Collaborative Research: An Empirical Study of Broadband Internet Service

宽带互联网接入现在很普遍；美国有大约2.1亿用户，而在2000年，这一数字约为800万。这些用户平均每月使用互联网的时间为30小时，而在1995年，这一数字仅为1小时。令人惊讶的是，人们对互联网接入需求的基本面知之甚少，比如每GB数据的付费意愿，或者网络拥堵造成的外部性的重要性。造成这一差距的一个关键原因是缺乏有关宽带使用的详细数据。该项目使用一组北美互联网服务提供商提供的订户级别服务计划和每小时使用量的独特数据集来研究宽带服务的定价和使用情况。研究小组将编制关于宽带使用的描述性统计数据，估计宽带接入的需求和付费意愿，评估基于使用的定价的福利影响，并分析网络拥塞对需求的影响。结果将使PI团队能够评估拥堵收费在这个市场上可能产生的影响。PI团队将使用详细数据和影子价格的变化来估计需求，这些影子价格是由一些供应商使用的基于使用的定价产生的。然后，需求估计被用来计算基于使用的定价对消费者福利的影响。此外，通过利用网络配置中的外生变化，团队可以估计拥塞对使用率和福利的影响。这使得他们能够计算拥塞定价的均衡效应。这项研究的结果形成了一个关于互联网定价和使用的重要窗口，为最近关于互联网监管的政策辩论提供了新的视角。

Broadband internet access is now widespread; there about 210 million users in the United States, up from roughly eight million in 2000. These users spend an average of 30 hours a month using the Internet, up from just one hour per month in 1995. Surprisingly little is known about the fundamentals of demand for internet access, such as the willingness-to-pay per gigabyte of data or the importance of externalities created by network congestions. A key reason for this gap is the lack of detailed data on broadband useage.This project studies pricing and usage of broadband service using a unique data set of subscriber-level service plans and hourly usage from a group of North American internet service providers. The research team will produce descriptive statistics on broadband usage, estimate demand and willingness-to-pay for broadband access, evaluate the welfare implications of usage based pricing, and analyze the impact of network congestion on demand. The results will allow the PI team to evaluate the possible effects of congestion pricing in this market. The PI team will use both the detailed data and variation in the shadow price created by the usage-based pricing used by some providers to estimate demand. The demand estimates are then used to compute the effects of usage based pricing on consumer welfare. Furthermore, by exploiting exogenous variation in the configuration of the network the team can estimate the impact of congestion on both usage and welfare. This allows them to compute the equilibrium effect of congestion pricing.The results of this research form an important window on internet pricing and usage that sheds new light on recent policy debates on regulation of the internet.