Pemodelan Durasi Pencarian Kerja di Jawa Timur Menggunakan Model Subdistribution Hazard dengan Multivariate Frailties

Atyasa, Abiyyu Rakha (2026) Pemodelan Durasi Pencarian Kerja di Jawa Timur Menggunakan Model Subdistribution Hazard dengan Multivariate Frailties. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5006221059-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Pengangguran menjadi tantangan besar bagi banyak negara berkembang tanpa terkecuali Indonesia. Badan Pusat Statistik (BPS) pada Februari 2024 mencatat Tingkat Pengangguran Terbuka (TPT) Indonesia dari Survei Angkatan Kerja Nasional (SAKERNAS) sebesar 4,82%, meski pada Agustus 2024 mengalami kenaikan menjadi 4,91%, pada Februari 2025 TPT Indonesia berada di angka 4,76%. Kejadian seseorang mendapatkan pekerjaan di sektor formal dan informal dipertimbangkan sebagai kejadian saling bersaing (competing risks). Klasterisasi kabupaten/kota dari domisili tempat tinggal pencari kerja diduga memiliki heterogenitas akibat efek laten tak teramati yang mendistorsi eksposur risiko dan peluang dalam mendapatkan pekerjaan. Penelitian ini mengaplikasikan model Subdistribution Hazard (Fine-Gray), Subdistribution Hazard Frailty, dan Cause-Specific Hazard Frailty Bivariate Normal. Penambahan efek acak (frailty) digunakan untuk mengukur profil risiko tak teramati antar klaster wilayah. Estimasi parameter diselesaikan menggunakan hierarchical likelihood yang menggabungkan efek tetap (usia, jenis kelamin, status pernikahan, pendidikan, pelatihan, pengalaman kerja) dan efek acak kewilayahan Jawa Timur. Hasil evaluasi model membuktikan bahwa penambahan komponen frailty secara signifikan meningkatkan kecocokan model. Berdasarkan struktur kriteria informasi cAIC dan rAIC, model Subdistribution Hazard Frailty dan Cause-Specific Hazard Frailty disimpulkan tidak dapat diperbandingkan secara langsung akibat perbedaan fundamental pada struktur pembentukan modelnya. Namun demikian, kedua model tersebut terbukti tangguh dan komplementer untuk memodelkan durasi pencarian kerja di Jawa Timur dengan hasil interpretasi analitik yang sama baiknya. Estimasi varians frailty menunjukkan adanya ketimpangan serapan kerja spasial, di mana varians pada sektor formal (0,1169) terestimasi lebih besar dibandingkan sektor informal (0,0455). Pada tingkat kovariat, probabilitas penyerapan sektor formal didorong oleh usia muda dan pendidikan tinggi. Sebaliknya, pada sektor informal berlaku efek substitusi risiko, di mana pendidikan tinggi justru memperpanjang durasi tunggu penyerapan dan peluang lebih dominan didorong oleh usia yang lebih tua, status pernikahan, serta keterampilan praktis. Temuan penelitian ini diharapkan dapat menjadi landasan perumusan kebijakan ketenagakerjaan berbasis penilaian risiko wilayah, khususnya dalam penyusunan program pelatihan vokasi, penguatan sektor informal sebagai jaring pengaman sosial alami, serta optimalisasi aktuaria pada desain Jaminan Kehilangan Pekerjaan (JKP) untuk mempercepat transisi penduduk ke pasar kerja di Jawa Timur.=========================================================================================================================================
Unemployment remains a major challenge for many developing countries, including Indonesia. In February 2024, Statistics Indonesia (BPS) recorded Indonesia's Open Unemployment Rate (TPT) from the National Labor Force Survey (SAKERNAS) at 4.82%. Although it experienced an increase to 4.91% in August 2024, the TPT stood at 4.76% in February 2025. The event of an individual securing employment in the formal versus informal sector is considered as competing risks. The clustering of job seekers' residential cities and regencies is suspected to exhibit heterogeneity due to unobserved latent effects that distort risk exposure and employment probabilities. This study applies the Subdistribution Hazard (Fine-Gray), Subdistribution Hazard Frailty, and Cause-Specific Hazard Frailty Bivariate Normal models. The inclusion of random effects (frailty) is utilized to measure unobserved risk profiles across regional clusters. Parameter estimation is conducted using hierarchical likelihood, which integrates fixed effects (age, gender, marital status, highest education completed, training, and work experience) with the regional random effects of East Java. Model evaluation results prove that incorporating the frailty component significantly improves the model fit. Based on the information criteria structure cAIC and rAIC, it is concluded that the Subdistribution Hazard Frailty and Cause-Specific Hazard Frailty models cannot be directly compared due to fundamental differences in their structural formulation. Nevertheless, both models prove to be equally robust and complementary in modeling job search durations in East Java, providing equally excellent analytical interpretations. The frailty variance estimation reveals spatial disparities in employment absorption, where the variance in the formal sector (0.1169) is estimated to be larger than in the informal sector (0.0455). At the covariate level, the probability of absorption into the formal sector is driven by younger age and higher education. Conversely, a risk substitution effect occurs in the informal sector, where higher education actually prolongs the expected waiting duration, and the employment probability is predominantly driven by older age, marital status, and practical skills. The findings of this study are expected to serve as a foundation for formulating region-based labor policies, particularly in designing vocational training programs, strengthening the informal sector as a natural social safety net, and optimizing the actuarial design of the Job Loss Guarantee (JKP) to accelerate the population's transition into the labor market in East Java.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Frailty, Kejadian Bersaing, Maximum Hierarchical Likelihood, Pengangguran, Subdistribution Hazard,Competing Risk, Frailty, Maximum Hierarchical Likelihood, Subdistribution Hazard, Unemployment
Subjects:	H Social Sciences > HA Statistics Q Science > Q Science (General) > Q180.55.M38 Mathematical models Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) Q Science > QA Mathematics > QA278 Cluster Analysis. Multivariate analysis. Correspondence analysis (Statistics)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Actuaria > 94203-(S1) Undergraduate Thesis
Depositing User:	Abiyyu Rakha Atyasa
Date Deposited:	16 Jul 2026 08:09
Last Modified:	16 Jul 2026 08:09
URI:	http://repository.its.ac.id/id/eprint/135199

Actions (login required)

View Item