The effectiveness of movie trailer advertising

Abstract

Prior to a movie release in theaters, trailer advertising provides valuable information that can help viewers and investors form expectations about the movie's future success. While previous research has looked at the financial implications of movie advertising budgets, the effects of trailers' creative characteristics on abnormal returns have not yet been investigated. Using a sample of movie trailers, results from our event study and cross-sectional analysis show that the appeal of the movie plot revealed in the trailer, the number of scene cuts and the inclusion of violent, sexual, or humorous scenes influence the movie's abnormal returns. However, the use of special effects in the movie trailer does not impact investors. Results also suggest that investors react more strongly to first than to follow-up trailers released for the movie, and that early release of the first positively impacts the movie's returns.

Keywords:

• advertising effectiveness
• movies
• event study
• regression
• trailer advertising

Notes

1. As defined by the World Health Organization, violence is “the intentional use of physical force or power, threatened or actual, against oneself, another person, or against a group or community that either results in or has a high likelihood of resulting in injury, death, psychological harm, mal-development or deprivation” (Kug et al. 2002). Sexual content in a trailer refers to scenes in which persons engage in sexual behavior, appear in attire designed to evoke sex appeal or are portrayed as sex objects (Oliver and Kalyanaraman 2002).

2. The results from a one-sample Kolmogorov-Smirnov test (p = .771) also supports normality for CAAR.

3. The t-tests were performed on the box cox transformed SCAAR in order to reach normality at 95% interval for L where the transformed SCAAR is (SCAAR^L-1)/L, and L = .049 and −.42 for the event windows [0,1] and [0,0] respectively.

4. A regression including only control variables as predictors shows that it is significant (F(11, 128) = 1.95, R² = .143, p < .05) after verifying for homogeneity of standard errors using White's (p = .347) and Breusch-Pagan/Cook-Weisberg tests (p = .818). However, a model specification link test for single-equation models (Pregibon 1980; Tukey 1949) indicates that such control model is not properly specified (p > .1), and, therefore, that additional significant predictors can be added to the model. This is further confirmed by an adjusted R² of the full model that is over 50% higher than for the control model.

5. All reported values are estimated standardized coefficients (see Table 4).

6. We also considered that the effect of the number of scenes on CAAR could be curvilinear and performed a regression where we add the mean-centred squared value to the model. The result shows that the coefficient for the added variable is not significant (b_SCENES^2 =−.042, p =.589). All remaining variables have similar effects to those discussed in the paper.