Demand estimation is at the core of economic analysis in many different areas. Examples include cost benefit analysis, antitrust evaluation, and computation of price indices. Considerable progress has been made developing new tool for demand estimation, but most of these tools and their applications have been in static analysis. Demand for most goods, either due to product storability or durability, is dynamic. Namely, intertemporal considerations play an important role, possibly compromising static estimates.
Indeed, intertemporal considerations are present in many decisions like schooling, migration, as well as purchasing. In durable good markets falling prices and improved quality provide incentives to delay purchases. In storable goods markets, temporary price reductions create incentives to anticipate future needs. While these dynamic effects are known, and in some cases the bias from ignoring them is well understood, the use of the dynamic models has been limited. Most dynamic methods (specially, those that can be applied to inventory problems) are typically data demanding (require household level data) and computationally intensive. As a result, in academic work as well as in policy applications, dynamics are often neglected. However, neglecting dynamics even when they are not the essence of the question can contaminate the estimates.
This award funds the development of new tools to study demand dynamics, as well as optimal seller behavior given such demand. The PIs have a tractable model of consumer demand that can be estimated using market level data. The proposed framework captures dynamic consumer behavior but, unlike the typical inventory model, its predictions are very simple to characterize (and thus, to estimate). The estimation method offered can be done using standard software packages, without the need for time consuming parameter searches.
The results of this study have an impact in the field of industrial organization, where demand estimation plays a key role both in academic and policy work. The results also have implications for other fields including macro and trade where recent work has focused on understanding micro-level prices and how they respond to shocks such as exchange rate fluctuations and monetary policy. Both trade and macro work has paid significant attention to pricing in scanner data. One of the key issues faced by all these studies, when using disaggregated high frequency data, is how to deal with temporary price reductions.
To show the performance of the method, the model is applied to the soft-drinks industry and used to measure the long-run preference parameters. The proposed demand framework leads to a simple solution to the sellers' pricing problem. Preliminary estimates suggest that the typical pricing patterns of storables, with constant prices and sporadic price reductions, are dictated by intertemproal price discrimination. Intertemporal pricing captures up to 40% of the potential gains from third degree price discrimination between more and less price sensitive buyers. The estimates suggest that both price discrimination and intertemporal price discrimination are welfare enhancing, in a total surplus sense. Less price sensitive consumers are worse off, but more price sensitive ones (shoppers) and sellers gain more than the loss by non-price sensitive consumers.