Administrative Simplification in the Mexican Social Security Institute by OECD

Author:OECD
Language: eng
Format: epub
Tags: economics/governance
Publisher: OECD Publishing
Published: 2019-05-15T00:00:00+00:00

For example: is composed of the following sub-variables:

is the regression error.

This regression would pose two challenges: multicollinearity and heterocedasticity. The way both problems were corrected is explained below.

ii) Principal components analysis to eliminate multicollinearity

Aiming to reduce the regression’s multicollinearity problem an algebraic transformation of exogenous variables was implemented, based on the existing correlation between them. This transformation intended to reduce the dimensionality of data and consequently a principal components analysis was conducted. This type of analysis reduces the number of variables, with no significant loss on information quality and thus eliminates multicollinearity (Jolliffe, 2002[9]), (Perez, 2017[10]).

The principal components analysis is useful when there is a high correlation between study variables. The information obtained from IMSS procedures, according to the user’s classification and the new or existing condition, shows that there was indeed a high correlation between the 10 exogenous variables.

Statistical analysis shows that the three first components, based on the information breakdown, accounted for 92% of data variability —see Table 1.A.16. On the other hand, the first two components account for 80% of the variability. Therefore, regarding regression transformation, tests were made using up to four components, respectively, to determine the best option. However, the best alternative was the use of two components, since this specification eliminated the multicollinearity problem and the regressions’ parameter turned out to be significant.

Annex Table ‎1.A.16 shows the components results for each variable. It only includes the first six components since, based on previous analysis, components 7 to 9 do not provide relevant information. That is, the variability in the information is explained with six components.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.