Researchers from the University of Georgia and the U.S. Department of Justice were awarded $530,000 over three years by the National Science Foundation and CableLabs to study demand for residential Internet service.

John Turner and Jonathan Williams, both faculty in the department of economics at UGA's Terry College of Business, and Aviv Nevo, of Northwestern University who is currently serving as the chief economist at the DOJ, will use the grants to collect and analyze data from more than 1 million Internet users. In studying how Americans make choices online when facing overage prices and network congestion, their work promises to shed light on the future of residential broadband.

At the heart of the project are the unique data secured from a group of North American Internet Service Providers. The data include plan characteristics such as prices-per-gigabyte of usage above an allowance, as well as hourly levels of household usage and network congestion. With these data, the researchers can precisely estimate willingness-to-pay for Internet usage.

"Think about it like a cellular plan," Williams said. "If you've ever run up against your usage allowance, you get text or email messages that say ‘You're at 90 percent of your data' or ‘You're at 95 percent of your data allowance' and because of that you might change your behavior. Now we get to see that same data for residential broadband service. We can literally see you weigh what you're doing against an overage cost and determine demand."

The researchers' models are designed to generate hypothetical predictions about how usage-based pricing affects usage and welfare. They plan to use these predictions to inform public policy, where Internet usage already has become a political issue.

"Right now there are senators who are rolling out usage-based pricing policies concerning what providers should be able to do, and so forth," Williams said. "The FCC is trying to make standards about what they should do, and basically there is nothing out there in terms of empirical estimates to inform these decisions."

As more Americans work and watch TV online, ISPs are concerned about network congestion and the costly network expansions necessary to avoid it.

"There are an increasing number of people who aren't watching traditional TV anymore, and Netflix uses a lot of bandwidth," Turner said. "So in the current structure there's going to be network congestion problems that you might see as buffering or pixelated video."

ISPs are split over the best method of dealing with increasing demand and limited bandwidth. One option is for ISPs to continue to rely on current coaxial-cable based network architecture, but charge users for the amount of data they use. Another is for ISPs to run fiber optic cables to households. Fiber-to-the-home offers virtually unlimited data use, but is expensive to install.

"If you upgrade the infrastructure to avoid network congestion, you have to spend a lot of money," Turner said. "In a perfect world for ISPs, the ISPs wouldn't do that. They would just find a way to charge customers for their use. The real question is which way is going to win out? Are the economics such that the cost of the technology is declining so fast that what we're eventually going to see is fiber-to-the-home everywhere?"

Much like interstate traffic, Internet data usage doesn't decline as more bandwidth becomes available. When the road widens, more commuters use the highway.

"One of our central goals," Williams said, "is to identify what the effect of these externalities is on consumers. How much does network congestion affect you? Then we can talk about models of congestion pricing. From 7-9 p.m., everybody's on the Internet. They have to build the network to accommodate that. But if you're using it at 2 in the afternoon, there's nobody else there. So if you use it then, should you pay a lower price?"

About 210 million Americans are using the Internet today, up from 8 million in 2000. And along with their numbers, their consumption increases every year.

"The median users are double their usage every year," Turner said. The guy in the top 10 percent might be increasing his usage by 30 percent. But that's still a lot."