The Empirical Analysis Of Price Dispersion And The Reputation Of Sellers On The Internet

The market on internet is a rising market and has a great development in recently decade. The characters of Convenience, speediness, and the abundance of information on internet give us a great of help, especial in e-market. But, the efficiency of e-market is a problem that scholar usually argued. The price dispersion is a token that can evaluate the efficiency of market. Traditional theory suggests that, the price dispersion should be decrease with the development of e-market. In fact, the price dispersion doesn't decrease, but increase with the development of e-market. This is a famous"Price Paradox"on e-market.This paper adds reputation of sellers to a simple principle (the manufacturer)–agent (the sellers) model, and use this model to explain the phenomenon of price dispersion for the homogeneous products. We use the ratings of different sellers on www.bizrate.com as an instrument variable for the reputation of sellers on the internet to explain the price dispersion on the internet. For simply, we use two Independent variables: the number of sells and the reputation of sellers on internet and attempt to compare which method are better, linear model or nonparametric method or Partial linear nonparametric model. Results show that parametric model and nonparametric model both give predictions on the price dispersion which is consistent with the theory and previous research result by others: the variation of reputation has a positive effect on the price dispersion (coefficient of variation) and the number of sellers has a negative effect on the price dispersion.On the base of the analysis of previously models research, we adopt the conclusion of Xing Pan,Ratchford and Shankar(2001), make a regression for the price dispersion by using the variable of the characters of market and sells. The results are significant at 10% critical value, indicating that reputation can be help to explain the price dispersion on internet. Also my research found some results which are consistent with former research. In the end this paper gives us some limitation and the directions for further research.