No decree, no transfer of title, and yet control has changed hands. This is the most frequent situation in investment insurance and the hardest to work up. The classification module set the principle: classification decides the date, and the date decides which policy year applies. This one gives the method, because the problem is not proving that something happened, the investor knows that better than anyone, it is designating which of the acts caused control to pass. The policy needs a day, not a period.
That requirement surprises people, and it is better understood in advance than suffered. Cover has a period, annual or multi-year, the waiting period runs from a triggering event, the duty to notify runs from it too, and the applicable terms are those of the period in which it falls. A loss that happened over eighteen months is not a loss the contract knows how to process. All the work therefore consists in converting a process into an event, and that conversion is not an observation, it is an argument that has to hold up before someone with an interest in making a different one.
The criterion arbitral practice has developed, and which underwriting borrows, fits in a sentence: the cumulative effect of the measures deprives the investor of the economic substance of its investment. What counts is therefore neither the number of measures nor their individual severity, but the moment after which the investor can no longer do what it came to do. Two questions locate it in practice. Could it still operate? Could it still put things right? As long as both answers are yes, control has not been lost, however brutal the latest decision may be.
A distinction then governs most files, and a file that does not address it is not a file, it is a complaint. A state may tax, regulate, impose environmental or social standards: that is the exercise of its sovereignty, it is not insured, and international law recognizes the right to expropriate while framing it with conditions of public purpose, due process, non-discrimination and prompt, adequate and effective compensation, under the formula American Secretary of State Cordell Hull stated in 1938 in his correspondence with Mexico, never codified and still contested. What tips a measure toward compensable expropriation is therefore something other than its harshness: discrimination, where it targets this investor or this class, disproportion, where the effect bears no relation to the aim invoked, and the absence of any effective remedy. Naming which of the three is at work is what marks out a serious file.
The chronology is the tool, and it is built in columns. For each measure: its date, the legal instrument carrying it or the absence of one, its effect on operations expressed in figures rather than adjectives, and what the investor did in the weeks that followed. That last column is the one people skip and the one that decides. A measure absorbed without any change in the conduct of the business argues that control had not yet been lost; a measure followed by an emergency board meeting, a halt to investment or the repatriation of expatriate staff argues the opposite, and it argues it with dated documents the investor holds.
What cannot be done after the fact is to obtain from the state a dated record of what it decided. No administration issues a document attesting that it removed control of a business without saying so. The practical consequence is severe: a creeping expropriation reconstructed two years later rests entirely on documents the investor produced itself, and their value depends on their having been contemporaneous with the facts. This is why this module is the companion of the one on evidence, and why the first reflex, from the second measure onward, is to open a dated file rather than to wait until one is sure.
That leaves the case where two dates are equally defensible, and it is common. They then often fall in two different policy years, therefore under two sets of terms, sometimes on either side of a renewal that narrowed the cover, sometimes on either side of inception itself. It has to be said plainly: on this question the insurer and the insured do not have symmetrical interests, and each will argue for the date that suits it. The only thing that shifts the balance is the quality of the chronology, because a documented date can be defended and a chosen one cannot.
The practical rule follows, and it is counterintuitive: retain the date you can prove, not the one that pays best. A date chosen for its yield collapses at the first serious examination, and it collapses after the notification deadlines have run on all the others, which leaves a file with no date at all. An indemnity argued over on a solid date is worth more than an indemnity refused on a convenient one.
A mining company holds an operating licence. Over eighteen months: in January an exceptional extraction tax, in March a cap on dividend repatriation, in June a requirement for a majority local partner, in September a refusal to approve the proposed chief executive, in December the suspension of the export licence, and the following June the non-renewal of the operating licence. The policy was renewed on 1 May on narrower terms. Which date should be retained, and what turns on it?
Two dates are defensible, and they do not fall in the same policy year. June of the first year, the majority local partner requirement, is the moment the investor stops being able to decide alone. December, the export licence suspension, is the moment it can no longer sell what it produces. The temptation is to retain the most visible act, the non-renewal the following June, and that is probably the most expensive error in the file: the non-renewal is the culmination rather than the triggering event, and by that date the notification deadlines have long since run on everything else. The first question is therefore not which date is most favorable but which is documented: for each of the six measures, does a dated instrument exist, published or notified, and what did the company do in the following weeks? A measure absorbed with no change of conduct argues that control had not yet been lost; a measure followed by an emergency board meeting and a halt to investment argues the opposite. The second question concerns the 1 May renewal: a date retained before it places the loss under the older and wider terms, a date retained after it places it under the new ones. The insurer will argue for whichever suits it, and the only counterweight is a chronology built while the facts were unfolding.
- 01The policy needs a day, not a period: all the work is converting a process into an event, and that is an argument rather than an observation.
- 02The criterion is the cumulative effect that strips economic substance, and two questions locate it: could you still operate, could you still put it right?
- 03A state may tax and regulate without that being insurable; what tips a measure over is discrimination, disproportion or the absence of a remedy, and which of the three applies must be named.
- 04Retain the date you can prove, not the one that pays best: a convenient date collapses after the deadlines have run on all the others.