![IconGems-Print](images/IconGems-Print.png)
208 – Survey Estimation
Maximum Constrained Pseudo-Likelihood Estimation of Income Distributions, Combining Sources
Alfredo Bustos
INEGI-Mexico
Accurate determination of the distribution of income from data collected in households through sample surveys is difficult due to limitations presented by this data source. Among them are the under-declaration of income throughout the entire sample; as well as the lack of representatives, in it, of the small very high-income household population group. The simultaneous use of this source with others, such as fiscal records or household income national accounts aggregates, contributes to reducing, albeit partially, its limitations. Similarly, limitations of supplemental sources will also see their importance reduced. In this note, we seek to extend results, which simultaneously use several data sources to improve the estimation of income distributions, to include the years 2010, 2012, 2014 and 2016, both for Mexico and for the Mexican states. The above will allow us to establish new comparisons with sample results to confirm or rectify the behavior of various trends during the period. In the same way, we will carry out comparisons for the states over time, and between them. We will show that the adjustments made give rise to an increasing trend in the Gini at the national level, which contrasts with the decrease in the value of the same indicator for Mexico’s income and expenditure survey (ENIGH). In other words, our results indicate that inequality increased during the period. Despite this, inequality within states does not seem to have undergone a significant evolution. This seems to suggest that what has driven the growth of inequality in the country is an increase in inequality between states.