AUTOCORRELATION: WHAT HAPPENS IF THE ERROR OR DISTURBANCE TERMS ARE CORRELATED IN TIME-SERIES DATA

Abstract

This research explores the econometric investigation of emblematic and well-acknowledged subjects over the years, such as serial correlation, also known as autocorrelation, where error terms in a time series data transfer from one-time period to another. This inquiry explores a few major and central foundations of autocorrelation terminologies. Furthermore, this study will enlighten the cause and remedies for the classical assumption named autocorrelation. This problematic term in time series data analysis arises because exogenous seems dissimilar dimensionally with Predict, but errors are not independently acting. It violates the assumption of linear regression where model indicators or one of the explanatory variables is the lagged value of the dependent with the same quantified amount of errors from a one-time period shifted in pattern to the subsequent. Correspondingly omitted variables or essential exogenous are absent or deleted cause their effect on the dependent variable becomes part of the error term. Hence, this paper has employed suitable heteroscedasticity-consistent (HC) and heteroscedasticity, cross correlogram, and autocorrelation consistent (HAC) estimators. In this inquiry, Eviews for statistical computation is utilized and show how variables' suggested functioning and behaviour. This study employed five economic indicators to show the autocorrelation cause and its remedial measure (2003 to 2020).