Increasingly, firms have the ability to make high-quality, microlevel predictions of demand for their products, which improves their ability to target advertising. In spite of this, firms may choose to target advertising at a higher level of aggregation than their predictions allow to benefit from the significant discounts that often accompany mass advertising purchases. We argue that firms making such a choice generate “advertising spillovers” that are quasi-random and can be used to identify the response to advertising. These advertising spillovers occur when local levels of advertising are higher or lower than locally optimal because of the influence of other markets or individuals on the mass advertising decision. We formalize the supply-side conditions that incentivize firms to generate these spillovers as part of their optimization strategy, present an empirical strategy for exploiting these conditions, and apply the strategy to multiple product categories and brands. Estimates from this “spillover strategy” agree with recent literature that suggests many standard approaches to estimating the response to advertising may produce biased results because of unobservables; our estimates also suggest that some recent empirical strategies, such as the DMA-border strategy, can produce biased estimates for seasonal products.

中文翻译：

---

大规模广告溢出：一种识别策略

公司越来越有能力对产品的需求进行高质量的微观预测，从而提高了针对广告的能力。尽管如此，企业可能会选择将广告定位在更高的汇总水平上，而其预测却无法从通常伴随着大量广告购买而带来的巨大折扣中受益。我们认为，做出这种选择的公司会产生“广告溢出”现象，这是准随机的，可以用来识别对广告的反应。当由于其他市场或个人对大规模广告决策的影响而导致本地广告水平高于或低于本地最佳水平时，就会发生这些广告溢出。我们将供应方条件形式化，以激励公司产生溢出效应，作为其优化策略的一部分，提出了利用这些条件的经验策略，并将该策略应用于多个产品类别和品牌。这种“溢出策略”的估算结果与最近的文献相吻合，后者表明，由于无法观察，许多标准的方法来估计对广告的响应可能会产生偏差的结果。我们的估算还表明，一些最新的经验策略（例如DMA边界策略）可能会产生季节性产品估算偏差。