Movie Stars and Authors As Brand Names: Measuring Brand Equity in Experiential Products

Aron M. Levin, Washburn University
Irwin P. Levin, University of Iowa
C. Edward Heath, University of Kentucky
ABSTRACT - Using experimental designs analogous to those employed with standard products, brand equity effects were measured for two types of experiential products: movies and novels. Descriptions of the plots of new movies (Experiment 1) and new novels (Experiment 2) were ascribed to well-known or unknown movie stars or authors. Not only were consumer reactions more favorable with well-known movie stars/authors, but interaction effects in each experiment revealed that the effects of critics’ comments were attenuated with products ascribed to familiar and trusted stars. However, line extension effects were shown to be negligible for authors associated with unfamiliar genres.
[ to cite ]:
Aron M. Levin, Irwin P. Levin, and C. Edward Heath (1997) ,"Movie Stars and Authors As Brand Names: Measuring Brand Equity in Experiential Products", in NA - Advances in Consumer Research Volume 24, eds. Merrie Brucks and Deborah J. MacInnis, Provo, UT : Association for Consumer Research, Pages: 175-181.

Advances in Consumer Research Volume 24, 1997      Pages 175-181


Aron M. Levin, Washburn University

Irwin P. Levin, University of Iowa

C. Edward Heath, University of Kentucky


Using experimental designs analogous to those employed with standard products, brand equity effects were measured for two types of experiential products: movies and novels. Descriptions of the plots of new movies (Experiment 1) and new novels (Experiment 2) were ascribed to well-known or unknown movie stars or authors. Not only were consumer reactions more favorable with well-known movie stars/authors, but interaction effects in each experiment revealed that the effects of critics’ comments were attenuated with products ascribed to familiar and trusted stars. However, line extension effects were shown to be negligible for authors associated with unfamiliar genres.

More than a decade ago, Holbrook and Hirschman (1982) urged consumer behavior researchers to break from the traditional information processing paradigm and examine consumer behavior from an experiential perspective. In particular, they suggested that marketers interested in consumer search behavior study entertainment, the arts and leisure activities such as musical recordings, novels, paintings, and movies: "Hence, applications of multivariate methods may be more valid with this type of produt than with some low-involvement consumer nondurables, such as detergents or canned peas, for which consumers may be unable to make valid perceptual or affective distinctions among more than a few different brands" (Holbrook & Hirschman 1982, p. 134). Since the publication of Holbrook and Hirschman’s influential article, a number of studies have examined experiential consumption and consumers’ choice of experiential products. Some of the forms of entertainment that have been explored are: comic books (Belk 1987), television programs (Barwise & Ehrenberg 1987; Hirschman 1988), sports and games (Holbrook, Chestnut, Oliva, & Greenleaf 1984), and motion pictures (Holbrook & Grayson 1986).

The purpose of this paper is to describe two experiments that build onto previous research on consumers’ choice of experiential products. Specifically, we are interested in whether consumers use "brand names" (in this case well-known actors or authors) for experiential products such as movies and novels in a similar manner to their search for low involvement, tangible products. Our analog of "star" as "brand" is based on the notion that popular movie actors and authors, like popular brands, have name recognition, have developed a certain image, and, furthermore, are associated with particular types of products. Consider, for example, the public images of Clint Eastwood and Stephen King.

In an industry where multi-million dollar films are released to compete with one another head-to-head, the presence of a star acts as a beacon that makes the movie stand out and entices larger audiences. A quote by Gorham Kindem in Wallace, Seigerman, and Holbrook (1993) points out the value of the star:

" the Hollywood star system is a business strategy designed to generate large audiences and differentiate entertainment programs and products, and has been used for over seventy years to provide increasing returns on production investments (Kindem 1982, p.79)."

Thus, the value of including stars in a movie or promoting novels with big name authors may be analogous to the signaling effect of brands. In other words the presence of a recognizable star signals quality to the prospective consumer much as a trusted brand name implies quality in a product.

