Putting a number on political risk

Most political-risk writing fails an investment committee for one of two opposite reasons. Either it stays entirely in prose, "elevated tensions," "a deteriorating environment," and never produces anything a model can ingest. Or it overcorrects into false precision, handing over a single risk score with two decimal places and no honest account of where it came from. The first is useless to a decision. The second is worse, because it hides its own uncertainty.

Start with the decision, not the country

The discipline that makes country risk usable is the same one that governs a loan-loss reserve: you do not estimate "risk" in the abstract, you estimate the distribution of outcomes for a specific exposure over a specific horizon. Before writing a word about the country, name the decision, whether to enter, hold, exit, hedge or price, and the exposure it puts at stake. Everything downstream is in service of that.

A risk read that isn't attached to an exposure and a horizon isn't a risk read. It's a mood.

Build scenarios, then weight them

From there the work is scenario construction: three to five coherent, mutually exclusive paths the situation could take over the horizon, each described concretely enough that a reader could tell whether it had happened. Assign each a probability, subjective but explicit and defensible, and a financial consequence for the exposure in question. The weighted result is your input. The spread across scenarios is your uncertainty, and you report both.

Two West African sovereigns, scored on the same five pillars

Pillar score, 0 (benign) to 100 (severe)

The two overall scores differ by fifteen points, which on its own tells a committee almost nothing. The shape is the useful object. These sovereigns are separated mainly on macro and fiscal ground, and they are far closer on external and sanctions exposure than the headline gap implies, which is exactly the sort of thing a single number is built to hide and a decomposed one is built to show.

This is deliberately close to how a credit desk stress-tests a portfolio, because the underlying problem is identical: an uncertain future, a specific book, and a committee that needs a number it can stand behind. The geopolitics changes the scenarios. It does not change the machinery.

What makes a number arguable

A read built that way can be disagreed with precisely. Someone can accept the framework and reject one weight, or accept the weights and challenge a stale input, and the conversation stays about the substance.

A single figure with a paragraph of narrative behind it can only be accepted or dismissed. In practice it gets dismissed by anyone who was going to disagree and accepted without inspection by everyone else, which are the same two failures the machinery was supposed to prevent.

Whether the machinery makes anyone more accurate is a separate question, and an open one. Over a small number of decisions it probably does not; a well-informed person guessing carefully will often land closer. What it changes is what happens after the answer turns out to be wrong: whether there is a specific input to revisit, or only a judgement to lose confidence in.