Productivity reform needs a clock and an owner

“Improve the business environment” cannot be timed with a stopwatch. Register a title, clear a shipment, connect power or obtain a standards certificate, and the wait has a start, an owner and a cost.

The IMF's April 2026 outlook calls for reforms that raise productivity and support private-sector-led growth. The World Bank's April update describes low investment and weak productivity as structural constraints, then argues for policy matched to implementation capacity. Both point past the announcement. Reform has to change an action performed by a firm or agency.

The queue is observable.

Choose the transaction

A useful scorecard begins with a service name and a defined user. “Customs reform” is too broad. “Release a correctly declared container of specified goods at a named port” can be measured. So can “issue a low-voltage commercial connection after a complete application” or “refund an undisputed value-added-tax credit.”

Define the start and finish before collecting time data. Does port clearance begin when the vessel arrives, when the declaration is filed or when the container is discharged? Does it end at customs release or gate exit? Different choices can produce different results without any process changing.

Segment the cases. A first-time applicant should not be averaged with a renewal. A shipment selected for inspection should not be compared with a green-lane declaration unless the dashboard shows both. Publication of a single average can hide a long tail where firms carry most of the cost.

The scorecard needs volume as well as time. Cutting the median for a small pilot while the majority remains on the old system is not economy-wide reform.

Name the person who can remove the delay

Processes cross agencies, and each handoff can become an excuse. Map the legal owner of the service, the official responsible for the operating target and the other bodies whose approval is required. A committee can coordinate; it cannot replace accountability.

Record the reason for each paused day. The applicant may submit incomplete documents. An agency may wait for a payment confirmation. A system may fail. Another regulator may hold the file. This breakdown prevents every delay from being attributed to the same institution and exposes requirements that cause repeat work.

Ownership includes the power to change the cause. A service-desk manager cannot amend a regulation. A minister may approve a rule while lacking control of a vendor's software release. Assign actions at the level where budget and authority sit.

Public contact details matter. A user needs an escalation route with a case number and response time. Counting complaints without closing them adds another queue.

Count money, not only days

Time becomes an economic measure through cost. An imported input held for five extra days can incur storage, demurrage, financing and production delay. A licence renewal can require staff travel, agent fees and lost selling time. Firms should capture those amounts instead of relying on a perception survey alone.

The cost calculation must state its assumptions. Financing cost depends on the cash tied up and the firm's actual funding rate. Lost production depends on whether the delayed input had a substitute and whether customer demand existed. Do not assign revenue to every idle hour.

Official fees belong in the same table as unofficial or avoidable process costs, but the evidence standard differs. A published fee can be verified. A report of an informal payment needs safe collection and careful interpretation. Anonymous surveys can reveal a pattern without proving misconduct in a named case.

Productivity appears when the same labour and capital produce more useful output. A faster permit can contribute if it brings an asset into service earlier. It cannot prove economy-wide productivity growth by itself. The scorecard should claim the transaction result it measured.

Put elapsed time beside the reform budget

Digital projects are often reported through money spent, modules delivered or users registered. Add elapsed time at the front end. If a portal moves an application online but users still carry paper to an office, count the physical step. If the agency requests the same data twice, count the rework.

For each quarter, publish:

• completed case volume by segment;
• median and 90th-percentile elapsed time;
• applicant visits and document resubmissions;
• official charge and estimated user cost under a stated method;
• system availability during service hours;
• unresolved cases older than the legal limit.

Six rows are enough to show whether the common case improved and whether difficult cases were abandoned. The 90th percentile matters because a reform can lower the average while a minority of firms remains trapped.

Place the programme's operating cost and capital spending beside the service result. A more expensive service can be worthwhile if it saves users far more, but the comparison has to be made. Savings claimed across all firms need a defensible transaction count.

Stop rewarding the launch date

A legal amendment, portal launch or one-stop office marks an input. The first useful review date comes after enough real cases have passed through the new process. Publish the baseline before launch so no one can choose a favourable comparison later.

Independent verification can sample case records and interview users. The verifier should disclose sampling, exclusions and whether agencies knew which cases would be checked. Company associations can contribute transaction logs if they protect commercially sensitive information.

A reform can produce an early setback. Staff learn a system, old and new cases overlap, or applicants misunderstand a requirement. Report the transition rather than resetting the baseline. A temporary increase in time can be honest; a permanent absence of data cannot.

Read a national claim from the bottom up

The World Bank's ecosystem approach recognises that infrastructure, skills, finance and institutions interact. A customs portal cannot compensate for a broken port gate. A faster company registration does not supply power. That complexity argues for linked transaction measures, not a vague composite score.

An operator can build a private version. Log each government transaction, dates, documents, visits, fees and employee hours. Compare cases over time and share aggregated evidence through a trade body. One firm's records won't establish national performance, but they can challenge an agency with a reproducible case.

The next reform document may promise efficiency, investment and jobs. Look for the service, baseline, responsible official, user cost and first publication date. If those fields are blank, the work has not reached the clock yet.

Sources

• International Monetary Fund, 16 April 2026: Regional Economic Outlook for Sub-Saharan Africa, April 2026
• World Bank, 8 April 2026: Africa Economic Update April 2026