Stochastic Search Variable Selection Pada Model Conditional Value-At-Risk Berbasis Quantile Regression Dan Quantile Autoregressive Untuk Penentuan Faktor Risiko Saham Perbankan

Almas, Luqyana Zakiya (2026) Stochastic Search Variable Selection Pada Model Conditional Value-At-Risk Berbasis Quantile Regression Dan Quantile Autoregressive Untuk Penentuan Faktor Risiko Saham Perbankan. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6003241031-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Stabilitas sektor keuangan merupakan aspek penting dalam menjaga ketahanan ekonomi, terutama sektor perbankan. Risiko sistemik yang muncul akibat keterkaitan antar institusi keuangan membutuhkan pengukuran risiko yang mampu menangkap dinamika hubungan tersebut. Value-at-Risk (VaR) sering digunakan untuk mengukur risiko kerugian pada tingkat institusi atau portofolio individual, namun belum mampu merepresentasikan kontribusi risiko tersebut terhadap stabilitas sistem keuangan secara makro atau risiko sistemik. Oleh karena itu, Conditional Value-at-Risk (CoVaR) dengan dikembangkan untuk mengukur risiko suatu institusi yang dikondisikan pada institusi lain yang mengalami tekanan ekstrem. Penelitian ini bertujuan membangun model CoVaR berbasis pendekatan Quantile Autoregressive (QAR), serta mengidentifikasi faktor risiko paling signifikan menggunakan metode Stochastic Search Variable Selection (SSVS). Pendekatan QAR digunakan karena kemampuannya menangkap distribusi return yang asimetris dan volatil, sedangkan metode SSVS memungkinkan proses seleksi variabel dan estimasi parameter dilakukan secara simultan dalam satu kerangka pemodelan, tanpa memerlukan estimasi ulang model secara terpisah untuk setiap kombinasi variabel. Data merupakan harga saham harian bank dan variabel makroekonomi yang relevan. Penelitian ini meliputi pemodelan VaR-QAR, estimasi CoVaR menggunakan regresi kuantil, seleksi faktor risiko dengan SSVS, serta validasi performa model menggunakan uji Kupiec Proportion of Failures (POF). Hasil penelitian menunjukkan bahwa model QAR menghasilkan estimasi VaR yang seluruhnya bernilai negatif pada kuantil 5% dan 1%, mengindikasikan potensi kerugian pada kondisi pasar ekstrem, dengan risiko tertinggi pada bank berkarakter volatilitas tinggi seperti ARTO, BBHI, dan PNBN. Estimasi CoVaR juga menunjukkan nilai negatif dan mengungkapkan risiko sistemik yang lebih besar pada bank menengah dan digital dibandingkan bank besar. Seleksi variabel menggunakan SSVS mengidentifikasi IHSG, Indeks LQ45, dan Indeks Sektor Keuangan sebagai faktor risiko paling signifikan. Uji Kupiec POF menunjukkan bahwa model CoVaR-SSVS memiliki validitas yang lebih baik dibandingkan model CoVaR konvensional.
======================================================================================================================================
Financial sector stability is a crucial aspect in maintaining economic resilience, particularly in the banking sector. Systemic risk arising from interconnectedness among financial institutions requires risk measurement approaches capable of capturing the dynamics of such relationships. Value-at-Risk (VaR) is commonly used to measure potential losses at the level of individual institutions or portfolios; however, it is insufficient to represent the contribution of individual risk to macro-level financial stability or systemic risk. Therefore, Conditional Value-at-Risk (CoVaR) is developed to measure the risk of a financial institution conditional on another institution experiencing extreme distress. This study aims to construct a CoVaR model based on the Quantile Autoregressive (QAR) approach and to identify the most significant risk factors using the Stochastic Search Variable Selection (SSVS) method. The QAR approach is employed due to its ability to capture asymmetric and volatile return distributions, while SSVS enables variable selection and parameter estimation to be conducted simultaneously within a unified modeling framework, without requiring repeated re-estimation of separate models. The data consist of daily stock prices of banking institutions and relevant macroeconomic variables. The research framework includes VaR–QAR modeling, CoVaR estimation using quantile regression, risk factor selection via SSVS, and model performance validation using the Kupiec Proportion of Failures (POF) test. The results indicate that the QAR-based model produces VaR estimates that are entirely negative at the 5% and 1% quantiles, reflecting potential losses under extreme market conditions, with the highest risk observed in highly volatile banks such as ARTO, BBHI, and PNBN. CoVaR estimates are also negative and reveal greater systemic risk exposure among medium-sized and digital banks compared to large banks. Variable selection using SSVS identifies the Composite Stock Price Index (IHSG), the LQ45 Index, and the Financial Sector Index as the most significant risk factors. The Kupiec POF test demonstrates that the CoVaR–SSVS model exhibits superior validity compared to the conventional CoVaR model.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Risiko Sistemik, Conditional Value-at-Risk (CoVaR), Quantile Autoregressive (QAR), Stochastic Search Variable Selection (SSVS)
Subjects:	H Social Sciences > HA Statistics > HA30.3 Time-series analysis H Social Sciences > HG Finance > HG4529 Investment analysis H Social Sciences > HG Finance > HG4910 Investments H Social Sciences > HG Finance > HG4915 Stocks--Prices
Divisions:	Faculty of Mathematics, Computation, and Data Science > Statistics > 49101-(S2) Master Thesis
Depositing User:	Luqyana Zakiya Almas
Date Deposited:	27 Jan 2026 06:36
Last Modified:	27 Jan 2026 06:36
URI:	http://repository.its.ac.id/id/eprint/130483

Actions (login required)

View Item