Where has Price Endogeneity in Consumer Prices gone? A Discrete Choice Analysis of the Market for Detergents with Daily Data on Purchases and Advertising

incomplete draft- comments welcome This work studies five causes for the well-known price endogene-ity bias in brand choice and demand estimation models pioneered by Berry (1994). Previous work tackled the problem by suggesting price endogeneity corrections, but never accounted for all of the possible causes for the endogeneity. I propose to treat all five causes simulta-neously. Using a novel German nationwide panel dataset that com-prises daily consumer purchase and TV viewing history, I estimate discrete choice models augmented by additional information provided from the dataset. The results suggest that the traditional endogene-ity corrections only work well in restrictive models, that cannot be detailed further due to crude data and cannot address all five causes simultaneously. However, if we have sufficiently precise data and can control for the five causes, the endogeneity corrections are rendered unnecessary. This result implies that endogeneity corrections are only necessary in the case of incomplete data, that does not allow to esti-mate models that address all five causes. Furthermore, TV advertising does not affect the consumer purchase decision in the short run, but seems to be fully dominated by long run brand fixed effects.