An earlier module showed how a tax measure can cease to be a tax and become a taking dressed as one. This one takes the same ground from the other end, and it is the hardest exercise in the certification: saying no. An investor whose returns collapse because a state has raised a royalty comes looking for an indemnity and has good subjective reasons; helping it often consists in explaining why it will not get one. A professional who can only defend favourable files has no doctrine, it has a leaning.
The starting point is an obvious thing forgotten under the pressure of a particular case: a state has the right to set its taxes, and that right is the very content of fiscal sovereignty. Treaties reserve it, policies exclude it, and that convergence is no drafting coincidence. It reflects a long held choice of public policy: if every change in taxation opened a compensation claim to foreign investors, no state could reform its levies without buying the consent of every company on its soil. The exclusion protects a competence, not a budget.
Four features must then be named which keep a measure within ordinary sovereignty, and they are held together rather than ticked off one by one. It is general, that is it targets an objective category and not designated persons. It is prospective, applying to future years and not to closed ones. It is proportionate, leaving a viable operation rather than making it impossible. And it is regular, adopted under the procedure the state gave itself. A measure showing all four is a tax, whatever pain it inflicts, and pain is not a legal test.
Here comes the objection every investor makes, to be treated seriously rather than brushed aside: the return calculation that justified the investment rested on the tax regime then in force, so changing that regime destroys the value one came to create. The objection is economically accurate and legally weak, for a reason that must be stated without evasion. The law does not protect an expectation of return, it protects property and undertakings. An investor wanting to freeze taxation had an instrument made for that, the stabilisation clause negotiated in the concession contract, and if it did not obtain one, that is a risk it accepted when it signed, not a loss for insurance to repair.
That stabilisation clause deserves being told what it is actually worth, since it is often presented as absolute protection and never is. It binds the state contractually, so a breach of it is a sovereign contract breach, which belongs to a distinct cover with its own regime of proof. It does not make the measure unlawful, it makes it compensable between the parties. And its reach is read in its text: some freeze the rate, others only the regime, others still provide for an economic rebalancing that leaves the state free to change its taxation subject to compensating. Those three drafts do not have the same effect and they are routinely confused.
An analytical trap waits at this stage and must be named. A general measure may in fact reach almost only foreign operators, not because it targets them but because they alone carry on the activity concerned. A royalty on mineral extraction in a country where every mine is foreign owned strikes foreigners in practice, and it remains general in the legal sense. Discrimination is shown by differential treatment of comparable situations, not by the composition of a sector. An adviser building a claim on that confusion loses, and loses late, after spending.
Finally one must say what actually moves a file, because the boundary is not a closed door. A cluster exists and begins to weigh when several features give way together: retroactive application, a threshold cut to measure, a departure from ordinary procedure, an exemption reserved to domestic operators, a rate exceeding the activity's capacity. None suffices alone, three together open a serious discussion. The useful skill consists in counting those features without indulgence, and in telling a client its file shows one when it shows one.
What must be doable at the end of this module states itself plainly. Recognise a measure of ordinary sovereignty, name it as such to a disappointed client, and tell it at what date it could have protected itself. The answer is almost always the same: in negotiating the concession contract, through a stabilisation clause whose drafting says what it freezes. It is advice that earns nothing for the one who gives it and is worth, to the one who receives it, a great deal more than a claim it would have lost.
A European group has operated an iron ore concession in a West African country since 2018, under a contract containing no tax stabilisation clause. The budget law passed on 14 December 2025, applicable to financial years opening on or after 1 January 2026, raises the mining royalty from 3 to 8 per cent of mine gate value. The measure covers all metallic substances and all holders of mining titles, domestic and foreign alike; it happens that all six holders in the country are foreign groups. The parliamentary debate and the explanatory memorandum invoke the funding of a road programme. The project's internal rate of return falls from 14 to 6 per cent, below the group's cost of capital. Management asks to call on the political risk policy. What is the answer, and where should effort go?
The answer is no, and it must be given with its demonstration rather than with regrets, because a client who does not understand a refusal will have it reviewed elsewhere for six months. The four features of ordinary sovereignty are present here, and it is their combination that closes the file. The measure is general: it covers all metallic substances and all holders, an objective category. It is prospective: passed on 14 December for years opening on or after 1 January, touching no closed year. It is regular: a budget law, a parliamentary debate, an explanatory memorandum invoking identifiable public expenditure. It is proportionate in the legal sense, which is the point to be precise about: a return reduced from 14 to 6 per cent is a worse operation, not an impossible one, and the test is not the group's cost of capital but the viability of the activity. The fact that all six holders are foreign is the argument management will want to use, and it must be set aside explicitly rather than left alive: discrimination is shown by differential treatment of comparable situations, and no domestic operator is treated differently, there are none. The composition of a sector is not discrimination. It remains to say where effort usefully goes, and it does not go to insurance. Three things are worth doing. Check whether the concession contract contains an economic rebalancing undertaking, even absent a stabilisation clause proper, since that drafting opens a contractual negotiation rather than a claim. Follow the implementing decrees, because a base defined by regulation may introduce a threshold or an exemption that would itself be discriminatory, and that will be the only moment at which a new fact can arise. And take the subject where it is handled, that is into the renegotiation of the title, knowing that the missing protection was a stabilisation clause nobody asked for in 2018.
- 01The tax exclusion protects a competence and not a budget: if every tax reform opened a claim, no state could reform without buying every foreign investor's consent.
- 02Four features keep a measure within ordinary sovereignty and hold together: general, prospective, proportionate, regular. The pain inflicted is not a test.
- 03The law protects property and undertakings, not an expectation of return: the instrument that freezes taxation is the stabilisation clause, and not obtaining one is an accepted risk.
- 04A measure that in fact reaches only foreigners remains general: discrimination is shown by differential treatment of comparable situations, not by the composition of a sector.
- 05Three features giving way together open a serious discussion; one opens nothing, and being able to say so to a disappointed client is part of the job.