Tuesday, May 14, 2013

Gigabit Networks are a Real Challenge in Rural Areas


Unless something fairly dramatic happens, gigabit networks will be unevenly deployed around the United States. No matter what might be done in suburban and urban markets, rural areas still have very long loop lengths (which raises network cost) and low-density populations, which both raises network cost and limits revenue potential.

Frontier Communications, which operates rural telco networks in a number of states, reports that 80 percent of its users buy 6 Mbps Internet access services.

About 75 percent of all locations can buy 6 Mbps service. About 54 percent can buy 12 Mbps service. About 42 percent of locations can buy 20 Mbps service.

And that roughly hints at the magnitude of any decision to upgrade networks more in line with gigabit networks that will be the future standard.

There are both “demand” issues and “cost of supply” issues. Current demand seems to be gravitate to 6 Mbps, even when 12 Mbps and 20 Mbps services are available. So demand for gigabit networks might seem a stretch.

One might argue that the take rates for higher-speed services would be higher if the price were lower, and that is true. Whether Frontier, with its current cost structure and revenue requirements, would find that an easy matter is quite unclear.

But cost of supply is the other huge issue. Some studies suggest fiber to the home cost can range from $5,000 per site to $20,000 or more, per site.  

The Broadband Availability Gap study suggested serving seven million underserved U.S. homes would require an inducement of about $23.5 billion, over and above the actual costs of construction, to create incentives for service providers to make the investment, estimated at $32.4 billion.

If one really wanted to build fiber to home networks for 130 million U.S. homes, with active bandwidth of 100 Mbps, the cost could range up to $670 billion, the study suggested. Gigabit networks, offering an order of magnitude more bandwidth, would cost more.

And Frontier has better prospects than most independent rural telcos. It is larger and generates more actual customer revenue than the typical rural telco.

About 2005 Frontier made a decision not to rely so heavily on “regulatory revenues” such as universal service, and to instead focus on actual customer revenue.

Since then, Frontier’s contribution from non-customer sources, which was as high as 60 percent, has fallen to less than 10 percent. Much of that change, one might argue was a decision to focus more heavily on business customer revenues, in or out of region.

At December 31, 2011, Frontier had 3,103,800 residential customers and 309,900 business customers.

Business revenue tells the story, though. For the six-month period ending Dec. 31, 2011, Frontier earned $692 million in business customer revenue, and $544 million in consumer revenue.  In other words, business customers generated 56 percent of total revenue.

Or, to put it another way, the nine percent of business customers generate 56 percent of Frontier revenues.

For Windstream, a firm that is similar in some ways, the reliance on business customer revenue is even more pronounced. Business revenue in the fourth quarter of 2011 was $888 million, while consumer revenues were $118 million. So Windstream generated 88 percent of revenue from business customers, not consumers.

Those results are driven by acquisitions. Windstream bought PAETEC, a major business-focused competitive local exchange carrier. Frontier bought a huge chunk of former Verizon Wireless assets.

Both acquisitions had the effect of boosting business customers and business revenue to unusual levels for any rural telco.

The point is that whatever problems policymakers and service providers might think they face now, just trying to get broadband of 4 Mbps to 10 Mbps to rural homes, the problems will be tougher when the standard gets moved to a gigabit.

No comments:

Many Winners and Losers from Generative AI

Perhaps there is no contradiction between low historical total factor annual productivity gains and high expected generative artificial inte...