This paper seeks to model the effect that different patterns of household composition have on the observed levels of fertility in the municipalities of Mexico in the year 2020; it is a quantitative cross-sectional cohort research based on the application of spatial Bayesian methods. The hypothesis is that the presence of a higher percentage of family households should have an impact on higher municipal fertility rates. The methodology involves the implementation of two latent Gaussian models. One null model, which seeks to determine whether the observed fertility patterns were generated by some socio-demographic mechanism or, on the contrary, arose randomly, and two, a model with covariates whose objective is to replicate the behavior of fertility by evaluating the effect of the proportion of nuclear, extended and compound households present in municipalities. The results obtained from estimation of null model confirm the existence of a direct relationship between increase in the proportion of nuclear and extended households and the increase of municipal fertility. However, it can be concluded that the level of replacement fertility reached by Mexico in the year 2020 is the product of marked differences between municipalities; differences originated by the presence of a heterogeneous typology of households immersed in disparate geographic, social and cultural contexts.

Keywords: Total fertility rate, household typology, INLA