Thursday, September 02, 2010

Internet Speeds Vary by City -- Why?

Internet Speeds Vary by City -- Why?

The US city you live in strongly predicts your Internet connection speed. We have been poring over Ookla [1] speed test data to understand city-by-city differences in consumer Internet access performance. Why, for example, are average access speeds for San Jose residents three times faster than those in Kansas City? Our preliminary analysis indicates socio-economic factors are at work. We uncovered a strong correlation between a metropolitan area's median per capita income and education level, and the Internet speed its residents experience.

We analyzed Ookla performance data for 50 US cities from which Ookla collected test results for at least 75,000 unique IP addresses from February 2008 to August 2010. You can find data for 30 of these cities here [2]. Ookla captured both download and upload speeds, and since both measurements are important to the user experience, we summed them to create a combined metric we call ISP bandwidth. This is the average bandwidth delivered (i.e., bandwidth measured), not the average advertised bandwidth (i.e., bandwidth purchased). We then set out to correlate ISP bandwidth information to demographic data in the US Census Bureau's "2010 Statistical Abstract [3]". The Census Bureau tallies information by metropolitan statistical areas (MSAs), some of which lump contiguous cities like San Francisco and Oakland together. So to compare apples to apples we did the same, resulting in a final count of 43 MSAs.

Here's what we found. As the following chart shows, per capita income for a metropolitan area correlates strongly to the average Internet connection speed its residents experience. Income has a positive correlation of 0.11, which means that for every $1K increase in an MSA's per capita income, residents experience 110Kbps faster Internet service. The trend line starts at 10 Mbps and ends at 13.5--for a significant 35% difference. We found the same 0.11 positive correlation between an MSA's ISP bandwidth, and the percentage of residents over 25 with a college degree.

The question is why do people in cities with higher income and/or education levels tend to experience higher Internet access speeds? Here are some possible explanations:

1) Consumers in higher income cities are more likely to purchase higher speed service packages.

2) ISPs offer higher bandwidth service packages in more affluent cities than they do in poorer cities.

3) ISPs deliver better service in more affluent cities than they do in poorer ones.

What do you think explains this phenomenon?

No comments: