when I want to fit a GARCH model to study volatility of the agricultural commodity price, I have included lag forms of the change in original variable ( for example p (t)- p (t-4 ). Am I right ? The variance equation of GARCH model can contain the lag terms of the original dependent variable?