Read it yourself, on a schedule, from the source. A central bank circular, a finance act clause or a regulator’s directive can reset your unit economics in a single PDF, and it does so on a published date that usually sits weeks ahead of when the change reaches your customers. The founders who lose are not slower operators. They are later readers.
On 6 May 2024 the Central Bank of Nigeria ordered every bank, mobile money operator and payment service provider to deduct a 0.5 percent cybersecurity levy on the value of all electronic transactions, remitted to the National Cybersecurity Fund, applied at the point of transfer origination, live within two weeks, with a fine of at least 2 percent of annual turnover for non-compliance.1 For any business whose margin lives inside the transfer, that circular did not report a change. It was the change. A founder who read it on the sixth had two weeks to model the hit and move. A founder who heard it from an angry customer had days.
Reading it also meant reading it closely. The rate itself was contested. The 2024 Act printed the levy as 0.5 percent, while a 2018 CBN circular had set an earlier version at 0.005 percent, and the new text carried the old number in brackets, as “0.5% (0.005),” which is exactly the confusion that followed.8 0.5 percent is one hundred times 0.005 percent. On a charge applied to every transaction, that decimal was the whole business case.
Why these three models, and not one
One lens flatters you. Three fight. This decision has three separate questions inside it, and each needs a different tool.
Which circulars are even worth reading is a question about decision-relevance, so the first model prices information by whether it can change what you do. When a rule reaches you, and how late, is a question about diffusion through a population, so the second model treats regulatory news as a contagion with a transmission lag. What the lead time is actually worth is a question about the value of a right you hold under uncertainty, so the third model treats early reading as a written option. Equilibrium, cycle, and randomness. Three outcome types, three different errors, which is the point of running an ensemble instead of one favourite.
1. The filter: which circulars can move your number
The value of an input is the most you would pay to know it before you decide.2 The corollary is the useful part. If a piece of information cannot change any action you would take, its value is zero, no matter how loud it is. Most regulatory noise is exactly this: real, published, and irrelevant to your specific take rate.
So the discipline is selective. Know the short list of instruments that can actually move a number you act on, and read those. For a payments company, the levy on a transfer moves the take rate directly. For a lender, a rate cap moves the spread. For a health or education startup holding personal data, a data-protection commencement moves your registration and audit cost. The Nigeria Data Protection Act, signed 12 June 2023, established a commission and duties that turned “we hold customer records” into a line item.3 Read the eight instruments that price your business. Skip the rest without guilt.
Assumes: you have a decision the input could change, and you can weigh the cost of reading against the cost of being wrong.
Fits because: a circular matters only if it moves a number you act on. Take rate, float, duty, cost per transaction. Most instruments touch none of them.
Breaks when: the rule changes a decision you did not know you had. A category you did not know you were in, a licence you assumed was permanent.
Counteracts: the urge to read all of it, which buries the one instrument that matters under a hundred that do not.
May reinforce: watchlist blindness, where you track only what has already hit you and miss the first-time exposure.
2. The wave: how a rule reaches you, and when
Information spreads the way an infection does. Daley and Kendall showed in 1964 that a rumour moves through a population by contact, splitting people into ignorants, spreaders and stiflers, with the same mathematics an epidemic uses.4 A circular travels the same curve. It starts at the source. It reaches law firms and tax advisers who publish notes within a day. It reaches competitors. It reaches your customers, who feel the deduction. Then it reaches you, if you were waiting to be told.
Your position on that curve is your lead time, and lead time is the entire asset here. A circular is next month’s price of doing business, published in advance. The operator who reads at the source is at the front of the wave. The operator who runs on newsletters is one full transmission step back, and in a two-week implementation window one step is most of the runway.
Assumes: information moves by contact through a population with a stable transmission structure and a measurable lag.
Fits because: a rule flows from gazette to advisers to competitors to customers to you. Where you sit on that chain sets how many days of warning you get.
Breaks when: the source broadcasts directly, an SMS to every account holder or a headline everyone reads at once, collapsing the lag the model prices.
Counteracts: the comforting belief that you and the market learn together. You do not. Someone always reads first.
May reinforce: false confidence from being early, if you reach the source first and then misread what it says.
