Wednesday, November 12, 2014

Ahead, Apace or Behind: Methodology Matters

Though few pay much attention, study methodology matters when trying to determine current status of high speed Internet access or income inequality. That likely is one reason recent studies come to such vastly-different conclusions about the state of high speed access in the United States, compared to Europe, for example.

A study by the World Economic Forum ranked the United States 34th in terms of Internet bandwidth available per user. A study by Ookla ranks the United States 26th in terms of typical access speed.

But a study led by Christopher Yoo of the University of Pennsylvania Center for Technology, Innovation and Competition found a far greater percentage of US households had access to next generation networks (25 Mbps) than in Europe.

This was true whether one considered coverage for the entire nation (82 percent in the United States and 54 percent) in Europe, or for rural areas (48 percent in the United States compared to 12 percent in Europe).

The United States had better coverage for fiber-to-the-premises (FTTP) (23 percent compared to 12 percent in Europe).

The U.S. broadband industry invested more than two times more capital per household than the European broadband industry every year from 2007 to 2012.

In 2012, for example, the US industry invested US$ 562 per household, while EU providers invested only US$ 244 per household.

U.S. download speeds during peak times (weekday evenings) averaged 15 Mbps in 2012, which was below the European average of 19 Mbps. But during peak hours, U.S. actual download speeds were 96 percent of what was advertised, compared to Europe where consumers received only 74 percent of advertised download speeds.

The United States also fared better in terms of advertised compared to  actual upload speeds, latency, and packet loss.

Pricing comparisons always are difficult. U.S. high speed access was cheaper than European broadband for all speed tiers below 12 Mbps, though U.S. prices were higher for higher speed tiers/

Still, the picture is complicated. U.S. Internet users on average consumed 50 percent more bandwidth than their European counterparts. To the extent there is a broad but direct connection between consumption and price, that matters.

The point is that assumptions matter.

The choice of specific retail service plans, and the percentage of people in any nation that purchase those plans, make a difference when trying to compare “typical prices” for any good or service, across national boundaries.

In other words, if most people buy high speed access as a feature of a triple-play bundle, then the posted prices for “stand-alone” high speed access are misleading.

That is why one can come to very different conclusions on the state of U.S. high speed access--how fast and how costly--compared to other nations.

Recent research on U.S. levels and trends in income inequality also vary quite substantially based on how these studies measure income. To be sure, there are important social mobility and income inequality issues to be faced in the United States, as elsewhere.

But a fact-based approach remains important, and methodology affects the “facts.” Economists Philip Armour and Richard V. Burkhauser of Cornell University and Jeff Larrimore of Congress’s Joint Committee on Taxation recently studied “income” including all public and private in-kind benefits, taxes, Social Security payments and accounting for household size.

The bottom quintile of U.S. residents saw a 31 percent increase in income from 1979 to 2007 instead of a 33 percent decline as measured by the Piketty-Saez market-income measure alone.

The income of the second quintile, often referred to as the working class, rose by 32 percent, not 0.7 percent. The income of the middle quintile, America’s middle class, increased by 37 percent, not 2.2 percent.

Those significant differences occur because one study omits Social Security, Medicare and Medicaid payments, for example, as well as employer-provided health insurance and capital gains on homes.

Also, some of the apparent income inequality is the result of a definitional change.  In 1992 the U.S. Census Bureau began to collect more in-depth data on high-income individuals. This change in survey technique alone, causing a one-time upward shift in the measured income of high-income individuals, is the source of almost 30 percent of the total growth of inequality in the United States since 1979.

Again, the point is not the importance of any developing or existing changes in opportunity for social mobility, or gaps in income or wealth. The point is that methodology matters.

That is as much true for how we measure progress on high speed access or social mobility.

No comments:

The Downside of Multi-Purpose IP Networks

By now, virtually all observers agree that direct revenue generated by fixed networks will shift to supplying broadband access, while some o...