Feasible Generalized Least Squares Based on Minimum Variance Quadratic Unbiased Estimation for an Unbalanced One-Way Error-Component Panel Model
Abstract
Panel data combine cross-sectional and time-series information, but incomplete firm-year observations produce an unbalanced structure that requires an estimation procedure accommodating unequal observation counts. This study applies a one-way random-effects error-component model to 42 observations from 10 consumer-goods manufacturing firms listed on the Indonesia Stock Exchange during 2017–2021. The firm-specific and idiosyncratic variance components are estimated using Minimum Variance Quadratic Unbiased Estimation (MIVQUE), and the resulting covariance matrix is used to estimate the regression coefficients through feasible generalized least squares (FGLS). The estimated variance components are 0.0063 for the firm-specific effect and 0.0805 for the idiosyncratic disturbance, indicating that approximately 7.26% of the composite disturbance variance is attributable to persistent differences among firms. The fitted model is Yit = 0.0657 − 0.0229DERit+).2089NPMit. The debt-to-equity ratio is negatively associated with stock returns, whereas the net profit margin is positively associated with stock returns. However, neither coefficient is statistically significant at the 5% level, with p-values of 0.0938 and 0.5165, respectively. Therefore, the estimated relationships are interpreted as conditional associations under the assumed random-effects specification rather than as causal effects. This study provides an empirical implementation of MIVQUE-based covariance estimation combined with FGLS for an unbalanced one-way error-component panel model.
Downloads
Copyright (c) 2026 Refa Joyce Semida, Raupong Raupong, Siswanto Siswanto, A. Muthiah Nur Angriany

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.














.jpg)





