Trading-Day Variation: Theory and Implications for Monthly Meat Demand

We consider the problem Of fitting regression models with data containing trading-day variation. Multiplicative and additive expressions for trading-day variation are presented. Multiplicative adjustment is more reasonable than the additive but is also more complex. Expressions are derived for biases which trading-day variation introduces into least squares estimators. Estimated retail demands for beef, pork, and chicken show the biases are large and in the directions predicted when monthly data are used, but are small when quarterly data are used. Multiplicative adjustment is statistically superior to additive adjustment, although practical differences are small.