Empirically Distinguishing Informative and Prestige Effects of - - PowerPoint PPT Presentation

Introduction Data Model Results Conclusions

Empirically Distinguishing Informative and Prestige Effects of Advertising

(Ackerberg, 2001 RAND)

Panagiotis Adamopoulos

Department of Information, Operations and Management Sciences Stern School of Business, NYU

padamopo@stern.nyu.edu

11/14/2011 - Industrial Organization (Ph.D.)

Empirically Distinguishing Informative and Prestige Effects of Advertising Panagiotis Adamopoulos

Introduction Data Model Results Conclusions

◮ How advertisements affect rational consumers who view them? ◮ Argument:

◮ Advertisements that give consumers product information should

primarily affect consumers who have never tried the brand.

◮ Advertisements that create prestige or image effects should affect

both inexperienced and experienced users.

◮ Techniques to empirically distinguish and measure different

effects of brand advertising in nondurable, experience-goods markets.

◮ Important implications for optimal advertising behavior.

Empirically Distinguishing Informative and Prestige Effects of Advertising Panagiotis Adamopoulos

Introduction Data Model Results Conclusions

There are two different types of consumers:

◮ Inexperienced:

◮ people who have never tried the brand or product, and

◮ Experienced:

◮ those who have consumed the brand at some point in the past.

Empirically Distinguishing Informative and Prestige Effects of Advertising Panagiotis Adamopoulos

Introduction Data Model Results Conclusions

There are three different types of advertising:

◮ Information on product existence & search characteristics

◮ Informs consumers of a brand’s existence and search

characteristics.

◮ Such advertising will affect only consumers who do not already

know of the brand’s existence or those search characteristics.

◮ Information on experience characteristics

◮ Informs consumers about a brand’s experience characteristics. ◮ Consumers who have consumed the brand more times should

know more about these characteristics and be less affected.

◮ Prestige and image effects

◮ Directly affect the utility a consumer derives from consuming the

brand (either increase or decrease).

Empirically Distinguishing Informative and Prestige Effects of Advertising Panagiotis Adamopoulos

Introduction Data Model Results Conclusions

The data

Publicly1 available consumer-level panel data on yogurt purchases and television advertising exposures to empirically examine the previous arguments.

◮ 2000 households ◮ “Scanner data” ◮ Two geographic markets ◮ Three years (1986-1988) ◮ Weekly prices ◮ Household TV advertising exposures ◮ Newly introduced brand (Yoplait 150)

1A.C. Nielson data

Empirically Distinguishing Informative and Prestige Effects of Advertising Panagiotis Adamopoulos

Introduction Data Model Results Conclusions

Why Yoplait 150?

◮ Short shelf life (no stockpiling behavior) ◮ Newly introduced brand (no experienced consumers)

◮ Broadly advertised

◮ Competitors

Empirically Distinguishing Informative and Prestige Effects of Advertising Panagiotis Adamopoulos

Introduction Data Model Results Conclusions

Quantity and Price Series Figure: Time Series.

Empirically Distinguishing Informative and Prestige Effects of Advertising Panagiotis Adamopoulos

Introduction Data Model Results Conclusions

Advertising Series Figure: Advertising Series.

Empirically Distinguishing Informative and Prestige Effects of Advertising Panagiotis Adamopoulos

Introduction Data Model Results Conclusions

Results from regression Figure: Results from regression.

Empirically Distinguishing Informative and Prestige Effects of Advertising Panagiotis Adamopoulos

Introduction Data Model Results Conclusions

The empirical model

Reduced form of discrete choice model: cit =

• 1

iff Xitβ1 −γpit +ε1it > Zitβ2 +ε2it

• therwise,

where cit, indicates whether consumer i purchased Yoplait 150 on shopping trip t.

Empirically Distinguishing Informative and Prestige Effects of Advertising Panagiotis Adamopoulos

Introduction Data Model Results Conclusions

Econometric results

Parameter Simple Logit Normal Random Effect Advertising*Inexperienced (1) 2.04073 2.30566 (.723113) (.77561) Advertising*Experienced (2) .90371 .43304 (Image & Prestige Effects) (.63504) (1.21180) (1) - (2) 1.13702 1.87262 (Informative Effects)

Empirically Distinguishing Informative and Prestige Effects of Advertising Panagiotis Adamopoulos

Introduction Data Model Results Conclusions

Econometric results (2)

◮ An additional 30-second commercial every week for the average

inexperienced consumer has the same effect to the purchase probability as a 10-cent price decrease.

◮ Advertising elasticity of demand for Yoplait 150 is around .15. ◮ The marginal effect of advertising is going down as a consumer’s

number of previous purchases increases.

Empirically Distinguishing Informative and Prestige Effects of Advertising Panagiotis Adamopoulos

Introduction Data Model Results Conclusions

Conclusions

◮ Yoplait 150 advertisements were influencing consumer behavior

primarily by informing them about search and experience characteristics, not by creating prestige or associating the product with favorable images.

◮ This approach can fairly easily be applied to other products or

industries.

Empirically Distinguishing Informative and Prestige Effects of Advertising Panagiotis Adamopoulos

Appendix

Summary Statistics Figure: Summary Statistics.

Empirically Distinguishing Informative and Prestige Effects of Advertising Panagiotis Adamopoulos

Appendix

Number of Shopping Trip Figure: Number of Shopping Trips.

Empirically Distinguishing Informative and Prestige Effects of Advertising Panagiotis Adamopoulos

Appendix

Purchases Figure: Purchases.

Empirically Distinguishing Informative and Prestige Effects of Advertising Panagiotis Adamopoulos

Appendix

Explanatory Variable Descriptive Statistics Figure: Explanatory Variable Descriptive Statistics.

Empirically Distinguishing Informative and Prestige Effects of Advertising Panagiotis Adamopoulos

Appendix

Endogeneity Problem

A consumer has a high initial intercept because she saw an advertisement or simply because she happened to like the brand Yoplait already?

◮ Instrumental Variable: of Mean advertising intensity.

Empirically Distinguishing Informative and Prestige Effects of Advertising Panagiotis Adamopoulos