From a conceptual standpoint, we are interested in whether "brands" in experiential products are capable of possessing the same type of brand equity that they do in non-experiential products. Our operationalization of brand equity is consistent with Keller’s (1993) definition of customer-based brand equity: "a brand is said to have positive (negative) customer-based brand equity if consumers react more (less) favorably to the product, price, promotion, or distribution of the brand than they do to the same marketing mix element when it is attributed to a fictitiously named or unnamed version of the product or service". We test the applicability of this definition to movie actors in Experiment 1 and to authors in Experiment 2 by comparing consumers’ evaluations of unfamiliar movies or novels when the same descriptions are associated with either popular or fictitious actors/authors. Differences in evaluations between conditions will then represent the effects of "star power".

In our study of brand equity effects possessed by star actors and authors, we not only extend the study of brand equity effects to experiential products, we also examine how brand equity can attenuate the effects of other sources of information, in this case the comments of movie or book critics. Like many other consumer decisions, the decision of which movie to see or novel to read has multiple determinantsCwhat we have read about them, promotional material, what friends have told us, how much we like the stars/authors, and so forth. Specifically, in this study we focus on how consumers’ decisions are influenced by the following sources of information: identification of "stars" and criti comments. These factors are manipulated experimentally by using plot descriptions of new movies or novels, identifying them in some cases with well-known actors/authors and in other cases using fictitious names, and creating favorable and unfavorable critic comments. Measured reactions to each product include expected degree of liking of the product and the likelihood of patronage.

Critics’ comments are particularly interesting sources of social influence for many experiential products because they provide a vicarious learning experience for the consumer. Critics are credible communicators because of their expertise and lack of vested interest in the product. A viable model for consumer preference for experiential products such as movies, plays and novels must assume a role for critics’ comments. (See, for example, the model of Reddy, Swaminathan, and Motley, 1996.) Nevertheless, the exact manner in which this source of information is integrated with others is not well understood.

There is a clear expectation of preference for movies/novels with popular stars/authors and preference for products with favorable critic comments. A more interesting prediction is that these two sources will have configural effects. Just as reliance on a trusted brand name has been shown to reduce the influence of other sources of information on consumer behavior (Aaker & Keller 1990; Jacoby, Chestnut, & Fisher 1978; Payne, Bettman, & Johnson 1988), the influence of critic comments is predicted to be reduced for movies/novels featuring popular stars/authors compared to products with unknown stars/authors. Such a result would have important implications for marketing experiential products contingent on whether they feature familiar or unknown artists.


There are both economic and psychological reasons for studying movie consumption. The motion picture industry is a multi-billion dollar industry in the United States and around the world. It has been estimated that over $9 billion per year are spent by Americans going to first run movies and renting videos and that a sizable proportion of Americans go to the movie theater at least 20 times every year. [These figures were obtained from: "The Study of Media and Markets," (1992), Simmons Market Research Bureau, Inc,: and "Critics' Choice and Studio Briefing," American On-Line, Inc.] Many more rent or buy videos within a year after a movie is released to the theaters. Underlying these gross figures, of course, are a series of specific decisions made by individual consumers. There has been much research showing the large impact that movies have on individual viewers’ attitudes, mood, physiological and affective states, and even their outlook on life (Erber & Tesser 1992; Forgas & Moylan 1987; Hirschman & Holbrook 1982; Isen & Gorgolione 1983; Linz, Fuson, & Donnerstein 1990; Palmgren & Lawrence 1989). The enormous economic and psychological significance of movie-going decisions should lead to increased interest in research on this particular form of consumer choice.

Experiment 1 tests the following hypotheses: 1) If brand equity effects extend to experiential products and if our conceptualization of "star as brand" is correct, then the same movie description will elicit more favorable responses when identified with a popular star than when associated with an unknown (fictitious) actor. 2) If "star as brand" is a correct conceptualization and if a trusted brand name serves as a heuristic for reducing reliance on other information, then the effects of critics’ comments will be attenuated for movies with popular stars. In addition to testing these hypotheses, this experiment will include a follow-up phase to identify key mediators of brand equity effects attributable to movie stars.


Participants. Participants were 62 undergraduate students in marketing classes at a large southeastern university. They were randomly assigned to a critic comments condition (N=31) and a no comments condition (N=31). In response to questins about their movie-going behavior during the past three months, these groups were found to be comparable on number of movies seen in the theater (means=2.90 and 3.48), number of movies watched on cable TV (means=11.32 and 11.00), and number of video rentals (means=7.68 and 7.10).

