Abstract: Emerging tracking data allow precise predictions of individuals'
reservation values. However, firms are reluctant to conspicuously implement personalized pricing because of
concerns about consumer and regulatory reprisals. This paper proposes and applies a method which disguises
personalized pricing as dynamic pricing. Specifically, a firm can sometimes tailor the ''posted'' price for
the arriving consumer but privately commits to change price infrequently. Note this personalized pricing
strategy should arise---possibly unintentionally---through algorithmic pricing when some employed variables
reflect characteristics of the arriving consumer. I examine outcomes in four contexts: one empirical and
three hypothetical distributions of consumer valuations. While one may expect this strategy to be most
profitable for low popularity items, I find, counterintuitively, that this strategy raises profits most for
medium popularity products. Moreover, typically observable measures of price discrimination suggest it is
most intense for these products. Furthermore, improvements in the precision of individual-level demand
estimates raise both the popularity-level where absolute profit gains peak and the range of popularities
this strategy can be profitably applied to. I conclude that this is an auspicious strategy for online
platforms, if not already secretly in use.
Abstract: We investigate the impact of intentionally coarsening ratings in
the context of automobile safety ratings. First, we construct a novel univariate continuous crashworthiness
rating from crash test measurements and observed fatality rates. We then estimate a random coefficient model
of vehicle demand under status quo coarse ratings and simulate outcomes under counterfactual continuous
ratings. We find that consumers alter vehicle choices, thereby reducing fatalities by 7.4% --- implying 1850
fewer US fatalities annually. Finally, we explore whether incentives to produce crashworthy vehicles are
reduced enough to offset benefits of finer information. We conclude that a continuous rating format would
reduce fatalities.
Abstract: Since its founding, Amazon has established a reputation for being
consumer friendly by consistently offering low prices. However, recent antitrust concerns about dominant
online platforms have revived questions about whether Amazon uses its market share to exploit consumers.
Using the sudden U.S. exit of Toys R Us as a natural experiment, we find that Amazon's prices increased by
almost 5% in the wake of the exit, with larger increases for popular products most likely stocked by Toys R
Us. Thus, despite Amazon's long-standing reputation for low prices, it may exploit increases in market power
as traditional retailers cease operating.
Abstract: A consumer's web-browsing history, now readily available, may be
much more useful than demographics for both targeting advertisements and personalizing prices. Using a
method that combines economic modeling and machine learning methods, I find a striking difference.
Personalizing prices based on web-browsing histories increases profits by 12.99%. Using demographics alone
to personalize prices raises profits by only 0.25%, suggesting the percent profit gain from personalized
pricing has increased 50-fold. I then investigate whether regulations intended to prevent price gouging
increase aggregate consumer surplus. Two feasible regulations considered offer at best modest improvements.
Abstract: Concerns about anti-competitive effects of proprietary data
collection have motivated recent European data portability laws. We investigate such concerns and search for
evidence of direct benefits of data collection in the context of Pay How You Drive (PHYD) auto insurance,
which offers tailored discounts to drivers monitored by telematics devices. We exploit the staggered entry
of PHYD insurance across states and insurers in a difference-in-differences framework, and we replicate the
main findings using state insurance regulations as instruments for entry timing. We find a meaningful impact
of PHYD programs on fatal accidents, but we find no evidence of antitrust concerns.
Abstract:Ad blocking software allows Internet users to obtain information
without generating ad revenue for site owners, potentially undermining investments in content. We explore
the impact of site-level ad blocker usage on website quality, as inferred from traffic. We find that each
additional percentage point of site visitors blocking ads reduces its traffic by 0.67% over 35 months.
Impacted sites provide less content over time, providing corroboration for the mechanism. Effects on revenue
are compounded; ad blocking reduces visits, and remaining visitors blocking ads do not generate revenue. We
conclude that ad blocking poses a threat to the ad-supported web.
Abstract: An existing theoretical literature finds that frictionless resale
markets cannot reduce profits of monopolist producers of perfectly durable goods. This paper starts by
presenting logical arguments suggesting this finding does not hold for goods consumers tire of with use,
implying the impact of resale is an empirical question. The empirical impact is then estimated in the market
for video games, one of many markets in which producers may soon legally prevent resale by distributing
their products digitally as downloads or streamed rentals. Estimation proceeds in two steps. First, demand
parameters are estimated using a dynamic discrete choice model in a market with allowed resale, using data
on new sales and used trade-ins. Then, using these parameter estimates, prices, profits, and consumer
welfare are simulated under counterfactual environments. When resale is allowed, firms are unable to prevent
their goods from selling for low prices in later periods. The ability to do so by restricting resale
outright yields significant profit increases. Renting, however, does not raise profits as much due to a
revenue extraction problem.
Abstract: Although bundling can substantially increase profits relative to
standalone pricing, particularly for zero‐marginal‐cost information products, it has one major problem:
bundling produces revenue that is not readily attributable to particular pieces of intellectual property,
creating a revenue division problem. We evaluate several possible solutions using unique song valuation
survey data. We find the Shapley value, a well‐motivated theoretical solution, is universally incentive
compatible (all bundle elements fare better inside the bundle than under standalone pricing), but
revenue‐sharing schemes feasible with readily available consumption data are not. Among feasible schemes,
Ginsburgh and Zang's modified Shapley value performs best.
Abstract: With digital music as its context, this paper quantifies how much
money would be made using alternatives to uniform pricing. Using survey-based data on nearly 1,000 students'
valuations of 100 popular songs in early 2008 and early 2009, we find that various alternatives can raise
both producer and consumer surplus. Digital music revenue could be raised by between a sixth and a third
relative to profit-maximizing uniform pricing. While person-specific uniform pricing can raise revenue by
over 50 per cent, none of the non-discriminatory schemes raise revenue's share of surplus above 40 per cent
of total surplus.