Bivariate Recursive Semiparametric Probit Regression Model with Random Effects
Nabilla Rida Tri Nisa (a*), Purhadi (a), Achmad Choiruddin (a)

a) Department of Statistics,
Institut Teknologi Sepuluh Nopember
Kampus ITS - Sukolilo, Surabaya 60111, Indonesia
*6003212015[at]mhs.its.ac.id

Abstract

The utilization of healthcare services is the effort to utilize healthcare facilities to improve health, prevent and treat diseases, and restore health. One of the factors that influences individuals to utilize healthcare facilities is health insurance ownership. The utilization of healthcare facilities and health insurance ownership can lead to endogeneity, resulting in inconsistent or biased parameter estimation results. Therefore, we apply a semiparametric recursive bivariate probit model with random effects to examine the relationship between healthcare utilization and health insurance ownership in Probolinggo City in 2021. The data used in this study is from the National Socio-Economic Survey (SUSENAS) of Probolinggo City in 2021. Parameter estimation in the RPBRS-RE model is conducted using the Maximum Likelihood Estimation (MLE) and Fisher Scoring Algorithm. The resulting information shows that gender, employment status, smoking status, and marital status significantly influence health insurance ownership. Health insurance ownership and employment status significantly affect healthcare utilization. The bivariate mass points generated are (12.07, 69) and (-22.72, 102.53) with respective probabilities of 0.126 and 0.874.

Keywords: Penalized Likelihood, Probit, Semiparametric

Topic: Mathematics and Statistics