ANALISIS REGRESI DATA PANEL DENGAN PENDEKATAN COMMON EFFECT MODEL (CEM), FIXED EFFECT MODEL (FEM) DAN RANDOM EFFECT MODEL (REM) (Studi Kasus: IPM Kalimantan Selatan Periode 2010-2016)

Abstract

Analisis regresi data panel merupakan penggabungan antara data cross section dan time series. Penggunaan data panel dapat menjelaskan dua macam informasi yakni informasi antar unit dan antar waktu. Analisis regresi data panel dalam bidang ekonomi biasanya digunakan untuk data Indeks Pembangunan Manusia (IPM). IPM adalah sebagai indikator penting untuk mengukur keberhasilan dalam upaya membangun kualitas hidup manusia, IPM juga dapat menentukan peringkat atau level pembangunan suatu wilayah/negara. Dalam regresi data panel terdapat tiga estimasi model, yakni CEM, FEM dan REM. Metode CEM adalah metode yang berasumsi bahwa intercept dan slope pada setiap subjek dan setiap waktu adalah sama, metode FEM berasumsi bahwa intercept berbeda antar subjek dan slope sama antar subjek, sedangkan metode REM berasumsi bahwa variabel residual memiliki hubungan antar waktu dan antar subjek. Hasil dari penelitian ini, model regresi data panel terbaik ialah menggunakan Random Effect Model (REM) dengan efek individu. Variabel angka harapan hidup, rata-rata lama sekolah, harapan lama sekolah, pengeluaran perkapita disesuaikan, pendapatan asli daerah dan dana alokasi umum mampu menjelaskan IPM di Kalimantan Selatan sebesar 99,89%. Persamaan regresi data panelnya adalah: IPMit = (2,217+β_0i) + 0,495 AHHit + 1,217 RLSit + 1,067 HLSit + 9,018x10-4 PPit – 5,862x10-10 PADit + 4,831x10-10 DAUit

Panel data regression analysis is a combination of cross section data and time series. The use of panel data can explain two kinds of information ie information between units and between time. Regression analysis of panel data in the economic field is usually used for Human Development Index (HDI) data. HDI is an important indicator to measure success in the effort to build the quality of human life, HDI can also determine the ranking or level of development of a region/country. In panel data regression there are three estimation models, namely CEM, FEM and REM. The CEM method is a method that assumes that the intercept and slope on each subject and every time are the same, the FEM method assumes that intercept differs between subjects and slopes equally between subjects, whereas REM method assumes that residual variables have inter-time and inter-subject relationships. Result of this research, best panel data regression model is using Random Effect Model (REM) with individual effect. Variables of life expectancy at birth, mean years of schooling, expected years of schooling, adjusted per capita expenditure, local revenues and general allocation funds were able to explain HDI in South Kalimantan of 99,89%. The panel data regression equation is: IPMit = (2,217+β_0i) + 0,495 AHHit + 1,217 RLSit + 1,067 HLSit + 9,018x10-4 PPit – 5,862x10-10 PADit + 4,831x10-10 DAUit