Experimental design and material. Each subject received eight movie descriptions. Subjects in the no comments condition were given some descriptions that included the actual well-known stars of the movie (e.g., Billy Crystal, Ted Danson, Tommy Lee Jones) and some descriptions with fictitious actors’ names. Critics’ comments were added for subjects in the critic comments condition. Both movies with and without familiar star names were assigned either favorable comments (e.g., "A pleasant surprise......Very enjoyable") or unfavorable comments (e.g., "Boring plot......I barely even let out a chuckle"). Except for the critic comments, movie descriptions and star designations were matched in the two conditions. Actual movie descriptions were obtained approximately one month prior to their release and before information was available to the public.

Procedure. The study was described as one in which we are interested in how people make decisions about which movies to see. Subjects were told that on each page of their response booklet they would be given a short plot summary of an actual movie that will be released in the next 1-3 months. Subjects in the critic comments condition were also told that they would be shown some comments on the movie from the film critic of a major newspaper who had previewed the movie.

Based on this information, they were asked to indicate their likelihood of: a) Going to see the movie in the theater; b) Waiting to see the movie on cable TV; c) Waiting to rent the video; or d) Deciding not to see the movie at all. Subjects were instructed to assign a value between 0 and 100 to each choice and to make sure that they total to 100. Although we are primarily interested in the decision of whether or not to see a movie, we provided multiple options to make the task more realistic and thus obtain more reliable and valid data. Subjects were also asked to rate on a 5-point scale "How much do you think you would like the movie?", with 1="like a lot" and 5="dislike a lot."


The main data concern the rated likelihood of seeing the movie as a function of type of critic comments (positive, negative or none) and whether or not there are known stars. Data from the category "decide not to see the movie at all" were converted to "likelihood of seeing the movie" by subtracting from 100, which is also equivalent to the sum of the stated likelihoods of seeing the movie by each of the designated options (in the theater, through video rental and on cable TV). Figure 1 shows the likelihood of seeing the movie as a function of critic comments and whether or not a well-known star was identified. There was a wide range of mean values, from an 86% likelihood of seeing a movie with stars and positive critic comments to a 32% likelihood of seeing a movie without stars and with negative critic comments. Figure 1 reveals that movies with stars are more desirable than movies without starsCconfirming the brand equity effectCand that positive critic comments make the movie more desirable than negative comments or no comments.

Perhaps the most interesting result seen in Figure 1 is that the lines differ in slope such that critic comments have a much greater effect for movies without stars than for movies with stars. For example, positive critic comments are especially helpful to movies without stars. Specific contrasts between the means displayed in Figure 1 were conducted to provide statistical tests of the hypotheses of interest. Because one group of subjects received both positive and negative critic comments and another group received no comments, some of the following are within-subject comparisons and some are between-subjects, with different degrees of freedom for eac. Positive critic comments led to higher likelihood ratings than negative comments, t(30)=8.46, p<.01; positive comments led to higher likelihood ratings than no comments, t(60)=3.97, p<.01; and negative comments led to lower ratings than no comments, t(60)=1.79, p<.10. Supporting the observation that critic comments had a reduced effect for movies with stars compared to movies without stars, the interaction between comments and stars was statistically significant for comparison between positive and negative comments, F(1,30)=38.40, p<.01; comparison between positive and no comments, F(1,60)=5.97, p<.05; and comparison between negative and no comments, F(1,60)=4.28, p<.05. While movies with stars produced higher likelihood ratings than movies without stars, this effect was statistically significant with negative comments, t(30)=7.25, p<.01, and with no comments, t(30)=4.57, p<.05, but was not significant with positive comments, t(30)=.053.



An additional dependent variable of interest was "How much do you think you would like this movie?" Results here, summarized in Figure 2, are very much like those for rated likelihood of seeing the movie (Figure 1), except that the scale is reversed where low numbers represent more favorable ratings. As in Figure 1, the lines systematically converge. (The systematic convergence of lines in both Figures 1 and 2, without a flattening at the top or bottom, rules out ceiling or floor effects as the cause of the observed interaction.) Statistical comparisons were virtually identical for "like" ratings and rated likelihood of seeing the movie.


In order to further our understanding of the process of deciding which movies to spend one’s money on, we modeled the joint effects of critic comments and "star power." Consistent with our conceptualization of movie stars as possessing brand equity, the same unknown film was more attractive when associated with a well-known star than with a fictitious name. Just as the Intel Pentium microprocessor can transfer its brand equity to an unknown brand of personal computer, the name of Clint Eastwood can enhance our desire to see the new movie he’s in.

