Loading...

Fuel Smuggling Detection and Empirical Analysis of Its Demand Determinants in Iran

Mostafa, Ali | 2023

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 58198 (44)
  4. University: Sharif University of Technology
  5. Department: Management and Economics
  6. Advisor(s): Rahmati, Mohammad Hossein
  7. Abstract:
  8. One of the most important commodities frequently smuggled due to imposed tariffs is fuel, particularly gasoline and diesel. The policy of strict energy price controls has led to a disparity in energy prices domestically and internationally, creating the primary driver of smuggling. In this thesis, we aim to analyze big data on gasoline consumption through exploratory data analysis and employ anomaly detection algorithms in machine learning to identify suspicious demand related to smuggling. Additionally, we estimate the elasticity of fuel smuggling demand in Iran. We utilize a unique dataset of all gasoline transactions in Iran over an eight-month period, centered around an unexpected exogenous price change, to detect smuggling. First, using unsupervised machine learning, we identify suspicious cases potentially linked to smuggling. Then, we examine key factors influencing smuggling, such as fuel price differentials with neighboring countries, distance from borders, and regional fixed effects, quantifying the impact of each factor. Since this study is the first to use machine learning algorithms on consumer-level big data to detect smuggling with no similar studies in the existing literature we refer to fraud detection research in the financial industry, particularly credit card transactions, for analogous empirical approaches. This is because the dataset closely resembles credit card transaction data
  9. Keywords:
  10. Big Data ; Gasoline Smuggling ; Machine Learning ; Energy Subside ; Unsupervised Machine Learning Algorithms ; Anomaly Detection ; Gasoline Consumption

 Digital Object List

 Bookmark

No TOC