Administrative data and the decisions it does not reach

It is the last week of the quarter in a district finance office, and a revenue officer is assembling the figures the council will see on Monday. The trading licences and market dues are already in the revenue system, entered receipt by receipt across three months by the people who collected them. The council paper wants them in a different shape entirely, so he opens a blank spreadsheet, reads totals off one screen, types them into another, and squares the two by hand until they agree. It takes most of two days. Nobody in the meeting will ask him what the figures mean.

Multiply that desk across the country and the shape of the problem comes into view. Uganda's institutions record an enormous quantity of routine administrative data — every licence issued, every pupil enrolled, every transfer released, every line on a payroll — and then decide very little with any of it. Collection and use have come apart, and the gap between them is now wide enough to be measured.

The collection deserves to be credited first, because it is real work and it was expensive. The Auditor General's report to Parliament for the financial year ended 30 June 2023 found that revenue collected at local governments running the Integrated Revenue Administration System rose from UGX 31.67 billion in FY2019/20 to UGX 56.35 billion in FY2022/23, an increase of 78 per cent the audit attributed to the rollout and use of the system itself. Recording things properly changed what districts were able to collect, and it did so within four years of the system reaching them.

THE SYSTEM ARRIVED; THE USE DID NOT

What happened next is the part worth studying. The same audit reports that UGX 20.07 billion had been spent putting the revenue system into 124 sites, then sets out how many of those sites were actually running on it: eleven had fully embraced the system, 107 were using it only partially, and six were not using it at all despite having been trained.

The audit is precise about why. Staff at only 52 of the 124 sites had been trained in the system's use and application. Of nine sites the auditors visited, not one had an approved IT staff structure or any IT personnel, and 39 of the activated local governments had not been supplied with enough of the phones, laptops, printers and point-of-sale machines the work needs. A revenue clerk does not abandon a system because she prefers paper; she abandons it because nobody trained her on it, the printer is somebody else's budget line, and the person who could fix the login does not exist on the establishment.

TWO SYSTEMS ASKING THE SAME QUESTION

Fragmentation compounds it, and here the audit is blunt. Two systems built by different vendors were being rolled out across local governments to do the same job: E-LogRev, owned by the Ministry of Local Government, and the revenue system, owned by the Local Government Finance Commission. The Auditor General put the cost of rolling out E-LogRev at UGX 250 million a site and the other at UGX 269 million, and observed that operating two identical systems wastes money on development, support and maintenance. Neither was integrated with the Integrated Financial Management System, so information moved between them manually, which is to say by somebody retyping it.

Beneath both sits a thinner base than the rollout figures suggest. Reviewing the tax register expansion programme, the same report found that of 134 district local governments only 86 had information systems, leaving 48 on paper, and that eight of 34 urban municipal councils had none at all. Partner agencies, the audit adds, held unclean data found to be unreliable, and lacked any unique identifier that would let a single client be recognised across their separate registers. Records that cannot be joined to each other cannot be analysed either, whatever software they happen to sit in.

REPORTING IS NOT THE SAME AS DECIDING

The National Planning Authority reached the same conclusion from the other direction. The Fourth National Development Plan, published in March 2025, states plainly that administrative data systems remain manual and disjointed, that data is captured by hand at service delivery points, and that the statistical system suffers from fragmented administrative data systems and limited utilisation of what it produces. It records the Community Information System, the instrument meant to carry data at local government level, as non-functional.

The plan is sharper still on what the reporting costs. Oversight institutions — the Office of the Prime Minister, the finance ministry, the Planning Authority, the Auditor General, the local government ministry and others — each demand different reports from the same institutions, duplicating effort and causing what the plan calls reporting fatigue, after which those reports are underutilised for decision-making. Only 29 per cent of the expected results of the previous plan had been delivered by 2022/23. The remedies proposed are unglamorous and telling. Reinstate the post of statistician in the staffing structures of ministries and local governments. Provide conditional grants for statistics. Audit administrative data routinely for accuracy, completeness and relevance.

WHAT RWANDA WROTE INTO THE PERFORMANCE CONTRACT

Rwanda took the same problem and attached it to something officials are already measured on. The National Institute of Statistics of Rwanda issued guidelines for the quality assessment of administrative data in June 2018, setting out the dimensions against which a ministry's routine records would be judged. Its National Data Governance Framework, published in 2025, goes further: implementation of data governance must be written into each institution's annual performance contract, the imihigo; institutions report key performance indicators quarterly to the statistics institute, which consolidates them into an annual national report; and every ministry and agency is periodically rated on a five-level maturity model across people, process, technology and measurement. The World Bank's Statistical Performance Indicators scored Rwanda 56.3 on data sources in 2024, against Uganda's 45.1.

Rwanda's own 2018 guidelines name the risk this creates, which is worth repeating rather than hiding: where data is used to evaluate imihigo, there is a risk of misreporting and over-reporting. Tying records to a scorecard changes what gets written down, and the answer to that is independent quality assessment rather than looser scoring.

WHAT IS THE WAY FORWARD?

Government should do what its own plan recommends, beginning with the statistician posts and the conditional grants for statistics, because a system with nobody whose job is the data gets abandoned whatever it cost. The two revenue systems should be consolidated and connected to the financial management system, and the missing unique identifier settled nationally rather than reinvented register by register. Oversight institutions should agree a single reporting set between themselves; the fatigue the plan describes is of their own making. Local governments should name one person accountable for the data, and budget the gadgets and the training as part of the system. Providers of data engineering and analysis work should be selling the decision rather than the dashboard: taking records an institution already holds, joining them, and returning something a council can act on in the week it meets.

None of it requires a new survey. It requires the figures the revenue officer already has to reach Monday's meeting as an argument rather than an attachment, so that somebody in the room finally asks him what they mean.