3. The option: what the lead time is actually worth
Reading early is cheap and reversible. Acting on a rule, repricing a product, changing a routing, filing for a status, is costly and hard to undo. Bernanke’s result on irreversible investment is that uncertainty increases the value of waiting for information, because the wait preserves your ability to move once you know more.5 Early reading is how you hold that wait open. It buys the right, not the obligation, to reprice or reposition before the rule bites.
That right is a written option, and its value scales with two things: how volatile the rule set is, and how long your lead. African regulatory environments are high on the first. The cybersecurity levy is the proof. It was withdrawn on 17 May 2024, eleven days after it landed, after public outcry and a suspension from the Federal Executive Council.6 An operator who read the first circular repriced. When the withdrawal came, they repriced again, ahead of a market still catching up to the first move. High volatility is precisely the condition under which the reading is worth the most.
Assumes: reading is cheap and reversible, acting is costly and less so, and the underlying rule set moves enough to matter.
Fits because: the lead time you buy by reading early is the right to reprice or reposition before the change reaches your customers.
Breaks when: the lead time buys no move. If the rule is instant and universal and nothing you do changes your exposure, the option is empty.
Counteracts: the instinct to wait for certainty before reading. Volatility raises what the option is worth, and you only hold it by reading now.
May reinforce: paralysis, holding the option past the point where acting plainly dominates.
Behavioral and governance layers are folded rather than shipped as separate cards. The behavioral effect here is information avoidance, the pull to skip a boring official PDF, and its fix is identical to the value-of-information fix: lower the cost of reading until avoidance has nothing to feed on. Governance is a single sentence and lives in the first lever below: one named owner, one recurring slot.
GEER: the levers, cheapest and most reversible first
Start with the move that costs nothing and undoes cleanly.
Build the watchlist. Name the six to eight primary pages that can price your business. The central bank circulars page, the tax authority, the data-protection commission, your sector regulator. Not aggregators. The pages the rule appears on first. This is an afternoon of work and it is the whole foundation.
Put reading on a schedule with an owner. One person, one recurring slot, weekly. A schedule without a name is a wish. This is the governance layer, and it is one line.
Model the hit the day you read. Keep a one-line template that turns any instrument into its effect on your number. The levy example: 0.5 percent of every transaction against your take rate tells you in a minute whether it halves your margin or dents it. Reading without modelling is just being worried earlier.
Pre-commit the irreversible moves. The repricing, the routing change, the filing you will not make today. Decide the trigger now, so you act on the rule and not on the panic.
RADAR: what to have in place before the next levy lands
DO NOW, by T+3 (reversible, dominant in every scenario). Name the six to eight primary sources. Assign one owner. Set the recurring reading slot. This is pure upside and costs an afternoon.
HEDGE, by T+14 (cheap insurance against the next surprise). Build the one-line impact template. Then run a drill: take the last real circular in your sector and time how long it takes you to produce the number. If it takes a week, that is your current lag, and it is the thing to compress.
DEFER AND TRIGGER, by T+28 (irreversible, so pre-commit the signal). The repricing and repositioning moves you will not make now. Write the observable trigger next to each: a specific instrument reference, a signed levy, a commencement notice. When the trigger fires, you already know the response.
If you are writing the cheques, not the product. In diligence, ask which primary sources the team reads and how fast they turned the last rule change into a number. A team that learns its own regulations from customers is carrying an unpriced regulatory exposure, and you are about to own a slice of it. The reading discipline is a line in your risk memo, not a nicety.
CHAIN: what tends to happen to businesses standing here
Match the reference class on structure, not surface: any business whose unit economics rest on a rule a single authority can change by publication. Payments, lending, mobile money, cross-border transfer, data-heavy health and education. The Electronic Money Transfer Levy is the pattern in miniature, a flat 50 naira on transfers of 10,000 naira and above, introduced by the Finance Act 2020 and live from January 2021, which quietly reshaped how sub-10,000 transactions were priced and structured across the market.7
The base rate in this class: rule changes that move the take rate arrive with a short published lead, and most operators act only once the effect shows in the numbers, which is already a step too late. Present-state modifiers push the wrong way. Fiscal pressure raises the rate of new levies, and revenue authorities reach for the electronic channel because it is the easiest to meter.
Subtract the counterfactual before you draw a lesson. When a competitor stumbled after a levy, some of that was the rule and not your superior product, and some of your own survival was the rule sparing your category, not your resilience. Credit the reading, not a story you prefer.
