Paul Rappoport

Residential demand for access to the Internet

The focus in this paper is on the residential demand for access to the Internet, and represents an extension of earlier work on Internet access demand by Rappoport, Taylor, Kridel, and Serad (1998), Kridel, Rappoport, and Taylor (1999), Kridel, Rappoport, and Taylor (2000), and Duffy-Deno (2000). The analysis of broadband demand has been studied by Eisner and Waldon (1999), Madden, Savage, and Coble-Neal (1999) and Madden and Simpspn (1997). With the aggressive marketing of cable modems and ADSL service, a growing number of residential households in the U.S. now have a choice regarding how they access the Internet. The choice set available, however, is not uniform. In some areas, the only form of access is through dial-up modem, while in other areas various forms of high-speed access (cable modems or ADSL) are available as well. This paper reports the results from a set of models of Internet access where the models are differentiated by the availability of Internet access options. The models are based on the analysis of surveys submitted by over 20,000 households during the period January – March, 2000. Among other things, we are able to report broadband penetration rates and compare those to Internet access estimates presented in the NTIA Report, Falling Through the Net: Toward Digital Inclusion. In addition, we present a more complete set of estimated price elasticities for both basic and high-speed access to the Internet.

AN ECONOMETRIC STUDY OF THE DEMAND FOR ACCESS TO THE INTERNET

The increased popularity and explosive growth of the Internet have been widely tracked. A recent Business Week survey estimates that 40 million people are internet users, which is in line with an April 1997 FIND/SVP survey estimate that there are among 40-45 million adult users. 1 These numbers are up sharply from estimates obtained from surveys conducted in the fall of 1996. A November Louis Harris & Associates’ survey, for example, estimated that there were 35 million adult users, while IDC Research estimated that adult users in October 1996 were 31.4 million.2 PNR and Associates’ most recent estimate of households that use the Internet is 16%, which compares with an estimate of 14.8% in 1996. 3 These percentages are consistent with the Harris estimates. Thus, there is ample evidence to support the increased popularity of the Internet and the resulting demand in Internet applications.

Toll price elasticities estimated from a sample of U.S. residential telephone bills

Abstract Demand models for four categories of toll calls are estimated from a sample of U.S. residential telephone bills. Besides the novelty of the data set, innovations in the analysis include the use of a call concentration index constructed from a households itemized individual calls, plus variables representing choice of IX carrier and the presence or absence of an optional calling plan. A principal finding is that price elasticity and call concentration are inversely related.

Relative efficiencies of some simple Bayes estimators of coefficients in dynamic models -- I

Abstract In this paper we consider some approximations to Bayes estimators of coefficients in simple autoregressive models and give an example of a Monte Carlo experiment where these approximate Bayes estimators yield a substantial improvement over the usual sampling theory or quasi-Bayesian estimators. The practical situation is represented by the case where the coefficient vector is known to lie in or on a hypersphere of radius r with center at 0 . We show that arbitrariness in the choice of the value of r is often not catastrophic if r is sufficiently large, but finite.

Optimal Pricing with Sunk Cost and Uncertainty

When static models of the firm are considered regulators make serious errors in the determination of the proper wholesale price as the opportunity cost of the delay option is neglected. For an incumbent, the option is exercised and represents an opportunity cost. For a potential entrant the delay option need not be exercised should the regulator allow the purchase access at below economic cost, i.e., operating cost plus the delay options cost. Thus service-based entry is excessive and facilities-based entry remains suboptimal. To summarize, in a static-regulated paradigm: wholesale prices are below their economic cost; inefficient service entry occurs as prices are not set at the correct marginal cost while an entrant is not required to exercise a delay option; social welfare is suboptimal; incumbent firm’s valuation by the financial markets is less; and the cost of capital for incumbent firms is higher. When a regulated paradigm is dynamic the converse holds: optimal prices are higher than for static calculations; only efficient service entry occurs as prices are set at the correct economic marginal cost; facilities-based entrants receive correct price signals; social welfare is maximal; an incumbent’s valuation by financial markets is higher; and the cost of capital is lower than for a regulated paradigm. This chapter has demonstrated the profound implications for regulatory practices. In particular, dynamics and uncertainty matter for the realization of optimal regulatory pricing outcomes. However, much work remains to develop and quantify precise models to fit the telecommunications context.

IntraLATA long-distance demand: carrier choice, usage demand and price elasticities

Abstract Competition in the long-distance market in the US continues to intensify; the 1996 Telecommunications Act has led to increased competition in long-distance telephony especially as the Regional Bell Operating Companies have begun to gain entry to long-haul, long-distance markets. In order to better understand the implications of having increased service offerings, models of how customers choose between carriers (and the impact of this choice on subsequent usage) will be useful. We develop the first publicly available models that simultaneously estimate choice and usage for intraLATA long-distance in the US. Utilizing a generalized Tobit model, the price responsiveness of usage and carrier choice are estimated. The results are generally consistent with expectations both in terms of theory and of practical experience in the industry.

The Demand for High-Speed Access to the Internet:

The demand for Internet access continues to grow at a dramatic rate. While the primary mode of access has been dial-up service, recent innovations have allowed even residential subscribers the option of obtaining affordable highs-peed or broadband access. In particular, cable television companies have begun to offer cable modems and many local exchange telephone companies are beginning to offer various forms of digital subscriber line (DSL) service. This chapter analyzes the demand for broadband access to the Internet by US households. Using a very large data set of US households with over 32,000 observations, we estimate a discrete-choice model for the demand for Cable Modems. Preliminary elasticity estimates indicate cable-modem Internet access demand to be price-elastic.

Relative efficiencies of some simple Bayes estimators of coefficients in a dynamic equation with serially correlated errors - II

Abstract Generalized least squares estimators, with estimated variance-covariance matrices, and maximum likelihood estimators have been proposed in the literature to deal with the problem of estimating autoregressive models with autocorrelated disturbances. In this paper we compare the small sample efficiencies of these estimators with those of some approximate Bayes estimators. The comparison is done with the help of a sampling experiment applied to a model specification. Though these Bayes estimators utilize very weak prior information, they out-perform the sampling theory estimators in every case we consider.

Forecasting Video Cord-Cutting: The Bypass of Traditional Pay Television

Following the substitution of mobile phones for fixed-line phones (“voice cord-cutting”), a similar transition is now occurring for video services. In the United States, traditional pay television service providers are experiencing some revenue losses and slowdowns. Also, consumers are increasingly streaming or downloading long-form video programming (mainly movies and TV shows). This phenomenon—described as “video cord-cutting” or “over-the-top (OTT) bypass”—suggests that the business models of traditional TV service providers are under threat. There is, however, considerable debate about the severity of that threat.

Competition in intraLATA long distance: carrier choice models estimated from residential telephone bills

Abstract Competition in the long distance market in the US is at a crossroads. The recent passage of the Telecommunications Act will (eventually) significantly increase competition in long distance telephony. In order to better understand the implications of having more service offerings, models of how customers choose between carriers will be useful. We develop the first publicly available models of intraLATA carrier choice in the US. We find that the results are consistent with expectations — both in terms of theory and practical experience in the industry.