an application of entry and exit modelling to the internet service provider market

Open Access
Carle, Joshua David
Area of Honors:
Bachelor of Science
Document Type:
Thesis Supervisors:
  • Charles Theodore Murry, Thesis Supervisor
  • Russell Paul Chuderewicz, Honors Advisor
  • Internet Service Provider
  • ISP
  • Entry and Exit
  • Ordered Probit
  • FCC
  • Competition
This paper explores different types of modelling, all of which are applied to the Internet Service Provider (ISP) market. The primary goal of this paper is to further understand the ISP market structure. In all models, the dependent variable is the number of ISPs located in a given zip code. A zip code is assumed to represent a geographically isolated ISP market. Using the models presented in this paper, I first analyze how population, income, and geographic variables affect market concentration. Then, I use a different modelling technique to determine that almost all of the increase in competition in the ISP market appears with the addition of the first additional firm, or the 4th firm. Because of the format of the ISP data that was acquired by the FCC, markets with between 1 and 3 firms are all counted as a 3 firm market. With regards to modeling techniques, the first model is a linear regression that explores which variables are significant in predicting market concentration. I also edit this model to add fixed effects that account for which state the ISP market is located in. The next model is an ordered probit model that uses theory from the Bresnahan and Reiss paper (1991). I begin by using an ordered probit model to model the demand in the ISP region, which finds that demand increases the most with the addition of the 4th firm. I continue by implementing a model that is a very close version of the Bresnahan and Reiss model. This version models variable profits, fixed costs, and market sizes with varying numbers of competitors. I then use these components to determine entry threshold ratios for each market size. These ratios are used to determine how much competition in the market increases as each additional firm enters the market. The results are similar to what Bresnahan and Reiss find, meaning Model 2.2 finds that the majority of the increase in competition takes place when the first additional firm, or the 4th firm, enters the market. However, Model 2.2 finds that the initial increase in competition is more extreme than in the Bresnahan and Reiss model. This could be due to the FCC data grouping 1-3 firm markets together and because both papers are studying different markets. In the model that uses theory from the Bresnahan and Reiss paper, some variables have very high standard errors. This could be attributed to irregularities in the data such as not differentiating between 1-3 firm markets or having a large proportion of the markets being very small. The result would be data that does not fit the model very closely. However, there is also a relatively large standard error in the Bresnahan and Reiss model, so the direct application of their model may simply be prone to higher standard errors.