Author Archive

Research Roundup: File-sharing vs. music sales, Craigslist vs. newspapers, and more

Wednesday, December 8th, 2010

In this edition of the Research Roundup we highlight a paper by Julie Holland Mortimer and Chris Nosko of Harvard and Alan Sorensen of Stanford GSB that analyzes music file-sharing as it relates to a well-known off-line complement: concert performances. The authors observe that

“Redistribution of the digital good [music downloads] may increase demand for the complementary good [concert tickets], partially offsetting the losses due to illegal redistribution of the digital good. The implication… is that public policy aimed at promoting innovation should not ignore the impact of an innovation on goods or assets that are complementary to it.” (p 2)

The team undertakes to investigate such a relationship in the recent collapse of recorded music sales and concurrent boom in the demand for live music, depicted in their Figure 1, which is reproduced below.

Album Sales and Concerts, 1995-2004

Their study concludes that

“sales of recorded music declined precipitously with the entry of Napster and large-scale file-sharing. While file-sharing may have substantially displaced album sales, it also facilitated a broader distribution of music, which appears to have expanded awareness of smaller artists and increased demand for their live concert performances. Concert revenues for large artists, however, appear to have been largely unaffected by file-sharing.” (p 19)

These conclusions may be somewhat shaky. For example, the analysis employs little data on actual file-sharing activity (except in one ancillary point of analysis); Mortimer, Nosko, and Sorensen instead estimate the effect of file-sharing through regression with a “Napster” indicator marking whether observations are from “years following the entry of Napster (i.e., 2000 to 2002)” (p 15). This analysis is odd in that the authors themselves warn “time trends alone cannot establish any causal link” (p 10) , and since they do not identify and account for fixed effects (other than the intercept). Their attempt to estimate results with an eye for variation in broadband penetration is promising; unfortunately the data are too coarse for the tests to yield meaningful results.

In spite of any shortcomings, the work is a commendable attempt at understanding the early period of a trend that is often discussed but rarely studied. Still, for right policymaking we should require clearer conclusions.

(Click through to the full post to see the abstract and link for this paper and others in the Roundup.)


25,000 Reasons Why Broadband Price Data Matter

Tuesday, November 23rd, 2010

On Friday, TPI released Part 1 of an ongoing study of broadband prices in OECD countries.  This part of our study uses extensive data and econometric tools to test the effects of different pricing tactics.  Perhaps even more important than our results are our methods, which mark a significant step toward more mature understandings of “what we pay” for broadband in the United States and elsewhere.

Our study acknowledges that “broadband” refers to service that varies widely across multiple dimensions:  Internet access is offered at different download and upload speeds, may be available with contracts of varying length, and may be bundled with complementary services such as voice, television, or both (the well-known “triple play”).  In addition, “price” is more complicated than a simple monthly payment for such a service: many ISPs offer temporary promotional prices or other discounts, and impose other necessary charges such as start-up and equipment fees.  All of these must be considered in order to reach a complete understanding of costs to consumers.

The FCC’s National Broadband Plan, in light of these complications, decries “a dearth of consistent, comprehensive, and detailed price data.”  With this project we work towards filling that void.  We assemble a dataset of about 25,000 residential and business broadband offerings from providers within the OECD from 2007-2009, keeping in mind the crucial nuances noted above.

With this improved data in hand, in Part 1 we econometrically examine several classes of broadband to decompose observed prices into their “determinants,” namely quality measures (e.g. download speed and limits on data), contract commitments, and fixed effects based on when and where the plan was offered.

Our findings (available in full here) are in some ways to be expected: bit caps—also referred to as data caps or metered billing—are associated with lower prices for all classes of users (on average), and our model finds that prices increase as bit cap size (i.e. the number of gigabytes allowed in a month) increases.  Contract commitments are also associated with lower prices, though generally without statistical certainty.  We do find an unexpected result regarding business consumers, who appear to pay more for longer contracts, other things being equal.  In addition, we see that the price of triple play plans is significantly influenced by the number of television channels included.

