Experimental Evidence on Structural State Dependence in Demand

Research in marketing and economics has documented structural state dependence in demand, that is, a causal effect of current on future brand purchases. We re-evaluate the credibility of this literature using novel data from a large-scale natural field experiment. Randomized coupon exposure serves as an encouragement to manipulate brand choices, identifying the local average treatment effect (LATE) of current on future purchases. Our estimates are nonparametric and free from selection and confounding biases. Analyzing 223 brands, we find positive structural state dependence for most brands using a principled evidence aggregation approach. On average, the purchase of a brand increases the repurchase of the same brand by 10 percentage points (conditional on a category purchase), equivalent to a 74% lift in sales. Moreover, structural brand choice models produce LATEs consistent with those from the encouragement design. These findings lend credibility to the structural choice model approach in our setting. This credibility is practically important because structural models can be estimated in settings without explicitly randomized data, yield more precise estimates, and are designed for policy evaluation. More generally, our results contribute to the credibility debate around structural models in industrial organization.

History: Catherine Tucker served as the senior editor.

Funding: S. Gabel acknowledges funding from the Dutch Research Council (NWO) under a Veni grant (VI.Veni.221E.073). D. Guhl gratefully acknowledges financial support from the German Research Foundation (DFG) through CRC TRR 190 “Rationality and Competition” (project number 280092119). None of the authors has any financial or non-financial interests that are directly or indirectly related to the work discussed in this manuscript.

Supplemental Material: The online appendices and data files are available at https://doi.org/10.1287/mksc.2024.1274.