Externalities across advertising markets

ABSTRACT

This paper analyzes externalities generated by offline advertising campaigns on the performance of online ads. Using advertising data on a panel of firms in the hotel industry, we estimate how a firm’s offline, display, and competing ad campaigns impact the effectiveness of Google and Facebook advertising. We find a positive effect of traditional mass-media campaigns on Google clicks. Advertising from competitors does not affect Google ad performance but it increases advertising prices, suggesting keyword poaching. Further analyses hint that Google’s monopoly power and auction system allow free-riding on advertising externalities. Although we find similar positive effects on Facebook ads, they are not significant.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ This phenomenon is known as last-touch attribution.

² In addition to a low level of spending, the firm does not advertise offline in Germany at all. This reduced advertising activity is due to a low number of hotels supplied by the chain abroad.

³ A similar plot by brand reveals some heterogenity between low-cost and middle range brands (Figure A1)

⁴ Dinner et al. (2014) found a negative non-significant effect of a luxury clothing retailer’s display ads on the click rate for search ads. In contrast, Kireyev et al. (2016)’s study on a bank showed that display impressions often increased search effectiveness while search ads decreased display performance. The study by R. Lewis and Nguyen (2015) also showed that display advertising could trigger consumers’ search for competing brands, explaining these ambiguous findings.

⁵ The query index, retrieved from the Google Trends service is normalized between 0 and 1 on the study period.

⁶ The reason for choosing t − 2 instead of t − 1 is that firms are likely to adjust their advertising budget on a year-on-year basis. This is especially true for offline ads which are typically bought far in advance (6 months to a year), before advertisers have observed search ads’ performance. Thus, the full effect of Google’s performance on the media budget share may be observed at least two semesters after the performance has been reported. A model run with log(Y_bc,t−1) shows similar but less significant results (Figure A2), suggesting that a semester is not the appropriate time span to observe a substitution effect.

⁷ The amount of the marginal offline adstock increases is comprised between e978 and e1148 but is hardly determinable because of a very high standard deviation.

⁸ In 2019, the entire Facebook group controlled less than half of display ad spendings (Statista, 2021b).

⁹ As our results suggest, online advertisement generates a distinct complement that has different effects on online searches.

Additional information

Notes on contributors

Rémi Devaux

Rémi Devaux is a PhD student in economics at Mines Paris - PSL. His work focuses on advertising and digital economics.

Olivier Bomsel

Olivier Bomsel is professor at Mines Paris PSL. He is the holder of the Chair of Media & Brand Economics.