Robust estimation of inverse Pareto distribution: insights into Malaysian lower-income groupsAnggaran teguh taburan Pareto songsang: pandangan tentang kumpulan berpendapatan rendah di Malaysia

The lower tail data of the income distribution can be well described by the inverse Pareto distribution, which is a useful statistical model. Researchers often use the maximum likelihood approach in evaluating the model’s shape parameter. However, this approach might be unreliable as extreme values data points can easily affect it negatively, which may cause inaccurate estimates. This paper introduces a robust estimator known as repeated median estimator for the inverse Pareto distribution. We also demonstrate its efficiency by measuring its asymptotic relative efficiency. The robustness of the method is assessed by the breakdown point and the influence function. The Monte Carlo simulation studies are then conducted to measure its performance as compared to its counterparts. It is found that our proposed estimator outperforms the maximum likelihood, method of moments, and method of product spacings based on the simulation studies. We then apply the inverse Pareto distribution, leveraging the repeated median estimator, to model the lower-income data from the Household Income Surveys in Malaysia for the years 2012, 2014, 2016, 2019, and 2022. To compute the income disparity of low-income households in Malaysia, the parametric Lorenz curve is fitted using the inverse Pareto model, and the Gini coefficient is estimated. This confirms that the proposed approach achieves meaningful practical performance improvements over conventional estimation techniques in the presence of outliers.

Text 125431.pdf - Published Version Download (1MB) Official URL or Download Paper: https://www.ukm.my/jsm/pdf_files/SM-PDF-55-3-2026/...

Actions (login required)

View Item