Many controversies that beset the digital economy turn on the role of advertising and its use of personal data. We document several new stylised facts about the global digital advertising marketplace and examine the trade-off between privacy and ad targeting accuracy from the advertisers’ perspective. We exploit a novel dataset with billions of observations of online ads spanning multiple countries, advertisers, and websites. Our focus is to estimate the impact of Apple’s gradual restriction and ultimate abolition of ad tracking in its Safari browser called Intelligent Tracking Prevention (ITP). We analyse how much advertisers are willing to pay for third-party cookies and how tightening privacy policies affects market outcomes. Our empirical strategy treats Apple’s policy changes as exogenous shocks to the supply of tracking opportunities and uses a series of event study models to estimate their causal impact. We find that the estimated treatment effects around the ITP introduction dates are small in magnitude on average but differ markedly across countries, advertising campaigns, and type of marketplace. These finding are consistent with a theoretical literature showing that changes in ad targeting have ambiguous general equilibrium effects. Moreover, our results suggest that markets failed to adjust immediately to new, more privacy-sensitive equilibria.
cemmap
WP 04/23,
SNF
WP A02/23
This paper measures the value of information mortgage brokers provide to UK households using an application of a structural model of search. Aided by administrative loans data, we document the existence of a substantial degree of unexplained price dispersion, and observe that while mortgages obtained from brokers are cheaper, borrowers who use intermediaries pay more once commissions are factored in. However, our results also show that broker presence exerts negative pressure on lenders’ market power. Compared to a world where broker advice is unavailable, brokers reduce average monthly mortgage costs by 21% and eliminate a part of welfare losses arising from costly search. In other words, eliminating brokers from the market would lead to even higher borrowing costs from the group of borrowers who need it most: those with the highest search costs. We also find that regulation in support of market centralization halves lenders’ markups and lowers monthly costs of an average mortgage by 4.4%.
We propose a dynamic oligopoly pricing model, in which consumers’ choices exhibit inertia and firms face costly price adjustments. The primitives of the model are estimated using scanner data from the UK butter and margarine industry. We evaluate the effects of frictions on price dynamics, profits and consumer welfare. We find that price adjustment costs are substantial and represent between 24-34% of net margins. Our model predicts that absence of these costs reduces persistence in prices, increases firms’ profits but has little effect on consumer surplus. The effects of consumer inertia on prices are much more pronounced than when firms cannot adjust prices freely.
We propose a model of nonsequential consumer search where consumers and firms differ in search and production costs respectively. We characterize the equilibrium of the game. We first show the distribution of search cost can be identified by market shares and prices. Subsequently, we identify the production cost distribution using a similar strategy to Guerre, Perrigne and Vuong (2000) as the firms’ decision problems resemble bidders’ problems in a particular procurement auction. We prove the firms’ cost density can be estimated at the same convergence rate as the optimal rate in Guerre et al. uniformly over any fixed subset on the interior of the support. The uniform convergence rate over any expanding support is slower due to a pole in the price pdf that is a feature of the equilibrium. Our simulation study confirms the theoretical features of the model. Our identification and convergence rate results also apply to two generalizations of the baseline search model that allow for: (i) vertically differentiated products; (ii) an intermediary. We apply the latter model to study loan search using UK mortgage data.
We use rich data on prices and characteristics of billions of digital ads over several years to document the rise of private marketplaces (PMPs)—invitation-only ad auctions that have overtaken open auctions (RTB) as the dominant method for trading display ads. We then develop a structural model in which publishers decide whether to use PMPs and how to allocate ad slots between the two types of exchanges.
We study the use of tax instruments to correct for vehicle pollution from both exhaust and non-exhaust sources, where the latter can be substantial and relevant for electric vehicles. We exploit detailed Norwegian register data to estimate a model where households decide on both driving and vehicle purchases and use the estimates to compare a range of counterfactual tax regimes. Absent strategic price setting, we find that a Pigouvian outcome can be approximated reasonably well with registration taxes/subsidies based on vehicle characteristics combined with fuel taxation. When we endow car manufacturers with market power, the oligopoly markups are shown to exceed the social marginal cost, and a corrective driving tax only exacerbates the distortions without further measures to address the reductions in new car sales due to market power.
… and several more projects with tentative titles
The National Infrastructure Commission (NIC) commissioned a team of academics and researchers at the IFS and UCL to create a software tool that estimates how land values respond to changes in land purpose or infrastructure improvements.
| © 2018-2023 Mateusz Mysliwski. Template by Bootswatch. |