Gibrat's law in marketing: The case of liquor brand sales

We test the applicability of Gibrat's Law in the liquor brand market. Basically, we model annual changes in the unit sales of the top fifty liquor brands as white noise. Our results reject this model, but we do find that changes in sales are independent of starting market sales. This leads to the interpretation that brands with above average market share do not tend to gain market share, i.e., initial market share does not affect the subsequent change in market share. Furthermore, brands with above average sales do not have more stable sales than do firms with below average sales. Changes in sales appear highly positively correlated between periods, i.e., brands that gain sales in one period tend to gain sales in the next. Finally, no major liquor type or manufacturer had consistently and significantly greater or lower success across our various annual time periods.