This analysis is relevant to current debates, in which Internet Service Providers who have attempted to introduce bitcaps or metered billing have been met with fierce resistance.  Our research shows that low- and mid-use consumers (and, in fact, the median U.S. user according to the FCC), could see savings from limited data use, at least if such plans were instituted in line with the average experience in other developed markets.

Research Roundup: Entrepreneurship, Clusters, Competition, and more

Monday, October 25th, 2010

If you only have time for one article from today’s Research Roundup—and especially if you don’t have the patience for something technical—make it the framework for thinking about entrepreneurship laid out by Henrekson and Sanandaji in the introductory chapter to Institutional Entrepreneurship (2011).  Here the editors set up a taxonomy that hinges on whether entrepreneurial activities actually create wealth (are “productive”), as well as whether they adhere to, evade, or alter prevailing institutions.  The definitions are rich enough to classify Washington insiders (including some who might frequent this blog) as “entrepreneurs” whose innovations include not business models but “altering” political contributions (read: lobbying) for destructive rent-seeking or broad social benefit.  Hanrekson and Sanandaji illustrate their framework with a diverse group of articles discussing archetypical “productive abiding entrepreneurs” (think Silicon Valley), the oligarchs of Russia, and even the Sicilian mafia.  We look forward to the release of their volume.

Another paper on entrepreneurship, from Temple’s Delgado, Harvard’s Porter, and Northwestern’s Stern, dives into the impact of geographic/industry clusters on the level of new business creation and start-up employment.  And a wealth of papers—ranging from the highly theoretical to a concrete discussion of the Oracle/Sun decision in the European Union—tackle competition.  Click through to the full post to see the set of abstracts on these and other topics.


Research Roundup: Small Businesses and Net Employment Growth, and more

Friday, September 17th, 2010

In Washington, pretty much everyone agrees that private sector job growth comes mostly from small businesses.  Every president since Ronald Reagan has echoed this proposition in a major speech, supported by literature showing an inverse relationship between job growth and firm size dating back to 1979.  However, according to Haltiwanger, Jarmin, and Miranda, this popular maxim of economic policy just isn’t true.  The authors employ a novel dataset from the Census Bureau to find that when controlling for firm age there is “no systematic relationship between net growth rates and firm size.”  For the visual learners among us, the adjustment looks like this:

Haltiwanger, Jarmin, & Miranda (2010) p 39

Source: Haltiwanger, Jarmin, & Miranda (2010) p. 39

The pink and green lines plot results that might be expected from popular wisdom, while the red and blue lines are gleaned from regressions incorporating a variable for firm age.  The results differ based on whether firms are classed by their size in a base year or by the current-year average (we’ll spare you the full methodology but note that the authors prefer the latter). Under either classification the famed inverse relationship is upended upon factoring in age.

As TPI’s Scott Wallsten points out in a recent commentary, measurement difficulties confound economic analysis of high tech markets.  Haltiwanger, Jarmin, and Miranda show that even metrics as banal as firm age can cast orthodox understandings across all business sectors into doubt when better data become available.

(Click through to the full post to see the abstract and link for this paper and over 30 others on green tech, the developing world, cyber security, and more.) (more…)

Research Roundup #5: Virtual Economics, E-Commerce Regulation, and more

Wednesday, August 18th, 2010

Today’s Research Roundup includes thirty studies on economics, law, and technology. Of these we highlight two papers that undertake complementary analyses of some implications of moving our lives increasingly online.

Vili Lehdonvirta of the Helsinki Institute for Information Technology offers an impressively lengthy look into virtual consumption, that is, the purchase of a unique class of not-quite-goods, not-quite-services that exist only within the confines of an online environment (items for sale in a computer game are the classic example).  With a decidedly sociological approach, Lehdonvirta seeks to answer “why do people spend real money on virtual goods?”