Our results are consistent with the findings of Wallace, Seigerman, and Holbrook (1993). Specifically, using correlational data, they found that presence of well known stars explained 32% of the variance in film rentals, while controlling for various factors including the film’s rating, the cost, the length of the film, etc. The current study builds on the findings of Wallace et. al by uncovering similar effects in a controlled experiment.

Supporting our notion that identification of familiar stars provides movie-goers with heuristic devices for making the decision to see a new movie, the factors "critic comments" and "stars" combined configurally, with the identification of a popular star serving to modify the critic effect. The effects of critic comments on the desire to see a new movie and the expected enjoyment of that movie, while large for movies with unknown stars, were reduced substantially when the movie description included identification of a popular star. Conversely, when a movie receives a "thumbs-up" from the critics, it is less in need of the additional boost provided by a trusted star. These results underscore the need to study such decisions using multi-factor designs rather than assuming that factors can be studied individually as though they had independent effects. The present results support the view that mere identification of a familiar star, like the recognition of a trusted brand name, provides a heuristic for movie-goers to select a new movie without need for much additional information.

A natural question is then: Why do stars have heuristic value? To further understand the reasons behind the "star power" heuristic effect, we asked a new group of 70 students to rate their agreement-disagreement with each of a series of statements concerning the influence of stars. Table 1 summarizes the results of this survey. Oe possibility that we explored was that the movie-going public would enjoy watching their favorite stars no matter what the quality of the movie. This item was generally not endorsed. They did, however, strongly endorse the item "Top stars are usually in good movies." Eighty percent agreed or strongly agreed with this statement, suggesting that movie-goers do indeed make strong assumptions about the quality of a movie based on identification of the movie’s star. (Linton & Petrovich, 1988, came to a similar conclusion and Bello & Holbrook, in press, concluded that perceived quality is the key mediator of brand equity effects.) Our subjects also endorsed the item "Top stars have 'sex appeal’."






Experiment 2 had two major purposes: (1) to extend the findings of Experiment 1 to a different experiential productCnovels; and (2) to examine "line extension" effects with experiential products. The first purpose was achieved by comparing reactions to the same plot descriptions provided by known vs. unknown authors and by manipulating the favorability of critics’ comments. The second purpose was achieved by including a condition in which familiar authors were associated with different genres than they’re known for.


Participants were 138 students in marketing classes at the same university as in Experiment 1. Participants were assigned at random to three experimental conditions. In the Known Authors, Same Genre condition, participants evaluated four novels each said to be written by a famous authorCMichael Crichton, John Grisham, Judith Krantz, and Danielle Steel. Plot descriptions ascribed to each author were constructed to be consistent with that author’s usual genre. In the Known Authors, Different Genre condition the same authors and plot descriptions were used but the first two authors were interchanged with the second two authors. In the Unknown Authors condition the same plot descriptions were used but with fictitious author namesCtwo male names to replace Michael Crichton’s and John Grisham’s assignments in the first condition and two female names to replace Judith Krantz and Danielle Steel.

Plot descriptions were said to be supplied by publicists for the book’s publishers and consisted of brief (fictitious) summaries such as "A tribunal conducted aboard a nuclear submarine implicates three generations of a powerful military family in acts of sabotage that go back to the bombing of Pearl Harbor in 1941. The outcome could discredit the first woman candidate for President of the United States."

Critics’ comments were said to be supplied by professional book critics and, for a given novel, consisted of three positive statements (e.g., Surprise ending. Suspenseful. Interesting historical background.) or three negative statements (e.g., Inaccurate reconstruction of history. Predictable ending. Slow). Each participant in each condition received two novel descriptions with positive comments and two with negative comments. Within each condition half the participants received positive comments for a given novel and half received negative comments.

After reading the plot description, author’s name, and critics’ comments for a given novel, the participant was asked to rate (on a scale of 1 to 10) how much he or she thought they would like the novel, the likelihood (from 0 to 100%) that they would read the novel, and the likelihood that they would buy the novel.

