The case for realism, and its limits

I read state behaviour through a realist lens by default, and I want to be precise about why, because the label gets used loosely enough that it has come to mean either cynicism or apologetics depending on who is saying it. It is neither. It is a claim about which variables carry the most weight when you have to forecast under uncertainty, and it earns its place because of what it does for you, not because it is a complete description of the world.

What the frame actually asserts

The structural version is narrow. There is no authority above states, so security cannot be outsourced. States therefore behave in ways shaped by the distribution of capability around them, and they behave that way largely regardless of their internal character. Great powers seek to prevent the emergence of a peer in their neighbourhood. Weaker states balance or bandwagon depending on proximity and available alternatives. Alliances follow threat, and they loosen when the threat recedes.

The distinguishing move is what gets deliberately left out. Domestic politics, ideology, the personality of the leader, the stated intentions of the government: all set aside, not because they are unreal but because as forecasting inputs they are noisy, and because the structural variables are visible and slow-moving.

That is the trade. You give up resolution to get durability.

Why it pays for the work I do

Three properties make it the right starting point when you have to put a number on something.

It is stable under a change of government. If your read on a country's external behaviour rests on who won the last election, you are rewriting your model every few years and you will be wrong across the transitions. If it rests on where that country sits relative to a larger neighbour and what its options are, the read survives. For anything with a five-year horizon, which is most credit exposure, that durability is worth more than resolution.

It is falsifiable in a useful direction. The frame produces expectations about behaviour, and behaviour is observable. Alignment claims can be checked against arms procurement, basing, voting patterns, trade routing and who holds whose debt. When those revealed choices diverge from the stated position, the frame tells you which one to believe.

It disciplines against the wishful reading. The dominant failure in country analysis is not excessive cynicism. It is assuming that because an outcome is manifestly bad for everyone, it will not happen. Realism has no such assumption. It expects states to accept large costs for security, and it is uncomfortable in the specific way that has repeatedly turned out to be correct.

The frame is not right because states are cynical. It is useful because it is stable, checkable, and immune to the wishful reading that ruins most forecasts.

Where it fails, and it does fail

An honest defence has to name the failures rather than absorb them.

It underdetermines timing. This is the big one for anyone with a position. Structural pressure explains why something is likely eventually; it says almost nothing about whether it happens this quarter or in eight years. A structurally correct read that is a decade early is indistinguishable from being wrong if you are carrying the exposure.

It mishandles the leader who genuinely does not optimise. The theory assumes a state that responds to incentives roughly rationally. Some do not, for reasons ranging from ideological commitment to information environments that have degraded to the point where the leadership is working from a false picture. Realism reads these as anomalies. They are not anomalies to the person holding the exposure.

It treats domestic politics as noise when sometimes it is the signal. A regime whose survival depends on a particular faction, revenue stream, or external patron will do structurally irrational things to preserve that arrangement. If you cannot see the internal coalition, you will misread the external behaviour.

It has almost nothing to say about economics. Realism explains why a state might seek control of a resource. It does not explain the price of it, the financing behind it, or how the exposure transmits to a balance sheet. That gap is precisely where the actual work happens.

A default, not a truth

I use the structural read as a prior rather than a conclusion. It establishes a baseline: given position, capability and available options, what would we expect this state to do. Everything else enters as a deviation from it that has to be argued for.

That places a burden of proof on claims requiring a state to act against its structural interest, which is the discipline that keeps the analysis from drifting into whatever the prevailing narrative happens to be. It also, unavoidably, biases against genuine novelty, and some of the things that matter most are novel.

The alternative frames are worth reading and occasionally decisive. Liberal institutionalism explains cooperation that persists where realism says it should not; constructivism explains identity-driven behaviour structure alone cannot reach. What I have not found is a rule for knowing in advance which frame a given situation belongs to, which is probably the honest limit of the whole enterprise, and the reason the output should be a scenario set rather than a forecast.