Miriam Cherry of the University of the Pacific looks at the virtual individual as producer by cataloging the phenomenon of “cyberwork.”  As opposed to Lehdonvirta’s more abstract academic exercise, Cherry considers for-pay online activity with an eye for how it relates to existing protections and legal frameworks in the offline labor market, including anti-discrimination laws, wage standards, and even unionization (for example, she believes virtual work will “encourage unions, out of necessity, to adopt a more global perspective”).

Neither study has many immediate policy implications.  Nevertheless, their consideration of such novel—and perhaps someday fundamental—elements of the economy remind us there is a lot left to learn about the digital age.

(Click through to the full post to see the list of papers and abstract excerpts)


Satellite Broadband: Line-of-Sight, Not Out of Mind

Wednesday, August 11th, 2010

The National Broadband Plan (NBP) estimates that firms would need subsidies totaling $23.5 billion to invest in the infrastructure necessary for universal broadband coverage in the United States (Exhibit 1-A, click to enlarge).[1]

Base-case broadband availability gap

The problem with the Plan’s estimate is that it includes only DSL and 4G wireless and omits broadband-over-satellite, which is by far the cheapest option for serving the most costly areas.  Thus this “base case” grossly overstates the necessary costs of achieving 100% broadband availability.

The Broadband Plan notes that “while satellite is capable of delivering speeds that meet the National Broadband Availability Target, satellite capacity can meet only a small portion of broadband demand in unserved areas for the foreseeable future….[w]hile satellite can serve any given household, satellite capacity does not appear sufficient to serve every unserved household.” (p 137)

But satellite need not serve all “unserved” households.  Serving only the highest-cost households would yield enormous savings.

The 250,000 housing units (0.2% of the U.S. total) with the highest costs account for $13.4 billion of the claimed investment gap (OBI Technical Paper No. 1, p 41).  This eye-popping estimate reflects hypothetical decisions such as one to build out DSL to a single house in Orange County, NY for $366,126, which exceeds the county’s median home value, and to 30 dwellings in Kauai County, Hawaii at an average cost of $205,890 each, or about half of that county’s median.[2] Exhibit 3-H graphs the steep “hockey stick” costs implied by the base-case model.

In its technical supplement explaining the investment gap, the Broadband Team estimates that using satellite (with minor federal support) to serve those 250,000 homes would reduce the gap by at least $11.4 billion, or almost 50%.[3]

The authors have clearly considered the tremendous efficiencies afforded by satellite access, and acknowledge the adequacy of broadband-over-satellite at meeting the NBP requirement for connection quality.[4] Recommendation 8.13 urges the FCC to consider “alternative approaches, such as satellite broadband, for addressing the most costly areas of the country” (p 150).  As such, the “broadband availability gap” as calculated should not be considered a strict endorsement of the technologies assumed (DSL and 4G), but rather a starting point for comparing the costs and benefits of alternative proposals.

To be sure, broadband-over-satellite has some drawbacks compared to other technologies.  Existing satellite broadband plans offer slower download and upload speeds than most wireline or other wireless technologies, are more expensive, and exhibit higher latency due to extreme length of the “last mile” (more than 20,000 miles) to orbiting geostationary satellites.  Speeds will become less of an issue with two new satellites expected to go into service in the next two years, both offering up to 10 Mbps downstream to homes; Hughes says it will even sell business plans of up to 25 Mbps.

The question then becomes whether it is worth spending an additional $12 billion to give those households a DSL or 4G wireless broadband option.  To put that in perspective, consider that the U.S. government (NIH) budgets $50 million for discovery and development of drugs for “rare diseases”—defined as those affecting 200,000 or fewer people.[5] Many of those illnesses are deadly.  Does it make sense to spend billions to allow 250,000 households the option of reducing delays in their Internet transmissions by half a second?

