Market design, borders, and gravity in the virtual world

This thesis consists of three separate papers which examine different aspects of the economics of online commerce. The first paper studies a natural experiment in the release of a new ad targeting feature onto an online advertising platform. The experiment affects the specificity of advertising assets in certain geographic ad markets. The paper finds evidence that the additional specificity negatively affects auction participation in the treated areas, an effect that has not been anticipated by the incumbent theoretical literature. The paper also finds evidence that despite negatively affecting auction participation, the additional specificity leads to higher revenue growth for the online platform in the treated areas. The paperâs results highlight the importance of considering entry and exit decisions in theoretical models of specificity choices by market designers. The second paper uses a proprietary data set from Google to find that online trade between two US states or two Canadian provinces is 6.7 times higher than trade between a US state and a Canadian province. This finding is surprising given that in the online environment, information costs and business-to-business transactions involving intermediate inputs are largely absent. When disaggregating the data by sectors of economic activity, the study finds that the largest US-Canada border effects occur for services whose consumption is tied to a particular location and goods that face large regulatory hurdles at the border. The third paper analyzes geographical patterns of cross-country Internet transactions using proprietary data from Google. The paper finds the effect of distance on online trade to be around -0.53. The study also finds that cultural characteristics, such as shared languages or religions, have a large impact on e-commerce, while economic ties, such as a common currency, have an insignificant effect. The paper underlines the importance of accounting for selection into trade in worldwide gravity estimations and identifies two exclusion restrictions that can be used when examining online trade flows.