Estimation of Hedonic Pricing Model for the Ukrainian Market of Laptops

The market for computers has been for almost 40 years already. Without any doubt, these products reshaped the world and bring efficiency to numerous industries. The progress in computer manufacturing is overwhelming, disruptive innovation dramatically increased the speed of evolving technology and the industry is highly competitive. Nowadays there a wide variety of products derived from the computer industry: laptops, desktops, tablets, server computers, microcontrollers are the main ones. For instance, we can divide laptops into more specific product types as Chromebooks, workstations, convertible (2 in 1), ultrabooks, gaming, and many others. Though basic text editor software will run on each of those laptops, they are very different types of products design for very different purposes. Even more complicated to navigate in hardware specs as laptops are highly complicated multi-component products. In this work, I will focus only on the laptops. The aim of this work is (i) to find the characteristics that influence the most on a laptop price, (ii) to estimate the marginal effects of those features on a laptop price, and (iii) to determine the combinations of the features that will offset the fewer CPU cores in laptops. The hypothesis of this paper is next: the brand of the laptop and processor with additional RAM going to offset fewer CPU cores For model estimation used a data set that includes 1956 distinct laptop models. Data is web-scraped from Ukrainian e-commerce retailer “Rozetka”. Web-scrapping was carried out using the programming language R. The data set contains 11 variables: laptop prices, screen refresh rate, CPU type, RAM, integrated graphics card or not, GPU size, weight, laptop brand, battery capacity, paid OS or not, storage volume. For model estimation, I use hedonic pricing regression with log-level functional form. Haan and Diewert (2013) recommend a log-level functional form for high-tech goods. This model shows how much the percent change in price if we change the continuous explanatory variable by one unit. If an explanatory variable is categorical then the model shows a percentage change in price when moving to another level of a categorical variable compared to the baseline.