Matrix-break flag. If an authority moves to deduct-at-source with no lead and no exemptions, the lead-time lever collapses, because the wave and the effect arrive together. The 2024 levy had 16 exempted transaction types listed in an appendix, so whether it even touched you depended on reading that appendix.8 Remove the lead and the exemptions and the only edge left is structural: which category you chose to be in. That is a different article, and a slower lever.
What this ensemble is blind to
Three things it cannot see, and none of them lets you off the reading.
Whether the rule survives. The cybersecurity levy was law on the sixth and withdrawn on the seventeenth. Reading early tells you a rule exists. It does not tell you it will last, and acting hard on a rule that evaporates in eleven days has its own cost.
Enforcement discretion. A circular is the ceiling of what could happen, not a forecast of what will. Between the text and the audit sits a regulator’s judgement you cannot read off the page.
The rule that changes a decision you did not know you had. The value-of-information filter only prices instruments you already know can touch you. The one that reclassifies your entire category is, by construction, the one your watchlist was not watching.
So do not wait for certainty about any single circular, because certainty is the one thing the source never ships. Build the schedule anyway. By T+3, name your six to eight sources and the owner. The discipline pays even when a specific rule is reversed, because the reversal is also a circular, published on its own date, and the operator who read the first one is standing at the front of the wave to read the second. Reading first is cheap, and it seats you at the front of the wave to read whatever comes next.
Sources and notes
- Andersen (Nigeria), “CBN Directs Financial Institutions to Implement 0.5% National Cybersecurity Levy Collection and Remittance,” May 2024. Confirms the 0.5 percent levy on the value of all electronic transactions, remittance to the National Cybersecurity Fund, application at the point of transfer origination, the two-week implementation window, and the minimum fine of 2 percent of annual turnover for non-compliance. Andersen note.
- “Value of information,” Wikipedia. Standard decision-analysis definition: the value of information is the highest price a decision-maker is willing to pay to know an uncertainty before making the decision it bears on. The corollary used here, that information with no power to change the action has zero value, follows directly. Entry.
- KPMG (Nigeria), “The Nigeria Data Protection Act, 2023.” Confirms the Act was signed on 12 June 2023 and establishes the Nigeria Data Protection Commission with registration, audit and breach-notification duties for data controllers and processors. KPMG insight.
- D. J. Daley and D. G. Kendall, “Epidemics and Rumours,” Nature 204 (1964): 1118. The original rumour-as-epidemic model, dividing a population into ignorants, spreaders and stiflers. Nature paywalls the note; the structure and 1964 attribution are reproduced in an open-access treatment: G. F. de Arruda et al., arXiv:2508.07099, which sets out the Daley-Kendall and Maki-Thompson models. Open-access mirror (PDF).
- Ben S. Bernanke, “Irreversibility, Uncertainty, and Cyclical Investment,” NBER Working Paper No. 502 (1980), published in the Quarterly Journal of Economics 98, no. 1 (1983): 85-106. The paper’s core: uncertainty increases the value of waiting for new information, because irreversible commitments forgo the option to move once more is known. NBER working-paper PDF.
- Andersen (Nigeria), “CBN Withdraws Circular on the Collection and Remittance of National Cybersecurity Levy,” May 2024. Confirms the 6 May 2024 circular imposing the 0.5 percent levy was withdrawn on 17 May 2024 following public outcry and the Federal Executive Council’s decision to suspend implementation. Andersen note.
- Forvis Mazars (Nigeria), “Finance Act 2020: The Shift from Stamp Duties Levy to Electronic Money Transfer Levy.” Confirms the Electronic Money Transfer Levy is a 50 naira charge on electronic transfers of 10,000 naira or more, introduced by Section 48 of the Finance Act 2020 (inserting Section 89A into the Stamp Duties Act), with a January 2021 commencement. Forvis Mazars publication.
- Nairametrics, “Here are 16 transaction types exempted from CBN’s cybersecurity levy,” 7 May 2024. Confirms the circular carried an appendix listing 16 exempted transaction types, so a business’s exposure depended on reading that list. See also Nairametrics, “Explainer: Is CBN’s cybersecurity levy 0.5% or 0.005%?”, 8 May 2024, on the decimal ambiguity between the 0.5 percent in the 2024 Act and the 0.005 percent in a 2018 CBN circular. Exemptions list, rate explainer.