ABSTRACT
Serial correlation, also known as autocorrelation, is when the error terms from a regression are related to one another. A common way in which the error terms can be related is a first-order Markov scheme. Autocorrelation is a time-series problem that occurs frequently with economic data. Second-order serial correlation is much less frequently encountered. Serial correlation is a direct violation of the Classical Linear Regression Model. There are many techniques available to detect serial correlation. The most popular technique for detecting serial correlation is the Durbin-Watson test. Lagrange multiplier tests for serial correlation are increasingly popular since they overcome all the debilities of the Durbin-Watson test: they are fine for small samples, autoregressive models, and models without a constant term. Serial correlation has more serious consequences in autoregressive models and it is tricky to remedy. If an autoregressive model suffers from serial correlation then the structural parameters are biased as well as the residual statistics.
