ABSTRACT
As per the Assumption 11 of Classical Linear Regression Model, the successive values of the random error term are independent i.e., the ‘µt’ value in any one period is independent of ‘µt-1’ value in previous period (t-1) or with earlier (lagged) periods, of the same variable. This assumption states that, the covariance and correlation between different disturbances are all zero and this is referred as serial independence. Some applications of regression involve regressor and regressand variables that have a natural sequential order over time. Autocorrelation is sometimes called as ‘Lagged correlation’ or ‘Serial correlation’. This autocorrelation is considered as a problem because, its presence means that, useful information is missing from the model. The presence of autocorrelation means that, a model is making a similar mistake over and over as it attempts to explain movement in the dependent variable, Y.
