This post reports the results from a bit of economic sleuthing about broadband prices. It quotes a research paper. The paper is called Evidence of a Modest Price Decline in US Broadband Services, which gives away the punch line in the title. Ryan McDevitt and I wrote the paper.
Let me explain why we did it.
The Internet mass market has grown in two waves. The first wave involved dial-up access to households, typically at $20 per month, give or take.
The second wave involved broadband, involving either DSL or Cable architectures for delivering such speeds. The cost of access in the second wave averages about twice as much, roughly $40 per month, give or take.
In some research we did last year Ryan and I tried to figure out how much households benefited from an upgrade, namely, moving from the first wave to the second one. That is, what is the gain to a household from going from dial-up to broadband?
We expressed the gains in terms of an equivalent decline in price. We figured out that the gain to all users was equivalent to about a 1.6% to 2.2% decrease in price each year between 1999 and 2006 (the last year we had data).
However, we did not examine the experience after the upgrade, nor estimate whether price increased or declined, or quality improved. As more households switch to broadband, the prices for such expenditure becomes more important.
Said another way: broadband is now part of a recurring household expenditure that exceeds $500 a year in many households. According to the latest survey of the Pew Internet and American Life Project, fewer than 10% of US households had dial-up Internet connections by April 2009, while 63% of US households had broadband. That over 70 million households by now.
So we ask: What change in prices did a typical US household experience after installing broadband in their home?
◦ Did prices go up or down in general? By what percentage, if at all?
◦ What is the answer if the prices are adjusted for qualitative improvement?
◦ Did DSL and cable users have a similar or different price experience?
There is a standard economic approach for addressing such questions: construct a price index. So we did just that. This post reports the results. If you want to know all the gory detail, then read the paper.
What is the answer?
We did this index because there is not another we could use. Though the Bureau of Labor Statistics (BLS) maintains a price index for Internet access as part of the Consumer Price Index (CPI) – this index is called “Internet Services and Electronic Information Service Providers.” – the BLS index combines price data for both dial-up and household Internet-access services, using lagged expenditure surveys to weight price movements. As a result, the price for dial-up service dominates the index for most of the first decade of the millennium, which renders the BLS index almost useless for assessing the effectiveness of US broadband policy.
So Ryan and I constructed a price index for broadband services in the United States between 2004 and 2009. We analyzed over 1500 service contracts offered by DSL and cable providers in the United States.
We did not do anything very fancy. We employed standard quality-adjusting techniques – a mix of matched-model methods and hedonic price index estimations. BLS uses the same techniques.
We looked for conclusive evidence of trends, namely, widespread dramatic declines in quality-adjusted prices or a widespread lack of change. In general, we find neither. Most of the evidence points towards modest price declines at most.
More precisely, we divide prices into three categories – standalone prices, bundled prices, or households switching between these two forms – and we present evidence about the first two.
We conclude there is no evidence of widespread dramatic price declines. Instead, we find some evidence of modest declines in quality-adjusted prices. We bound it between a 3% and 10% decline in quality-adjusted prices in just under six years. Lack of access to (confidential) data about the market share of services prevents us from being more precise.
Our evidence also pointed towards different Cable and DSL experience. There seemed to be more quality-adjusted price decline in cable than DSL. In DSL there was almost none.
These estimates contribute important new findings to several ongoing debates. First, despite rapid growth in adoption and revenue in broadband over the period we examine, 2004-2009, the lack of any dramatic price decline in broadband rejects any claim that US policy generated dramatic change in the price of broadband services.
Second, although the price declines are modest, our index declines faster than the BLS price index for Internet access over a period that allows us to make a direct comparison. About the nicest thing we can say is this: There were constant or declining nominal prices in a period of mild inflation for all the CPI, so the real price of broadband did not go up while the price of other goods did.
Third, and more broadly, it suggests this market looks nothing like other parts of electronics, such as CPUs, lap tops, printers or storage devices, where the quality-adjusted price declines regularly exceed double digits per year. That raises many questions about differences between mass-market Internet services and other parts of electronics in the determinants of prices, as well as questions about the role of market structure and demand.
It also has me worried about growth in the next decade. In the prior decade the upgrade to broadband has generated some growth externalities. For example, Web2.0 (You Tube, Facebook, Flickr), P2P (Bit-Torrent), Music downloading (iTunes), and Efficient search (Google) are all better on broadband.
Would some of this growth have had happened with dial-up anyway? Some perhaps, but not all.
Similarly, electronic retailing (Amazon) and advertising-supported media (blogs, news) are better with broadband. Once again, some perhaps, but not all.
If quality adjusted prices do not decline in near future… what are implications for online media supported by ads & for electronic commerce? That is a provocative and open question.