Decisions that last: state capacity and time horizons in Argentine policy

Public debate in Argentina about long-term policy is usually framed as a problem of agreement: if parties could only strike a pact, decisions would last. The data assembled for this report suggest otherwise. What makes a policy survive is not the way it was born, but the administrative infrastructure that sustains it and the existence of a predictable procedure for changing it.

We built a database of 240 national policies enacted between 1994 and 2024 across six areas: social security, energy, science and technology, education, health and taxation. For each we recorded the legal instrument, the legislative majority at origin, the implementing agency, the existence of a permanent technical corps, the involvement of provincial governments, and the status of the policy after four, eight and twelve years.

Consensus at origin is not enough

The most counter-intuitive finding is that the breadth of the coalition that approved a policy does not predict its persistence. Thirty-four of the ninety-one discontinued policies had been enacted with more than two-thirds of the vote in both chambers. In several cases that breadth was precisely the product of vague drafting: the title was agreed and the substance deferred to secondary regulation, where agreement never existed.

A bureaucracy with a career of its own

The most robust predictor of survival is the existence, within the implementing agency, of a technical corps with regular competitive recruitment, its own career ladder and continuity beyond the electoral cycle. Controlling for subject matter and origin majority, the presence of such a corps multiplies by 2.7 the probability that the policy remains in force twelve years later.

“A policy without a technical corps is not a policy: it is an announcement with an expiry date.”

From chapter 3, “Administration as memory”

The mechanism is prosaic. A stable corps accumulates operational memory: it knows what was tried before, where the data are, how the budget is executed, which actors must be convened. When the political leadership changes, that knowledge lowers the cost of continuing and raises the cost of starting over. Where it does not exist, every administration faces the same incentive: announcing a new policy is cheaper than learning the old one.

Predictable revision instead of irreversibility

Much of Argentine institutional design tries to protect decisions by making them hard to change: supermajorities, earmarked revenues, budgetary ring-fencing. The survey suggests this strategy backfires. The more rigid a policy, the more likely the next coalition is to abolish it outright rather than correct it.

• Mandatory evaluation with a deadline and a responsible body defined in the norm itself.
• Advance publication of the data feeding the evaluation, in reusable formats.
• Explicit renewal clauses that force a decision rather than allow quiet lapse.
• A procedure for amendment that is simpler than the procedure for repeal.

Federalism as an anchor

Coordination with provinces has an obvious cost in speed: policies that incorporate it take on average fourteen months longer to reach full implementation. But they also more than halve the probability of unilateral reversal. A decision already negotiated with twenty-four jurisdictions is expensive to undo from a single office.

Methodology

The database includes every national policy enacted by statute or emergency decree between 1 January 1994 and 31 December 2024 in the six areas listed, with identifiable budget allocation. Coding was performed independently by two researchers, with 89% initial agreement and joint resolution of discrepancies. Survival models are proportional hazards, with standard errors clustered by policy area.

The full database, codebook and replication code are available under an open licence in the Data section.