Promoting innovation and the deployment of advanced telecommunications services to businesses

Contemporary Economic Policy, April, 2008 by David J. Gabel, Kenneth Guang-Lih Huang

Finally, where previous studies have addressed the data aggregation issue noted above (Prieger, 2003), the level of observation studied (the ZIP code) is too disaggregated to correspond to the level of decision making by firms--the wire center in the case of DSL and the franchise area in the case of the cable modem. This study uses data at the wire center level where provisioning decisions are made.

The remainder of this section describes the data.

A. Service Availability

The availability of various ATS by wire center was obtained from the National Exchange Carrier Association, Inc. (2001) federal tariff. (A wire center is a hub that terminates cables that connect to retail customers. A tariff is a published schedule of prices and terms of service.) The National Exchange Carrier Association, Inc. (NECA) tariff identifies the wire centers where different services are and are not made available. The NECA availability data were used to create binary variables. For example, regarding packet switching, the corresponding observation for each wire center was given a value of 1 if frame relay and/or ATM was available and a value of 0 if these capabilities did not exist. While these data accurately represent the services of small companies because of their use of the NECA filing, we increased the accuracy for large companies where tariffing is voluntary by dropping companies where the deployment of advanced services (dependent variable) is less than 5%.

B. Competition

To examine the effect of competitive entry on the deployment of ATS, the authors determined the number of CLECs with facilities located within three miles of a wire center served by an ILEC. The authors used three miles because CLECs need to be located relatively close to ILECs to avoid substantial transport costs. (2)

Our data set shows where packet switching was available in 2001 (Packet 2001). The lagged value of competition, that is, the level of competition from 1999, was used as the instrument for the level of competition (see Tables 6-8). The level of competition from 1999 serves as an appropriate instrument as it is predetermined and therefore not correlated with the availability of packet switching. The instrument was highly significant, and its F value exceeds the first-stage F statistic of 10 rule of thumb proposed by Staiger and Stock (1997). To provide support for its validity, a univariate regression was performed on Packet 2001 including the instrument as an independent variable. The instrument is not statistically correlated with Packet 2001 (t value of -0.068) and provides further support for the proposition that the lagged value is a valid instrument. (3)

Although the number of CLECs within a three-mile radius of an ILEC's wire center was identified, the authors did not control for the type of services offered by CLECs or the size of the market served. There are no publicly available data on the type of services offered by CLECs throughout the country. Therefore, the resulting measurement of competition reflects the impact of all potential competitors rather than the effects of rivals offering identical services.