Working Papers
On Anticompetitive Third-Degree Price Discrimination
Robert Bork argued against the enforcement of the Robinson-Patman Act based on the economic theory of the 1970s where output and welfare effects of third-degree price discrimination are driven by local demand curvature conditions. Although known for nearly a century, these curvature conditions have never been evaluated empirically. I first use demand manifold invariance results to show that most econometric specifications predetermine demand curvature behavior and therefore, the predicted output and welfare effects of price discrimination relative to uniform pricing. Second, I use supermarket scanner data to evaluate demand curvature conditions nonparametrically for thousands of chain-store-product combinations and show that, more often than not, third-degree price discrimination (local store pricing) decreases output and welfare relative to uniform pricing (chain-store pricing). Furthermore, I show that using output as a proxy for welfare as Bork suggested overstates potential gains and understates potential damages of price discrimination. Zone pricing might thus serve as an effective tool for firms with market power to profit by restricting total sales. This anticompetitive effect of price discrimination could be reversed by the existence of economies of scale or the opening of new markets.
(Available in
December 2024) Additional material:
Online Appendix
Trade Protection, Industrial Policy, and the Shaping of Local Preferences
Industrial policy might enhance welfare if trade does not happen among competitive economies and/or if there are potential dynamic economies of scale or consumer learning to be accounted for. We conduct a retrospective analysis of the Spanish automobile industrial policy that overcomes many of the common empirical challenges: inconsistent enforcement of long-lived policies affecting not only domestic production but also welfare. The policy had four pillars: the state-sponsored creation of SEAT, the domestic manufacturer; a restrictive maximum share of foreign capital ownership; a stringent domestic minimum component requirement for foreign manufacturers to locate in Spain; and a de facto prohibition of imports until Spain joined the European Union in 1986. Spanish automobile production grew from non-existing in the 1950s to become the fifth largest in the world in the mid 1990s. We use historical province-level automobile registration data to highlight inefficiencies that are otherwise ignored when focusing on output growth alone. If given a chance, many Spaniards preferred cheaper, higher quality imports. Furthermore, entry of foreign manufacturers triggers a sudden but long-lasting increase in domestic sales of these brands, which is always more intense in the local market where the new assembly plant is located. Thirty years after import restrictions were abandoned, the strong preference for brands with a domestic production facility survives, resulting in a serious regional segmentation of the Spanish automobile market that softens competition despite being fully integrated in Europe. We show that intranational home bias accounts for the equivalent of a 5.5% import tariff, compared to a nominal import tariff into the European Union of 10.3%. Seventy years after its creation, SEAT is still the largest beneficiary of local biased preferences.
(Available upon request)
Elasticity and Curvature of Discrete Choice Demand Models
We study the determinants of cost pass-through in differentiated product markets. Random utility models of demand, such as mixed logit, are attractive because they place limited restrictions on customer substitution patterns. We show the shape of the distributions of customer preferences determines cost pass-through. Common functional form assumptions for these distributions lead to biased estimates of both pass-through and substitution. We offer a flexible and parsimonious unit-demand specification that accommodates both log-concave demands (incomplete pass-through) and log-convex demands (over-shifted pass-through) up to CES demand. Instruments and estimation are straightforward, and Monte Carlo analysis validates our ability to recover the underlying demand curvature. Using automobile data, we find the bias from ex-ante shape restrictions is large. Flexibly estimating cost pass-through, therefore, has important implications for designing and evaluating trade and subsidy policies. Earlier version:
CEPR DP 18310
Amazon HQ2: A Tale of Shocks to Housing Price Expectations
The measurement of the contribution of expectations to house prices is unresolved in the
macro-housing literature. We leverage a novel quasi-natural experiment using Amazon's unan-
ticipated split location decision for its second headquarters to identify the impact of this
expectations shock on local house prices, seller expectations and market liquidity. We find that
listed and transacted prices increased on average 7.9% and 7.5%, respectively in the six months
following the announcement. Furthermore, price gains were common across all market segments
and the announcement had no effect on rents. We develop a tractable general equilibrium
macro-housing model featuring mortgages and endogenous housing supply able to replicate the
response of the price-rent ratio to an expectations shock. The model quantifies the differences
between credit and expectations shocks and generates testable predictions for identifying the
nature of a housing price shock. Our empirical and theoretical results provide a benchmark test
for structural models that attempt to incorporate shocks to price expectations. Earlier version:
CEPR DP 18314