Bachelor Thesis (Data Science and Economics)

This is Julian Valdman's Bachelor Thesis spring 2022

Thesis Information

Major: Data Science and Economics

Title: Semi- and nonparametric estimation of price dispersion and auction heterogeneity with eBay auctions

Where: University of Copenhagen

When: June 2022

Grade: 12 (A)

Table of Content of the Report

1. Introduction
2. The eBay marketplace and data
3. Maximum Likelihood Estimation
4. Artificial Neural Networks and Multi-class classfication
5. Discussion and Conclusion

Methods

The analysis comprises 2 different methods of price analysis:

• Maximum Likelihood with Hermite Series: Using flexible polynomial to estimate density distribution of prices
• Artificial Neural Network (ANN): Using non-parametric deep learning model to exhaust predictive power of the auction attributes

Conclusions

From the analysis and including microeconomic theory, the report concludes:

• Some price dispersion can be explained by auction heterogeneity
• Most auction features have smaller economic effect on end price
• Most of the auction feature effects are aligned with economic theory and similar analysis