As a manipulation check for our designation of familiar vs. unknown authors, at the end of the session participants were asked to indicate which of the authors they had heard of, which they had read before, and how much they liked that author’s work. [These manipulation checks confirmed that practically all participants had heard of our "well-known" authors, and many had read their work. Female respondents were more apt than males to have read and liked the female authors and vice versa for males. Occasionally, a participant "recognized" the name of our fictitious authors. (In a later survey, we showed that popular movie stars generally have greater familiarity than best-selling authors.)]

Results and Discussion

The three dependent variablesCexpected liking of the novel, likelihood of reading it, and likelihood of buying itCwere highly correlated with each other and showed the same statistical results. Figure 3 displays the results for ratings of how much respondents thought they would like the novel.

First compare the Known Authors-Same Genre condition with the Unknown Authors condition. The pattern here is very much like that for movies (Figure 1). Responses were significantly higher for positive comments than for negative comments, t(91)=3.56, p<.01. The difference in the magnitude of the critics’ comments effect was less in the Known Authors-Same Genre condition than in the Unknown Authors condition, but this difference was of only marginal statistical significance, t(90)=1.60, p<.06 (one-tailed test). This probably reflects the fact that the authors in Experiment 2 had not attained the same level of star power as the actors in Experiment 1.

Now compare the Known Authors-Different Genre condition to the other two conditions. The pattern of results is almost identical for Known Authors-Different Genre and Unknown Authors. These two conditions were not significantly different from each other although each was different from the Known Authors-Same Genre condition for negative comments. This demonstrates that the "star power" possessed by our well-known authors did not extend to situations in which these authors were associated with a different genre. This is consistent with research with standard products where line extensions may suffer from lack of fit if they don’t meet consumers’ expectations of the brand (Park, Milberg, & Lawson, 1991; Tauber, 1988).

In summary, Experiment 2 showed that the analog of "star as brand name" with experiential products extends from popular movie actors to popular authors, albeit to a lesser degree. Stephen King, after first publishing a novel under the pseudonym, Richard Bachman, was quoted as saying, "But the fact that Thinner did 28,000 copies when Bachman was the author and 280,000 copies when Steve King became the author, might tell you something, huh?" [From The Stephen King Companion, 1989, by George and Mary Beahm, Andrews and McMeel: Kansas City, MO.] However, Experiment 2 showed that, at least in the case of authors, "line extensions" to a new type of product meet with limited success.


Following the suggestion of Holbrook and Hirschman (1982) multivariate procedures common to the study of brand equity effects with traditional products were applied to experiential products. The same product descriptions (movies in Experiment 1 and novels in Experiment 2) were ascribed to either well known actors/authors or to fictitious actors/authors. Confirming predictions, brand equity effects were established with experiential products by showing that consumer responses were more favorable in the "star" condition than in the "no star" condition. This held for both popular movie actors and popular authors.

Also consistent with the prediction that both popular actors and popular authors possess "star power", the effects of critics’ comments were reduced when the product was ascribed to a popular actor or author. The same form of interaction was found in both experiments but was of lesser magnitude in Experiment 2. [The magnitude of the "star power" effect undoubtedly differs across types of experiential products and constituents. Compare, for example, pop music artists and opera singers and, alas, authors of popular fiction and authors of journal articles.] In particular, consumers’ reactions to negative movie critics’ comments were greatly attenuated when the product was ascribed to a popular actor.

Beyond these general findings common to both movies and novels as experiential products, each experiment provided a unique insight. Experiment 1 showed that mediating the effect of "star power" in the movies was a perception that top stars, analogous to trusted brand names, are associated with high quality products. Nevertheless, Experiment 2 suggested that a star of the literary world has limited influence outside of his or her standard dmain.

The theoretical significance of these results is that they show that theoretical concepts such as brand equity which have proven useful in the analysis of consumer preferences for standard goods are also useful for analyzing consumer preferences for experiential products. Future studies of experiential products can expand on these notions to examine brand extensions and brand combinations (e.g., the effects of having several popular stars in the same movie) with experiential products.

The primary finding in each experiment of reduced critics’ comments effects when stars are identified, has direct marketing implications. When an experiential product such as a movie, play, novel or musical composition can be ascribed to an established and trusted star, this feature should be highlighted, especially when the product involves the star’s usual genre. When the artist is unknown to the public or is working in an unfamiliar domain (e.g., Clint Eastwood starring in a comedy or musical movie, recording a song, or writing a murder mystery), then additional information such as favorable critics’ comments or expanded "previews" is needed.




