Research

An Empirical Framework for Internet Policy: Quantifying the Effect of Net Neutrality Regulation

Slides

I estimate the welfare effect of imposing "net neutrality" regulation on the US market for fixed broadband Internet service. Net neutrality is a ban on price discrimination by network intermediaries, specifically packet-based pricing by ISPs or backbone providers that treats data differently according to its characteristics like originating content provider or time sensitivity. From an ex-ante theoretical perspective, the welfare effects of net neutrality are ambiguous, thus the actual welfare effect is an empirical question which depends on the ownership structure of the physical Internet, how firms on the shortest path split the surplus, and model parameters. I find that going from a world with perfect price discrimination by ISPs and backbone providers to a world under net neutrality rules where providers are constrained to charge link-specific prices which are otherwise non-contingent generates an overall welfare gain for society of $US 2.74 billion per year, or 7.62% of the market for fixed broadband.

I also consider the distribution of winners and losers under net neutrality. As groups, consumers, ISPs, and backbone providers all gain under net neutrality - profits are higher and consumers pay less individually. However, there still exist firms and consumers at some locations that experience welfare losses under a policy of net neutrality. Although net neutrality is beneficial for society as a whole, it does create a geographic distribution of welfare losses based on position relative to the location of key pieces of the physical Internet.



The Effect of Permanent Infrastructure Loss: Evidence from the UK Beeching Report Rail Cuts with Stephen Gibbons (LSE) and Stephan Heblich (University of Bristol)

Slides

We estimate the effect of railroad service cuts by British Rail in the period 1965 to 1980 on outcomes in UK census data, the so-called "Beeching cuts" that proposed closing 9700 km of lines and 2400 stations. We use averted rail cuts, cuts which were proposed in the original Beeching Report but not actually executed in practice by 1980, as an alternate control group. We find that the Beeching cuts had a persistent negative effect on localities which were impacted by the cuts in terms of increased travel time to destinations across the UK. Thus the Beeching cuts were responsible for a least some of the currently observed inequality between regions within the UK, and partially explain the divergence between large metropolitan areas and the rural periphery.



The Effect of Infrastructure Investment in a Low-Growth Environment: Evidence from the Great Depression with Miguel Morin (University of Cambridge) and Nicholas Ziebarth (University of Iowa)

Slides

We develop a new dataset on the US road network in the 1930s which includes smaller local roads that were not accounted for previously. We examine the effect of road construction and improved market access on US counties in the 1930-1940 period. As expected, there is no effect on industries that sell goods that are hard to trade, but for tradable industries we find an increase in the number of firms, employment, and value added while labor productivity falls. This effect is due to gains in market access to destinations across the US, not local road construction within a county.



Transport Networks and Internal Trade Costs: Quantifying the Gains from Repealing the Jones Act with Woan Foong Wong (University of Wisconsin-Madison)

Slides

We estimate the cost savings of repealing the Jones Act in terms of the resultant reduction in total shipping costs faced by US firms. The Jones Act, which is currently in effect, bans entry by foreign shipping firms for port-to-port and inland waterways shipments contained entirely within the territory of the US. Repeal of the Act results in an annual cost savings of $US 1.91 billion for US firms since both foreign ships and crew are substantially less expensive relative to their current domestic counterparts. We characterize the distribution of welfare gains across regions within the US, with California, Illinois, and Texas the gaining the most.



Reassessing Railroads and Growth: Accounting for Transport Network Endogeneity

Paper
Slides

Motivated by the seminal work of Robert Fogel on U.S. railroads, I reformulate Fogel's original counterfactual history question on 19th century U.S. economic growth without railroads by treating the transport network as an endogenous equilibrium object. I quantify the effect of the railroad on U.S. growth from its introduction in 1830 to 1861. Specifically, I estimate the output loss in a counterfactual world without the technology to build railroads, but retaining the ability to construct the next-best alternative of canals. My main contribution is to endogenize the counterfactual canal network through a decentralized network formation game played by profit-maximizing transport firms. I perform a similar exercise in a world without canals. My counterfactual differs from Fogel's in three main ways: I develop a structural model of transport link costs that takes heterogeneity in geography into account to determine the cost of unobserved links, the output distribution is determined in the model as a function of transport costs, and the transport network is endogenized as a stable result of a particular network formation game.

I find that railroads and canals are strategic complements, not strategic substitutes. Therefore, the welfare loss can be quite acute when one or the other is missing from the economy. Such a stark welfare loss is due to two main mechanisms: inefficiency of the decentralized equilibrium due to network externalities and complementarities due to spatial heterogeneity in costs across the two transport modes.



The Historical Evolution of Financial Exchanges with Bjorn N. Jorgensen (LSE) and Kenneth A. Kavajecz (Syracuse University)

Paper
Data Appendix

The historical dynamics of entry and exit in the financial exchange industry are analyzed for a panel of 741 exchanges in 52 countries from 1855 through 2012. We focus on economic, technological, and regulatory factors. Using novel panel data evidence, we empirically test whether these factors are consistent with existing financial theories. We find that US exchanges are 4.6% more likely to exit per year after the passage of the Securities Exchange Act. The telephone, literacy, and regulation are robust predictors of financial exchange dynamics. The upward trend in literacy is an important driver of exchange entry.



Information Aggregation and Product Reviews in the Entertainment Industry

Executive Summary
Short Paper
Paper

Most review aggregation websites use a transparent, observable method to derive their aggregate score from individual reviewer scores. However, Metacritic does not publicly disclose their weighted average methodology, specifically how they construct their own derived aggregate score from reviews using critic weights. I propose a simple economic theory to explain Metacritic's choice of weights: optimal weights are assigned proportional to a particular measure of review quality. Proceeding under the assumption that Metacritic uses a weighted average scheme with weights assigned to each publication, I estimate critic weights using a dataset of aggregate and individual scores. A data transformation method is used for estimation. I find that the most prolific publication in a genre is not necessarily assigned the most weight. Metacritic's assigned weight depends very weakly on critic age, number of reviews, and review length. The weighted average scheme in use at Metacritic weakly gives lower weight to less accurate reviewers in terms of the bias and noise embedded in their reviews. However, according to the economic theory they do not sufficiently penalize critics with reviews of low quality. Metacritic can do better by making their weighting system more responsive to review quality.