Economics of Digitization

A conference on "Economics of Digitization" took place in Stanford on March 1-2. Research Associate Shane Greenstein of Harvard University, the NBER Productivity, Innovation, and Entrepreneurship Program Director Josh Lerner of Harvard University, and Research Associate Scott Stern of MIT organized the meeting. These researchers' papers were presented and discussed:

Ananya Sen, MIT, and Catherine Tucker, MIT and NBER

Information Shocks and Internet Silos: Evidence from Creationist Friendly Curriculum

How the Internet affects the ability of its users to seek out information which either supports or contradicts their existing beliefs remains an open question. To examine this, Sen and Tucker analyze the effect of the Louisiana Science Education Act (2008), which allowed the teaching of creationism as an alternative 'theory' to evolution in Louisiana schools, on students' science test performance in nationally administered tests. Using detailed data on Louisiana schools, the researchers employ a difference-in-differences strategy to document that science test achievement declined after the law relative to schools in neighboring Texas. The effect of the law was driven by regions with high Internet penetration and low parental education levels. After the change in policy, Louisiana students were more likely to seek out information on the Internet using search terms which led them to web pages that reinforced a creationist message. Moreover, in line with the baseline results, the researchers find that increased search intensity comes from low-education areas in Louisiana and that it persists outside the school calendar (during test preparation months), implying a continuing effect on students.

Elizabeth Lyons, University of California at San Diego, and Laurina Zhang, Georgia Institute of Technology

Research as Leisure: Experimental Evidence on Voluntary Contributions to Science

Organizations that depend on voluntary contributions of time and money face unique managerial challenges. Lyons and Zhang investigate the impact of two distinct approaches organizations can use to attract and motivate impure altruistic volunteer labor using both a field and survey experiment on voluntary contributions to science. In both experiments, the researchers examine whether the salience of project outputs (i.e., project outcome) or project inputs (i.e., research hours) affect the quantity and quality of contributions. They find that increasing the salience of both input and output value decreases voluntary participation, but increases the match quality between the task and the volunteer. Furthermore, the researchers find that individuals that select out of volunteering in response to the type of information provided substitute volunteering time by donating money from wage work.

Daniel Bjorkegren, Brown University, and Darrell Grissen

Behavior Revealed in Mobile Phone Usage Predicts Loan Repayment

Many households in developing countries lack formal financial histories, making it difficult for banks to extend loans, and for potential borrowers to receive them. However, many of these households have mobile phones, which generate rich data about behavior. Bjorkegren and Grissen show that behavioral signatures in mobile phone data predict loan default. They evaluate their approach using call records matched to loan outcomes in two settings. In a middle income South American country, individuals in the highest quartile of risk by their measure are 7.4 times more likely to default than those in the lowest quartile. The researchers benchmark performance on their sample of individuals with (thin) financial histories. Their method performs no worse than models using credit bureau information. Bjorkegren and Grissen also test their method among microfinance borrowers in a much poorer Caribbean country. Their partner could have increased return on assets by at least 5% using the researchers' method. The method is already forming the basis for new forms of lending that reach the unbanked.

Neil Thompson, MIT, and Douglas Hanley, University of Pittsburgh

Science is Shaped by Wikipedia: Evidence From a Randomized Control Trial

"I sometimes think that general and popular treatises are almost as important for the progress of science as original work." -- Charles Darwin, 1865
As the largest encyclopedia in the world, it is not surprising that Wikipedia reflects the state of scientific knowledge. However, Wikipedia is also one of the most accessed websites in the world, including by scientists, which suggests that it also has the potential to shape science. Thompson and Hanley show that it does. Incorporating ideas into Wikipedia leads to those ideas being used more in the scientific literature. The researchers provide correlational evidence of this across thousands of Wikipedia articles and causal evidence of it through a randomized control trial where they add new scientific content to Wikipedia. The researchers find that the causal impact is strong, with Wikipedia influencing roughly one in every ~830 words in related scientific journal articles. They also find causal evidence that the scientific articles referenced in Wikipedia receive more citations, suggesting that Wikipedia complements the traditional journal system by pointing researchers to key underlying scientific articles. The researchers' findings speak not only to the influence of Wikipedia, but more broadly to the influence of repositories of scientific knowledge and the role that they play in the creation of scientific knowledge.

Christian W. Peukert, UCP Catolica-Lisbon, and Imke C. Reimers, Northeastern University

Digital Disintermediation and the Market for Ideas

Digitization has allowed inventors to circumvent traditional intermediaries and directly reach consumers, which may affect licensing outcomes and efficiency in the market for ideas. Peukert and Reimers study these impacts theoretically, and empirically in the book publishing industry, where the number of new books available to consumers has almost doubled due to the advent of digital self publishing platforms. Using data on 90,000 license deals between authors and publishers from 2002 to 2015, they identify disintermediation-related changes in this market from quasi-experimental variation across product types over time. Consistent with digital self-publishing improving the author's bargaining position, the researchers find that authors get substantially more favorable license deals. They further show that ex-ante license fees reflect ex-post demand more precisely. This is consistent with the idea that additional entry due to self-publishing creates more information about the realized appeal of specific product types. In markets in which product appeal is difficult to predict, such improvements in the information environment can have a large impact on efficiency.