Aaker, D. A., & K. L. Keller (1990). Consumer evaluations of brand extensions. Journal of Marketing, 54, 27-41.

Barwise, T. Patrick, & Andre S. C. Ehrenberg (1987). The liking and viewing of regular TV series. Journal of Consumer Research, 14, 63-70.

Belk, Russel W. (1987). Material values in the comics: A content analysis of comic books featuring themes of wealth. Journal of Consumer Research, 14, 26-42.

Bello, David C., & Holbrook, Morris B. (in press). Does an absence of brand equity generalize across product classes? Journal of Business Research.

Erber, R., & A. Tesser (1992). Task effort and the regulation of mood: The absorption hypothesis. Journal of Experimental Social Psychology, 28, 339-359.

Forgas, J. P., & S. Moylan (1987). After the movies: Transient mood and social judgments. Personality and Social Psychology Bulletin, 13, 467-477.

Hirschman, Elizabeth C. (1988). The ideology of consumption: A structural-syntactical analysis of "Dallas" and "Dynasty". Journal of Consumer Research, 15, 344-359.

Hirschman, Elizabeth C., & Morris B. Holbrook (1982). Hedonic consumption: Emerging concepts, methods, and propositions. Journal of Marketing, 46, 92-101.

Holbrook, Morris B., Robert W. Chestnut, Terence A. Oliva, & Eric A. Greenleaf (1984). Play as a consumption experience: The roles of emotions, performance, and personality in the enjoyment of games. Journal of Consumer Research, 11, 728-739.

Holbrook, Morris B., & Mark W. Grayson (1986). The semiology of cinematic consumption: Symbolic consumer behavior in Out of Africa. Journal of Consumer Research, 13, 374-381.

Holbrook, Morris B., & Elizabeth C. Hirschman (1982). The experiential aspects of consumption: Consumer fantasies, feelings, and fun. Journal of Consumer Research, 9, 132-140.

Isen, A. M., & J. M. Gorgoglione (1983). Some specific effects of four affect-induction procedures. Personality and Social Psychology Bulletin, 9, 136-143.

Jacoby, J., R. W. Chestnut, & W. A. Fisher (1978). A behavioral process approach to information acquisition in nondurable purchasing. Journal of Marketing Research, 15, 532-544.

Keller, Kevin Lane (1993). Conceptualizing, measuring, and managing customer-based brand equity. Journal of Marketing,57, 1-22.

Kindem, Gorham, "Hollywood’s Movie Star System: A Historical Overview," in The American Movie Industry: The Business of Motion Pictures, ed. Gorham Kindem, Carbondale: Southern Illinois University Press, 1982, pp. 79-94.

Linton, James M., & Joseph A. Petrovich (1988). The application of the consumer information acquisition approach to movie selection: An exploratory study. In Bruce A. Austin (Ed.), Current Research in Film: Audiences, Economics and Law, 4, 24-45.

Linz, D., I. K. Fuson, & E. L. Donnerstein (1990). Mitigating the negative effects of sexually violent mass communications through pre-exposure briefings. Communication Research, 17, 641-674.

Palmgren, P., & P. Lawrence (1989). Avoidances, gratifications, and consumption of theatrical films: The rest of the story. In B. Austin (Ed.), Current Research in Film 5. Norwood, NJ: Ablex.

Park, C. Whan, Sandra Milberg, & Robert Lawson (1991). Evaluation of brand extensions: The role of product level similarity and brand concept consistency. Journal of Consumer Research, 18, 185-193.

Payne, John W., James R. Bettman, & Eric J. Johnson (1988). Adaptive strategy selection in decision making. Journal of Experimental Psychology: Learning, Memory, and Cognition, 14, 534-552.

Reddy, Srinivas K., Vanitha Swaminathan, & Carol M. Motley (1996). Hits on Broadway: The role of critics’ reviews in the success of an experiential good. Working paper. University of Georgia, Athens.

Tauber, Edward M. (1988) Brand leverage: Strategy for growth in a cost-controlled world. Journal of Advertising Research, 28, 26-30.

Wallace, W. Timothy, Alan Seigerman, & Morris B. Holbrook (1993). The Role of Actors and Actresses in the Success of Films: How Much is a Movie Star Worth? Journal of Cultural Economics, 17, 1-28.