Abstract / Summary
Background. Since ca. 2010, period fertility has fallen in many middle- and high-income countries by more than conventional socioeconomic determinants predict. For Argentina, earlier studies documented a plateau until 2014, a subsequent decline arising within social groups, and a shift from postponement to reduced quantum. Its international position, its candidate explanations, the performance of existing projections and its demographic consequences have not been quantified. Objective. To characterise Argentina's recent fertility decline, identify which candidate explanations are and are not consistent with the evidence, and quantify its demographic consequences. Methods. We combined vital registration (2005–2024) with census-2022-adjusted denominators, 20 annual waves of the Permanent Household Survey (2004–2024, excluding 2007), MICS 2019–20, monthly registrations for Buenos Aires province to July 2026, and cross-national data for 145 countries, Chile and 4,598 Brazilian municipalities. Associations were grouped into six families corresponding to distinct hypotheses; we controlled the false discovery rate within families and estimated covariate effects with empirical-Bayes partial pooling. Futures were explored with stochastic Lee–Carter and scenario projections, back-tested on 2020–2024, and translated into population with a single-year cohort-component model including mortality and migration. Results. The total fertility rate halved from 2.32 in 2014 to 1.16 in 2024, and registrations in Buenos Aires province fell further in 2025–2026. The 2013–2023 decline (−43%) was among the three largest of 237 countries and territories (tied with Hong Kong, after Curaçao) and larger than in any OECD country (median −13%). Unlike the OECD, where rates rose at ages 35–44, Argentine rates fell at every age; Argentine cohorts nevertheless remain ahead of every Human Fertility Database population at ages 25–40. A model of 2003–2013 determinants failed out of sample across 145 countries (R² = −0.84). Within Argentina, common year effects absorbed 95–97% of provincial variation, and provincial TFR declines began between 2015 and 2018, earlier in higher-income provinces (ρ = −0.65, q = 0.009); spatial clustering of break years was suggestive but did not survive correction (Moran's I q = 0.07). About 85% of the decline occurred within union-status groups. Consumer confidence was associated with the monthly timing of registrations in 2020–2026 but not with their annual level, and no association survived correction for contraceptive supply among adults, the adolescent-pregnancy plan, government orientation or provincial economic conditions. Existing projections did not anticipate the decline: INDEC's official 2025 provincial projections exceed registered 2024 rates by 0.69–1.29 children, the UN estimate for 2023 (1.50) exceeds the registered rate (1.30), and every method fitted to data up to 2019 overpredicted 2024. Two contrasting models, continuation of the 2014–2024 trend and damping calibrated on 68 historical fast declines, give TFRs of 0.79 and 1.01 in 2030; provisional rates lie within the 80% intervals of both, close to the trend model in 2025 and closer to the calibrated model in the partial 2026 data. By 2035 the population aged 5–19 will be 27–30% smaller and that aged 65 and over 1.2 million larger under all empirical scenarios; people aged 65 and over will outnumber children under 15 from 2032–2034, and natural decrease begins between 2027 and 2043. Conclusions. Argentina's decline is among the largest ten-year falls recorded from a starting level below 2.5, and it differs from the postponement-led declines of high-income countries. The determinants of the preceding decade did not predict it and official projections did not anticipate it; planning based on those projections overstates future births. The contraction of the school-age population and the growth of the older population over the next decade are already largely determined; the longer term depends on whether fertility stabilises, which provisional data for 2025–2026 do not yet show.