Research Question The prevalence of smoking and its effect on the community have given rise to decades of anti-smoking public health initiatives in the United States. One of the ways in which organizations such as the CDC and the FDA have tried to reduce smoking rates is by restricting tobacco brands' ability to reach mass audiences. Most advertising and sponsorship activities have been banned, so one of the only remaining options for reaching a wide audience is product placement on American TV shows and movies. There are varied opinions as to how much this product placement affects sales of cigarettes. Anti-smoking advocates and marketing consultants have provided back-of-theenvelope estimates which indicate that "Mad Men" caused Lucky Strike sales to rise by 43%. On the other hand, trade groups representing TV and film producers have argued that there is no causal relationship between on-screen smoking and real-world smoking behavior. In this research, our goal is to quantify how TV product placement affects sales of cigarettes. Method and Data We estimate a demand model that quantifies the effect of product placement on retail sales for 15 cigarette brands. Our sales data comes from IRI, and we observe 2,078 stores in 91 designated market areas (DMAs) that sell cigarettes. The product placement data is collected from Nielsen PlaceViews, which is a proprietary data set that gathers information on branded product placement. Our sample of the PlaceViews data starts in December 2003 and ends in July 2006, and it contains all cigarette brand product placement that occurred on network television (ABC, CBS, FOX, and NBC) during that time period. We model product placement using weighted gross rating points (wGRPs) which account for both the duration of the product placement (i.e., how many seconds the product placement was shown on screen) and how many people viewed the product placement. The unit of analysis is a brand-store-week, and we have 2,740,781 such observations in the data. The demand model is a log-log demand model with log(quantity sold) as the dependent variable. Our independent variables include own-brand product placement wGRPs, competitor-brand product placement wGRPs, price, feature, and display. We also include high-dimensional fixed effects to better isolate the effect of product placement on sales. Summary of Findings Our main finding is that TV product placement has an own-brand elasticity of about 0.02 for the cigarette brands in our data. This estimate is statistically significant and also in line with recent elasticity estimates for conventional (interstitial) TV advertising among non-tobacco CPG brands. Our second notable finding is that the competitor-brand elasticity is also about 0.02 - this implies that Marlboro product placement helps increase Marlboro sales but also helps its competitors by roughly the same amount. Both of these elasticity estimates are robust to alternative assumptions regarding the identification strategy as well as functional form assumptions regarding the week-to-week carryover of advertising goodwill. The fact that product placement spills over to direct competitors by expanding demand for the product category indicates that a blanket ban on cigarette brand names and logos would have very limited effect on cigarette purchases on its own. For instance, if all branded cigarettes displayed in TV product placement were replaced by generic unbranded cigarettes on screen, our calculations suggest that overall cigarette sales would be reduced by less than 2 percent. On the other hand, a stronger measure such as a ban on all on-screen smoking activity would be about 4 times more effective. Statement of Key Contributions We find that product placement for tobacco brands is roughly as effective as TV advertising in other categories; this demonstrates the efficacy of TV product placement as a promotional tool. For government regulators like the FTC and the FCC that have purview over TV product placement, this result also suggests that it would be helpful to regulate cigarette TV product placement just like they have regulated cigarette TV advertising in recent decades. Tobacco companies have previously argued (including in Congressional testimony) that government restrictions on their promotional activity are unnecessary, because their promotional activity is about brand-stealing rather than category-expansion. However, our results demonstrate that product placement has a category-expansion effect for the cigarette category. Therefore, this finding is relevant for anti-smoking advocates as well as regulators that have a vested interest in reducing smoking rates in the United States. We also examine what kinds of product placement regulation would be most helpful in reducing cigarette use. Currently, the government limits (but does not completely eliminate) tobacco brands' ability to pay for product placement. Our results suggests that additional regulation would be more helpful if it were focused on TV producers rather than tobacco brands. [ABSTRACT FROM AUTHOR]