the upgrade provided a benefit “ equivalent” to a decline in price of $15.35, or a 30% decline.

The last row in the table illustrates our study’s estimate of a price index adjusted for upgrades. We follow the norms for a transactional index. Because only a small percentage of households a year upgraded to broadband, a price index that emphasizes transactions cannot give a huge weight to the upgrades in any given year. Upon initial assessment the gains appear modest even in our largest estimate, displayed in the last row of the table, around 2.2% decline per a year, each year prior by 2006.

Yet, it adds up over time. By 2006 the price index should have declined 16.4% more than it did (( 73. 1 – 63. 1)/73.1). Our study experimented with many other different ways to count it. Depending on how it is counted, we estimate that somewhere between 4. 8 and 6. 7 billion dollars of benefit went uncounted in 2006 alone.

Notice one other feature. Our new estimate displays a big difference in the timing of the recorded price decline. Accounting for the upgrade when users upgraded would have realized the benefits several years sooner than BLS recorded in its official index.

Overall, our study suggests there are many unresolved issues in the price indices for a wide range of improving electronic goods where users realize discrete gains from upgrading to a new good. After all, broadband is far from the only improving service a user can upgrade. What was true of broadband is likely to be true for smartphones, Netbooks, digital cam-corders, and even digital advertising. BLS has its hands full.

Policy

Our study also illustrates why a price index constructed for the CPI does not necessarily meet the needs of a policymaker interested in broadband policy, such as the Federal Communications Commission (FCC). We considered the gains from retirement of second phone lines at households. Once again, we experimented with numerous ways to estimate the gains from such retirement. We found that the savings on a second phone line accounts for anywhere from 30%–40% of the total savings in

how fast would
prices have to
come down to reflect
the value created
from the replacement
of dial-up access
with broadband?

2002–2006, or 21%–28% of the savings for 1999–2006. However, BLS price indices do not normally count the savings of expenditure in one category (on a second telephone line) as an input into calculating the price index for another (Internet access). While this procedural norm is fine for the CPI, it is misleading for broadband policy.

That would not have to be a problem if the FCC kept its own price index for its own purposes. However, the FCC has steadfastly refused to keep a broadband price index for many years, even while it tracks many other aspects of broadband. How can the FCC make informed competitive policy when it has to rely on the BLS for a price index that does not reflect the FCC’s policy priorities? Our study directs attention at this problem.

Indeed, the FCC has much to do. Its index would have to account for qualitative improvements, as well as the bundling of broadband pricing with other information services, such as cable television and telephony. The FCC also probably would want to measure changes to the frontier more quickly than BLS, watching prices of new access modes (smartphones), as well as the gains from Skype and other VoIP (using other phone services). It might be challenging to figure out the gains from Twitter over other text messaging, but that must be a component of the mix too.

 

Shane Greenstein ( greenstein@kellogg.northwestern. edu) is the elinor and Wendell hobbs Professor in the Kellogg School of Management at northwestern University, evanston, iL.

Copyright held by author.

Calendar
of Events

November 16–20
ieee/aCM international
Conference on automated
software engineering,
auckland, New Zealand,
sponsored: sigar T, sigsoFT,
Contact: John grundy,
Phone: 64-9-3737-599,
email: john-g@
cs.auckland.ac.nz

November 17–19
Kuwait First e-Conference,
Kuwait,
Contact: saleh Kassem,
email: saleh.kassem@
yahoo.com

November 18–20
asian internet engineering
Conference,
Bangkok, Thailand,
Contact: Kanchana
Kanchanasut,
email: kk@cs.ait.ac.th

November 22–25
8th international Conference
on Mobile and ubiquitous
Multimedia,
Cambridge, uK,
Contact: Jonna hakkila,
email: jonna.hakkila@
nokia.com

November 23–24
Netgames ’09: Network and
systems support for games,
Paris, France,
Contact: Maha abdallah,
email: maha.abdallah@lip6.fr

November 23–25 The aCM international Conference on interactive Tabletops and surfaces, Calgary, aB Canada, Contact: gerald Morrison, email: geraldm@ smarttech.com

November 30–december 4
10th international Middleware
Conference,
Champaign, iL,
Contact: roy h. Campbell,
email: rhc@uiuc.edu

december 1–4
Conference on emerging
Network experiments and
Technologies,
rome, italy,
sponsored: sigCoMM,
Contact: george Ventre,
Phone: 39-081-7683908,
email: giorgio@upina.it