Price Dispersion in the Online Book Market

Despite that the book industry is threatened by the growing digital sources of information in the United States annual sales of printed exceed 675 million .1 E-commerce introduction and other Internet-related facilites were expected to strengthen the competition and reduce price differention (Levin, 2012). Price dispersion stands for the differences in prices across sellers of exactly the same goods (for example, books sold at Amazon and Blackwells could be sold for different prices). Even taking into consideration distinctions in product or service quality this kind of phenomennon still holds (Ba, Stallaert and Zhang et al, 2011). Such observations also lead to the point of view that law of one price does not hold (Varian, 1980). Economic theory explains dispersion through the cost of searching. According to this explanation consumers do not have all of the necessary information about prices which let sellers to charge higher prices (Baye et al, 2005). Understanding the price differentials and their determinants is important for pricing strategy development and higher margins. This emprical analysis uses hedonic price model in which price dispersion is calculated for each item and is used as a dependent variable, while shipping time, availability in stock, number of pages and other attributes are treated as independent variables. Data consists of two separate datasets for Ukrainian and US market which include prices and characterstics of books that are sold by online retailers. Data was collected using web-scraping techniques using R programming language and ParseHub application. Dataset for Ukraine is collected using the most recognizable online bookstores for the Ukrainian market (Book-ye, Yakaboo, Rozetka etc) by several foreign online bookstores. Dataset for the US includes data from Amazon, Barnes and Noble and AbeBooks. The important implication of this work is that despite prices indeed can differ among marketplaces in Ukraine as well as in the US, price dispersion can be hardly determined by the book trim size. Instead of this we can observe for certain book genres (Fantasy and Movie Based in the US). We can also conclude that models that use the range of prices and standard deviation as a price dispersion measure give substantially different results from the dispersion measured as a skewness. Models used books length, width, number of pages, cover and genre as an independent variables.