CONTENTS

Advertising's Relative Impact on Milk Consumption

Have Expenditures to Advertise California Almonds been Effective?

Editor's Notes

Director’s Corner

Next Meeting

Advertising's Relative Impact on Milk Consumption

by Henry W. Kinnucan

Advertising expenditures for non-alcoholic beverages in the United States totaled $1.1 billion. Given these expenditures, one might think that advertising has a significant impact on consumption, especially since soft drinks, which account for nearly half the expenditures have seen tremendous consumption growth. In fact, the best available evidence suggests that soft drink advertising, indeed all advertising, has little effect on consumption at the aggregate (total industry) level. This is especially true when advertising effects are contrasted with other factors that affect consumption.

To illustrate this, I have summarized in table 1 (see page 2) the results from a recent study of U.S. non-alcoholic beverage demand (Xiao, Kinnucan, and Kaiser). This study included all known determinants of milk consumption--namely income, relative prices, advertising, and demographics. It was based on annual time series data for the period 1970-94. The table reports the study's elasticities for fluid milk, and the percentage change in the model's variables for 1990-94, the last five years of the sample. By combining these two pieces of information--elasticity and percentage change--one can quantify each variable's effect on milk consumption. For example, real per capita income in the United States between 1990 and 1994 increased by 3.44 percent. Multiplying 3.44 percent by 0.084, the study's income elasticity results in 0.29 percent. This 0.29 percent quantifies income's effect or "importance." For example, a similar calculation for milk price results in 1.67 percent. This means that milk price is 5.75 times more important than income in explaining milk consumption over the indicated time interval. (The milk price impact is positive since milk prices over the evaluation interval declined in real terms.)

With the foregoing interpretation in mind, consider soft drink advertising. Its importance is 0.39 percent--higher than income's, but substantially below milk price's. Thus, relative to milk price, soft drink advertising's impact is very small. Note, too, that soft drink's impact is due largely to the percentage change, and not the elasticity. Specifically, soft drink advertising decreased 22 percent in real terms, compared to a decrease of 8.7 percent in real milk prices for the same period. If soft drink advertising had decreased by the same percentage as milk price, its effect would be revised downward to 0.16 percent, which is 10.5 times smaller than the milk price effect.

Let us consider milk advertising now. Between 1990 and 1994, milk advertising expenditures more than doubled in real terms. Based on this, one might expect a substantial demand effect. In fact, the importance coefficient is 0.24 percent--even smaller for soft drinks. This pattern of relatively large percentage changes and miniscule elasticities is repeated across all the advertising effects. For example, coffee and tea advertising expenditures decreased 30 percent in real terms, yet the impact on milk demand is a mere 0.61 percent due to the small elasticity (-0.02). For juices, the only beverage whose advertising caused a reduction in milk demand, the effect is -0.38 percent. In view of the actual reduction in per capita consumption of 3.89 percent, advertising's impact is minor.

This conclusion is consistent with studies on alcoholic beverage advertising in the United Kindgom conducted by Duffy. The original study results, which were confirmed in later replications using alternative models and data sets, indicated that advertising, despite large expenditures, had little detectable effect on alcoholic beverage consumption. As a consequence, Duffy concluded that government regulation of advertising would do little to control alcoholism. Duffy's studies, it may be noted, refer strictly to brand advertising, whereas table 1 is based on a study that includes generic advertising, which can be expected to have a larger effect on aggregate demand. Notwithstanding these differences, the message seems clear: the decline in per capita milk consumption is not likely to be reversed by advertising. To see why, note from table 1 that economic factors, especially income and advertising, have a much smaller effects than demographic factors, which have unsympathetic trends. For example, the "demographic variable fafh, which represents the ratio of food-away-from-home to food-at-home expenditures, has an importance coefficient of -2.83 percent. Comparing this coefficient with the milk-advertising coefficient (0.24 percent), the dining-out effect is 11.8 times larger than the milk advertising effect, despite the 135 percent increase in expenditures. The primacy of demographic effects over advertising effects echoes an earlier study based on New York City data (Kinnucan). The estimates imply that to neutralize the effect of increased dining out, the milk industry would have been forced to increase its advertising expenditures by at least 1,593 percent. (The "at least" qualification is necessary because the law of diniminshing returns specifies a smaller demand shift than predicted by the elasticity given such a large percentage change.) At current levels of milk advertising, this would require an annual investment of over $1 billion, more than for all non-alcoholic beverages combined!

Further perspective can be gained from table 2, which presents the results from table 1 in cumulative form. The primacy of demographic factors in explaining milk consumption is evident. In particular, the total demographic effect is -7.87 percent, compared to a total advertising effect of 0.97 percent. Moreover, most of the total advertising effect is attributable to competing beverage advertising, not to milk advertising, whose individual effect is a mere 0.24 percent. Although advertising attenuated the milk consumption decline (it would have declined by an estimated 4.86 percent without advertising, compared to the actual decline of 3.89 percent) cost of the attenuation is due to reductions in competing beverage advertising, not to increases in milk advertising. Overall, economic factors, including advertising, had a beneficial effect on milk consumption, which helped offset the adverse effects of the demographic factors.

Given the minute effect, are we to conclude that milk advertising is unprofitable? Not necessarily. The reason is that the investment is very small in relation to milk value. For example, in 1994 Americans spent $15.2 billion on fluid milk against farmer investment in advertising of $79 million. With such a huge discrepancy between value and investment, even after subtracting the middlemen's 60 percent share, it does not take much of a demand shift to generate sufficient revenue to cover costs. Still, it should be noted that the milk advertising effect in table 1 is statistically insignificant. Although an alternative model specification produced a "significant" advertising effect, statistical tests indicated that the model's restrictions were incompatible with the data. Thus, we are hesitant to make any firm statements about the profitability of milk advertising based on this study, since the estimated effects are fragile.

A caveat in intrepreting study results is that the data period terminates in 1994, prior to the milk processors' "milk mustache" campaign. Thus, this research has no bearing on the effectiveness of the expanded campaign, which might be high. To test that, we plan to replicate the study using more recent data. If the milk-moustache campaign is indeed effective, as its proponents assert, the estimated coefficient for milk advertising should be positive and robust, i.e., not sensitive to model specification.