Apparently the Omnibus Broadband team didn’t think so.  And thanks to the recent stimulus package, the USDA (the federal government’s longstanding supporter of rural broadband) is increasingly on board.  It’s time we unite to make satellite broadband a priority in proposals for access in America’s most remote communities.

[1] The chart is actually taken from the corresponding technical supplement.  In the Broadband Plan itself (See Section 8.1) the gap is referred to as the “broadband availability gap” and was pegged at $24.3 billion before the estimate was revised.

[2] Estimated costs of buildout reflect net cost (initial capex and ongoing support less revenue) with a 20-year time horizon and 11.25% discount rate (the NBP standard). Data on gap by county available at

[3] The team reports the gap to be $10.1 billion—that is, reduced by a full $13.4 billion—when factoring in satellite “even with a potential buy-down” (p 41).  It appears they have not factored in their estimate of a $800 million-$2 billion buy down, a program in which the government would subsidize subscriptions to existing (planned) satellite capacity to bring the expected high subscription charges to a level approaching terrestrial service (p 93-94).  If necessitated, this cost would rightfully be subtracted from the savings, yielding the possible low of $11.4B reported above.

[4] That is, they acknowledge that satellite broadband will be sufficient for the “actual” 4 Mbps download, 1 Mbps upload minimum (NBP p 137).

[5] NIH requested this amount for the Therapeutic Rare and Neglected Diseases Initiative (see FY 2011 budget, p 5).  The amount overstates the magnitude of spending per patient because the program also covers neglected diseases, from which very few Americans suffer, and because it includes more than 6,800 diseases classified as rare, which together afflict an estimated 25-30 million Americans.

Research Roundup #4

Friday, July 16th, 2010

We’re well into another sweltering summer here in the District, with policy debates that seem to raise the mercury even higher.  Take a few minutes to cool down with a crisp, refreshing study from the authors below.  Don’t miss the Vigdor and Ladd piece, which provocatively argues that “students who gain access to a home computer between 5th and 8th grade tend to witness a persistent decline in reading and math test scores.”  Readers with a mind for engineering over economics may prefer Bauer, Clark, and Lehr’s thorough analysis of “speed” definition and measurement in “high-speed Internet access.”  And of course there are a few articles for the lawyer in all of us, including two on intellectual property rights.

(Click through to the full post to see the list of papers and abstract excerpts)


Research Roundup #3

Monday, June 21st, 2010

Today’s Roundup features a variety of papers from authors at organizations ranging from law schools and university economics departments to the research arm of a major investment bank.  Note the paper by Grimes and Ren, which offers a rare empirical analysis of high speed Internet access and firm productivity.  In addition, two articles discuss web search and social network data as tools for economic study (under the heading below of “Tech and Macroeconomics”).

(Click through to the full post to see the list of papers and abstract excerpts)


Research Roundup #2

Wednesday, May 26th, 2010

It’s time for another edition of the Research Roundup, with a fresh set of papers by authors outside of TPI.  In recent weeks we’ve seen a wealth of articles relating to patent and copyright, so today’s group is a bit heavy on intellectual property.  Additionally, a few interesting papers deal with online business models.  Take a look.

(Click through to the full post to see the list of papers and abstract excerpts)


Research Roundup #1

Thursday, May 13th, 2010

Today we kick off the Research Roundup, a new feature on the TPI blog. This semi-regular series highlights recently published articles and working papers related to our research interests by authors outside of the Technology Policy Institute.

TPI scholars do not necessarily agree with all the articles featured here—in fact, some of us may disagree completely with authors’ arguments or conclusions. Nevertheless, we believe that promoting literature on different sides of these debates encourages well-rounded and informed discussion.

These papers are typically culled from new postings on SSRN, but we will include new papers from multiple sources. If you are the author of a newly authored research paper relating to technology policy and wish to be included in a future entry, please send it our way.

The articles are organized loosely by category and are listed only once though they may relate to multiple topics. Enjoy!