What`s the Internet infrastructure worth to the U.S.? My best guess of the value of the Internet infrastructure to the U.S.: $174 billion/year (with a 90% confidence interval of $44-264 billion). Why so precise/imprecise? Read on.

I heard a Freakonomics pod cast recently that described the Internet as an infrastructure project but with no “price”:

And it took some time to realize that there is no people who run the Internet, and therefore there is no bill, … it’s infrastructure that everybody pays for and everybody gets to use.[1]

I hadn’t thought about the Internet as infrastructure in a while but doing so has some lessons for valuing other infrastructure. The podcast was about regulation of the Internet but also touched on value:

DUBNER: … do you have any kind of … dollar figure in terms of overall economic value? SHIRKY: I don’t.
DUBNER: But we’re talking trillions of dollars.
SHIRKY: Easily.[2]

It reminded me of a presentation I prepared on the value of the Internet. Economists have tackled the value of the Internet by asking “How has the Internet changed people’s lives”. An article in the Economist newspaper[3] documents some of the methods used and these illustrate how economists value big infrastructure projects that have multiple benefits.

“Measuring the economic impact of all the ways the internet has changed people’s lives is devilishly difficult because so much of it has no price.”[4]

Monetary Value vs. Full Value

One approach to determining value is to use the monetary transactions. Gains and losses are measured in GDP, but this is not full value or welfare. If consumers are willing to pay $50/month for high speed Internet and only pay $20, The difference, $30, is a non-monetary benefit called “consumers surplus”.

The Internet activity that is measured in GDP excludes this consumer surplus. For example: Google’s ad sales significantly understate the contribution of the Internet to welfare and the losses to encyclopedia publishers from Wikipedia does not capture its value to users.

Four Approaches to Full Value

Build A Demand Curve And Measure Consumer Surplus

• Shane Greenstein (Northwestern U.) & Ryan McDevitt ( U. of Rochester)
  • Consumer surplus from spread of broadband access
  • Price in 1999 = $20; Price in 2006 = $17
  • Consumer surplus $3 => $5-7 billion per year
  • Based on its share Wikipedia accounted for $50m
  • Assumes internet in 1999 = internet in 2006
  • Likely an underestimate

Ask People What They Are Willing to Pay

• McKinsey for IAB Europe
  • Survey 3,360 consumers in 6 countries what they would pay for 16 services that are now paid for by ads
  • On average households would pay €38 ($50) per month
  • Subtract from this the cost of intrusive ads and reduced privacy gives €32 billion consumer surplus in America ($42 billion) and €69 billion in Europe.
  • E-mail 16%; Search 15%; Social Networks 11% of total

Calculate Time Saved From Internet

• Yan Chen, Grace YongJoo Jeon, and Yong-Mi Kim (U. of Michigan) for Google
  • Experiment to measure time to answer questions using: a) search engine, b) U of M library
  • Internet: 7 minutes; Library: 22 minutes
  • Hal Varian (Google chief economist) calculated savings = 3.75 minutes/user/day
  • At $22/hour (average wage in U.S.) = $500/user/year
  • $65-150 billion per year for the U.S.

Value The Leisure Time Spent On The Web

• Erik Brynjolsson and Joo Hee Oh (MIT)
  • Leisure time spent on internet in U.S.:
  • 2002 – 3 hours per week; 2011 – 5.8 hours per week
  • People value time on internet more than alternatives => growing consumer surplus
  • $2,600 per user per year or $564 billion in 2011

Dealing With Differences of Opinion

While these methods are illustrative and there are other ways, if you are a policy maker and want to decide how much bang for your buck sponsoring rural Internet gets you what number(s) do you use to help you decide?

The range is quite large – $3 billion to $564 billion. In another post (see Can Guidance Be Provided on Risky or Contentious Input Values?) I have discussed how, when you have lots of studies and estimates, you can use meta-analysis and risk analysis. Even when you have little information (say the four estimates above), risk analysis can come to the rescue if you are willing to make a few assumptions. One approach is to use all of them and to discount the outliers (Consumer Surplus and the Value of Leisure estimates ) and give more weight to the middle of the range (Survey and Time Saved estimates).

So here is my best guess of the value of the Internet infrastructure to the U.S.: $174 billion/year (with a 90% confidence interval of $44-264 billion):

[1] http://freakonomics.com/2013/11/14/who-runs-the-internet-full-transcript/

[2] Ibid

[3] “How to quantify the gains the internet has brought to consumers” – The Economist March 9th 2013 http://www.economist.com/news/finance-and-economics/21573091-how-quantify-gains-internet-has-brought-consumers-net-benefits

[4] Ibid