Comparative modelling of parameter estimation methods with Nigerian financial sector statistics

Abstract

The study focused on the performance of the parameter estimation methods on financial sector statistics in the Nigerian economy and identified the sampling distribution of the test statistic θ. Three general methods of parameter estimation (least-squares estimation, maximum likelihood estimation and method of moments) were compared using their standard errors. Secondary quarterly data collected from Central Bank of Nigeria Statistical Bulletin 2017 spanning from 1988-2017. Datasets on financial sector statistics were used as the basis for defining the population and the true standard errors. The sampling distribution of the financial sector statistics was found to be a Chi-square distribution and was confirmed using a resampling method. The stability of the test statistic θ was also ascertained. In addition, other diagnostic checks, Akaike Information criteria; Schwart Bayesian Information criterion, Hannan-Quinn Information criterion were used and they confirmed that when the financial sector statistics was resampled, the parametric resampled model (R311) turned out to be the best model under several assessment conditions.

Keywords: Information criteria, resampling, parametric models, parameter estimation, kernel density, quantile-quantile plots