Brett W. Hollenbeck, University of California at Los Angeles; Davide Proserpio, University of Southern California; and Sridhar Moorthy, University of Toronto

Advertising Strategy in the Presence of Reviews: an Empirical Analysis

Over the last fifteen years, one of the major developments online has been the growth and proliferation of review websites such as TripAdvisor. The ready availability of independent information from past users poses interesting questions for marketing strategy. What role does advertising play in the new environment? How should firms adjust their advertising strategy to the presence of reviews? Hollenbeck, Proserpio, and Moorthy address these questions in the context of the hotel industry. Using a data set of TripAdvisor hotel reviews and another describing hotels' advertising expenditures, they show, first, that overall ad spending decreased from 2002 to 2015, suggesting that online reviews have had the effect of displacing advertising. Second, there is a negative causal relationship between TripAdvisor ratings and advertising spending in the cross-section: hotels with higher ratings spend less. This suggests that user ratings and advertising are substitutes, not complements. Third, this relationship is stronger for independent hotels than for chains, and stronger in competitive markets than in noncompetitive markets. The former suggests that a strong brand name provides some immunity to reviews, and the latter suggests that when ratings are pivotal, the advertising response might be particularly strong. Finally, the researchers show that the relationship between user ratings and advertising has strengthened over time, as websites such as TripAdvisor have become more influential. This provides further confirmation that the effect of online ratings on advertising operates through the demand side, and not the supply side. Hotels seem to react to reviews if and only if consumers react to them.

Chiara Farronato, Harvard University and NBER, and Georgios Zervas, Boston University

Consumer Reviews and Regulation: Evidence from NYC Restaurants

Farronato and Zervas investigate how two signals of restaurant quality, health inspections and consumers reviews, jointly affect consumer choice and restaurants' incentives to comply with hygiene standards. They first examine whether consumer reviews can be used to detect hygiene violations that health inspectors look for. To do so, the researchers use machine learning methods to isolate information contained in consumers reviews specifically pertaining to each type of violation, which they use to predict violations. The researchers find substantial heterogeneity in prediction accuracy. Violations relating to food temperature and pests are more accurately predicted than facility maintenance violations. Next, they estimate the supply and demand effects of violation-specific information contained in consumer reviews. On the demand side, the researchers find that the disclosure of hygiene conditions in reviews affects where consumers choose to eat. On the supply side, they find that relative to restaurants not on Yelp, restaurants reviewed on Yelp score better on hygiene dimensions that are predictable from consumer reviews compared to less predictable aspects of hygiene. The researchers' results have implications for the design of regulation when consumers rate their service experiences online.

Ariel Dora Stern, Harvard University, and Cirrus Foroughi, NBER Research Assistant

Digital Innovation in a Regulated Industry: Evidence from Software-Driven Medical Devices

Does technological opportunity enable the rise of new entrants or reinforce the position of incumbents? Research on this classic topic rarely considers the unique features of regulated industries. Stern and Foroughi offer a novel approach to this question in the context of regulated medical technology, where the introduction of software is of growing importance and has created fresh opportunities for new product development. Pioneering a new application of supervised document classification, they consider over 35,000 new medical devices that came to market in the United States from 2002-2016 in order to identify predictors of digital innovation in this industry. The researchers consider the relative importance of key factors such as geographic and within-firm capabilities and the role of financial resources. They find that location in a region of concentrated expertise and prior firm commercialization experience reinforce one another in predicting digital innovation. While venture capital funding appears to play a role in supporting innovative entrants, closer analysis suggests that this funding selects on other variables that predict digital innovation. In this regulated industry, financial resources do not substitute for existing capabilities. The researchers conclude that incumbent firms have an advantage in innovating in this setting.

Megan MacGarvie, Boston University and NBER; Jeremy Watson, Boston University; and John McKeon, Edgeworth Economics

It was Fifty Years Ago Today: Recording Copyright Term and the Supply of Music

MacGarvie, Watson, and McKeon examine the effect of the expiry of recording copyright on the supply of music — in the form of re-releases, availability in streaming platforms, and concert performances — by artists popular in the UK in the 1960s. The term of recording copyright in the UK was extended from 50 to 70 years in 2013, implying that copyrights on recordings made in the late fifties and early sixties are no longer in force, while tracks recorded a few years later remain under copyright protection. In a sample of 13,238 tracks by 140 artists first released between 1928 and 1975, the researchers find that the expiry of recording copyright is associated with an approximately 141-247% increase in the number of re-releases, holding constant artist, age and year fixed effects. The effect is not significantly different for the most popular artists in their sample, and is not apparent in placebo regressions on a sample of US re-releases. Results on availability on the Spotify streaming music platform tell a different story: there is no significant difference in the availability of tracks recorded before 1963 in the UK where their recording copyrights have expired, compared to the US where their recording copyrights are still in force. However, when a track's original recording copyright expires, it becomes less likely to be performed in concert, particularly by UK-focused artists, after controlling for age, year and artist fixed effects. These results suggest that copyright term extensions may lead to fewer re-releases but more live performances of popular music first recorded approximately fifty years ago. They also point towards substantial heterogeneity in the effects of copyright on availability of cultural products across different distribution channels, and raise the question of whether the platform distribution model may moderate the negative effects of long copyright terms on availability.

NBER Videos

National Bureau of Economic Research, 1050 Massachusetts Ave., Cambridge, MA 02138; 617-868-3900; email:

Contact Us