Pemodelan Tingkat Kemiskinan di Jawa Timur Menggunakan Regresi Ridge Hoerl–Kennard–Baldwin (HKB)
DOI:
https://doi.org/10.29313/bcss.v6i2.26201Keywords:
Regresi Ridge, Estimator HKB, Kemiskinan Jawa Timur, MultikolinearitasAbstract
Poverty remains a multidimensional challenge in East Java, with a rate of 10.35% in 2023 and significant disparities across districts. This study applies Ridge Regression with the Hoerl-Kennard-Baldwin (HKB) estimator to model poverty determinants while addressing multicollinearity. Using secondary data from BPS East Java (38 districts), six predictors were analyzed: Mean Years of Schooling (X₁), Open Unemployment Rate (X₂), GRDP (X₃), Human Development Index (X₄), Literacy Rate (X₅), and Poverty Depth Index (X₆). OLS estimation revealed severe multicollinearity (VIF X₁=33.6, X₄=29.1). The HKB method produced a penalty parameter k=1.134964. The final Ridge model is Ŷ=31.8732−0.1023X₁−0.1655X₂+0.0041X₃−0.2185X₄−0.0918X₅+3.0626X₆. Partial significance tests (effective df=32.46) identified only X₄ and X₆ as significant at α=5%. X₄ negatively affects poverty (−0.2185%), while X₆ has the largest positive impact (3.0626%). Other variables were statistically insignificant. These findings confirm the effectiveness of Ridge-HKB in stabilizing coefficient estimates under multicollinearity. Policy priorities should focus on reducing the Poverty Depth Index and improving Human Development Index.
Kemiskinan merupakan tantangan multidimensi di Jawa Timur, dengan angka 10,35% pada tahun 2023 dan disparitas antarkabupaten/kota yang signifikan. Penelitian ini menerapkan Regresi Ridge dengan estimator Hoerl-Kennard-Baldwin (HKB) untuk memodelkan determinan kemiskinan sambil mengatasi multikolinearitas. Menggunakan data sekunder BPS Jawa Timur (38 kabupaten/kota), enam prediktor dianalisis: Rata-rata Lama Sekolah (X₁), Tingkat Pengangguran Terbuka (X₂), PDRB (X₃), Indeks Pembangunan Manusia (X₄), Angka Melek Huruf (X₅), dan Indeks Kedalaman Kemiskinan (X₆). Estimasi MKT menunjukkan multikolinearitas berat (VIF X₁=33,6; X₄=29,1). Metode HKB menghasilkan parameter penalti k=1,134964. Model Ridge akhir adalah Ŷ=31,8732−0,1023X₁−0,1655X₂+0,0041X₃−0,2185X₄−0,0918X₅+3,0626X₆. Uji signifikansi parsial (df efektif=32,46) mengidentifikasi hanya X₄ dan X₆ yang signifikan pada α=5%. X₄ berpengaruh negatif (−0,2185%), sedangkan X₆ memiliki dampak positif terbesar (3,0626%). Variabel lainnya tidak signifikan secara statistik. Temuan ini mengonfirmasi efektivitas Ridge-HKB dalam menstabilkan estimasi koefisien di bawah multikolinearitas. Prioritas kebijakan sebaiknya difokuskan pada penurunan Indeks Kedalaman Kemiskinan dan peningkatan Indeks Pembangunan Manusia.
References
Cule, E., Vineis, P., & De Iorio, M. (2011). Significance testing in ridge regression for genetic data. BMC Bioinformatics, 12(1). https://doi.org/10.1186/1471-2105-12-372
Gujarati, D. N., & Porter, D. C. (2009). Basic econometrics (5. ed). McGraw-Hill Irwin.
Hoerl, A. E., Kannard, R. W., & Baldwin, K. F. (1975). Ridge regression:some simulations. Communications in Statistics, 4(2), 105–123. https://doi.org/10.1080/03610927508827232
Hoerl, A. E., & Kennard, R. W. (1970). Ridge Regression: Biased Estimation for Nonorthogonal Problems.
James, G., Witten, D., Hastie, T., & Tibshirani, R. (Eds.). (2013). An introduction to statistical learning: With applications in R. Springer.
Kutner, M. H., Nachtsheim, C. J., Neter, J., & Li, W. (2005). Applied Linear Statistical Models (5th ed). McGraw-Hill/Irwin.
Montgomery, D. C., Peck, E. A., & Vining, G. G. (2012). Introduction to Linear Regression Analysis (5th ed.). John Wiley & Sons.
Sen, A. (1999). Development as Freedom. Oxford University Press.
Todaro, M. P., & Smith, S. C. (2015). Economic Development (12th ed.). Pearson Education.
