Fourth Quarter 2002 A newsletter from the National Institute for Commodity Promotion Research and Evaluation on program evaluation and related issues NICPRE QUARTERLY Vol. 8 No. 4 Changes in Advertising Elasticities Over Time Todd M. Schmit and Harry M. Kaiser Cornell University Most economic models used to evaluate the effects of generic advertising programs over time measure the average impacts of various factors on demand. These constant-parameter models can be problematic when the time period covered is relatively long and the marketing environment has sufficiently changed over time. For example, some of our analyses of national dairy advertising have been based on data spanning more than 20 years. As such, the effect of generic fluid milk and cheese advertising was measured as an average over this lengthy period. A mean-response model of generic advertising may not accurately convey the current degree of advertising effectiveness, given changes in market environments, population profiles, and eating behavior over time. In addition, advertising messages have changed, conversion of state to nationally run programs has occurred, and additional producer groups (e.g. milk processors) have contributed to the national program since the inception of the generic advertising programs. A more appropriate way to measure the impacts of advertising, given a long history of time series data, is to use a time-varying parameter model. These types of models measure how the impact of demand factors, including generic advertising, vary over time. In this issue of the NICPRE Quarterly, the results of a time-varying parameter model for the dairy industry are examined. Because there are many factors that influence the demand for fluid milk besides advertising, the following variables were included as factors influencing per capita fluid milk demand: the Consumer Price Index (CPI) for fluid milk, the CPI for non-alcoholic beverages used as a proxy for fluid milk substitutes, per capita disposable income, the percentage of the U.S. population less than six years old, the percentage of the U.S. population that is African American, variables to capture seasonality in fluid milk demand, a trend variable to capture changes in consumer tastes towards fluid milk over time, expenditures on brand fluid milk advertising, and expenditures on generic fluid milk advertising. The following variables were included as factors influencing per capita cheese demand: the CPI for cheese, the CPI for meat used as a proxy for cheese substitutes, per capita disposable income, per capita food away from home (FAFH) expenditures, the percentage of the U.S. population that is Asian/ Hispanic; the percentage of the U.S. population between 20 and 44 years old, variables to capture seasonality in cheese demand, a trend variable to capture changes in consumer tastes towards cheese over time, expenditures on brand cheese advertising, and expenditures on generic cheese advertising. The model was estimated with national, quarterly data from 1975 through 2001. To account for the impact of inflation, all prices and income were deflated. Brand and generic fluid milk and cheese advertising expenditures were deflated by the Media Cost Index. Because advertising has a carry-over effect on demand, past advertising expenditures also were included in the model as explanatory variables using a distributed-lag structure. Unlike most previously used constant- parameter models which measure the average impact each of the above factors has on milk and cheese demand, the time- varying parameter model measures each demand factor’s impact on a quarterly basis. Moreover, the model is able to identify what factors were most important to the variation of advertising response over time. This is advantageous since the model not only allows one to measure the magnitude of each factor on demand, but also how that magnitude has changed and what has impacted this change over time. The relative impacts of variables affecting demand can be represented with what economists call “elasticities.” These measure the percentage change in per capita demand given a 1-percent change in one of the identified demand factors. Table 1 presents the estimated elasticity values for the primary demand factors for fluid milk and cheese averaged over the most recent five-year period. For example, the price elasticity of demand for fluid milk equal to -0.14 means that a one-percent increase in the real, inflation adjusted, fluid milk price decreases per capita fluid milk demand by 0.14 percent. While table 1 presents these elasticities as 5-year averages, discussion in the text and figures that follow display how selected elasticities have varied annually over time. Due to space constraints, the principal focus here is on the generic advertising elasticities for fluid milk and cheese. Generic advertising was positive and significant for both fluid milk and cheese demand. The time-varying advertising elasticities show substantial variation over time, with both increasing considerably since the beginning of the sample period (figure 1). Since 1995,