ESTIMACIÓN DE LA PROBABILIDAD DE POBREZA POR INGRESOS EN ECUADOR MEDIANTE UN MODELO LOGIT,
2022
además, se identifican las variables que conducen a la pobreza por rentas y poseen gran
capacidad predictiva.
Palabras Clave: Pobreza, probabilidad, modelo probabilístico logístico, población, muestra,
parámetros, efectos marginales, OddsRatios, Curva ROC
ABSTRACT:Poverty is a social phenomenon that has affected the Ecuadorian population over
time. This research focuses on estimating the probability of individuals becoming poor in
Ecuador in 2022, using a logistic probability model (logit). The dependent variable of this model
is Income Poverty, and the independent variables are Head of Household, Rural Area, Per Capita
Income, and Age. For the mathematical and statistical calculations, the National Institute of
Statistics and Censuses (INEC) database, specifically the 2022 National Survey of Employment,
Unemployment, and Underemployment (ENEMDU), is used. Using the logit model, a sample
analysis is performed, and the estimated parameters are consistent with socio-economic
theory. The marginal effects of the predictor variables are also calculated, showing that they
are logical and statistically significant. Furthermore, the odds ratios of an event are evaluated,
demonstrating that the independent variables influence the probability of the event. Finally,
the ROC curve demonstrates the high predictive capacity of the sample logit model.
Additionally, a population analysis is performed by adding the atonement factor (a value
constructed by the INEC) to the sample model, generating a population logit model. This aims
to compare the sample results with the population results and observe the effectiveness of the
sample logit model, with the goal of obtaining similar estimates in both models. The new
population model indicates that two of the four explanatory variables in the sample study are
statistically significant: Per Capita Income and Age. Consequently, the two variables that make
up the population logit model show reasonable marginal effects, their odds ratios indicate
effects on poverty, and most importantly, their results are very similar to the sample logit
model. Thus, the models used in the study explain the probability that individuals have of being
income poor in Ecuador in 2022; in addition, the variables that lead to income poverty are
identified and have great predictive capacity.
Keywords: Poverty, probability, probabilistic logistic model, population, sample, parameters,
marginal effects, Odds Ratios, ROC Curve
INTRODUCCIÓN
Entender la pobreza ha sido un tema de debate a lo largo del tiempo en la sociedad. Los antiguos
filósofos planteaban que el Estado debe corregir la injustica que genera la pobreza con el objetivo
de disminuirla. La perspectiva económica clásica (1776) proponía que una sociedad es más rica
que otra si posee mayor mano de obra, a más individuos, mayor producción y menos pobreza. En
1800 se complementa la idea clásica con el argumento, a mayor riqueza, mayor satisfacción de
las necedades. El pensamiento económico neoclásico, exponía el crecimiento de la riqueza como
un proceso paulatino de pequeños cambios incrementales, creando la teoría de la utilidad. A
principios del siglo XIX se mira a la pobreza como la carencia de recursos materiales, comparando
545