The Andorra Scenario Engine

National plans are built from indicator trajectories with no mechanism a decision-maker can pull, and a small country cannot learn from a policy mistake after it has been made.

You can’t roll back a country.

Can a small team build a data-grounded, auditable national model that compares conditional futures instead of predicting one?

The model is calibrated on official statistics from 2010 to 2024 and run to 2049 under four scenarios.

Continuity grows GDP by 2.5% a year, Overgrowth by 6.0%, Degrowth shrinks it by 1.5% with an explicit migration restriction, and Density grows it by 3.0% with no building expansion.

Population follows GDP through a lagged labor-immigration elasticity (0.40 at one year, 0.10 at two). 500 Monte Carlo draws. Outputs are mapped on a 12,474-cell H3 grid.

Population follows the economy through a lagged labor-immigration link. In the September 2026 update that choice back-casts population worse than a straight line, 5.5% error against 1.7%, and I kept it, because a straight line has no lever a minister can pull.

In 2049 the population ranges from 72,268 (Degrowth) to 175,070 (Overgrowth); figures updated September 2026.

Correcting the model moved three 2049 endpoints by under 4%, while Density rose 19% and now edges past Continuity.

Published 5 August 2026. Conditional scenarios, not forecasts.

Figures use the corrected model run (updated September 2026), which supersedes some values in the published article; a formal correction is in preparation.

As listed in the publisher’s Author Contributions, I did the conceptualization (with Kent Larson), methodology (with Luis Alonso-Pastor), software (with Adrian Mora-Carrero), formal analysis, investigation, data curation, original draft and visualization.

Marcel Bartumeu Ramentol, Parfait Atchade-Adelomou, Adrian Mora-Carrero, Kent Larson, Luis Alonso-Pastor, Marta Domenech (Andorra Research + Innovation).

Partner: Andorra Research + Innovation.