Economics of Digitization
March 4, 2016
Alexander White, Tsinghua University, and Glen Weyl, Microsoft Research New England
Platforms often charge low prices until they have built up a "critical mass" to smooth user coordination. Such "penetration pricing" dampens the ability of one platform to undermine another's market position through aggressive competition, thereby increasing equilibrium profits and excessive entry. White and Weyl propose a general static model incorporating a reduced form of such realistic pricing strategies and show how they improve the tractability of platforms' optimization problems, allowing the researchers to analyze richer user heterogeneity. They then characterize the impact of competition on the social efficiency of network-effect provision in stylized but calibrated models of the U.S. video game and local newspaper industries.
Sree Ramaswamy, McKinsey Global Institute
Mitchell Hoffman, University of Toronto; Lisa B. Kahn, Yale University and NBER; and Danielle Li, Harvard University
Who should make hiring decisions? Hoffman, Kahn, and Li propose an empirical test for assessing whether firms should rely on hard metrics such as job test scores or grant managers discretion in making hiring decisions. The researchers implement their test in the context of the introduction of a valuable job test across 15 firms employing low-skill service sector workers. Their results suggest that firms can improve worker quality by limiting managerial discretion. This is because, when faced with similar applicant pools, managers who exercise more discretion (as measured by their likelihood of overruling job test recommendations) systematically end up with worse hires.
Garrett A. Johnson, Simon Business School, University of Rochester; Randall A. Lewis, Netflix, Inc.; and Elmar I. Nubbemeyer, Google, Inc.
To measure the effects of advertising, marketers must know how consumers would behave had they not seen the ads. Johnson, Lewis, and Nubbemeyer develop a methodology they call 'Ghost Ads,' which facilitates this comparison by identifying the control-group counterparts of the exposed consumers in a randomized experiment. They show that, relative to Public Service Announcement (PSA) and Intent-to-Treat A/B tests, 'Ghost Ads' can reduce the cost of experimentation, improve measurement precision, and work with modern ad platforms that optimize ad delivery in real-time. The researchers also describe a variant 'Predicted Ghost Ads' methodology that is compatible with online display advertising platforms; their implementation records more than 100 million predicted ghost ads per day. They demonstrate the methodology with an online retailer's display retargeting campaign, for which a PSA test would be severely biased. The authors show novel evidence that retargeting can work as the ads lifted website visits by 17% and purchases by 11%. Compared to Intent-to-Treat or PSA experiments, advertisers can measure ad lift just as precisely while spending at least an order of magnitude less.
Bo Cowgill, Columbia University
Jean-François Houde, University of Pennsylvania and NBER; Peter W. Newberry, Pennsylvania State University; and Katja Seim, University of Pennsylvania and NBER
Houde, Newberry, and Seim estimate the cost savings associated with the expansion of Amazon's fulfillment center network from 2006-2018. They first demonstrate that, in placing a fulfillment center in a new state, Amazon faces a trade-off between the revenue considerations from exposing local customers to sales tax and the cost savings from reducing the shipping distance to those customers. Using detailed data on online transactions, the researchers estimate a model of demand for retail goods and show that consumers' online shopping is sensitive to being charged sales tax. The authors then use the demand estimates and the spatial distribution of demand relative to Amazon's fulfillment centers to produce predicted revenues and shipping distances under the observed fulfillment center roll-out and under counterfactual roll-outs over this time period. Using a moment inequalities approach, the researchers infer the cost savings associated with being closer to customers that render the observed network roll-out optimal. They find that Amazon saves between $0.40 and $1.30 for every 100 miles of $100 of goods shipped.
Timothy F. Bresnahan, Stanford University and NBER, and Xing Li and Pai-Ling Yin, Stanford University
Success breeds success in many mass market industries, as well known products gain further consumer acceptance because of their visibility. However, new products must struggle to gain consumer's scarce attention and initiate that virtuous cycle. The newest mass market industry, mobile apps, has these features. Success among apps is highly concentrated, in part because the "top apps lists" recommend apps based on past success as measured by downloads. Consequently, in order to introduce themselves to users, new app developers attempt to gain a position on the top app lists by "buying downloads," i.e., paying a user to download the app onto her device. Bresnahan, Li, and Yin leverage a private dataset of one platform for buying downloads and identify the return from this investment. $100 invested will improve the ranking by 2.2%. To understand the investment rationale for buying downloads, the researchers build a model that accommodates (1) the impact of buying downloads on top list rank, (2) optimal investment in buying downloads, (3) an empirical distinction between app diffusion generated by bought downloads and diffusion from organic downloads, and (4) a rich set of app-specific heterogeneities. They quantify the app-specific structural coefficients by estimating the model using time-series ranking positions of 2,306 free iOS apps. They find the median value of one organic download is 70% of the cost of buying one download, implying a huge marginal cost of buying downloads. App developers lose money during the initial days after release. The coefficients are correlated with ex-post quality, measured by user ratings, but uncorrelated with ex-ante observed app characteristics, suggesting that developers face a great deal of ex-ante uncertainty about the outcome for their apps when they enter the market. The authors then employ their model to estimate the diffusion delay resulting from the visibility problem in the mobile app industry.
Rachel Soloveichik, Bureau of Economic Analysis, and Leonard Nakamura, Federal Reserve Bank of Philadelphia
In this paper, Soloveichik and Nakamura introduce an experimental GDP methodology which includes advertising-supported entertainment like Facebook in final output as part of personal consumption expenditures. They then use that experimental methodology to recalculate measured GDP back to 1998. Including 'free' apps in measured GDP has almost no impact on recent growth rates. Between 1998 and 2012, real GDP growth rises by only 0.009% per year. The researchers then recalculate total factor productivity (TFP) growth when free apps are included as both final output and business inputs. For example, Google Maps would be counted as final output when it is used by a consumer to plan vacation driving routes. On the other hand, the same website would be counted as a business input when it is used by a pizza restaurant to plan delivery routes. Measured TFP changes for both media companies and the rest of the business sector. Internet publishing companies are producers of free apps, so including free apps in the input-output accounts raises their TFP growth by 1% per year. The rest of the business sector uses free apps, so including free apps lowers their TFP growth. The net impact is an increase in business sector TFP growth of only 0.004% per year.