CONTENTS

Does Advertising Effectiveness Vary over Time?

Editor’s Notes

Manager’s Viewpoint

Asymmetric Response to Advertising

Director’s Corner

Next Meeting

Asymmetric Response to Advertising and Implications for Pulsing:
A Case Study of Fluid Milk Advertising in New York City

by Philip R. Vande Kamp and Harry M. Kaiser

Understanding consumers’ response to generic commodity promotion is a major concern for generic commodity promotion organizations and for producers who provide the financial support for promotion programs. The nature of the demand response to advertising is important in determining what advertising strategies are most effective in increasing demand.

How do consumers react to an increase in advertising compared to a decrease in advertising? Is this information useful for determining the best advertising strategy? Is a pulsed advertising strategy more effective in increasing demand than maintaining a fixed level of advertising? In this research, these questions and others are addressed in a case study of fluid milk advertising in New York City.

By applying a new estimation methodology, this study independently measures the demand impact from an increase in advertising and the impact from a decrease in advertising. Separate measurements are important for measuring the pace of “information decay” and the resulting decline in demand that occurs when advertising decreases, and the “absorption of information” and corresponding increased demand when advertising increases.

Consumers can respond to increases and decreases in advertising in many possible ways. One possibility, which is imposed in most advertising and promotion research, is consumer demand responds symmetrically to increases and decreases in advertising. That is, if advertising increases by one unit, demand immediately increases by a fixed amount, and similarly, if advertising decreases by one unit, demand immediately decreases by the same fixed amount. In another scenario, demand could respond asymmetrically to increases and decreases in advertising. In this case, the speed and magnitude of the impact of advertising differs for increases in advertising compared to equal decreases in advertising.

By using an econometric model, the effects of advertising, competing advertising, price of competing beverages, household income, demographics, and seasonal variation on demand for fluid milk are statistically estimated. The data used in this study come from the New York City fluid milk market. Carbonated beverages are considered a substitute for fluid milk and therefore carbonated beverage advertising and price were included in the economic model. Data on fluid milk demand and prices were obtained from the New York State Department of Agriculture and Markets. Regional fluid milk advertising expenditures were provided by advertising agencies handling the fluid milk account. Advertising was measured as total per capita real advertising expenditures for fluid milk and was obtained by summing regional and national per capita real advertising expenditures in New York City. Other relevant data were obtained from Leading National Advertising, Inc. and Bureau of Labor Statistics. Demographic data were obtained from 1997 State Profile: New Jersey and New York.

An advertising “goodwill” variable was created from a weighted distribution of past advertising where the weights were simultaneously estimated with the other parameters in the econometric model. The advertising goodwill measurement reflects a “stock of information” from current and past advertising that currently influences demand. Using a new econometric approach developed for this purpose, the influences of increases in advertising and decreases in advertising are estimated along with the other demand determinants listed above. As expected, results indicate that milk demand increases with higher levels of advertising. In addition, reduced levels of carbonated beverage advertising and increases in carbonated beverage prices increase fluid milk demand. These results are not new or surprising.

More interestingly, the results also showed that the response of demand to advertising is asymmetric. When advertising increases, consumers were found to increase demand fairly rapidly. When advertising decreased, however, consumers did not decrease demand immediately, and only decreased demand gradually over time. Specifically, results showed that, in the long term, a permanent one percent increase in advertising for milk provides a 0.049 percent increase in demand. Similarly, a permanent one percent decrease in advertising gives a 0.049 decrease in the demand in the long term. Conversely, in the short term, the response is different for an increase in advertising compared to a decrease. For a permanent one percent increase in advertising goodwill, demand immediately increases 0.049 percent and remains at that level. However, when advertising goodwill decreases, demand is not immediately affected. Our results show that for one month after advertising goodwill decreases, demand does not decline. For two to three months after a permanent one percent decrease in goodwill, demand only declines 0.015 percent. Finally, for more than three months after a permanent one-percent decrease in goodwill, demand declines the full 0.049 percent. The demand response to an increase and decrease in advertising goodwill is shown graphically in figure 1.

Figure 1. Asymmetric Demand Response to Advertising

This result is not completely new in advertising research. Using marketing tests, some authors have reported a similar result where the response to advertising is more gradual when advertising declines than when advertising increases. Authors have suggested that consumers are more likely to be aware of the appearance of an advertisement than the absence of an advertisement. Advertising is an active communication process which may stimulate consumers to action, but a decline in advertising likely results in “information decay,” that is, consumers gradually forget information provided in advertisement. The demand response, after a reduction in advertising may depend on habitual behavior, and will gradually decline.

Should this asymmetric response to advertising have an impact on the choice of advertising strategy? Our answer is yes. The carry-over effect of advertising when advertising decreases plays a role in how advertising should be allocated over time. To illustrate this point, the estimated economic model, which incorporates the asymmetric response to advertising, is used to simulate fluid milk demand for several alternative advertising strategies.

Using the last two years of monthly data in this study, simulations are performed which compare a uniform advertising strategy (constant level of advertising in all months) to two pulsing advertising policies. The actual total advertising expenditures over the two-year period is used and is the same for all three simulations. With the exception of the advertising data, all of the other demand determinants are actual data for each month. In the first simulation, the sales from a uniform advertising strategy are predicted by setting milk advertising to a constant level at the average monthly advertising for the two years. In the second simulation, a pulsing advertising strategy is used where a pattern of three months of zero advertising followed by three months of twice the monthly average level of advertising is repeated. The third simulation is similar, where a pattern of four months of zero advertising followed by four months of twice the monthly average level of advertising is repeated.

Average per capita sales per day resulting from the uniform advertising strategy were 0.5488 pounds. For the second and third simulations, the average sales were 0.5651 and 0.5664 pounds respectively which is a 2.97 and 3.21 percent increase over the uniform advertising strategy respectively. Similar pulsing intervals beyond four months showed lower average sales relative to the second and third simulations. The actual advertising strategy, which includes some pulsing, resulted in average sales of 0.5566 pounds which is 1.42 percent higher than the uniform advertising strategy, but lower relative to the second and third simulations. In summary, the pulsing adverting strategies were found to result in significantly higher sales than the uniform advertising strategy and the actual advertising strategy. Since the simulation results presented here are only examples of possible advertising strategies, on-going research will attempt to characterize the optimal adverting strategy when the advertising response is asymmetric.

In response to the questions posed at the beginning of this article, fluid milk demand in New York City is found to react asymmetrically to increases and decreases in advertising. Furthermore, this asymmetric response turns out to be important for choosing the best advertising strategy. Finally, given the asymmetric response to advertising discovered here, several pulsing advertising strategies were found to be more effective in increasing demand than a uniform advertising strategy.

Phil Vande Kamp and Harry M. Kaiser are research associate and associate professor, respectively, with the Department of Agricultural, Resource, and Managerial Economics, Cornell University