Resumo

ARGOTE CUSI, Milenka Linneth. Sensitivity analysis of population projections. Pap. poblac [online]. 2015, vol.21, n.84, pp.45-67. ISSN 2448-7147.

Is population projection sensitive to small changes in Global Fertility Rate (GFR)? Considering the GFR sampling distribution of Bolivia in a method of population projection, sampling distribution of population projected has been obtained. With both distributions sensitivity analysis is performed to check whether the proper butterfly effect of chaotic systems is presented. The latest National Survey of Demography and Health of Bolivia, available on line, was used. For population projection, mortality tables and other demography data are collected of National Institute of Statistics. Algorithms and distributions obtained by Argote in 2007 and 2009 were used to generate new sampling distributions. This research produces projection of the resident population using a cohort-component method and assumptions about demographics component of change. Finally, the sensitivity analysis is performed by target. Unlike the amplifier effect found in sampling distribution of births to small changes in the GFR (Argote, 2011) the sensitivity analysis of the projected population is a heterogeneous behavior; in some cases, reducing, stable or amplification effect is observed. This heterogeneous behavior shows us population dynamic complexity. The method has proven to be robust, consistent with the results and suitable for the analysis of the uncertainty in population projections.

Palavras-chave : Sensibility analysis; bootstrap; population projection; uncertainty.