anyway. He scored four goals in twelve matches. It reads as a football story, yet the structure repeats everywhere: the evidence existed, and the decision was simply filed somewhere else.

The same fault line runs through Sindh's tax survey. Teams will go out, records will be digitised, committees will meet, platforms will go live. But if the archive still misclassifies, digitisation has only accelerated an old error. At frame twelve I can see a pressing line. In a register I have to see which room each property sits in.

Contrarian angle: the risk isn't evasion, it's the label

Conventional wisdom frames the programme's enemy as political resistance or tax evasion. My reading puts the risk elsewhere. A mediocre register is more dangerous than a missing one, because it grants administrative authority to error. Nobody questions the list any more, because the list says so.

The second contrarian point concerns the success metric. Going from 900,000 to 4.2 million registered properties looks spectacular, but the figure measures input, not outcome. The number worth publishing is the match rate: how closely the register's classifications and ownership entries correspond to properties that physically exist. That rate rarely appears in monthly announcements, yet under PforR logic it is precisely the metric that should unlock disbursement. Verifiable, not celebrated.

The third point is the balance of safeguards. The engagement plan, grievance channels and gender-balanced citizen committees are detailed and careful — in places more mature than the fiscal design itself. The question is whether both layers move at the same speed. If participation is mature while data discipline is infantile, the meeting minutes will be tidy and the register will still be hollow.

Takeaway

Before the next tranche releases, watch two things. Is disbursement conditioned on registration counts or on classification accuracy? And in each new survey batch, which way is the label-versus-content match rate moving? A document that mislabels itself can mislabel its own property counts. The next milestone is not 4.2 million. It is what share of properties lands in the right room.